{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9996863729026186,
  "eval_steps": 500,
  "global_step": 3188,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006272541947624275,
      "grad_norm": 14.32480525970459,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 1.1353,
      "step": 1
    },
    {
      "epoch": 0.001254508389524855,
      "grad_norm": 16.002328872680664,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.1649,
      "step": 2
    },
    {
      "epoch": 0.0018817625842872823,
      "grad_norm": 15.3951416015625,
      "learning_rate": 1.25e-06,
      "loss": 1.1479,
      "step": 3
    },
    {
      "epoch": 0.00250901677904971,
      "grad_norm": 17.10584259033203,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.0933,
      "step": 4
    },
    {
      "epoch": 0.0031362709738121373,
      "grad_norm": 12.109578132629395,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 1.0699,
      "step": 5
    },
    {
      "epoch": 0.0037635251685745647,
      "grad_norm": 8.512700080871582,
      "learning_rate": 2.5e-06,
      "loss": 0.9948,
      "step": 6
    },
    {
      "epoch": 0.0043907793633369925,
      "grad_norm": 8.072388648986816,
      "learning_rate": 2.916666666666667e-06,
      "loss": 1.0097,
      "step": 7
    },
    {
      "epoch": 0.00501803355809942,
      "grad_norm": 4.751190662384033,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.9145,
      "step": 8
    },
    {
      "epoch": 0.005645287752861847,
      "grad_norm": 5.013059139251709,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.9029,
      "step": 9
    },
    {
      "epoch": 0.006272541947624275,
      "grad_norm": 5.35429048538208,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.8575,
      "step": 10
    },
    {
      "epoch": 0.006899796142386702,
      "grad_norm": 3.0817456245422363,
      "learning_rate": 4.583333333333333e-06,
      "loss": 0.807,
      "step": 11
    },
    {
      "epoch": 0.007527050337149129,
      "grad_norm": 2.9373440742492676,
      "learning_rate": 5e-06,
      "loss": 0.7738,
      "step": 12
    },
    {
      "epoch": 0.008154304531911558,
      "grad_norm": 3.1255123615264893,
      "learning_rate": 5.416666666666667e-06,
      "loss": 0.717,
      "step": 13
    },
    {
      "epoch": 0.008781558726673985,
      "grad_norm": 2.4374232292175293,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.6959,
      "step": 14
    },
    {
      "epoch": 0.009408812921436412,
      "grad_norm": 3.297788619995117,
      "learning_rate": 6.25e-06,
      "loss": 0.7199,
      "step": 15
    },
    {
      "epoch": 0.01003606711619884,
      "grad_norm": 3.0335464477539062,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.6926,
      "step": 16
    },
    {
      "epoch": 0.010663321310961267,
      "grad_norm": 2.9025216102600098,
      "learning_rate": 7.083333333333335e-06,
      "loss": 0.6634,
      "step": 17
    },
    {
      "epoch": 0.011290575505723694,
      "grad_norm": 2.294412136077881,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.6045,
      "step": 18
    },
    {
      "epoch": 0.011917829700486122,
      "grad_norm": 2.4451427459716797,
      "learning_rate": 7.916666666666667e-06,
      "loss": 0.6442,
      "step": 19
    },
    {
      "epoch": 0.01254508389524855,
      "grad_norm": 2.1892669200897217,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.6858,
      "step": 20
    },
    {
      "epoch": 0.013172338090010977,
      "grad_norm": 2.2129485607147217,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.6401,
      "step": 21
    },
    {
      "epoch": 0.013799592284773404,
      "grad_norm": 2.105389356613159,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.6603,
      "step": 22
    },
    {
      "epoch": 0.014426846479535831,
      "grad_norm": 2.1527175903320312,
      "learning_rate": 9.583333333333335e-06,
      "loss": 0.5991,
      "step": 23
    },
    {
      "epoch": 0.015054100674298259,
      "grad_norm": 2.1053991317749023,
      "learning_rate": 1e-05,
      "loss": 0.6137,
      "step": 24
    },
    {
      "epoch": 0.015681354869060686,
      "grad_norm": 2.17954683303833,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 0.6025,
      "step": 25
    },
    {
      "epoch": 0.016308609063823115,
      "grad_norm": 1.9039045572280884,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 0.5373,
      "step": 26
    },
    {
      "epoch": 0.01693586325858554,
      "grad_norm": 3.1504151821136475,
      "learning_rate": 1.125e-05,
      "loss": 0.5966,
      "step": 27
    },
    {
      "epoch": 0.01756311745334797,
      "grad_norm": 2.2774252891540527,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.6107,
      "step": 28
    },
    {
      "epoch": 0.018190371648110396,
      "grad_norm": 1.9291549921035767,
      "learning_rate": 1.2083333333333333e-05,
      "loss": 0.5637,
      "step": 29
    },
    {
      "epoch": 0.018817625842872825,
      "grad_norm": 2.0674891471862793,
      "learning_rate": 1.25e-05,
      "loss": 0.5801,
      "step": 30
    },
    {
      "epoch": 0.01944488003763525,
      "grad_norm": 2.1760141849517822,
      "learning_rate": 1.2916666666666668e-05,
      "loss": 0.5665,
      "step": 31
    },
    {
      "epoch": 0.02007213423239768,
      "grad_norm": 2.295644998550415,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.6056,
      "step": 32
    },
    {
      "epoch": 0.020699388427160105,
      "grad_norm": 2.258507013320923,
      "learning_rate": 1.375e-05,
      "loss": 0.5676,
      "step": 33
    },
    {
      "epoch": 0.021326642621922534,
      "grad_norm": 2.356708526611328,
      "learning_rate": 1.416666666666667e-05,
      "loss": 0.5353,
      "step": 34
    },
    {
      "epoch": 0.021953896816684963,
      "grad_norm": 2.042067766189575,
      "learning_rate": 1.4583333333333333e-05,
      "loss": 0.5116,
      "step": 35
    },
    {
      "epoch": 0.02258115101144739,
      "grad_norm": 2.107545852661133,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.5721,
      "step": 36
    },
    {
      "epoch": 0.023208405206209818,
      "grad_norm": 2.1624722480773926,
      "learning_rate": 1.5416666666666668e-05,
      "loss": 0.5174,
      "step": 37
    },
    {
      "epoch": 0.023835659400972244,
      "grad_norm": 2.340648889541626,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 0.5976,
      "step": 38
    },
    {
      "epoch": 0.024462913595734673,
      "grad_norm": 2.218324661254883,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 0.5678,
      "step": 39
    },
    {
      "epoch": 0.0250901677904971,
      "grad_norm": 2.8050646781921387,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.594,
      "step": 40
    },
    {
      "epoch": 0.025717421985259527,
      "grad_norm": 2.405874013900757,
      "learning_rate": 1.7083333333333333e-05,
      "loss": 0.5582,
      "step": 41
    },
    {
      "epoch": 0.026344676180021953,
      "grad_norm": 2.1661484241485596,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.5674,
      "step": 42
    },
    {
      "epoch": 0.026971930374784382,
      "grad_norm": 2.0582382678985596,
      "learning_rate": 1.7916666666666667e-05,
      "loss": 0.5784,
      "step": 43
    },
    {
      "epoch": 0.027599184569546808,
      "grad_norm": 2.0946953296661377,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.5104,
      "step": 44
    },
    {
      "epoch": 0.028226438764309237,
      "grad_norm": 2.11618971824646,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.5508,
      "step": 45
    },
    {
      "epoch": 0.028853692959071663,
      "grad_norm": 2.330487012863159,
      "learning_rate": 1.916666666666667e-05,
      "loss": 0.5452,
      "step": 46
    },
    {
      "epoch": 0.02948094715383409,
      "grad_norm": 2.367173194885254,
      "learning_rate": 1.9583333333333333e-05,
      "loss": 0.5368,
      "step": 47
    },
    {
      "epoch": 0.030108201348596517,
      "grad_norm": 2.389894723892212,
      "learning_rate": 2e-05,
      "loss": 0.5093,
      "step": 48
    },
    {
      "epoch": 0.030735455543358946,
      "grad_norm": 2.0740127563476562,
      "learning_rate": 2.0416666666666667e-05,
      "loss": 0.5411,
      "step": 49
    },
    {
      "epoch": 0.03136270973812137,
      "grad_norm": 2.1907505989074707,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.5201,
      "step": 50
    },
    {
      "epoch": 0.0319899639328838,
      "grad_norm": 2.1322927474975586,
      "learning_rate": 2.125e-05,
      "loss": 0.5064,
      "step": 51
    },
    {
      "epoch": 0.03261721812764623,
      "grad_norm": 2.274514675140381,
      "learning_rate": 2.1666666666666667e-05,
      "loss": 0.5398,
      "step": 52
    },
    {
      "epoch": 0.03324447232240866,
      "grad_norm": 2.8248438835144043,
      "learning_rate": 2.2083333333333336e-05,
      "loss": 0.5083,
      "step": 53
    },
    {
      "epoch": 0.03387172651717108,
      "grad_norm": 2.0930676460266113,
      "learning_rate": 2.25e-05,
      "loss": 0.5392,
      "step": 54
    },
    {
      "epoch": 0.03449898071193351,
      "grad_norm": 2.262333869934082,
      "learning_rate": 2.2916666666666667e-05,
      "loss": 0.5642,
      "step": 55
    },
    {
      "epoch": 0.03512623490669594,
      "grad_norm": 2.224430799484253,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.5438,
      "step": 56
    },
    {
      "epoch": 0.03575348910145837,
      "grad_norm": 2.2733497619628906,
      "learning_rate": 2.375e-05,
      "loss": 0.5178,
      "step": 57
    },
    {
      "epoch": 0.03638074329622079,
      "grad_norm": 2.2176079750061035,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 0.5658,
      "step": 58
    },
    {
      "epoch": 0.03700799749098322,
      "grad_norm": 2.0688469409942627,
      "learning_rate": 2.4583333333333336e-05,
      "loss": 0.5296,
      "step": 59
    },
    {
      "epoch": 0.03763525168574565,
      "grad_norm": 2.167177677154541,
      "learning_rate": 2.5e-05,
      "loss": 0.5691,
      "step": 60
    },
    {
      "epoch": 0.03826250588050808,
      "grad_norm": 2.094510078430176,
      "learning_rate": 2.5416666666666667e-05,
      "loss": 0.5326,
      "step": 61
    },
    {
      "epoch": 0.0388897600752705,
      "grad_norm": 2.416938066482544,
      "learning_rate": 2.5833333333333336e-05,
      "loss": 0.547,
      "step": 62
    },
    {
      "epoch": 0.03951701427003293,
      "grad_norm": 2.5658774375915527,
      "learning_rate": 2.625e-05,
      "loss": 0.5641,
      "step": 63
    },
    {
      "epoch": 0.04014426846479536,
      "grad_norm": 3.0288596153259277,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.532,
      "step": 64
    },
    {
      "epoch": 0.04077152265955779,
      "grad_norm": 2.6260979175567627,
      "learning_rate": 2.7083333333333335e-05,
      "loss": 0.52,
      "step": 65
    },
    {
      "epoch": 0.04139877685432021,
      "grad_norm": 2.6695799827575684,
      "learning_rate": 2.75e-05,
      "loss": 0.5547,
      "step": 66
    },
    {
      "epoch": 0.04202603104908264,
      "grad_norm": 2.1383681297302246,
      "learning_rate": 2.7916666666666666e-05,
      "loss": 0.5475,
      "step": 67
    },
    {
      "epoch": 0.04265328524384507,
      "grad_norm": 2.1226611137390137,
      "learning_rate": 2.833333333333334e-05,
      "loss": 0.5785,
      "step": 68
    },
    {
      "epoch": 0.0432805394386075,
      "grad_norm": 2.173794746398926,
      "learning_rate": 2.875e-05,
      "loss": 0.5368,
      "step": 69
    },
    {
      "epoch": 0.043907793633369926,
      "grad_norm": 1.7689262628555298,
      "learning_rate": 2.9166666666666666e-05,
      "loss": 0.5106,
      "step": 70
    },
    {
      "epoch": 0.04453504782813235,
      "grad_norm": 2.331559419631958,
      "learning_rate": 2.958333333333334e-05,
      "loss": 0.5657,
      "step": 71
    },
    {
      "epoch": 0.04516230202289478,
      "grad_norm": 2.389425754547119,
      "learning_rate": 3.0000000000000004e-05,
      "loss": 0.5645,
      "step": 72
    },
    {
      "epoch": 0.04578955621765721,
      "grad_norm": 2.035191297531128,
      "learning_rate": 3.0416666666666666e-05,
      "loss": 0.5482,
      "step": 73
    },
    {
      "epoch": 0.046416810412419636,
      "grad_norm": 2.112544059753418,
      "learning_rate": 3.0833333333333335e-05,
      "loss": 0.577,
      "step": 74
    },
    {
      "epoch": 0.04704406460718206,
      "grad_norm": 1.9131213426589966,
      "learning_rate": 3.125e-05,
      "loss": 0.5463,
      "step": 75
    },
    {
      "epoch": 0.04767131880194449,
      "grad_norm": 1.978047490119934,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 0.5304,
      "step": 76
    },
    {
      "epoch": 0.048298572996706916,
      "grad_norm": 2.5002808570861816,
      "learning_rate": 3.208333333333334e-05,
      "loss": 0.5329,
      "step": 77
    },
    {
      "epoch": 0.048925827191469345,
      "grad_norm": 2.056502103805542,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.5628,
      "step": 78
    },
    {
      "epoch": 0.04955308138623177,
      "grad_norm": 2.020198106765747,
      "learning_rate": 3.291666666666667e-05,
      "loss": 0.569,
      "step": 79
    },
    {
      "epoch": 0.0501803355809942,
      "grad_norm": 2.3168647289276123,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.5478,
      "step": 80
    },
    {
      "epoch": 0.050807589775756626,
      "grad_norm": 2.414154529571533,
      "learning_rate": 3.375e-05,
      "loss": 0.5546,
      "step": 81
    },
    {
      "epoch": 0.051434843970519055,
      "grad_norm": 1.8864158391952515,
      "learning_rate": 3.4166666666666666e-05,
      "loss": 0.5245,
      "step": 82
    },
    {
      "epoch": 0.05206209816528148,
      "grad_norm": 2.200151205062866,
      "learning_rate": 3.458333333333334e-05,
      "loss": 0.4861,
      "step": 83
    },
    {
      "epoch": 0.052689352360043906,
      "grad_norm": 2.844407558441162,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 0.6039,
      "step": 84
    },
    {
      "epoch": 0.053316606554806335,
      "grad_norm": 2.2177329063415527,
      "learning_rate": 3.541666666666667e-05,
      "loss": 0.5096,
      "step": 85
    },
    {
      "epoch": 0.053943860749568764,
      "grad_norm": 1.8765463829040527,
      "learning_rate": 3.5833333333333335e-05,
      "loss": 0.5152,
      "step": 86
    },
    {
      "epoch": 0.054571114944331194,
      "grad_norm": 2.0037007331848145,
      "learning_rate": 3.625e-05,
      "loss": 0.5499,
      "step": 87
    },
    {
      "epoch": 0.055198369139093616,
      "grad_norm": 2.07822322845459,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 0.5642,
      "step": 88
    },
    {
      "epoch": 0.055825623333856045,
      "grad_norm": 1.2361040115356445,
      "learning_rate": 3.708333333333334e-05,
      "loss": 0.4945,
      "step": 89
    },
    {
      "epoch": 0.056452877528618474,
      "grad_norm": 2.390566349029541,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.5,
      "step": 90
    },
    {
      "epoch": 0.0570801317233809,
      "grad_norm": 2.0272109508514404,
      "learning_rate": 3.791666666666667e-05,
      "loss": 0.5207,
      "step": 91
    },
    {
      "epoch": 0.057707385918143325,
      "grad_norm": 2.093296527862549,
      "learning_rate": 3.833333333333334e-05,
      "loss": 0.5015,
      "step": 92
    },
    {
      "epoch": 0.058334640112905754,
      "grad_norm": 1.8016709089279175,
      "learning_rate": 3.875e-05,
      "loss": 0.5522,
      "step": 93
    },
    {
      "epoch": 0.05896189430766818,
      "grad_norm": 1.7712875604629517,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 0.5239,
      "step": 94
    },
    {
      "epoch": 0.05958914850243061,
      "grad_norm": 2.0883615016937256,
      "learning_rate": 3.958333333333334e-05,
      "loss": 0.5488,
      "step": 95
    },
    {
      "epoch": 0.060216402697193035,
      "grad_norm": 1.9757493734359741,
      "learning_rate": 4e-05,
      "loss": 0.542,
      "step": 96
    },
    {
      "epoch": 0.060843656891955464,
      "grad_norm": 2.7352821826934814,
      "learning_rate": 3.999998967664797e-05,
      "loss": 0.5771,
      "step": 97
    },
    {
      "epoch": 0.06147091108671789,
      "grad_norm": 1.8274654150009155,
      "learning_rate": 3.999995870660251e-05,
      "loss": 0.5859,
      "step": 98
    },
    {
      "epoch": 0.06209816528148032,
      "grad_norm": 1.8539522886276245,
      "learning_rate": 3.99999070898956e-05,
      "loss": 0.5153,
      "step": 99
    },
    {
      "epoch": 0.06272541947624274,
      "grad_norm": 2.2877085208892822,
      "learning_rate": 3.999983482658054e-05,
      "loss": 0.5611,
      "step": 100
    },
    {
      "epoch": 0.06335267367100518,
      "grad_norm": 2.228724718093872,
      "learning_rate": 3.99997419167319e-05,
      "loss": 0.5685,
      "step": 101
    },
    {
      "epoch": 0.0639799278657676,
      "grad_norm": 2.013684034347534,
      "learning_rate": 3.999962836044563e-05,
      "loss": 0.4979,
      "step": 102
    },
    {
      "epoch": 0.06460718206053002,
      "grad_norm": 2.16192889213562,
      "learning_rate": 3.999949415783893e-05,
      "loss": 0.5113,
      "step": 103
    },
    {
      "epoch": 0.06523443625529246,
      "grad_norm": 1.7808058261871338,
      "learning_rate": 3.9999339309050354e-05,
      "loss": 0.5392,
      "step": 104
    },
    {
      "epoch": 0.06586169045005488,
      "grad_norm": 1.9338231086730957,
      "learning_rate": 3.999916381423976e-05,
      "loss": 0.5294,
      "step": 105
    },
    {
      "epoch": 0.06648894464481732,
      "grad_norm": 1.8520095348358154,
      "learning_rate": 3.999896767358832e-05,
      "loss": 0.5084,
      "step": 106
    },
    {
      "epoch": 0.06711619883957974,
      "grad_norm": 2.6990466117858887,
      "learning_rate": 3.99987508872985e-05,
      "loss": 0.5662,
      "step": 107
    },
    {
      "epoch": 0.06774345303434216,
      "grad_norm": 2.1028778553009033,
      "learning_rate": 3.99985134555941e-05,
      "loss": 0.5066,
      "step": 108
    },
    {
      "epoch": 0.0683707072291046,
      "grad_norm": 2.1288583278656006,
      "learning_rate": 3.999825537872025e-05,
      "loss": 0.5326,
      "step": 109
    },
    {
      "epoch": 0.06899796142386702,
      "grad_norm": 2.033177375793457,
      "learning_rate": 3.999797665694335e-05,
      "loss": 0.5255,
      "step": 110
    },
    {
      "epoch": 0.06962521561862944,
      "grad_norm": 1.5184273719787598,
      "learning_rate": 3.999767729055115e-05,
      "loss": 0.5342,
      "step": 111
    },
    {
      "epoch": 0.07025246981339188,
      "grad_norm": 2.1982336044311523,
      "learning_rate": 3.999735727985268e-05,
      "loss": 0.5264,
      "step": 112
    },
    {
      "epoch": 0.0708797240081543,
      "grad_norm": 2.133439302444458,
      "learning_rate": 3.999701662517831e-05,
      "loss": 0.5293,
      "step": 113
    },
    {
      "epoch": 0.07150697820291674,
      "grad_norm": 2.059997320175171,
      "learning_rate": 3.99966553268797e-05,
      "loss": 0.5311,
      "step": 114
    },
    {
      "epoch": 0.07213423239767916,
      "grad_norm": 2.0948469638824463,
      "learning_rate": 3.9996273385329845e-05,
      "loss": 0.5141,
      "step": 115
    },
    {
      "epoch": 0.07276148659244158,
      "grad_norm": 1.8727495670318604,
      "learning_rate": 3.999587080092303e-05,
      "loss": 0.4775,
      "step": 116
    },
    {
      "epoch": 0.07338874078720402,
      "grad_norm": 1.7768906354904175,
      "learning_rate": 3.999544757407485e-05,
      "loss": 0.4938,
      "step": 117
    },
    {
      "epoch": 0.07401599498196644,
      "grad_norm": 1.797930121421814,
      "learning_rate": 3.9995003705222224e-05,
      "loss": 0.4991,
      "step": 118
    },
    {
      "epoch": 0.07464324917672886,
      "grad_norm": 1.623299479484558,
      "learning_rate": 3.9994539194823376e-05,
      "loss": 0.4888,
      "step": 119
    },
    {
      "epoch": 0.0752705033714913,
      "grad_norm": 1.712273359298706,
      "learning_rate": 3.999405404335783e-05,
      "loss": 0.4958,
      "step": 120
    },
    {
      "epoch": 0.07589775756625372,
      "grad_norm": 1.6322563886642456,
      "learning_rate": 3.999354825132644e-05,
      "loss": 0.5035,
      "step": 121
    },
    {
      "epoch": 0.07652501176101616,
      "grad_norm": 1.5734409093856812,
      "learning_rate": 3.999302181925133e-05,
      "loss": 0.5249,
      "step": 122
    },
    {
      "epoch": 0.07715226595577858,
      "grad_norm": 2.058497428894043,
      "learning_rate": 3.9992474747675965e-05,
      "loss": 0.5335,
      "step": 123
    },
    {
      "epoch": 0.077779520150541,
      "grad_norm": 1.8265260457992554,
      "learning_rate": 3.999190703716511e-05,
      "loss": 0.5389,
      "step": 124
    },
    {
      "epoch": 0.07840677434530344,
      "grad_norm": 1.323286533355713,
      "learning_rate": 3.999131868830482e-05,
      "loss": 0.4437,
      "step": 125
    },
    {
      "epoch": 0.07903402854006586,
      "grad_norm": 1.5716819763183594,
      "learning_rate": 3.999070970170249e-05,
      "loss": 0.5257,
      "step": 126
    },
    {
      "epoch": 0.0796612827348283,
      "grad_norm": 1.7783918380737305,
      "learning_rate": 3.999008007798678e-05,
      "loss": 0.5242,
      "step": 127
    },
    {
      "epoch": 0.08028853692959072,
      "grad_norm": 1.7182682752609253,
      "learning_rate": 3.9989429817807674e-05,
      "loss": 0.5095,
      "step": 128
    },
    {
      "epoch": 0.08091579112435314,
      "grad_norm": 1.639614462852478,
      "learning_rate": 3.998875892183647e-05,
      "loss": 0.5386,
      "step": 129
    },
    {
      "epoch": 0.08154304531911558,
      "grad_norm": 1.953802466392517,
      "learning_rate": 3.998806739076575e-05,
      "loss": 0.5062,
      "step": 130
    },
    {
      "epoch": 0.082170299513878,
      "grad_norm": 1.7502291202545166,
      "learning_rate": 3.998735522530941e-05,
      "loss": 0.5126,
      "step": 131
    },
    {
      "epoch": 0.08279755370864042,
      "grad_norm": 1.5992299318313599,
      "learning_rate": 3.9986622426202626e-05,
      "loss": 0.4931,
      "step": 132
    },
    {
      "epoch": 0.08342480790340286,
      "grad_norm": 1.9821819067001343,
      "learning_rate": 3.998586899420192e-05,
      "loss": 0.5361,
      "step": 133
    },
    {
      "epoch": 0.08405206209816528,
      "grad_norm": 1.9804344177246094,
      "learning_rate": 3.998509493008506e-05,
      "loss": 0.5656,
      "step": 134
    },
    {
      "epoch": 0.08467931629292771,
      "grad_norm": 2.0142171382904053,
      "learning_rate": 3.998430023465116e-05,
      "loss": 0.5339,
      "step": 135
    },
    {
      "epoch": 0.08530657048769014,
      "grad_norm": 1.8212288618087769,
      "learning_rate": 3.99834849087206e-05,
      "loss": 0.5231,
      "step": 136
    },
    {
      "epoch": 0.08593382468245256,
      "grad_norm": 2.456505298614502,
      "learning_rate": 3.9982648953135076e-05,
      "loss": 0.5488,
      "step": 137
    },
    {
      "epoch": 0.086561078877215,
      "grad_norm": 2.4705069065093994,
      "learning_rate": 3.998179236875757e-05,
      "loss": 0.5157,
      "step": 138
    },
    {
      "epoch": 0.08718833307197742,
      "grad_norm": 2.2800309658050537,
      "learning_rate": 3.9980915156472366e-05,
      "loss": 0.4889,
      "step": 139
    },
    {
      "epoch": 0.08781558726673985,
      "grad_norm": 1.8949161767959595,
      "learning_rate": 3.998001731718505e-05,
      "loss": 0.5009,
      "step": 140
    },
    {
      "epoch": 0.08844284146150228,
      "grad_norm": 1.7371482849121094,
      "learning_rate": 3.997909885182248e-05,
      "loss": 0.5241,
      "step": 141
    },
    {
      "epoch": 0.0890700956562647,
      "grad_norm": 2.2259998321533203,
      "learning_rate": 3.997815976133282e-05,
      "loss": 0.4674,
      "step": 142
    },
    {
      "epoch": 0.08969734985102713,
      "grad_norm": 2.1841187477111816,
      "learning_rate": 3.9977200046685536e-05,
      "loss": 0.4897,
      "step": 143
    },
    {
      "epoch": 0.09032460404578956,
      "grad_norm": 1.913406491279602,
      "learning_rate": 3.997621970887137e-05,
      "loss": 0.504,
      "step": 144
    },
    {
      "epoch": 0.09095185824055198,
      "grad_norm": 1.6738669872283936,
      "learning_rate": 3.9975218748902356e-05,
      "loss": 0.5081,
      "step": 145
    },
    {
      "epoch": 0.09157911243531441,
      "grad_norm": 2.280686378479004,
      "learning_rate": 3.997419716781183e-05,
      "loss": 0.5569,
      "step": 146
    },
    {
      "epoch": 0.09220636663007684,
      "grad_norm": 2.0637691020965576,
      "learning_rate": 3.9973154966654396e-05,
      "loss": 0.5153,
      "step": 147
    },
    {
      "epoch": 0.09283362082483927,
      "grad_norm": 1.6154152154922485,
      "learning_rate": 3.997209214650595e-05,
      "loss": 0.5022,
      "step": 148
    },
    {
      "epoch": 0.0934608750196017,
      "grad_norm": 1.67239511013031,
      "learning_rate": 3.99710087084637e-05,
      "loss": 0.4846,
      "step": 149
    },
    {
      "epoch": 0.09408812921436412,
      "grad_norm": 1.6062970161437988,
      "learning_rate": 3.99699046536461e-05,
      "loss": 0.5185,
      "step": 150
    },
    {
      "epoch": 0.09471538340912655,
      "grad_norm": 1.733093500137329,
      "learning_rate": 3.99687799831929e-05,
      "loss": 0.5387,
      "step": 151
    },
    {
      "epoch": 0.09534263760388897,
      "grad_norm": 1.7517509460449219,
      "learning_rate": 3.996763469826516e-05,
      "loss": 0.5322,
      "step": 152
    },
    {
      "epoch": 0.0959698917986514,
      "grad_norm": 1.6479005813598633,
      "learning_rate": 3.996646880004518e-05,
      "loss": 0.5165,
      "step": 153
    },
    {
      "epoch": 0.09659714599341383,
      "grad_norm": 1.7011340856552124,
      "learning_rate": 3.9965282289736557e-05,
      "loss": 0.5157,
      "step": 154
    },
    {
      "epoch": 0.09722440018817625,
      "grad_norm": 1.596718668937683,
      "learning_rate": 3.996407516856418e-05,
      "loss": 0.517,
      "step": 155
    },
    {
      "epoch": 0.09785165438293869,
      "grad_norm": 2.0577502250671387,
      "learning_rate": 3.996284743777419e-05,
      "loss": 0.5421,
      "step": 156
    },
    {
      "epoch": 0.09847890857770111,
      "grad_norm": 1.754554033279419,
      "learning_rate": 3.996159909863402e-05,
      "loss": 0.493,
      "step": 157
    },
    {
      "epoch": 0.09910616277246354,
      "grad_norm": 1.9648163318634033,
      "learning_rate": 3.996033015243238e-05,
      "loss": 0.5173,
      "step": 158
    },
    {
      "epoch": 0.09973341696722597,
      "grad_norm": 2.0128462314605713,
      "learning_rate": 3.995904060047924e-05,
      "loss": 0.4702,
      "step": 159
    },
    {
      "epoch": 0.1003606711619884,
      "grad_norm": 1.6790200471878052,
      "learning_rate": 3.995773044410585e-05,
      "loss": 0.4902,
      "step": 160
    },
    {
      "epoch": 0.10098792535675083,
      "grad_norm": 1.7891554832458496,
      "learning_rate": 3.995639968466475e-05,
      "loss": 0.4998,
      "step": 161
    },
    {
      "epoch": 0.10161517955151325,
      "grad_norm": 1.5091252326965332,
      "learning_rate": 3.99550483235297e-05,
      "loss": 0.4433,
      "step": 162
    },
    {
      "epoch": 0.10224243374627567,
      "grad_norm": 1.6836857795715332,
      "learning_rate": 3.9953676362095774e-05,
      "loss": 0.5127,
      "step": 163
    },
    {
      "epoch": 0.10286968794103811,
      "grad_norm": 1.5248311758041382,
      "learning_rate": 3.99522838017793e-05,
      "loss": 0.4925,
      "step": 164
    },
    {
      "epoch": 0.10349694213580053,
      "grad_norm": 1.7744476795196533,
      "learning_rate": 3.9950870644017854e-05,
      "loss": 0.4807,
      "step": 165
    },
    {
      "epoch": 0.10412419633056295,
      "grad_norm": 2.1345975399017334,
      "learning_rate": 3.99494368902703e-05,
      "loss": 0.5226,
      "step": 166
    },
    {
      "epoch": 0.10475145052532539,
      "grad_norm": 1.655029535293579,
      "learning_rate": 3.9947982542016744e-05,
      "loss": 0.5134,
      "step": 167
    },
    {
      "epoch": 0.10537870472008781,
      "grad_norm": 1.9914706945419312,
      "learning_rate": 3.9946507600758565e-05,
      "loss": 0.5149,
      "step": 168
    },
    {
      "epoch": 0.10600595891485025,
      "grad_norm": 1.765305757522583,
      "learning_rate": 3.9945012068018396e-05,
      "loss": 0.4751,
      "step": 169
    },
    {
      "epoch": 0.10663321310961267,
      "grad_norm": 1.4296326637268066,
      "learning_rate": 3.9943495945340134e-05,
      "loss": 0.4881,
      "step": 170
    },
    {
      "epoch": 0.10726046730437509,
      "grad_norm": 1.626119613647461,
      "learning_rate": 3.994195923428891e-05,
      "loss": 0.4823,
      "step": 171
    },
    {
      "epoch": 0.10788772149913753,
      "grad_norm": 1.6655110120773315,
      "learning_rate": 3.994040193645114e-05,
      "loss": 0.4852,
      "step": 172
    },
    {
      "epoch": 0.10851497569389995,
      "grad_norm": 1.44282066822052,
      "learning_rate": 3.993882405343448e-05,
      "loss": 0.4823,
      "step": 173
    },
    {
      "epoch": 0.10914222988866239,
      "grad_norm": 1.7007629871368408,
      "learning_rate": 3.9937225586867824e-05,
      "loss": 0.4787,
      "step": 174
    },
    {
      "epoch": 0.10976948408342481,
      "grad_norm": 1.6929346323013306,
      "learning_rate": 3.993560653840133e-05,
      "loss": 0.4838,
      "step": 175
    },
    {
      "epoch": 0.11039673827818723,
      "grad_norm": 1.8040822744369507,
      "learning_rate": 3.993396690970639e-05,
      "loss": 0.5002,
      "step": 176
    },
    {
      "epoch": 0.11102399247294967,
      "grad_norm": 1.395953893661499,
      "learning_rate": 3.9932306702475655e-05,
      "loss": 0.4543,
      "step": 177
    },
    {
      "epoch": 0.11165124666771209,
      "grad_norm": 2.022174119949341,
      "learning_rate": 3.993062591842303e-05,
      "loss": 0.4643,
      "step": 178
    },
    {
      "epoch": 0.11227850086247451,
      "grad_norm": 1.7045681476593018,
      "learning_rate": 3.9928924559283626e-05,
      "loss": 0.5578,
      "step": 179
    },
    {
      "epoch": 0.11290575505723695,
      "grad_norm": 1.59603750705719,
      "learning_rate": 3.992720262681383e-05,
      "loss": 0.4858,
      "step": 180
    },
    {
      "epoch": 0.11353300925199937,
      "grad_norm": 1.837712287902832,
      "learning_rate": 3.992546012279125e-05,
      "loss": 0.5204,
      "step": 181
    },
    {
      "epoch": 0.1141602634467618,
      "grad_norm": 1.828675389289856,
      "learning_rate": 3.9923697049014724e-05,
      "loss": 0.4654,
      "step": 182
    },
    {
      "epoch": 0.11478751764152423,
      "grad_norm": 2.048175811767578,
      "learning_rate": 3.992191340730435e-05,
      "loss": 0.4729,
      "step": 183
    },
    {
      "epoch": 0.11541477183628665,
      "grad_norm": 1.7749079465866089,
      "learning_rate": 3.9920109199501435e-05,
      "loss": 0.4678,
      "step": 184
    },
    {
      "epoch": 0.11604202603104909,
      "grad_norm": 2.1206958293914795,
      "learning_rate": 3.9918284427468536e-05,
      "loss": 0.508,
      "step": 185
    },
    {
      "epoch": 0.11666928022581151,
      "grad_norm": 1.678009271621704,
      "learning_rate": 3.991643909308942e-05,
      "loss": 0.5016,
      "step": 186
    },
    {
      "epoch": 0.11729653442057393,
      "grad_norm": 1.4095009565353394,
      "learning_rate": 3.9914573198269086e-05,
      "loss": 0.4784,
      "step": 187
    },
    {
      "epoch": 0.11792378861533637,
      "grad_norm": 1.7792853116989136,
      "learning_rate": 3.991268674493378e-05,
      "loss": 0.4687,
      "step": 188
    },
    {
      "epoch": 0.11855104281009879,
      "grad_norm": 1.4806967973709106,
      "learning_rate": 3.991077973503094e-05,
      "loss": 0.4812,
      "step": 189
    },
    {
      "epoch": 0.11917829700486123,
      "grad_norm": 1.5337979793548584,
      "learning_rate": 3.990885217052925e-05,
      "loss": 0.4738,
      "step": 190
    },
    {
      "epoch": 0.11980555119962365,
      "grad_norm": 1.493370771408081,
      "learning_rate": 3.990690405341859e-05,
      "loss": 0.4698,
      "step": 191
    },
    {
      "epoch": 0.12043280539438607,
      "grad_norm": 1.6235061883926392,
      "learning_rate": 3.990493538571009e-05,
      "loss": 0.4647,
      "step": 192
    },
    {
      "epoch": 0.1210600595891485,
      "grad_norm": 1.65871000289917,
      "learning_rate": 3.9902946169436045e-05,
      "loss": 0.4461,
      "step": 193
    },
    {
      "epoch": 0.12168731378391093,
      "grad_norm": 1.5810375213623047,
      "learning_rate": 3.9900936406650016e-05,
      "loss": 0.4626,
      "step": 194
    },
    {
      "epoch": 0.12231456797867336,
      "grad_norm": 1.635001301765442,
      "learning_rate": 3.989890609942674e-05,
      "loss": 0.4851,
      "step": 195
    },
    {
      "epoch": 0.12294182217343579,
      "grad_norm": 1.480832576751709,
      "learning_rate": 3.989685524986219e-05,
      "loss": 0.465,
      "step": 196
    },
    {
      "epoch": 0.12356907636819821,
      "grad_norm": 1.6289597749710083,
      "learning_rate": 3.989478386007351e-05,
      "loss": 0.4996,
      "step": 197
    },
    {
      "epoch": 0.12419633056296064,
      "grad_norm": 1.557519793510437,
      "learning_rate": 3.989269193219909e-05,
      "loss": 0.425,
      "step": 198
    },
    {
      "epoch": 0.12482358475772307,
      "grad_norm": 1.946897029876709,
      "learning_rate": 3.989057946839848e-05,
      "loss": 0.4981,
      "step": 199
    },
    {
      "epoch": 0.1254508389524855,
      "grad_norm": 1.511824369430542,
      "learning_rate": 3.9888446470852464e-05,
      "loss": 0.4981,
      "step": 200
    },
    {
      "epoch": 0.1260780931472479,
      "grad_norm": 1.5527335405349731,
      "learning_rate": 3.988629294176301e-05,
      "loss": 0.479,
      "step": 201
    },
    {
      "epoch": 0.12670534734201036,
      "grad_norm": 1.5926644802093506,
      "learning_rate": 3.9884118883353275e-05,
      "loss": 0.438,
      "step": 202
    },
    {
      "epoch": 0.12733260153677278,
      "grad_norm": 1.6053894758224487,
      "learning_rate": 3.988192429786762e-05,
      "loss": 0.4984,
      "step": 203
    },
    {
      "epoch": 0.1279598557315352,
      "grad_norm": 1.9284647703170776,
      "learning_rate": 3.987970918757159e-05,
      "loss": 0.4753,
      "step": 204
    },
    {
      "epoch": 0.12858710992629763,
      "grad_norm": 1.6620078086853027,
      "learning_rate": 3.9877473554751935e-05,
      "loss": 0.4458,
      "step": 205
    },
    {
      "epoch": 0.12921436412106005,
      "grad_norm": 1.523679256439209,
      "learning_rate": 3.9875217401716557e-05,
      "loss": 0.477,
      "step": 206
    },
    {
      "epoch": 0.1298416183158225,
      "grad_norm": 1.6295181512832642,
      "learning_rate": 3.987294073079457e-05,
      "loss": 0.5062,
      "step": 207
    },
    {
      "epoch": 0.13046887251058492,
      "grad_norm": 1.52949059009552,
      "learning_rate": 3.987064354433627e-05,
      "loss": 0.4642,
      "step": 208
    },
    {
      "epoch": 0.13109612670534734,
      "grad_norm": 1.5415035486221313,
      "learning_rate": 3.9868325844713115e-05,
      "loss": 0.4537,
      "step": 209
    },
    {
      "epoch": 0.13172338090010977,
      "grad_norm": 1.7037349939346313,
      "learning_rate": 3.986598763431775e-05,
      "loss": 0.4638,
      "step": 210
    },
    {
      "epoch": 0.1323506350948722,
      "grad_norm": 1.8606200218200684,
      "learning_rate": 3.986362891556398e-05,
      "loss": 0.485,
      "step": 211
    },
    {
      "epoch": 0.13297788928963464,
      "grad_norm": 1.6160262823104858,
      "learning_rate": 3.986124969088682e-05,
      "loss": 0.4789,
      "step": 212
    },
    {
      "epoch": 0.13360514348439706,
      "grad_norm": 1.791393518447876,
      "learning_rate": 3.9858849962742404e-05,
      "loss": 0.4621,
      "step": 213
    },
    {
      "epoch": 0.13423239767915948,
      "grad_norm": 1.5315974950790405,
      "learning_rate": 3.9856429733608066e-05,
      "loss": 0.4739,
      "step": 214
    },
    {
      "epoch": 0.1348596518739219,
      "grad_norm": 1.6416422128677368,
      "learning_rate": 3.98539890059823e-05,
      "loss": 0.4703,
      "step": 215
    },
    {
      "epoch": 0.13548690606868433,
      "grad_norm": 1.4342625141143799,
      "learning_rate": 3.985152778238474e-05,
      "loss": 0.4601,
      "step": 216
    },
    {
      "epoch": 0.13611416026344675,
      "grad_norm": 1.5981355905532837,
      "learning_rate": 3.98490460653562e-05,
      "loss": 0.4549,
      "step": 217
    },
    {
      "epoch": 0.1367414144582092,
      "grad_norm": 1.8291752338409424,
      "learning_rate": 3.9846543857458645e-05,
      "loss": 0.4482,
      "step": 218
    },
    {
      "epoch": 0.13736866865297162,
      "grad_norm": 1.4503984451293945,
      "learning_rate": 3.9844021161275186e-05,
      "loss": 0.4647,
      "step": 219
    },
    {
      "epoch": 0.13799592284773404,
      "grad_norm": 1.7972581386566162,
      "learning_rate": 3.984147797941011e-05,
      "loss": 0.4419,
      "step": 220
    },
    {
      "epoch": 0.13862317704249646,
      "grad_norm": 1.6375170946121216,
      "learning_rate": 3.983891431448881e-05,
      "loss": 0.461,
      "step": 221
    },
    {
      "epoch": 0.1392504312372589,
      "grad_norm": 1.7466061115264893,
      "learning_rate": 3.983633016915786e-05,
      "loss": 0.4601,
      "step": 222
    },
    {
      "epoch": 0.13987768543202134,
      "grad_norm": 1.5807160139083862,
      "learning_rate": 3.9833725546084966e-05,
      "loss": 0.4961,
      "step": 223
    },
    {
      "epoch": 0.14050493962678376,
      "grad_norm": 1.974726676940918,
      "learning_rate": 3.9831100447958963e-05,
      "loss": 0.4462,
      "step": 224
    },
    {
      "epoch": 0.14113219382154618,
      "grad_norm": 1.9808381795883179,
      "learning_rate": 3.982845487748984e-05,
      "loss": 0.4914,
      "step": 225
    },
    {
      "epoch": 0.1417594480163086,
      "grad_norm": 1.7497327327728271,
      "learning_rate": 3.982578883740871e-05,
      "loss": 0.4705,
      "step": 226
    },
    {
      "epoch": 0.14238670221107103,
      "grad_norm": 2.0800588130950928,
      "learning_rate": 3.982310233046781e-05,
      "loss": 0.4661,
      "step": 227
    },
    {
      "epoch": 0.14301395640583348,
      "grad_norm": 1.6125692129135132,
      "learning_rate": 3.982039535944054e-05,
      "loss": 0.4418,
      "step": 228
    },
    {
      "epoch": 0.1436412106005959,
      "grad_norm": 1.6863831281661987,
      "learning_rate": 3.9817667927121376e-05,
      "loss": 0.4683,
      "step": 229
    },
    {
      "epoch": 0.14426846479535832,
      "grad_norm": 1.620818018913269,
      "learning_rate": 3.981492003632596e-05,
      "loss": 0.4375,
      "step": 230
    },
    {
      "epoch": 0.14489571899012074,
      "grad_norm": 1.3409901857376099,
      "learning_rate": 3.9812151689891026e-05,
      "loss": 0.4468,
      "step": 231
    },
    {
      "epoch": 0.14552297318488316,
      "grad_norm": 1.4613945484161377,
      "learning_rate": 3.980936289067444e-05,
      "loss": 0.4528,
      "step": 232
    },
    {
      "epoch": 0.14615022737964561,
      "grad_norm": 1.4099844694137573,
      "learning_rate": 3.980655364155518e-05,
      "loss": 0.454,
      "step": 233
    },
    {
      "epoch": 0.14677748157440804,
      "grad_norm": 1.5455256700515747,
      "learning_rate": 3.980372394543332e-05,
      "loss": 0.4494,
      "step": 234
    },
    {
      "epoch": 0.14740473576917046,
      "grad_norm": 1.351446509361267,
      "learning_rate": 3.9800873805230075e-05,
      "loss": 0.4479,
      "step": 235
    },
    {
      "epoch": 0.14803198996393288,
      "grad_norm": 1.8010048866271973,
      "learning_rate": 3.979800322388773e-05,
      "loss": 0.5094,
      "step": 236
    },
    {
      "epoch": 0.1486592441586953,
      "grad_norm": 1.400158166885376,
      "learning_rate": 3.979511220436969e-05,
      "loss": 0.4157,
      "step": 237
    },
    {
      "epoch": 0.14928649835345773,
      "grad_norm": 1.4748667478561401,
      "learning_rate": 3.979220074966046e-05,
      "loss": 0.4776,
      "step": 238
    },
    {
      "epoch": 0.14991375254822017,
      "grad_norm": 1.5479722023010254,
      "learning_rate": 3.978926886276564e-05,
      "loss": 0.4863,
      "step": 239
    },
    {
      "epoch": 0.1505410067429826,
      "grad_norm": 1.7338252067565918,
      "learning_rate": 3.978631654671191e-05,
      "loss": 0.4525,
      "step": 240
    },
    {
      "epoch": 0.15116826093774502,
      "grad_norm": 1.8159891366958618,
      "learning_rate": 3.9783343804547055e-05,
      "loss": 0.5029,
      "step": 241
    },
    {
      "epoch": 0.15179551513250744,
      "grad_norm": 1.8051979541778564,
      "learning_rate": 3.978035063933994e-05,
      "loss": 0.406,
      "step": 242
    },
    {
      "epoch": 0.15242276932726986,
      "grad_norm": 1.5267044305801392,
      "learning_rate": 3.977733705418052e-05,
      "loss": 0.4829,
      "step": 243
    },
    {
      "epoch": 0.1530500235220323,
      "grad_norm": 1.4523890018463135,
      "learning_rate": 3.9774303052179816e-05,
      "loss": 0.4392,
      "step": 244
    },
    {
      "epoch": 0.15367727771679474,
      "grad_norm": 1.6153901815414429,
      "learning_rate": 3.977124863646994e-05,
      "loss": 0.4849,
      "step": 245
    },
    {
      "epoch": 0.15430453191155716,
      "grad_norm": 1.575958013534546,
      "learning_rate": 3.976817381020408e-05,
      "loss": 0.4659,
      "step": 246
    },
    {
      "epoch": 0.15493178610631958,
      "grad_norm": 1.4412821531295776,
      "learning_rate": 3.976507857655648e-05,
      "loss": 0.4048,
      "step": 247
    },
    {
      "epoch": 0.155559040301082,
      "grad_norm": 1.5614181756973267,
      "learning_rate": 3.976196293872246e-05,
      "loss": 0.439,
      "step": 248
    },
    {
      "epoch": 0.15618629449584445,
      "grad_norm": 1.375407099723816,
      "learning_rate": 3.975882689991839e-05,
      "loss": 0.4249,
      "step": 249
    },
    {
      "epoch": 0.15681354869060687,
      "grad_norm": 1.4382860660552979,
      "learning_rate": 3.9755670463381735e-05,
      "loss": 0.4281,
      "step": 250
    },
    {
      "epoch": 0.1574408028853693,
      "grad_norm": 1.3078750371932983,
      "learning_rate": 3.9752493632370983e-05,
      "loss": 0.4415,
      "step": 251
    },
    {
      "epoch": 0.15806805708013172,
      "grad_norm": 1.4344347715377808,
      "learning_rate": 3.9749296410165694e-05,
      "loss": 0.4135,
      "step": 252
    },
    {
      "epoch": 0.15869531127489414,
      "grad_norm": 1.4680067300796509,
      "learning_rate": 3.974607880006647e-05,
      "loss": 0.4178,
      "step": 253
    },
    {
      "epoch": 0.1593225654696566,
      "grad_norm": 1.4680876731872559,
      "learning_rate": 3.974284080539496e-05,
      "loss": 0.4249,
      "step": 254
    },
    {
      "epoch": 0.159949819664419,
      "grad_norm": 1.4555891752243042,
      "learning_rate": 3.973958242949387e-05,
      "loss": 0.4534,
      "step": 255
    },
    {
      "epoch": 0.16057707385918144,
      "grad_norm": 1.4939229488372803,
      "learning_rate": 3.973630367572692e-05,
      "loss": 0.5196,
      "step": 256
    },
    {
      "epoch": 0.16120432805394386,
      "grad_norm": 1.591544270515442,
      "learning_rate": 3.9733004547478897e-05,
      "loss": 0.4704,
      "step": 257
    },
    {
      "epoch": 0.16183158224870628,
      "grad_norm": 1.500394582748413,
      "learning_rate": 3.9729685048155604e-05,
      "loss": 0.4432,
      "step": 258
    },
    {
      "epoch": 0.16245883644346873,
      "grad_norm": 1.663857102394104,
      "learning_rate": 3.972634518118387e-05,
      "loss": 0.4546,
      "step": 259
    },
    {
      "epoch": 0.16308609063823115,
      "grad_norm": 1.3415054082870483,
      "learning_rate": 3.972298495001157e-05,
      "loss": 0.4328,
      "step": 260
    },
    {
      "epoch": 0.16371334483299357,
      "grad_norm": 1.4470168352127075,
      "learning_rate": 3.9719604358107576e-05,
      "loss": 0.3941,
      "step": 261
    },
    {
      "epoch": 0.164340599027756,
      "grad_norm": 1.5569263696670532,
      "learning_rate": 3.97162034089618e-05,
      "loss": 0.4389,
      "step": 262
    },
    {
      "epoch": 0.16496785322251842,
      "grad_norm": 1.9183214902877808,
      "learning_rate": 3.971278210608516e-05,
      "loss": 0.4015,
      "step": 263
    },
    {
      "epoch": 0.16559510741728084,
      "grad_norm": 1.5554015636444092,
      "learning_rate": 3.9709340453009584e-05,
      "loss": 0.4753,
      "step": 264
    },
    {
      "epoch": 0.1662223616120433,
      "grad_norm": 1.5944366455078125,
      "learning_rate": 3.970587845328802e-05,
      "loss": 0.4813,
      "step": 265
    },
    {
      "epoch": 0.1668496158068057,
      "grad_norm": 1.6086550951004028,
      "learning_rate": 3.97023961104944e-05,
      "loss": 0.5004,
      "step": 266
    },
    {
      "epoch": 0.16747687000156813,
      "grad_norm": 1.234297752380371,
      "learning_rate": 3.9698893428223674e-05,
      "loss": 0.4305,
      "step": 267
    },
    {
      "epoch": 0.16810412419633056,
      "grad_norm": 1.2687492370605469,
      "learning_rate": 3.969537041009179e-05,
      "loss": 0.4292,
      "step": 268
    },
    {
      "epoch": 0.16873137839109298,
      "grad_norm": 1.5316917896270752,
      "learning_rate": 3.9691827059735676e-05,
      "loss": 0.4486,
      "step": 269
    },
    {
      "epoch": 0.16935863258585543,
      "grad_norm": 1.5229873657226562,
      "learning_rate": 3.968826338081327e-05,
      "loss": 0.4321,
      "step": 270
    },
    {
      "epoch": 0.16998588678061785,
      "grad_norm": 1.6713693141937256,
      "learning_rate": 3.968467937700347e-05,
      "loss": 0.4628,
      "step": 271
    },
    {
      "epoch": 0.17061314097538027,
      "grad_norm": 1.7388298511505127,
      "learning_rate": 3.968107505200617e-05,
      "loss": 0.4182,
      "step": 272
    },
    {
      "epoch": 0.1712403951701427,
      "grad_norm": 1.4563961029052734,
      "learning_rate": 3.9677450409542243e-05,
      "loss": 0.405,
      "step": 273
    },
    {
      "epoch": 0.17186764936490512,
      "grad_norm": 1.4333897829055786,
      "learning_rate": 3.967380545335354e-05,
      "loss": 0.42,
      "step": 274
    },
    {
      "epoch": 0.17249490355966757,
      "grad_norm": 1.6393811702728271,
      "learning_rate": 3.9670140187202875e-05,
      "loss": 0.4636,
      "step": 275
    },
    {
      "epoch": 0.17312215775443,
      "grad_norm": 1.5931137800216675,
      "learning_rate": 3.9666454614874035e-05,
      "loss": 0.4459,
      "step": 276
    },
    {
      "epoch": 0.1737494119491924,
      "grad_norm": 1.5716625452041626,
      "learning_rate": 3.966274874017176e-05,
      "loss": 0.4184,
      "step": 277
    },
    {
      "epoch": 0.17437666614395483,
      "grad_norm": 1.7171566486358643,
      "learning_rate": 3.9659022566921755e-05,
      "loss": 0.4256,
      "step": 278
    },
    {
      "epoch": 0.17500392033871726,
      "grad_norm": 1.461174488067627,
      "learning_rate": 3.965527609897068e-05,
      "loss": 0.3932,
      "step": 279
    },
    {
      "epoch": 0.1756311745334797,
      "grad_norm": 1.4326555728912354,
      "learning_rate": 3.965150934018614e-05,
      "loss": 0.3798,
      "step": 280
    },
    {
      "epoch": 0.17625842872824213,
      "grad_norm": 1.4033962488174438,
      "learning_rate": 3.964772229445671e-05,
      "loss": 0.3764,
      "step": 281
    },
    {
      "epoch": 0.17688568292300455,
      "grad_norm": 1.3416073322296143,
      "learning_rate": 3.964391496569188e-05,
      "loss": 0.4138,
      "step": 282
    },
    {
      "epoch": 0.17751293711776697,
      "grad_norm": 1.5007586479187012,
      "learning_rate": 3.964008735782209e-05,
      "loss": 0.4117,
      "step": 283
    },
    {
      "epoch": 0.1781401913125294,
      "grad_norm": 1.332431674003601,
      "learning_rate": 3.963623947479871e-05,
      "loss": 0.4387,
      "step": 284
    },
    {
      "epoch": 0.17876744550729182,
      "grad_norm": 1.4107450246810913,
      "learning_rate": 3.963237132059405e-05,
      "loss": 0.4079,
      "step": 285
    },
    {
      "epoch": 0.17939469970205427,
      "grad_norm": 1.5216684341430664,
      "learning_rate": 3.9628482899201347e-05,
      "loss": 0.4097,
      "step": 286
    },
    {
      "epoch": 0.1800219538968167,
      "grad_norm": 1.3541133403778076,
      "learning_rate": 3.962457421463475e-05,
      "loss": 0.4188,
      "step": 287
    },
    {
      "epoch": 0.1806492080915791,
      "grad_norm": 1.3088887929916382,
      "learning_rate": 3.9620645270929324e-05,
      "loss": 0.4739,
      "step": 288
    },
    {
      "epoch": 0.18127646228634153,
      "grad_norm": 1.384234070777893,
      "learning_rate": 3.961669607214106e-05,
      "loss": 0.3998,
      "step": 289
    },
    {
      "epoch": 0.18190371648110396,
      "grad_norm": 1.684230089187622,
      "learning_rate": 3.961272662234687e-05,
      "loss": 0.4014,
      "step": 290
    },
    {
      "epoch": 0.1825309706758664,
      "grad_norm": 1.3789516687393188,
      "learning_rate": 3.960873692564454e-05,
      "loss": 0.4188,
      "step": 291
    },
    {
      "epoch": 0.18315822487062883,
      "grad_norm": 1.4697010517120361,
      "learning_rate": 3.9604726986152784e-05,
      "loss": 0.4248,
      "step": 292
    },
    {
      "epoch": 0.18378547906539125,
      "grad_norm": 1.6977431774139404,
      "learning_rate": 3.9600696808011194e-05,
      "loss": 0.429,
      "step": 293
    },
    {
      "epoch": 0.18441273326015367,
      "grad_norm": 1.623642921447754,
      "learning_rate": 3.959664639538027e-05,
      "loss": 0.3813,
      "step": 294
    },
    {
      "epoch": 0.1850399874549161,
      "grad_norm": 1.4522424936294556,
      "learning_rate": 3.959257575244139e-05,
      "loss": 0.4441,
      "step": 295
    },
    {
      "epoch": 0.18566724164967854,
      "grad_norm": 1.3811010122299194,
      "learning_rate": 3.9588484883396836e-05,
      "loss": 0.4286,
      "step": 296
    },
    {
      "epoch": 0.18629449584444097,
      "grad_norm": 1.416077733039856,
      "learning_rate": 3.958437379246975e-05,
      "loss": 0.4021,
      "step": 297
    },
    {
      "epoch": 0.1869217500392034,
      "grad_norm": 1.5948553085327148,
      "learning_rate": 3.958024248390414e-05,
      "loss": 0.4049,
      "step": 298
    },
    {
      "epoch": 0.1875490042339658,
      "grad_norm": 1.7502264976501465,
      "learning_rate": 3.9576090961964925e-05,
      "loss": 0.458,
      "step": 299
    },
    {
      "epoch": 0.18817625842872823,
      "grad_norm": 1.3904632329940796,
      "learning_rate": 3.957191923093785e-05,
      "loss": 0.4346,
      "step": 300
    },
    {
      "epoch": 0.18880351262349068,
      "grad_norm": 1.64029061794281,
      "learning_rate": 3.956772729512955e-05,
      "loss": 0.4512,
      "step": 301
    },
    {
      "epoch": 0.1894307668182531,
      "grad_norm": 1.4691493511199951,
      "learning_rate": 3.956351515886751e-05,
      "loss": 0.4115,
      "step": 302
    },
    {
      "epoch": 0.19005802101301553,
      "grad_norm": 1.3712893724441528,
      "learning_rate": 3.955928282650005e-05,
      "loss": 0.3901,
      "step": 303
    },
    {
      "epoch": 0.19068527520777795,
      "grad_norm": 1.2790814638137817,
      "learning_rate": 3.955503030239638e-05,
      "loss": 0.4318,
      "step": 304
    },
    {
      "epoch": 0.19131252940254037,
      "grad_norm": 1.3729325532913208,
      "learning_rate": 3.9550757590946505e-05,
      "loss": 0.4505,
      "step": 305
    },
    {
      "epoch": 0.1919397835973028,
      "grad_norm": 1.6766351461410522,
      "learning_rate": 3.954646469656132e-05,
      "loss": 0.4333,
      "step": 306
    },
    {
      "epoch": 0.19256703779206524,
      "grad_norm": 1.444437861442566,
      "learning_rate": 3.95421516236725e-05,
      "loss": 0.4192,
      "step": 307
    },
    {
      "epoch": 0.19319429198682767,
      "grad_norm": 1.205487847328186,
      "learning_rate": 3.953781837673262e-05,
      "loss": 0.3647,
      "step": 308
    },
    {
      "epoch": 0.1938215461815901,
      "grad_norm": 1.4272749423980713,
      "learning_rate": 3.953346496021501e-05,
      "loss": 0.41,
      "step": 309
    },
    {
      "epoch": 0.1944488003763525,
      "grad_norm": 1.4003968238830566,
      "learning_rate": 3.952909137861388e-05,
      "loss": 0.3954,
      "step": 310
    },
    {
      "epoch": 0.19507605457111493,
      "grad_norm": 1.4121627807617188,
      "learning_rate": 3.952469763644421e-05,
      "loss": 0.3789,
      "step": 311
    },
    {
      "epoch": 0.19570330876587738,
      "grad_norm": 1.2762525081634521,
      "learning_rate": 3.952028373824183e-05,
      "loss": 0.4221,
      "step": 312
    },
    {
      "epoch": 0.1963305629606398,
      "grad_norm": 1.363707184791565,
      "learning_rate": 3.951584968856336e-05,
      "loss": 0.3754,
      "step": 313
    },
    {
      "epoch": 0.19695781715540223,
      "grad_norm": 1.5307010412216187,
      "learning_rate": 3.951139549198623e-05,
      "loss": 0.4095,
      "step": 314
    },
    {
      "epoch": 0.19758507135016465,
      "grad_norm": 1.5087469816207886,
      "learning_rate": 3.9506921153108656e-05,
      "loss": 0.4083,
      "step": 315
    },
    {
      "epoch": 0.19821232554492707,
      "grad_norm": 1.495043396949768,
      "learning_rate": 3.950242667654965e-05,
      "loss": 0.4412,
      "step": 316
    },
    {
      "epoch": 0.19883957973968952,
      "grad_norm": 1.4186872243881226,
      "learning_rate": 3.949791206694903e-05,
      "loss": 0.4067,
      "step": 317
    },
    {
      "epoch": 0.19946683393445194,
      "grad_norm": 1.3536276817321777,
      "learning_rate": 3.9493377328967384e-05,
      "loss": 0.4339,
      "step": 318
    },
    {
      "epoch": 0.20009408812921436,
      "grad_norm": 1.5129317045211792,
      "learning_rate": 3.948882246728608e-05,
      "loss": 0.5023,
      "step": 319
    },
    {
      "epoch": 0.2007213423239768,
      "grad_norm": 1.2687675952911377,
      "learning_rate": 3.948424748660726e-05,
      "loss": 0.4013,
      "step": 320
    },
    {
      "epoch": 0.2013485965187392,
      "grad_norm": 1.4301279783248901,
      "learning_rate": 3.9479652391653835e-05,
      "loss": 0.4344,
      "step": 321
    },
    {
      "epoch": 0.20197585071350166,
      "grad_norm": 1.4821714162826538,
      "learning_rate": 3.94750371871695e-05,
      "loss": 0.4155,
      "step": 322
    },
    {
      "epoch": 0.20260310490826408,
      "grad_norm": 1.252396821975708,
      "learning_rate": 3.947040187791867e-05,
      "loss": 0.3807,
      "step": 323
    },
    {
      "epoch": 0.2032303591030265,
      "grad_norm": 1.598359227180481,
      "learning_rate": 3.946574646868655e-05,
      "loss": 0.4269,
      "step": 324
    },
    {
      "epoch": 0.20385761329778893,
      "grad_norm": 1.3481076955795288,
      "learning_rate": 3.9461070964279084e-05,
      "loss": 0.4113,
      "step": 325
    },
    {
      "epoch": 0.20448486749255135,
      "grad_norm": 1.299596905708313,
      "learning_rate": 3.945637536952296e-05,
      "loss": 0.4432,
      "step": 326
    },
    {
      "epoch": 0.2051121216873138,
      "grad_norm": 1.4996416568756104,
      "learning_rate": 3.94516596892656e-05,
      "loss": 0.4161,
      "step": 327
    },
    {
      "epoch": 0.20573937588207622,
      "grad_norm": 1.520275592803955,
      "learning_rate": 3.9446923928375174e-05,
      "loss": 0.4482,
      "step": 328
    },
    {
      "epoch": 0.20636663007683864,
      "grad_norm": 1.2884626388549805,
      "learning_rate": 3.944216809174057e-05,
      "loss": 0.4304,
      "step": 329
    },
    {
      "epoch": 0.20699388427160106,
      "grad_norm": 1.6496778726577759,
      "learning_rate": 3.943739218427141e-05,
      "loss": 0.4208,
      "step": 330
    },
    {
      "epoch": 0.2076211384663635,
      "grad_norm": 1.434212565422058,
      "learning_rate": 3.943259621089801e-05,
      "loss": 0.4171,
      "step": 331
    },
    {
      "epoch": 0.2082483926611259,
      "grad_norm": 1.2932742834091187,
      "learning_rate": 3.942778017657146e-05,
      "loss": 0.377,
      "step": 332
    },
    {
      "epoch": 0.20887564685588836,
      "grad_norm": 1.3967646360397339,
      "learning_rate": 3.9422944086263494e-05,
      "loss": 0.4526,
      "step": 333
    },
    {
      "epoch": 0.20950290105065078,
      "grad_norm": 1.1630432605743408,
      "learning_rate": 3.941808794496659e-05,
      "loss": 0.3943,
      "step": 334
    },
    {
      "epoch": 0.2101301552454132,
      "grad_norm": 1.549743413925171,
      "learning_rate": 3.94132117576939e-05,
      "loss": 0.3819,
      "step": 335
    },
    {
      "epoch": 0.21075740944017562,
      "grad_norm": 1.455093264579773,
      "learning_rate": 3.940831552947931e-05,
      "loss": 0.3965,
      "step": 336
    },
    {
      "epoch": 0.21138466363493805,
      "grad_norm": 1.2519598007202148,
      "learning_rate": 3.940339926537734e-05,
      "loss": 0.4236,
      "step": 337
    },
    {
      "epoch": 0.2120119178297005,
      "grad_norm": 1.519441843032837,
      "learning_rate": 3.9398462970463234e-05,
      "loss": 0.4191,
      "step": 338
    },
    {
      "epoch": 0.21263917202446292,
      "grad_norm": 1.5170977115631104,
      "learning_rate": 3.93935066498329e-05,
      "loss": 0.4308,
      "step": 339
    },
    {
      "epoch": 0.21326642621922534,
      "grad_norm": 1.3977679014205933,
      "learning_rate": 3.9388530308602935e-05,
      "loss": 0.4476,
      "step": 340
    },
    {
      "epoch": 0.21389368041398776,
      "grad_norm": 1.6217604875564575,
      "learning_rate": 3.938353395191058e-05,
      "loss": 0.4158,
      "step": 341
    },
    {
      "epoch": 0.21452093460875019,
      "grad_norm": 1.3827109336853027,
      "learning_rate": 3.937851758491375e-05,
      "loss": 0.3565,
      "step": 342
    },
    {
      "epoch": 0.21514818880351264,
      "grad_norm": 1.4254240989685059,
      "learning_rate": 3.937348121279102e-05,
      "loss": 0.4092,
      "step": 343
    },
    {
      "epoch": 0.21577544299827506,
      "grad_norm": 1.8280800580978394,
      "learning_rate": 3.936842484074161e-05,
      "loss": 0.4093,
      "step": 344
    },
    {
      "epoch": 0.21640269719303748,
      "grad_norm": 1.4849143028259277,
      "learning_rate": 3.93633484739854e-05,
      "loss": 0.4213,
      "step": 345
    },
    {
      "epoch": 0.2170299513877999,
      "grad_norm": 1.3602157831192017,
      "learning_rate": 3.93582521177629e-05,
      "loss": 0.4088,
      "step": 346
    },
    {
      "epoch": 0.21765720558256232,
      "grad_norm": 1.8441216945648193,
      "learning_rate": 3.9353135777335244e-05,
      "loss": 0.4077,
      "step": 347
    },
    {
      "epoch": 0.21828445977732477,
      "grad_norm": 1.284321904182434,
      "learning_rate": 3.934799945798423e-05,
      "loss": 0.4113,
      "step": 348
    },
    {
      "epoch": 0.2189117139720872,
      "grad_norm": 1.2958706617355347,
      "learning_rate": 3.934284316501225e-05,
      "loss": 0.3932,
      "step": 349
    },
    {
      "epoch": 0.21953896816684962,
      "grad_norm": 1.341288447380066,
      "learning_rate": 3.933766690374232e-05,
      "loss": 0.41,
      "step": 350
    },
    {
      "epoch": 0.22016622236161204,
      "grad_norm": 1.4277088642120361,
      "learning_rate": 3.9332470679518095e-05,
      "loss": 0.4162,
      "step": 351
    },
    {
      "epoch": 0.22079347655637446,
      "grad_norm": 1.3362282514572144,
      "learning_rate": 3.9327254497703806e-05,
      "loss": 0.3986,
      "step": 352
    },
    {
      "epoch": 0.22142073075113688,
      "grad_norm": 1.4247506856918335,
      "learning_rate": 3.9322018363684306e-05,
      "loss": 0.3634,
      "step": 353
    },
    {
      "epoch": 0.22204798494589933,
      "grad_norm": 1.8625775575637817,
      "learning_rate": 3.9316762282865046e-05,
      "loss": 0.4288,
      "step": 354
    },
    {
      "epoch": 0.22267523914066176,
      "grad_norm": 1.6340423822402954,
      "learning_rate": 3.931148626067205e-05,
      "loss": 0.415,
      "step": 355
    },
    {
      "epoch": 0.22330249333542418,
      "grad_norm": 1.3341423273086548,
      "learning_rate": 3.930619030255196e-05,
      "loss": 0.3979,
      "step": 356
    },
    {
      "epoch": 0.2239297475301866,
      "grad_norm": 1.362868070602417,
      "learning_rate": 3.930087441397196e-05,
      "loss": 0.3724,
      "step": 357
    },
    {
      "epoch": 0.22455700172494902,
      "grad_norm": 1.2193236351013184,
      "learning_rate": 3.929553860041984e-05,
      "loss": 0.4193,
      "step": 358
    },
    {
      "epoch": 0.22518425591971147,
      "grad_norm": 1.515285849571228,
      "learning_rate": 3.929018286740394e-05,
      "loss": 0.4735,
      "step": 359
    },
    {
      "epoch": 0.2258115101144739,
      "grad_norm": 1.4257380962371826,
      "learning_rate": 3.928480722045319e-05,
      "loss": 0.4074,
      "step": 360
    },
    {
      "epoch": 0.22643876430923632,
      "grad_norm": 1.409763216972351,
      "learning_rate": 3.927941166511704e-05,
      "loss": 0.4456,
      "step": 361
    },
    {
      "epoch": 0.22706601850399874,
      "grad_norm": 1.384979009628296,
      "learning_rate": 3.927399620696553e-05,
      "loss": 0.3816,
      "step": 362
    },
    {
      "epoch": 0.22769327269876116,
      "grad_norm": 1.2846951484680176,
      "learning_rate": 3.92685608515892e-05,
      "loss": 0.4072,
      "step": 363
    },
    {
      "epoch": 0.2283205268935236,
      "grad_norm": 1.151328444480896,
      "learning_rate": 3.926310560459919e-05,
      "loss": 0.4332,
      "step": 364
    },
    {
      "epoch": 0.22894778108828603,
      "grad_norm": 1.4825246334075928,
      "learning_rate": 3.925763047162712e-05,
      "loss": 0.3971,
      "step": 365
    },
    {
      "epoch": 0.22957503528304846,
      "grad_norm": 1.3169598579406738,
      "learning_rate": 3.9252135458325176e-05,
      "loss": 0.3898,
      "step": 366
    },
    {
      "epoch": 0.23020228947781088,
      "grad_norm": 1.3232101202011108,
      "learning_rate": 3.9246620570366046e-05,
      "loss": 0.378,
      "step": 367
    },
    {
      "epoch": 0.2308295436725733,
      "grad_norm": 1.3413264751434326,
      "learning_rate": 3.924108581344295e-05,
      "loss": 0.4331,
      "step": 368
    },
    {
      "epoch": 0.23145679786733575,
      "grad_norm": 1.4002594947814941,
      "learning_rate": 3.9235531193269614e-05,
      "loss": 0.4,
      "step": 369
    },
    {
      "epoch": 0.23208405206209817,
      "grad_norm": 1.4046937227249146,
      "learning_rate": 3.922995671558026e-05,
      "loss": 0.3886,
      "step": 370
    },
    {
      "epoch": 0.2327113062568606,
      "grad_norm": 1.4203039407730103,
      "learning_rate": 3.922436238612962e-05,
      "loss": 0.402,
      "step": 371
    },
    {
      "epoch": 0.23333856045162302,
      "grad_norm": 1.3028756380081177,
      "learning_rate": 3.9218748210692916e-05,
      "loss": 0.3552,
      "step": 372
    },
    {
      "epoch": 0.23396581464638544,
      "grad_norm": 1.2306022644042969,
      "learning_rate": 3.921311419506586e-05,
      "loss": 0.411,
      "step": 373
    },
    {
      "epoch": 0.23459306884114786,
      "grad_norm": 1.2577680349349976,
      "learning_rate": 3.920746034506465e-05,
      "loss": 0.418,
      "step": 374
    },
    {
      "epoch": 0.2352203230359103,
      "grad_norm": 1.234844446182251,
      "learning_rate": 3.920178666652595e-05,
      "loss": 0.3706,
      "step": 375
    },
    {
      "epoch": 0.23584757723067273,
      "grad_norm": 1.4486712217330933,
      "learning_rate": 3.919609316530689e-05,
      "loss": 0.3818,
      "step": 376
    },
    {
      "epoch": 0.23647483142543516,
      "grad_norm": 1.2798314094543457,
      "learning_rate": 3.9190379847285085e-05,
      "loss": 0.4522,
      "step": 377
    },
    {
      "epoch": 0.23710208562019758,
      "grad_norm": 1.3837852478027344,
      "learning_rate": 3.918464671835858e-05,
      "loss": 0.3565,
      "step": 378
    },
    {
      "epoch": 0.23772933981496,
      "grad_norm": 1.4889065027236938,
      "learning_rate": 3.917889378444591e-05,
      "loss": 0.428,
      "step": 379
    },
    {
      "epoch": 0.23835659400972245,
      "grad_norm": 1.42939293384552,
      "learning_rate": 3.9173121051486014e-05,
      "loss": 0.416,
      "step": 380
    },
    {
      "epoch": 0.23898384820448487,
      "grad_norm": 1.412003755569458,
      "learning_rate": 3.9167328525438285e-05,
      "loss": 0.4457,
      "step": 381
    },
    {
      "epoch": 0.2396111023992473,
      "grad_norm": 1.4608865976333618,
      "learning_rate": 3.9161516212282555e-05,
      "loss": 0.4138,
      "step": 382
    },
    {
      "epoch": 0.24023835659400972,
      "grad_norm": 1.4364871978759766,
      "learning_rate": 3.915568411801908e-05,
      "loss": 0.371,
      "step": 383
    },
    {
      "epoch": 0.24086561078877214,
      "grad_norm": 1.456733226776123,
      "learning_rate": 3.914983224866854e-05,
      "loss": 0.4101,
      "step": 384
    },
    {
      "epoch": 0.2414928649835346,
      "grad_norm": 1.2100237607955933,
      "learning_rate": 3.914396061027202e-05,
      "loss": 0.3938,
      "step": 385
    },
    {
      "epoch": 0.242120119178297,
      "grad_norm": 1.3283522129058838,
      "learning_rate": 3.913806920889102e-05,
      "loss": 0.3727,
      "step": 386
    },
    {
      "epoch": 0.24274737337305943,
      "grad_norm": 1.5742121934890747,
      "learning_rate": 3.913215805060745e-05,
      "loss": 0.3825,
      "step": 387
    },
    {
      "epoch": 0.24337462756782186,
      "grad_norm": 1.551124095916748,
      "learning_rate": 3.91262271415236e-05,
      "loss": 0.4228,
      "step": 388
    },
    {
      "epoch": 0.24400188176258428,
      "grad_norm": 1.488900065422058,
      "learning_rate": 3.912027648776215e-05,
      "loss": 0.403,
      "step": 389
    },
    {
      "epoch": 0.24462913595734673,
      "grad_norm": 1.4520589113235474,
      "learning_rate": 3.9114306095466165e-05,
      "loss": 0.3894,
      "step": 390
    },
    {
      "epoch": 0.24525639015210915,
      "grad_norm": 1.4398682117462158,
      "learning_rate": 3.910831597079912e-05,
      "loss": 0.3537,
      "step": 391
    },
    {
      "epoch": 0.24588364434687157,
      "grad_norm": 1.324266791343689,
      "learning_rate": 3.9102306119944795e-05,
      "loss": 0.3257,
      "step": 392
    },
    {
      "epoch": 0.246510898541634,
      "grad_norm": 1.2131379842758179,
      "learning_rate": 3.90962765491074e-05,
      "loss": 0.4242,
      "step": 393
    },
    {
      "epoch": 0.24713815273639642,
      "grad_norm": 1.186140775680542,
      "learning_rate": 3.909022726451146e-05,
      "loss": 0.3969,
      "step": 394
    },
    {
      "epoch": 0.24776540693115887,
      "grad_norm": 1.2435379028320312,
      "learning_rate": 3.9084158272401865e-05,
      "loss": 0.4061,
      "step": 395
    },
    {
      "epoch": 0.2483926611259213,
      "grad_norm": 1.3766072988510132,
      "learning_rate": 3.907806957904385e-05,
      "loss": 0.3967,
      "step": 396
    },
    {
      "epoch": 0.2490199153206837,
      "grad_norm": 1.316792368888855,
      "learning_rate": 3.907196119072299e-05,
      "loss": 0.389,
      "step": 397
    },
    {
      "epoch": 0.24964716951544613,
      "grad_norm": 1.2080376148223877,
      "learning_rate": 3.906583311374518e-05,
      "loss": 0.3774,
      "step": 398
    },
    {
      "epoch": 0.2502744237102086,
      "grad_norm": 1.3271780014038086,
      "learning_rate": 3.905968535443666e-05,
      "loss": 0.4158,
      "step": 399
    },
    {
      "epoch": 0.250901677904971,
      "grad_norm": 1.442582130432129,
      "learning_rate": 3.905351791914398e-05,
      "loss": 0.3879,
      "step": 400
    },
    {
      "epoch": 0.2515289320997334,
      "grad_norm": 1.319524884223938,
      "learning_rate": 3.904733081423399e-05,
      "loss": 0.3404,
      "step": 401
    },
    {
      "epoch": 0.2521561862944958,
      "grad_norm": 1.310402750968933,
      "learning_rate": 3.9041124046093866e-05,
      "loss": 0.3603,
      "step": 402
    },
    {
      "epoch": 0.25278344048925827,
      "grad_norm": 1.2856608629226685,
      "learning_rate": 3.903489762113106e-05,
      "loss": 0.3566,
      "step": 403
    },
    {
      "epoch": 0.2534106946840207,
      "grad_norm": 1.3391993045806885,
      "learning_rate": 3.902865154577335e-05,
      "loss": 0.337,
      "step": 404
    },
    {
      "epoch": 0.2540379488787831,
      "grad_norm": 1.288784146308899,
      "learning_rate": 3.902238582646876e-05,
      "loss": 0.3885,
      "step": 405
    },
    {
      "epoch": 0.25466520307354557,
      "grad_norm": 1.1475400924682617,
      "learning_rate": 3.9016100469685625e-05,
      "loss": 0.3283,
      "step": 406
    },
    {
      "epoch": 0.25529245726830796,
      "grad_norm": 1.5950651168823242,
      "learning_rate": 3.900979548191254e-05,
      "loss": 0.4125,
      "step": 407
    },
    {
      "epoch": 0.2559197114630704,
      "grad_norm": 1.4678776264190674,
      "learning_rate": 3.900347086965835e-05,
      "loss": 0.3735,
      "step": 408
    },
    {
      "epoch": 0.25654696565783286,
      "grad_norm": 1.5201358795166016,
      "learning_rate": 3.899712663945219e-05,
      "loss": 0.4163,
      "step": 409
    },
    {
      "epoch": 0.25717421985259525,
      "grad_norm": 1.0794888734817505,
      "learning_rate": 3.8990762797843436e-05,
      "loss": 0.3985,
      "step": 410
    },
    {
      "epoch": 0.2578014740473577,
      "grad_norm": 1.3003852367401123,
      "learning_rate": 3.8984379351401696e-05,
      "loss": 0.3735,
      "step": 411
    },
    {
      "epoch": 0.2584287282421201,
      "grad_norm": 1.7244070768356323,
      "learning_rate": 3.897797630671682e-05,
      "loss": 0.4131,
      "step": 412
    },
    {
      "epoch": 0.25905598243688255,
      "grad_norm": 1.3861531019210815,
      "learning_rate": 3.897155367039892e-05,
      "loss": 0.3649,
      "step": 413
    },
    {
      "epoch": 0.259683236631645,
      "grad_norm": 1.6174381971359253,
      "learning_rate": 3.896511144907829e-05,
      "loss": 0.363,
      "step": 414
    },
    {
      "epoch": 0.2603104908264074,
      "grad_norm": 1.4827419519424438,
      "learning_rate": 3.895864964940546e-05,
      "loss": 0.3722,
      "step": 415
    },
    {
      "epoch": 0.26093774502116984,
      "grad_norm": 1.174606442451477,
      "learning_rate": 3.8952168278051184e-05,
      "loss": 0.3527,
      "step": 416
    },
    {
      "epoch": 0.26156499921593224,
      "grad_norm": 1.4897773265838623,
      "learning_rate": 3.8945667341706405e-05,
      "loss": 0.3602,
      "step": 417
    },
    {
      "epoch": 0.2621922534106947,
      "grad_norm": 1.1790823936462402,
      "learning_rate": 3.893914684708227e-05,
      "loss": 0.3857,
      "step": 418
    },
    {
      "epoch": 0.26281950760545714,
      "grad_norm": 1.4660543203353882,
      "learning_rate": 3.893260680091011e-05,
      "loss": 0.3808,
      "step": 419
    },
    {
      "epoch": 0.26344676180021953,
      "grad_norm": 1.4048151969909668,
      "learning_rate": 3.892604720994146e-05,
      "loss": 0.3589,
      "step": 420
    },
    {
      "epoch": 0.264074015994982,
      "grad_norm": 1.247186541557312,
      "learning_rate": 3.8919468080947996e-05,
      "loss": 0.3675,
      "step": 421
    },
    {
      "epoch": 0.2647012701897444,
      "grad_norm": 1.426533818244934,
      "learning_rate": 3.8912869420721604e-05,
      "loss": 0.3759,
      "step": 422
    },
    {
      "epoch": 0.2653285243845068,
      "grad_norm": 1.4955856800079346,
      "learning_rate": 3.890625123607429e-05,
      "loss": 0.3558,
      "step": 423
    },
    {
      "epoch": 0.2659557785792693,
      "grad_norm": 1.2410643100738525,
      "learning_rate": 3.8899613533838265e-05,
      "loss": 0.3567,
      "step": 424
    },
    {
      "epoch": 0.26658303277403167,
      "grad_norm": 1.9216816425323486,
      "learning_rate": 3.8892956320865846e-05,
      "loss": 0.4118,
      "step": 425
    },
    {
      "epoch": 0.2672102869687941,
      "grad_norm": 1.4261921644210815,
      "learning_rate": 3.888627960402952e-05,
      "loss": 0.3689,
      "step": 426
    },
    {
      "epoch": 0.2678375411635565,
      "grad_norm": 1.1982080936431885,
      "learning_rate": 3.887958339022188e-05,
      "loss": 0.3706,
      "step": 427
    },
    {
      "epoch": 0.26846479535831896,
      "grad_norm": 1.3578213453292847,
      "learning_rate": 3.887286768635568e-05,
      "loss": 0.37,
      "step": 428
    },
    {
      "epoch": 0.2690920495530814,
      "grad_norm": 1.3112714290618896,
      "learning_rate": 3.886613249936376e-05,
      "loss": 0.3861,
      "step": 429
    },
    {
      "epoch": 0.2697193037478438,
      "grad_norm": 1.2745850086212158,
      "learning_rate": 3.885937783619911e-05,
      "loss": 0.4104,
      "step": 430
    },
    {
      "epoch": 0.27034655794260626,
      "grad_norm": 1.239473819732666,
      "learning_rate": 3.885260370383479e-05,
      "loss": 0.3458,
      "step": 431
    },
    {
      "epoch": 0.27097381213736865,
      "grad_norm": 1.2031855583190918,
      "learning_rate": 3.884581010926399e-05,
      "loss": 0.3464,
      "step": 432
    },
    {
      "epoch": 0.2716010663321311,
      "grad_norm": 1.2598925828933716,
      "learning_rate": 3.883899705949997e-05,
      "loss": 0.3721,
      "step": 433
    },
    {
      "epoch": 0.2722283205268935,
      "grad_norm": 1.157799482345581,
      "learning_rate": 3.883216456157608e-05,
      "loss": 0.3513,
      "step": 434
    },
    {
      "epoch": 0.27285557472165595,
      "grad_norm": 1.3330152034759521,
      "learning_rate": 3.8825312622545745e-05,
      "loss": 0.4207,
      "step": 435
    },
    {
      "epoch": 0.2734828289164184,
      "grad_norm": 1.162471890449524,
      "learning_rate": 3.881844124948247e-05,
      "loss": 0.3748,
      "step": 436
    },
    {
      "epoch": 0.2741100831111808,
      "grad_norm": 1.442447543144226,
      "learning_rate": 3.881155044947981e-05,
      "loss": 0.372,
      "step": 437
    },
    {
      "epoch": 0.27473733730594324,
      "grad_norm": 1.4956692457199097,
      "learning_rate": 3.880464022965138e-05,
      "loss": 0.3877,
      "step": 438
    },
    {
      "epoch": 0.27536459150070564,
      "grad_norm": 1.2240872383117676,
      "learning_rate": 3.8797710597130855e-05,
      "loss": 0.3811,
      "step": 439
    },
    {
      "epoch": 0.2759918456954681,
      "grad_norm": 1.4835654497146606,
      "learning_rate": 3.879076155907193e-05,
      "loss": 0.3826,
      "step": 440
    },
    {
      "epoch": 0.27661909989023054,
      "grad_norm": 1.3184068202972412,
      "learning_rate": 3.878379312264833e-05,
      "loss": 0.3868,
      "step": 441
    },
    {
      "epoch": 0.27724635408499293,
      "grad_norm": 1.30062735080719,
      "learning_rate": 3.8776805295053835e-05,
      "loss": 0.3799,
      "step": 442
    },
    {
      "epoch": 0.2778736082797554,
      "grad_norm": 1.2152583599090576,
      "learning_rate": 3.876979808350222e-05,
      "loss": 0.3565,
      "step": 443
    },
    {
      "epoch": 0.2785008624745178,
      "grad_norm": 1.1691176891326904,
      "learning_rate": 3.876277149522727e-05,
      "loss": 0.343,
      "step": 444
    },
    {
      "epoch": 0.2791281166692802,
      "grad_norm": 1.3291245698928833,
      "learning_rate": 3.8755725537482785e-05,
      "loss": 0.336,
      "step": 445
    },
    {
      "epoch": 0.2797553708640427,
      "grad_norm": 1.3104655742645264,
      "learning_rate": 3.874866021754256e-05,
      "loss": 0.3728,
      "step": 446
    },
    {
      "epoch": 0.28038262505880507,
      "grad_norm": 1.2047230005264282,
      "learning_rate": 3.8741575542700366e-05,
      "loss": 0.3284,
      "step": 447
    },
    {
      "epoch": 0.2810098792535675,
      "grad_norm": 1.5223572254180908,
      "learning_rate": 3.873447152026996e-05,
      "loss": 0.4426,
      "step": 448
    },
    {
      "epoch": 0.2816371334483299,
      "grad_norm": 1.2537058591842651,
      "learning_rate": 3.872734815758508e-05,
      "loss": 0.3687,
      "step": 449
    },
    {
      "epoch": 0.28226438764309236,
      "grad_norm": 1.32411527633667,
      "learning_rate": 3.8720205461999426e-05,
      "loss": 0.3903,
      "step": 450
    },
    {
      "epoch": 0.2828916418378548,
      "grad_norm": 1.271552324295044,
      "learning_rate": 3.871304344088665e-05,
      "loss": 0.3738,
      "step": 451
    },
    {
      "epoch": 0.2835188960326172,
      "grad_norm": 1.0983104705810547,
      "learning_rate": 3.870586210164035e-05,
      "loss": 0.3529,
      "step": 452
    },
    {
      "epoch": 0.28414615022737966,
      "grad_norm": 1.3040975332260132,
      "learning_rate": 3.86986614516741e-05,
      "loss": 0.3524,
      "step": 453
    },
    {
      "epoch": 0.28477340442214205,
      "grad_norm": 1.2826478481292725,
      "learning_rate": 3.869144149842136e-05,
      "loss": 0.3693,
      "step": 454
    },
    {
      "epoch": 0.2854006586169045,
      "grad_norm": 1.199196457862854,
      "learning_rate": 3.868420224933555e-05,
      "loss": 0.3779,
      "step": 455
    },
    {
      "epoch": 0.28602791281166695,
      "grad_norm": 1.2873188257217407,
      "learning_rate": 3.8676943711890006e-05,
      "loss": 0.3436,
      "step": 456
    },
    {
      "epoch": 0.28665516700642935,
      "grad_norm": 1.2819852828979492,
      "learning_rate": 3.8669665893577975e-05,
      "loss": 0.3779,
      "step": 457
    },
    {
      "epoch": 0.2872824212011918,
      "grad_norm": 1.288419246673584,
      "learning_rate": 3.866236880191259e-05,
      "loss": 0.3918,
      "step": 458
    },
    {
      "epoch": 0.2879096753959542,
      "grad_norm": 1.165993332862854,
      "learning_rate": 3.865505244442691e-05,
      "loss": 0.3715,
      "step": 459
    },
    {
      "epoch": 0.28853692959071664,
      "grad_norm": 1.201463222503662,
      "learning_rate": 3.864771682867386e-05,
      "loss": 0.3701,
      "step": 460
    },
    {
      "epoch": 0.2891641837854791,
      "grad_norm": 1.2459325790405273,
      "learning_rate": 3.8640361962226254e-05,
      "loss": 0.3279,
      "step": 461
    },
    {
      "epoch": 0.2897914379802415,
      "grad_norm": 1.205305814743042,
      "learning_rate": 3.863298785267678e-05,
      "loss": 0.3772,
      "step": 462
    },
    {
      "epoch": 0.29041869217500393,
      "grad_norm": 1.2977550029754639,
      "learning_rate": 3.8625594507638e-05,
      "loss": 0.3818,
      "step": 463
    },
    {
      "epoch": 0.29104594636976633,
      "grad_norm": 1.1289427280426025,
      "learning_rate": 3.861818193474231e-05,
      "loss": 0.3551,
      "step": 464
    },
    {
      "epoch": 0.2916732005645288,
      "grad_norm": 1.2582975625991821,
      "learning_rate": 3.8610750141641984e-05,
      "loss": 0.3849,
      "step": 465
    },
    {
      "epoch": 0.29230045475929123,
      "grad_norm": 1.1075705289840698,
      "learning_rate": 3.860329913600912e-05,
      "loss": 0.3662,
      "step": 466
    },
    {
      "epoch": 0.2929277089540536,
      "grad_norm": 1.2868229150772095,
      "learning_rate": 3.8595828925535646e-05,
      "loss": 0.3765,
      "step": 467
    },
    {
      "epoch": 0.2935549631488161,
      "grad_norm": 1.1359583139419556,
      "learning_rate": 3.858833951793333e-05,
      "loss": 0.3182,
      "step": 468
    },
    {
      "epoch": 0.29418221734357847,
      "grad_norm": 1.207962155342102,
      "learning_rate": 3.858083092093375e-05,
      "loss": 0.3768,
      "step": 469
    },
    {
      "epoch": 0.2948094715383409,
      "grad_norm": 1.1784318685531616,
      "learning_rate": 3.8573303142288295e-05,
      "loss": 0.3576,
      "step": 470
    },
    {
      "epoch": 0.29543672573310337,
      "grad_norm": 1.2792519330978394,
      "learning_rate": 3.8565756189768146e-05,
      "loss": 0.3437,
      "step": 471
    },
    {
      "epoch": 0.29606397992786576,
      "grad_norm": 1.5046852827072144,
      "learning_rate": 3.8558190071164305e-05,
      "loss": 0.3556,
      "step": 472
    },
    {
      "epoch": 0.2966912341226282,
      "grad_norm": 1.2456021308898926,
      "learning_rate": 3.8550604794287536e-05,
      "loss": 0.3052,
      "step": 473
    },
    {
      "epoch": 0.2973184883173906,
      "grad_norm": 1.3958989381790161,
      "learning_rate": 3.8543000366968385e-05,
      "loss": 0.3994,
      "step": 474
    },
    {
      "epoch": 0.29794574251215306,
      "grad_norm": 1.4529577493667603,
      "learning_rate": 3.853537679705716e-05,
      "loss": 0.3377,
      "step": 475
    },
    {
      "epoch": 0.29857299670691545,
      "grad_norm": 1.3536865711212158,
      "learning_rate": 3.8527734092423966e-05,
      "loss": 0.3789,
      "step": 476
    },
    {
      "epoch": 0.2992002509016779,
      "grad_norm": 1.2169958353042603,
      "learning_rate": 3.852007226095861e-05,
      "loss": 0.3557,
      "step": 477
    },
    {
      "epoch": 0.29982750509644035,
      "grad_norm": 1.295535683631897,
      "learning_rate": 3.851239131057069e-05,
      "loss": 0.3687,
      "step": 478
    },
    {
      "epoch": 0.30045475929120274,
      "grad_norm": 1.296625018119812,
      "learning_rate": 3.850469124918951e-05,
      "loss": 0.3856,
      "step": 479
    },
    {
      "epoch": 0.3010820134859652,
      "grad_norm": 1.459418773651123,
      "learning_rate": 3.8496972084764116e-05,
      "loss": 0.3622,
      "step": 480
    },
    {
      "epoch": 0.3017092676807276,
      "grad_norm": 1.2736573219299316,
      "learning_rate": 3.848923382526327e-05,
      "loss": 0.3921,
      "step": 481
    },
    {
      "epoch": 0.30233652187549004,
      "grad_norm": 1.26237952709198,
      "learning_rate": 3.8481476478675464e-05,
      "loss": 0.3326,
      "step": 482
    },
    {
      "epoch": 0.3029637760702525,
      "grad_norm": 1.3334345817565918,
      "learning_rate": 3.847370005300887e-05,
      "loss": 0.4125,
      "step": 483
    },
    {
      "epoch": 0.3035910302650149,
      "grad_norm": 1.1309443712234497,
      "learning_rate": 3.8465904556291366e-05,
      "loss": 0.3206,
      "step": 484
    },
    {
      "epoch": 0.30421828445977733,
      "grad_norm": 1.3421224355697632,
      "learning_rate": 3.8458089996570516e-05,
      "loss": 0.3545,
      "step": 485
    },
    {
      "epoch": 0.3048455386545397,
      "grad_norm": 1.3264871835708618,
      "learning_rate": 3.845025638191357e-05,
      "loss": 0.3736,
      "step": 486
    },
    {
      "epoch": 0.3054727928493022,
      "grad_norm": 1.199151873588562,
      "learning_rate": 3.844240372040744e-05,
      "loss": 0.3658,
      "step": 487
    },
    {
      "epoch": 0.3061000470440646,
      "grad_norm": 1.4013934135437012,
      "learning_rate": 3.843453202015871e-05,
      "loss": 0.3631,
      "step": 488
    },
    {
      "epoch": 0.306727301238827,
      "grad_norm": 1.254109263420105,
      "learning_rate": 3.84266412892936e-05,
      "loss": 0.3534,
      "step": 489
    },
    {
      "epoch": 0.30735455543358947,
      "grad_norm": 1.218373417854309,
      "learning_rate": 3.841873153595801e-05,
      "loss": 0.3358,
      "step": 490
    },
    {
      "epoch": 0.30798180962835187,
      "grad_norm": 1.1666988134384155,
      "learning_rate": 3.841080276831744e-05,
      "loss": 0.3714,
      "step": 491
    },
    {
      "epoch": 0.3086090638231143,
      "grad_norm": 1.1505777835845947,
      "learning_rate": 3.8402854994557036e-05,
      "loss": 0.3689,
      "step": 492
    },
    {
      "epoch": 0.30923631801787677,
      "grad_norm": 1.1682237386703491,
      "learning_rate": 3.8394888222881576e-05,
      "loss": 0.365,
      "step": 493
    },
    {
      "epoch": 0.30986357221263916,
      "grad_norm": 1.2740191221237183,
      "learning_rate": 3.838690246151544e-05,
      "loss": 0.3514,
      "step": 494
    },
    {
      "epoch": 0.3104908264074016,
      "grad_norm": 1.1906490325927734,
      "learning_rate": 3.8378897718702595e-05,
      "loss": 0.3505,
      "step": 495
    },
    {
      "epoch": 0.311118080602164,
      "grad_norm": 1.2643780708312988,
      "learning_rate": 3.837087400270663e-05,
      "loss": 0.3209,
      "step": 496
    },
    {
      "epoch": 0.31174533479692645,
      "grad_norm": 1.479504942893982,
      "learning_rate": 3.83628313218107e-05,
      "loss": 0.4005,
      "step": 497
    },
    {
      "epoch": 0.3123725889916889,
      "grad_norm": 1.326695442199707,
      "learning_rate": 3.835476968431756e-05,
      "loss": 0.3244,
      "step": 498
    },
    {
      "epoch": 0.3129998431864513,
      "grad_norm": 1.3005268573760986,
      "learning_rate": 3.8346689098549525e-05,
      "loss": 0.3717,
      "step": 499
    },
    {
      "epoch": 0.31362709738121375,
      "grad_norm": 1.1932462453842163,
      "learning_rate": 3.833858957284845e-05,
      "loss": 0.3364,
      "step": 500
    },
    {
      "epoch": 0.31425435157597614,
      "grad_norm": 1.4240779876708984,
      "learning_rate": 3.833047111557578e-05,
      "loss": 0.371,
      "step": 501
    },
    {
      "epoch": 0.3148816057707386,
      "grad_norm": 1.4582301378250122,
      "learning_rate": 3.8322333735112466e-05,
      "loss": 0.389,
      "step": 502
    },
    {
      "epoch": 0.31550885996550104,
      "grad_norm": 1.1639416217803955,
      "learning_rate": 3.831417743985903e-05,
      "loss": 0.3803,
      "step": 503
    },
    {
      "epoch": 0.31613611416026344,
      "grad_norm": 1.6114230155944824,
      "learning_rate": 3.830600223823548e-05,
      "loss": 0.37,
      "step": 504
    },
    {
      "epoch": 0.3167633683550259,
      "grad_norm": 1.3899036645889282,
      "learning_rate": 3.829780813868139e-05,
      "loss": 0.4002,
      "step": 505
    },
    {
      "epoch": 0.3173906225497883,
      "grad_norm": 1.1674432754516602,
      "learning_rate": 3.82895951496558e-05,
      "loss": 0.3284,
      "step": 506
    },
    {
      "epoch": 0.31801787674455073,
      "grad_norm": 1.1973252296447754,
      "learning_rate": 3.8281363279637276e-05,
      "loss": 0.4293,
      "step": 507
    },
    {
      "epoch": 0.3186451309393132,
      "grad_norm": 1.3239046335220337,
      "learning_rate": 3.8273112537123864e-05,
      "loss": 0.3344,
      "step": 508
    },
    {
      "epoch": 0.3192723851340756,
      "grad_norm": 1.2565101385116577,
      "learning_rate": 3.8264842930633095e-05,
      "loss": 0.3603,
      "step": 509
    },
    {
      "epoch": 0.319899639328838,
      "grad_norm": 1.2703858613967896,
      "learning_rate": 3.8256554468701985e-05,
      "loss": 0.3771,
      "step": 510
    },
    {
      "epoch": 0.3205268935236004,
      "grad_norm": 1.3956307172775269,
      "learning_rate": 3.824824715988698e-05,
      "loss": 0.3476,
      "step": 511
    },
    {
      "epoch": 0.32115414771836287,
      "grad_norm": 1.2926727533340454,
      "learning_rate": 3.823992101276404e-05,
      "loss": 0.3706,
      "step": 512
    },
    {
      "epoch": 0.3217814019131253,
      "grad_norm": 1.1364754438400269,
      "learning_rate": 3.823157603592852e-05,
      "loss": 0.3034,
      "step": 513
    },
    {
      "epoch": 0.3224086561078877,
      "grad_norm": 1.1433966159820557,
      "learning_rate": 3.8223212237995235e-05,
      "loss": 0.3442,
      "step": 514
    },
    {
      "epoch": 0.32303591030265016,
      "grad_norm": 1.21794855594635,
      "learning_rate": 3.821482962759843e-05,
      "loss": 0.3795,
      "step": 515
    },
    {
      "epoch": 0.32366316449741256,
      "grad_norm": 1.1466041803359985,
      "learning_rate": 3.820642821339177e-05,
      "loss": 0.3156,
      "step": 516
    },
    {
      "epoch": 0.324290418692175,
      "grad_norm": 1.362508773803711,
      "learning_rate": 3.819800800404832e-05,
      "loss": 0.355,
      "step": 517
    },
    {
      "epoch": 0.32491767288693746,
      "grad_norm": 1.2425683736801147,
      "learning_rate": 3.818956900826058e-05,
      "loss": 0.3461,
      "step": 518
    },
    {
      "epoch": 0.32554492708169985,
      "grad_norm": 0.8988218307495117,
      "learning_rate": 3.818111123474041e-05,
      "loss": 0.3696,
      "step": 519
    },
    {
      "epoch": 0.3261721812764623,
      "grad_norm": 1.2538793087005615,
      "learning_rate": 3.817263469221907e-05,
      "loss": 0.358,
      "step": 520
    },
    {
      "epoch": 0.3267994354712247,
      "grad_norm": 1.2860736846923828,
      "learning_rate": 3.816413938944718e-05,
      "loss": 0.3074,
      "step": 521
    },
    {
      "epoch": 0.32742668966598715,
      "grad_norm": 1.2698616981506348,
      "learning_rate": 3.815562533519476e-05,
      "loss": 0.3474,
      "step": 522
    },
    {
      "epoch": 0.32805394386074954,
      "grad_norm": 1.1828845739364624,
      "learning_rate": 3.814709253825115e-05,
      "loss": 0.3791,
      "step": 523
    },
    {
      "epoch": 0.328681198055512,
      "grad_norm": 1.1985173225402832,
      "learning_rate": 3.813854100742507e-05,
      "loss": 0.3597,
      "step": 524
    },
    {
      "epoch": 0.32930845225027444,
      "grad_norm": 1.3471806049346924,
      "learning_rate": 3.812997075154457e-05,
      "loss": 0.3697,
      "step": 525
    },
    {
      "epoch": 0.32993570644503684,
      "grad_norm": 1.2784637212753296,
      "learning_rate": 3.812138177945701e-05,
      "loss": 0.3718,
      "step": 526
    },
    {
      "epoch": 0.3305629606397993,
      "grad_norm": 1.35215163230896,
      "learning_rate": 3.81127741000291e-05,
      "loss": 0.3958,
      "step": 527
    },
    {
      "epoch": 0.3311902148345617,
      "grad_norm": 1.283639907836914,
      "learning_rate": 3.810414772214685e-05,
      "loss": 0.3842,
      "step": 528
    },
    {
      "epoch": 0.33181746902932413,
      "grad_norm": 1.1455283164978027,
      "learning_rate": 3.809550265471557e-05,
      "loss": 0.3605,
      "step": 529
    },
    {
      "epoch": 0.3324447232240866,
      "grad_norm": 1.1047929525375366,
      "learning_rate": 3.8086838906659865e-05,
      "loss": 0.3275,
      "step": 530
    },
    {
      "epoch": 0.333071977418849,
      "grad_norm": 1.631423830986023,
      "learning_rate": 3.8078156486923636e-05,
      "loss": 0.3591,
      "step": 531
    },
    {
      "epoch": 0.3336992316136114,
      "grad_norm": 1.3591557741165161,
      "learning_rate": 3.8069455404470046e-05,
      "loss": 0.378,
      "step": 532
    },
    {
      "epoch": 0.3343264858083738,
      "grad_norm": 1.312411904335022,
      "learning_rate": 3.8060735668281527e-05,
      "loss": 0.3579,
      "step": 533
    },
    {
      "epoch": 0.33495374000313627,
      "grad_norm": 1.4502252340316772,
      "learning_rate": 3.805199728735977e-05,
      "loss": 0.3674,
      "step": 534
    },
    {
      "epoch": 0.3355809941978987,
      "grad_norm": 1.1842923164367676,
      "learning_rate": 3.8043240270725714e-05,
      "loss": 0.3137,
      "step": 535
    },
    {
      "epoch": 0.3362082483926611,
      "grad_norm": 1.2416174411773682,
      "learning_rate": 3.8034464627419535e-05,
      "loss": 0.3362,
      "step": 536
    },
    {
      "epoch": 0.33683550258742356,
      "grad_norm": 1.1419811248779297,
      "learning_rate": 3.8025670366500635e-05,
      "loss": 0.3369,
      "step": 537
    },
    {
      "epoch": 0.33746275678218596,
      "grad_norm": 1.2286869287490845,
      "learning_rate": 3.8016857497047655e-05,
      "loss": 0.317,
      "step": 538
    },
    {
      "epoch": 0.3380900109769484,
      "grad_norm": 1.616650938987732,
      "learning_rate": 3.80080260281584e-05,
      "loss": 0.3304,
      "step": 539
    },
    {
      "epoch": 0.33871726517171086,
      "grad_norm": 1.1498764753341675,
      "learning_rate": 3.799917596894994e-05,
      "loss": 0.3206,
      "step": 540
    },
    {
      "epoch": 0.33934451936647325,
      "grad_norm": 1.123672604560852,
      "learning_rate": 3.799030732855848e-05,
      "loss": 0.3231,
      "step": 541
    },
    {
      "epoch": 0.3399717735612357,
      "grad_norm": 1.2703033685684204,
      "learning_rate": 3.798142011613944e-05,
      "loss": 0.3143,
      "step": 542
    },
    {
      "epoch": 0.3405990277559981,
      "grad_norm": 1.1162161827087402,
      "learning_rate": 3.7972514340867404e-05,
      "loss": 0.3324,
      "step": 543
    },
    {
      "epoch": 0.34122628195076055,
      "grad_norm": 1.2812544107437134,
      "learning_rate": 3.796359001193611e-05,
      "loss": 0.3229,
      "step": 544
    },
    {
      "epoch": 0.341853536145523,
      "grad_norm": 1.1549571752548218,
      "learning_rate": 3.7954647138558455e-05,
      "loss": 0.3268,
      "step": 545
    },
    {
      "epoch": 0.3424807903402854,
      "grad_norm": 1.4987828731536865,
      "learning_rate": 3.794568572996649e-05,
      "loss": 0.364,
      "step": 546
    },
    {
      "epoch": 0.34310804453504784,
      "grad_norm": 1.5693942308425903,
      "learning_rate": 3.79367057954114e-05,
      "loss": 0.348,
      "step": 547
    },
    {
      "epoch": 0.34373529872981023,
      "grad_norm": 1.3781763315200806,
      "learning_rate": 3.7927707344163464e-05,
      "loss": 0.2911,
      "step": 548
    },
    {
      "epoch": 0.3443625529245727,
      "grad_norm": 1.1173834800720215,
      "learning_rate": 3.791869038551212e-05,
      "loss": 0.322,
      "step": 549
    },
    {
      "epoch": 0.34498980711933513,
      "grad_norm": 1.2679322957992554,
      "learning_rate": 3.7909654928765875e-05,
      "loss": 0.3907,
      "step": 550
    },
    {
      "epoch": 0.34561706131409753,
      "grad_norm": 1.3186674118041992,
      "learning_rate": 3.790060098325237e-05,
      "loss": 0.3744,
      "step": 551
    },
    {
      "epoch": 0.34624431550886,
      "grad_norm": 1.333404779434204,
      "learning_rate": 3.789152855831829e-05,
      "loss": 0.3881,
      "step": 552
    },
    {
      "epoch": 0.3468715697036224,
      "grad_norm": 1.2173197269439697,
      "learning_rate": 3.7882437663329434e-05,
      "loss": 0.3365,
      "step": 553
    },
    {
      "epoch": 0.3474988238983848,
      "grad_norm": 1.1697617769241333,
      "learning_rate": 3.787332830767064e-05,
      "loss": 0.3129,
      "step": 554
    },
    {
      "epoch": 0.3481260780931473,
      "grad_norm": 1.214918851852417,
      "learning_rate": 3.786420050074584e-05,
      "loss": 0.3574,
      "step": 555
    },
    {
      "epoch": 0.34875333228790967,
      "grad_norm": 1.1760424375534058,
      "learning_rate": 3.785505425197797e-05,
      "loss": 0.3307,
      "step": 556
    },
    {
      "epoch": 0.3493805864826721,
      "grad_norm": 1.1405760049819946,
      "learning_rate": 3.7845889570809025e-05,
      "loss": 0.359,
      "step": 557
    },
    {
      "epoch": 0.3500078406774345,
      "grad_norm": 1.1279023885726929,
      "learning_rate": 3.7836706466700036e-05,
      "loss": 0.3233,
      "step": 558
    },
    {
      "epoch": 0.35063509487219696,
      "grad_norm": 1.1536526679992676,
      "learning_rate": 3.782750494913104e-05,
      "loss": 0.3556,
      "step": 559
    },
    {
      "epoch": 0.3512623490669594,
      "grad_norm": 1.2649977207183838,
      "learning_rate": 3.7818285027601086e-05,
      "loss": 0.3147,
      "step": 560
    },
    {
      "epoch": 0.3518896032617218,
      "grad_norm": 1.190099835395813,
      "learning_rate": 3.780904671162823e-05,
      "loss": 0.2968,
      "step": 561
    },
    {
      "epoch": 0.35251685745648426,
      "grad_norm": 1.1292306184768677,
      "learning_rate": 3.779979001074951e-05,
      "loss": 0.3268,
      "step": 562
    },
    {
      "epoch": 0.35314411165124665,
      "grad_norm": 1.0524557828903198,
      "learning_rate": 3.7790514934520936e-05,
      "loss": 0.3412,
      "step": 563
    },
    {
      "epoch": 0.3537713658460091,
      "grad_norm": 1.3033021688461304,
      "learning_rate": 3.77812214925175e-05,
      "loss": 0.3541,
      "step": 564
    },
    {
      "epoch": 0.35439862004077155,
      "grad_norm": 1.1734102964401245,
      "learning_rate": 3.7771909694333156e-05,
      "loss": 0.3378,
      "step": 565
    },
    {
      "epoch": 0.35502587423553394,
      "grad_norm": 1.104713797569275,
      "learning_rate": 3.776257954958079e-05,
      "loss": 0.3196,
      "step": 566
    },
    {
      "epoch": 0.3556531284302964,
      "grad_norm": 1.1589550971984863,
      "learning_rate": 3.775323106789225e-05,
      "loss": 0.3302,
      "step": 567
    },
    {
      "epoch": 0.3562803826250588,
      "grad_norm": 1.1092143058776855,
      "learning_rate": 3.774386425891829e-05,
      "loss": 0.3228,
      "step": 568
    },
    {
      "epoch": 0.35690763681982124,
      "grad_norm": 1.1946264505386353,
      "learning_rate": 3.7734479132328615e-05,
      "loss": 0.3391,
      "step": 569
    },
    {
      "epoch": 0.35753489101458363,
      "grad_norm": 1.216552734375,
      "learning_rate": 3.77250756978118e-05,
      "loss": 0.3491,
      "step": 570
    },
    {
      "epoch": 0.3581621452093461,
      "grad_norm": 1.0963689088821411,
      "learning_rate": 3.771565396507535e-05,
      "loss": 0.3377,
      "step": 571
    },
    {
      "epoch": 0.35878939940410853,
      "grad_norm": 1.323681116104126,
      "learning_rate": 3.7706213943845666e-05,
      "loss": 0.3252,
      "step": 572
    },
    {
      "epoch": 0.3594166535988709,
      "grad_norm": 1.1954137086868286,
      "learning_rate": 3.7696755643867993e-05,
      "loss": 0.3651,
      "step": 573
    },
    {
      "epoch": 0.3600439077936334,
      "grad_norm": 1.2133947610855103,
      "learning_rate": 3.7687279074906474e-05,
      "loss": 0.3439,
      "step": 574
    },
    {
      "epoch": 0.36067116198839577,
      "grad_norm": 1.4137613773345947,
      "learning_rate": 3.76777842467441e-05,
      "loss": 0.3366,
      "step": 575
    },
    {
      "epoch": 0.3612984161831582,
      "grad_norm": 1.1659473180770874,
      "learning_rate": 3.7668271169182736e-05,
      "loss": 0.3335,
      "step": 576
    },
    {
      "epoch": 0.36192567037792067,
      "grad_norm": 1.2614892721176147,
      "learning_rate": 3.765873985204305e-05,
      "loss": 0.4005,
      "step": 577
    },
    {
      "epoch": 0.36255292457268307,
      "grad_norm": 1.0579726696014404,
      "learning_rate": 3.764919030516456e-05,
      "loss": 0.3114,
      "step": 578
    },
    {
      "epoch": 0.3631801787674455,
      "grad_norm": 1.2478663921356201,
      "learning_rate": 3.7639622538405595e-05,
      "loss": 0.3205,
      "step": 579
    },
    {
      "epoch": 0.3638074329622079,
      "grad_norm": 1.1137259006500244,
      "learning_rate": 3.7630036561643304e-05,
      "loss": 0.3056,
      "step": 580
    },
    {
      "epoch": 0.36443468715697036,
      "grad_norm": 1.081593632698059,
      "learning_rate": 3.762043238477363e-05,
      "loss": 0.327,
      "step": 581
    },
    {
      "epoch": 0.3650619413517328,
      "grad_norm": 1.2735402584075928,
      "learning_rate": 3.761081001771129e-05,
      "loss": 0.3226,
      "step": 582
    },
    {
      "epoch": 0.3656891955464952,
      "grad_norm": 1.0872917175292969,
      "learning_rate": 3.760116947038981e-05,
      "loss": 0.3428,
      "step": 583
    },
    {
      "epoch": 0.36631644974125765,
      "grad_norm": 1.2489984035491943,
      "learning_rate": 3.759151075276145e-05,
      "loss": 0.3488,
      "step": 584
    },
    {
      "epoch": 0.36694370393602005,
      "grad_norm": 1.4249989986419678,
      "learning_rate": 3.7581833874797254e-05,
      "loss": 0.3354,
      "step": 585
    },
    {
      "epoch": 0.3675709581307825,
      "grad_norm": 1.3429278135299683,
      "learning_rate": 3.7572138846487e-05,
      "loss": 0.3471,
      "step": 586
    },
    {
      "epoch": 0.36819821232554495,
      "grad_norm": 1.1182597875595093,
      "learning_rate": 3.756242567783921e-05,
      "loss": 0.3138,
      "step": 587
    },
    {
      "epoch": 0.36882546652030734,
      "grad_norm": 1.22892427444458,
      "learning_rate": 3.755269437888113e-05,
      "loss": 0.3459,
      "step": 588
    },
    {
      "epoch": 0.3694527207150698,
      "grad_norm": 1.56260347366333,
      "learning_rate": 3.754294495965872e-05,
      "loss": 0.348,
      "step": 589
    },
    {
      "epoch": 0.3700799749098322,
      "grad_norm": 1.3148378133773804,
      "learning_rate": 3.753317743023665e-05,
      "loss": 0.3643,
      "step": 590
    },
    {
      "epoch": 0.37070722910459464,
      "grad_norm": 1.1588764190673828,
      "learning_rate": 3.7523391800698276e-05,
      "loss": 0.3409,
      "step": 591
    },
    {
      "epoch": 0.3713344832993571,
      "grad_norm": 1.202447772026062,
      "learning_rate": 3.751358808114567e-05,
      "loss": 0.3276,
      "step": 592
    },
    {
      "epoch": 0.3719617374941195,
      "grad_norm": 1.1717889308929443,
      "learning_rate": 3.7503766281699535e-05,
      "loss": 0.3171,
      "step": 593
    },
    {
      "epoch": 0.37258899168888193,
      "grad_norm": 1.106449842453003,
      "learning_rate": 3.7493926412499267e-05,
      "loss": 0.311,
      "step": 594
    },
    {
      "epoch": 0.3732162458836443,
      "grad_norm": 1.0862950086593628,
      "learning_rate": 3.748406848370291e-05,
      "loss": 0.319,
      "step": 595
    },
    {
      "epoch": 0.3738435000784068,
      "grad_norm": 1.1050410270690918,
      "learning_rate": 3.747419250548715e-05,
      "loss": 0.3431,
      "step": 596
    },
    {
      "epoch": 0.3744707542731692,
      "grad_norm": 1.0442489385604858,
      "learning_rate": 3.7464298488047314e-05,
      "loss": 0.3031,
      "step": 597
    },
    {
      "epoch": 0.3750980084679316,
      "grad_norm": 1.0916913747787476,
      "learning_rate": 3.745438644159734e-05,
      "loss": 0.3401,
      "step": 598
    },
    {
      "epoch": 0.37572526266269407,
      "grad_norm": 1.247828483581543,
      "learning_rate": 3.744445637636978e-05,
      "loss": 0.3915,
      "step": 599
    },
    {
      "epoch": 0.37635251685745646,
      "grad_norm": 1.3874841928482056,
      "learning_rate": 3.743450830261578e-05,
      "loss": 0.3032,
      "step": 600
    },
    {
      "epoch": 0.3769797710522189,
      "grad_norm": 1.1481797695159912,
      "learning_rate": 3.7424542230605115e-05,
      "loss": 0.3731,
      "step": 601
    },
    {
      "epoch": 0.37760702524698136,
      "grad_norm": 1.1188069581985474,
      "learning_rate": 3.7414558170626084e-05,
      "loss": 0.3415,
      "step": 602
    },
    {
      "epoch": 0.37823427944174376,
      "grad_norm": 1.2879520654678345,
      "learning_rate": 3.74045561329856e-05,
      "loss": 0.3372,
      "step": 603
    },
    {
      "epoch": 0.3788615336365062,
      "grad_norm": 1.4175255298614502,
      "learning_rate": 3.739453612800911e-05,
      "loss": 0.34,
      "step": 604
    },
    {
      "epoch": 0.3794887878312686,
      "grad_norm": 1.1900228261947632,
      "learning_rate": 3.738449816604062e-05,
      "loss": 0.3712,
      "step": 605
    },
    {
      "epoch": 0.38011604202603105,
      "grad_norm": 1.2238755226135254,
      "learning_rate": 3.737444225744267e-05,
      "loss": 0.3031,
      "step": 606
    },
    {
      "epoch": 0.3807432962207935,
      "grad_norm": 1.266327977180481,
      "learning_rate": 3.736436841259634e-05,
      "loss": 0.352,
      "step": 607
    },
    {
      "epoch": 0.3813705504155559,
      "grad_norm": 1.0682594776153564,
      "learning_rate": 3.7354276641901195e-05,
      "loss": 0.3583,
      "step": 608
    },
    {
      "epoch": 0.38199780461031835,
      "grad_norm": 1.384995698928833,
      "learning_rate": 3.734416695577534e-05,
      "loss": 0.3207,
      "step": 609
    },
    {
      "epoch": 0.38262505880508074,
      "grad_norm": 1.1252288818359375,
      "learning_rate": 3.733403936465536e-05,
      "loss": 0.2987,
      "step": 610
    },
    {
      "epoch": 0.3832523129998432,
      "grad_norm": 1.1938388347625732,
      "learning_rate": 3.7323893878996316e-05,
      "loss": 0.3524,
      "step": 611
    },
    {
      "epoch": 0.3838795671946056,
      "grad_norm": 1.3328951597213745,
      "learning_rate": 3.731373050927175e-05,
      "loss": 0.3044,
      "step": 612
    },
    {
      "epoch": 0.38450682138936804,
      "grad_norm": 1.1329494714736938,
      "learning_rate": 3.730354926597368e-05,
      "loss": 0.323,
      "step": 613
    },
    {
      "epoch": 0.3851340755841305,
      "grad_norm": 1.1388083696365356,
      "learning_rate": 3.7293350159612545e-05,
      "loss": 0.3148,
      "step": 614
    },
    {
      "epoch": 0.3857613297788929,
      "grad_norm": 1.149870753288269,
      "learning_rate": 3.728313320071725e-05,
      "loss": 0.2885,
      "step": 615
    },
    {
      "epoch": 0.38638858397365533,
      "grad_norm": 1.1834301948547363,
      "learning_rate": 3.7272898399835114e-05,
      "loss": 0.3581,
      "step": 616
    },
    {
      "epoch": 0.3870158381684177,
      "grad_norm": 1.1635024547576904,
      "learning_rate": 3.726264576753189e-05,
      "loss": 0.3265,
      "step": 617
    },
    {
      "epoch": 0.3876430923631802,
      "grad_norm": 1.914668321609497,
      "learning_rate": 3.7252375314391736e-05,
      "loss": 0.3274,
      "step": 618
    },
    {
      "epoch": 0.3882703465579426,
      "grad_norm": 1.167209506034851,
      "learning_rate": 3.724208705101719e-05,
      "loss": 0.2844,
      "step": 619
    },
    {
      "epoch": 0.388897600752705,
      "grad_norm": 1.2498699426651,
      "learning_rate": 3.72317809880292e-05,
      "loss": 0.2896,
      "step": 620
    },
    {
      "epoch": 0.38952485494746747,
      "grad_norm": 1.3018776178359985,
      "learning_rate": 3.722145713606706e-05,
      "loss": 0.3243,
      "step": 621
    },
    {
      "epoch": 0.39015210914222986,
      "grad_norm": 1.1689425706863403,
      "learning_rate": 3.7211115505788467e-05,
      "loss": 0.3126,
      "step": 622
    },
    {
      "epoch": 0.3907793633369923,
      "grad_norm": 1.0691388845443726,
      "learning_rate": 3.720075610786943e-05,
      "loss": 0.2924,
      "step": 623
    },
    {
      "epoch": 0.39140661753175476,
      "grad_norm": 1.1361101865768433,
      "learning_rate": 3.719037895300434e-05,
      "loss": 0.3153,
      "step": 624
    },
    {
      "epoch": 0.39203387172651716,
      "grad_norm": 1.1004093885421753,
      "learning_rate": 3.717998405190589e-05,
      "loss": 0.3138,
      "step": 625
    },
    {
      "epoch": 0.3926611259212796,
      "grad_norm": 0.9791253209114075,
      "learning_rate": 3.7169571415305104e-05,
      "loss": 0.2875,
      "step": 626
    },
    {
      "epoch": 0.393288380116042,
      "grad_norm": 1.083272099494934,
      "learning_rate": 3.71591410539513e-05,
      "loss": 0.3178,
      "step": 627
    },
    {
      "epoch": 0.39391563431080445,
      "grad_norm": 1.0247498750686646,
      "learning_rate": 3.714869297861213e-05,
      "loss": 0.357,
      "step": 628
    },
    {
      "epoch": 0.3945428885055669,
      "grad_norm": 1.1475611925125122,
      "learning_rate": 3.7138227200073495e-05,
      "loss": 0.3464,
      "step": 629
    },
    {
      "epoch": 0.3951701427003293,
      "grad_norm": 1.127882480621338,
      "learning_rate": 3.712774372913959e-05,
      "loss": 0.3384,
      "step": 630
    },
    {
      "epoch": 0.39579739689509175,
      "grad_norm": 1.4782390594482422,
      "learning_rate": 3.711724257663287e-05,
      "loss": 0.3523,
      "step": 631
    },
    {
      "epoch": 0.39642465108985414,
      "grad_norm": 1.1467382907867432,
      "learning_rate": 3.710672375339405e-05,
      "loss": 0.3009,
      "step": 632
    },
    {
      "epoch": 0.3970519052846166,
      "grad_norm": 1.283112645149231,
      "learning_rate": 3.709618727028208e-05,
      "loss": 0.3304,
      "step": 633
    },
    {
      "epoch": 0.39767915947937904,
      "grad_norm": 1.074562668800354,
      "learning_rate": 3.708563313817413e-05,
      "loss": 0.3218,
      "step": 634
    },
    {
      "epoch": 0.39830641367414144,
      "grad_norm": 1.056430697441101,
      "learning_rate": 3.7075061367965614e-05,
      "loss": 0.3028,
      "step": 635
    },
    {
      "epoch": 0.3989336678689039,
      "grad_norm": 0.8145391345024109,
      "learning_rate": 3.7064471970570146e-05,
      "loss": 0.2704,
      "step": 636
    },
    {
      "epoch": 0.3995609220636663,
      "grad_norm": 1.1371313333511353,
      "learning_rate": 3.7053864956919523e-05,
      "loss": 0.329,
      "step": 637
    },
    {
      "epoch": 0.40018817625842873,
      "grad_norm": 1.0839051008224487,
      "learning_rate": 3.704324033796375e-05,
      "loss": 0.3107,
      "step": 638
    },
    {
      "epoch": 0.4008154304531912,
      "grad_norm": 1.1648067235946655,
      "learning_rate": 3.703259812467098e-05,
      "loss": 0.3164,
      "step": 639
    },
    {
      "epoch": 0.4014426846479536,
      "grad_norm": 1.1891517639160156,
      "learning_rate": 3.7021938328027566e-05,
      "loss": 0.3461,
      "step": 640
    },
    {
      "epoch": 0.402069938842716,
      "grad_norm": 1.3886860609054565,
      "learning_rate": 3.7011260959037964e-05,
      "loss": 0.3754,
      "step": 641
    },
    {
      "epoch": 0.4026971930374784,
      "grad_norm": 1.1208279132843018,
      "learning_rate": 3.7000566028724825e-05,
      "loss": 0.2906,
      "step": 642
    },
    {
      "epoch": 0.40332444723224087,
      "grad_norm": 1.2986044883728027,
      "learning_rate": 3.6989853548128886e-05,
      "loss": 0.2868,
      "step": 643
    },
    {
      "epoch": 0.4039517014270033,
      "grad_norm": 1.0989553928375244,
      "learning_rate": 3.697912352830902e-05,
      "loss": 0.3277,
      "step": 644
    },
    {
      "epoch": 0.4045789556217657,
      "grad_norm": 1.2516005039215088,
      "learning_rate": 3.6968375980342206e-05,
      "loss": 0.3303,
      "step": 645
    },
    {
      "epoch": 0.40520620981652816,
      "grad_norm": 1.2921992540359497,
      "learning_rate": 3.6957610915323516e-05,
      "loss": 0.3041,
      "step": 646
    },
    {
      "epoch": 0.40583346401129056,
      "grad_norm": 1.204262375831604,
      "learning_rate": 3.69468283443661e-05,
      "loss": 0.3059,
      "step": 647
    },
    {
      "epoch": 0.406460718206053,
      "grad_norm": 1.205397605895996,
      "learning_rate": 3.69360282786012e-05,
      "loss": 0.3184,
      "step": 648
    },
    {
      "epoch": 0.40708797240081546,
      "grad_norm": 1.1826282739639282,
      "learning_rate": 3.692521072917809e-05,
      "loss": 0.3203,
      "step": 649
    },
    {
      "epoch": 0.40771522659557785,
      "grad_norm": 1.1708602905273438,
      "learning_rate": 3.691437570726412e-05,
      "loss": 0.3165,
      "step": 650
    },
    {
      "epoch": 0.4083424807903403,
      "grad_norm": 1.2916769981384277,
      "learning_rate": 3.690352322404465e-05,
      "loss": 0.3326,
      "step": 651
    },
    {
      "epoch": 0.4089697349851027,
      "grad_norm": 1.1950420141220093,
      "learning_rate": 3.6892653290723085e-05,
      "loss": 0.305,
      "step": 652
    },
    {
      "epoch": 0.40959698917986515,
      "grad_norm": 1.3934024572372437,
      "learning_rate": 3.688176591852084e-05,
      "loss": 0.3076,
      "step": 653
    },
    {
      "epoch": 0.4102242433746276,
      "grad_norm": 1.1073112487792969,
      "learning_rate": 3.6870861118677345e-05,
      "loss": 0.3165,
      "step": 654
    },
    {
      "epoch": 0.41085149756939,
      "grad_norm": 1.1293233633041382,
      "learning_rate": 3.6859938902449996e-05,
      "loss": 0.3136,
      "step": 655
    },
    {
      "epoch": 0.41147875176415244,
      "grad_norm": 1.1260145902633667,
      "learning_rate": 3.684899928111418e-05,
      "loss": 0.352,
      "step": 656
    },
    {
      "epoch": 0.41210600595891483,
      "grad_norm": 1.3531520366668701,
      "learning_rate": 3.683804226596326e-05,
      "loss": 0.3678,
      "step": 657
    },
    {
      "epoch": 0.4127332601536773,
      "grad_norm": 1.3921016454696655,
      "learning_rate": 3.682706786830854e-05,
      "loss": 0.3163,
      "step": 658
    },
    {
      "epoch": 0.4133605143484397,
      "grad_norm": 1.1174161434173584,
      "learning_rate": 3.6816076099479286e-05,
      "loss": 0.3197,
      "step": 659
    },
    {
      "epoch": 0.41398776854320213,
      "grad_norm": 1.5480095148086548,
      "learning_rate": 3.680506697082269e-05,
      "loss": 0.3097,
      "step": 660
    },
    {
      "epoch": 0.4146150227379646,
      "grad_norm": 1.157529354095459,
      "learning_rate": 3.6794040493703856e-05,
      "loss": 0.3148,
      "step": 661
    },
    {
      "epoch": 0.415242276932727,
      "grad_norm": 1.011178970336914,
      "learning_rate": 3.6782996679505794e-05,
      "loss": 0.3122,
      "step": 662
    },
    {
      "epoch": 0.4158695311274894,
      "grad_norm": 1.20054292678833,
      "learning_rate": 3.677193553962944e-05,
      "loss": 0.3488,
      "step": 663
    },
    {
      "epoch": 0.4164967853222518,
      "grad_norm": 1.2420406341552734,
      "learning_rate": 3.6760857085493595e-05,
      "loss": 0.3387,
      "step": 664
    },
    {
      "epoch": 0.41712403951701427,
      "grad_norm": 1.2079081535339355,
      "learning_rate": 3.674976132853494e-05,
      "loss": 0.3474,
      "step": 665
    },
    {
      "epoch": 0.4177512937117767,
      "grad_norm": 1.1342418193817139,
      "learning_rate": 3.6738648280208e-05,
      "loss": 0.3176,
      "step": 666
    },
    {
      "epoch": 0.4183785479065391,
      "grad_norm": 1.09598970413208,
      "learning_rate": 3.6727517951985185e-05,
      "loss": 0.3147,
      "step": 667
    },
    {
      "epoch": 0.41900580210130156,
      "grad_norm": 1.0939102172851562,
      "learning_rate": 3.671637035535671e-05,
      "loss": 0.2975,
      "step": 668
    },
    {
      "epoch": 0.41963305629606396,
      "grad_norm": 1.0264432430267334,
      "learning_rate": 3.6705205501830635e-05,
      "loss": 0.3037,
      "step": 669
    },
    {
      "epoch": 0.4202603104908264,
      "grad_norm": 1.2257050275802612,
      "learning_rate": 3.6694023402932835e-05,
      "loss": 0.2768,
      "step": 670
    },
    {
      "epoch": 0.42088756468558886,
      "grad_norm": 1.299730896949768,
      "learning_rate": 3.668282407020699e-05,
      "loss": 0.3634,
      "step": 671
    },
    {
      "epoch": 0.42151481888035125,
      "grad_norm": 1.1695563793182373,
      "learning_rate": 3.667160751521455e-05,
      "loss": 0.3129,
      "step": 672
    },
    {
      "epoch": 0.4221420730751137,
      "grad_norm": 1.1521062850952148,
      "learning_rate": 3.666037374953477e-05,
      "loss": 0.3308,
      "step": 673
    },
    {
      "epoch": 0.4227693272698761,
      "grad_norm": 1.1733102798461914,
      "learning_rate": 3.664912278476466e-05,
      "loss": 0.2691,
      "step": 674
    },
    {
      "epoch": 0.42339658146463854,
      "grad_norm": 1.1485964059829712,
      "learning_rate": 3.6637854632518985e-05,
      "loss": 0.2979,
      "step": 675
    },
    {
      "epoch": 0.424023835659401,
      "grad_norm": 1.2730668783187866,
      "learning_rate": 3.662656930443026e-05,
      "loss": 0.3181,
      "step": 676
    },
    {
      "epoch": 0.4246510898541634,
      "grad_norm": 1.181434988975525,
      "learning_rate": 3.6615266812148724e-05,
      "loss": 0.3085,
      "step": 677
    },
    {
      "epoch": 0.42527834404892584,
      "grad_norm": 1.1413264274597168,
      "learning_rate": 3.6603947167342335e-05,
      "loss": 0.3222,
      "step": 678
    },
    {
      "epoch": 0.42590559824368823,
      "grad_norm": 1.2547311782836914,
      "learning_rate": 3.659261038169676e-05,
      "loss": 0.3595,
      "step": 679
    },
    {
      "epoch": 0.4265328524384507,
      "grad_norm": 1.1639899015426636,
      "learning_rate": 3.658125646691537e-05,
      "loss": 0.3257,
      "step": 680
    },
    {
      "epoch": 0.42716010663321313,
      "grad_norm": 1.4667763710021973,
      "learning_rate": 3.6569885434719205e-05,
      "loss": 0.3502,
      "step": 681
    },
    {
      "epoch": 0.4277873608279755,
      "grad_norm": 1.1822808980941772,
      "learning_rate": 3.6558497296846976e-05,
      "loss": 0.3213,
      "step": 682
    },
    {
      "epoch": 0.428414615022738,
      "grad_norm": 1.0532478094100952,
      "learning_rate": 3.654709206505507e-05,
      "loss": 0.3176,
      "step": 683
    },
    {
      "epoch": 0.42904186921750037,
      "grad_norm": 1.0396804809570312,
      "learning_rate": 3.65356697511175e-05,
      "loss": 0.2881,
      "step": 684
    },
    {
      "epoch": 0.4296691234122628,
      "grad_norm": 1.1286225318908691,
      "learning_rate": 3.6524230366825935e-05,
      "loss": 0.3311,
      "step": 685
    },
    {
      "epoch": 0.43029637760702527,
      "grad_norm": 1.092599630355835,
      "learning_rate": 3.6512773923989645e-05,
      "loss": 0.2964,
      "step": 686
    },
    {
      "epoch": 0.43092363180178767,
      "grad_norm": 1.198910117149353,
      "learning_rate": 3.650130043443552e-05,
      "loss": 0.3192,
      "step": 687
    },
    {
      "epoch": 0.4315508859965501,
      "grad_norm": 1.1863718032836914,
      "learning_rate": 3.648980991000804e-05,
      "loss": 0.2836,
      "step": 688
    },
    {
      "epoch": 0.4321781401913125,
      "grad_norm": 1.3961267471313477,
      "learning_rate": 3.6478302362569293e-05,
      "loss": 0.3608,
      "step": 689
    },
    {
      "epoch": 0.43280539438607496,
      "grad_norm": 1.1938984394073486,
      "learning_rate": 3.6466777803998915e-05,
      "loss": 0.3076,
      "step": 690
    },
    {
      "epoch": 0.4334326485808374,
      "grad_norm": 1.128846287727356,
      "learning_rate": 3.645523624619412e-05,
      "loss": 0.3223,
      "step": 691
    },
    {
      "epoch": 0.4340599027755998,
      "grad_norm": 1.1837422847747803,
      "learning_rate": 3.644367770106966e-05,
      "loss": 0.3188,
      "step": 692
    },
    {
      "epoch": 0.43468715697036225,
      "grad_norm": 1.122409462928772,
      "learning_rate": 3.643210218055783e-05,
      "loss": 0.2997,
      "step": 693
    },
    {
      "epoch": 0.43531441116512465,
      "grad_norm": 1.1411001682281494,
      "learning_rate": 3.642050969660844e-05,
      "loss": 0.337,
      "step": 694
    },
    {
      "epoch": 0.4359416653598871,
      "grad_norm": 1.231703758239746,
      "learning_rate": 3.640890026118883e-05,
      "loss": 0.3357,
      "step": 695
    },
    {
      "epoch": 0.43656891955464955,
      "grad_norm": 0.9981296062469482,
      "learning_rate": 3.639727388628383e-05,
      "loss": 0.2708,
      "step": 696
    },
    {
      "epoch": 0.43719617374941194,
      "grad_norm": 1.0697611570358276,
      "learning_rate": 3.638563058389574e-05,
      "loss": 0.3111,
      "step": 697
    },
    {
      "epoch": 0.4378234279441744,
      "grad_norm": 1.3875848054885864,
      "learning_rate": 3.6373970366044366e-05,
      "loss": 0.3417,
      "step": 698
    },
    {
      "epoch": 0.4384506821389368,
      "grad_norm": 1.3041980266571045,
      "learning_rate": 3.636229324476696e-05,
      "loss": 0.3391,
      "step": 699
    },
    {
      "epoch": 0.43907793633369924,
      "grad_norm": 1.058411955833435,
      "learning_rate": 3.635059923211822e-05,
      "loss": 0.3055,
      "step": 700
    },
    {
      "epoch": 0.4397051905284617,
      "grad_norm": 1.2041007280349731,
      "learning_rate": 3.633888834017029e-05,
      "loss": 0.3058,
      "step": 701
    },
    {
      "epoch": 0.4403324447232241,
      "grad_norm": 1.2628463506698608,
      "learning_rate": 3.632716058101273e-05,
      "loss": 0.3144,
      "step": 702
    },
    {
      "epoch": 0.44095969891798653,
      "grad_norm": 1.3504395484924316,
      "learning_rate": 3.631541596675253e-05,
      "loss": 0.3256,
      "step": 703
    },
    {
      "epoch": 0.4415869531127489,
      "grad_norm": 1.0061591863632202,
      "learning_rate": 3.630365450951406e-05,
      "loss": 0.2783,
      "step": 704
    },
    {
      "epoch": 0.4422142073075114,
      "grad_norm": 1.075989007949829,
      "learning_rate": 3.629187622143909e-05,
      "loss": 0.2889,
      "step": 705
    },
    {
      "epoch": 0.44284146150227377,
      "grad_norm": 1.089695930480957,
      "learning_rate": 3.6280081114686757e-05,
      "loss": 0.3028,
      "step": 706
    },
    {
      "epoch": 0.4434687156970362,
      "grad_norm": 1.1697717905044556,
      "learning_rate": 3.626826920143357e-05,
      "loss": 0.3198,
      "step": 707
    },
    {
      "epoch": 0.44409596989179867,
      "grad_norm": 1.14765465259552,
      "learning_rate": 3.6256440493873385e-05,
      "loss": 0.2959,
      "step": 708
    },
    {
      "epoch": 0.44472322408656106,
      "grad_norm": 1.1275452375411987,
      "learning_rate": 3.6244595004217385e-05,
      "loss": 0.3092,
      "step": 709
    },
    {
      "epoch": 0.4453504782813235,
      "grad_norm": 0.986223578453064,
      "learning_rate": 3.623273274469409e-05,
      "loss": 0.2817,
      "step": 710
    },
    {
      "epoch": 0.4459777324760859,
      "grad_norm": 1.11913001537323,
      "learning_rate": 3.622085372754933e-05,
      "loss": 0.2931,
      "step": 711
    },
    {
      "epoch": 0.44660498667084836,
      "grad_norm": 1.1649326086044312,
      "learning_rate": 3.6208957965046234e-05,
      "loss": 0.2926,
      "step": 712
    },
    {
      "epoch": 0.4472322408656108,
      "grad_norm": 1.0906825065612793,
      "learning_rate": 3.619704546946521e-05,
      "loss": 0.2339,
      "step": 713
    },
    {
      "epoch": 0.4478594950603732,
      "grad_norm": 1.333734154701233,
      "learning_rate": 3.618511625310395e-05,
      "loss": 0.3512,
      "step": 714
    },
    {
      "epoch": 0.44848674925513565,
      "grad_norm": 1.1435856819152832,
      "learning_rate": 3.617317032827741e-05,
      "loss": 0.2765,
      "step": 715
    },
    {
      "epoch": 0.44911400344989805,
      "grad_norm": 1.174895167350769,
      "learning_rate": 3.616120770731778e-05,
      "loss": 0.3099,
      "step": 716
    },
    {
      "epoch": 0.4497412576446605,
      "grad_norm": 1.352873682975769,
      "learning_rate": 3.6149228402574485e-05,
      "loss": 0.3625,
      "step": 717
    },
    {
      "epoch": 0.45036851183942295,
      "grad_norm": 1.1254205703735352,
      "learning_rate": 3.613723242641421e-05,
      "loss": 0.3131,
      "step": 718
    },
    {
      "epoch": 0.45099576603418534,
      "grad_norm": 1.3090929985046387,
      "learning_rate": 3.6125219791220804e-05,
      "loss": 0.3151,
      "step": 719
    },
    {
      "epoch": 0.4516230202289478,
      "grad_norm": 1.2363439798355103,
      "learning_rate": 3.611319050939534e-05,
      "loss": 0.293,
      "step": 720
    },
    {
      "epoch": 0.4522502744237102,
      "grad_norm": 1.2641264200210571,
      "learning_rate": 3.6101144593356066e-05,
      "loss": 0.3381,
      "step": 721
    },
    {
      "epoch": 0.45287752861847264,
      "grad_norm": 1.0665712356567383,
      "learning_rate": 3.608908205553841e-05,
      "loss": 0.3114,
      "step": 722
    },
    {
      "epoch": 0.4535047828132351,
      "grad_norm": 1.1418439149856567,
      "learning_rate": 3.607700290839495e-05,
      "loss": 0.2869,
      "step": 723
    },
    {
      "epoch": 0.4541320370079975,
      "grad_norm": 1.339607834815979,
      "learning_rate": 3.606490716439541e-05,
      "loss": 0.313,
      "step": 724
    },
    {
      "epoch": 0.45475929120275993,
      "grad_norm": 1.1564210653305054,
      "learning_rate": 3.605279483602666e-05,
      "loss": 0.3399,
      "step": 725
    },
    {
      "epoch": 0.4553865453975223,
      "grad_norm": 1.1793553829193115,
      "learning_rate": 3.604066593579269e-05,
      "loss": 0.2752,
      "step": 726
    },
    {
      "epoch": 0.4560137995922848,
      "grad_norm": 1.0112354755401611,
      "learning_rate": 3.602852047621457e-05,
      "loss": 0.3143,
      "step": 727
    },
    {
      "epoch": 0.4566410537870472,
      "grad_norm": 1.1206257343292236,
      "learning_rate": 3.60163584698305e-05,
      "loss": 0.3092,
      "step": 728
    },
    {
      "epoch": 0.4572683079818096,
      "grad_norm": 1.2401798963546753,
      "learning_rate": 3.600417992919575e-05,
      "loss": 0.3128,
      "step": 729
    },
    {
      "epoch": 0.45789556217657207,
      "grad_norm": 1.2118809223175049,
      "learning_rate": 3.599198486688264e-05,
      "loss": 0.3121,
      "step": 730
    },
    {
      "epoch": 0.45852281637133446,
      "grad_norm": 1.0759859085083008,
      "learning_rate": 3.597977329548059e-05,
      "loss": 0.3004,
      "step": 731
    },
    {
      "epoch": 0.4591500705660969,
      "grad_norm": 1.1380683183670044,
      "learning_rate": 3.596754522759601e-05,
      "loss": 0.3007,
      "step": 732
    },
    {
      "epoch": 0.45977732476085936,
      "grad_norm": 1.0690897703170776,
      "learning_rate": 3.595530067585237e-05,
      "loss": 0.2819,
      "step": 733
    },
    {
      "epoch": 0.46040457895562176,
      "grad_norm": 1.059333086013794,
      "learning_rate": 3.5943039652890154e-05,
      "loss": 0.3236,
      "step": 734
    },
    {
      "epoch": 0.4610318331503842,
      "grad_norm": 1.1993666887283325,
      "learning_rate": 3.5930762171366855e-05,
      "loss": 0.2923,
      "step": 735
    },
    {
      "epoch": 0.4616590873451466,
      "grad_norm": 1.2433502674102783,
      "learning_rate": 3.591846824395694e-05,
      "loss": 0.32,
      "step": 736
    },
    {
      "epoch": 0.46228634153990905,
      "grad_norm": 1.0564230680465698,
      "learning_rate": 3.5906157883351864e-05,
      "loss": 0.3052,
      "step": 737
    },
    {
      "epoch": 0.4629135957346715,
      "grad_norm": 1.0913139581680298,
      "learning_rate": 3.589383110226004e-05,
      "loss": 0.3229,
      "step": 738
    },
    {
      "epoch": 0.4635408499294339,
      "grad_norm": 1.049662709236145,
      "learning_rate": 3.588148791340686e-05,
      "loss": 0.2653,
      "step": 739
    },
    {
      "epoch": 0.46416810412419635,
      "grad_norm": 1.0421611070632935,
      "learning_rate": 3.586912832953461e-05,
      "loss": 0.3305,
      "step": 740
    },
    {
      "epoch": 0.46479535831895874,
      "grad_norm": 1.1456973552703857,
      "learning_rate": 3.585675236340254e-05,
      "loss": 0.2904,
      "step": 741
    },
    {
      "epoch": 0.4654226125137212,
      "grad_norm": 1.1543675661087036,
      "learning_rate": 3.5844360027786776e-05,
      "loss": 0.299,
      "step": 742
    },
    {
      "epoch": 0.46604986670848364,
      "grad_norm": 1.1671220064163208,
      "learning_rate": 3.583195133548038e-05,
      "loss": 0.3057,
      "step": 743
    },
    {
      "epoch": 0.46667712090324603,
      "grad_norm": 1.256575107574463,
      "learning_rate": 3.581952629929327e-05,
      "loss": 0.3138,
      "step": 744
    },
    {
      "epoch": 0.4673043750980085,
      "grad_norm": 1.1006120443344116,
      "learning_rate": 3.580708493205226e-05,
      "loss": 0.3048,
      "step": 745
    },
    {
      "epoch": 0.4679316292927709,
      "grad_norm": 1.2163777351379395,
      "learning_rate": 3.5794627246600996e-05,
      "loss": 0.3516,
      "step": 746
    },
    {
      "epoch": 0.46855888348753333,
      "grad_norm": 1.1042163372039795,
      "learning_rate": 3.57821532558e-05,
      "loss": 0.2687,
      "step": 747
    },
    {
      "epoch": 0.4691861376822957,
      "grad_norm": 1.1301376819610596,
      "learning_rate": 3.5769662972526604e-05,
      "loss": 0.3447,
      "step": 748
    },
    {
      "epoch": 0.4698133918770582,
      "grad_norm": 1.1545332670211792,
      "learning_rate": 3.5757156409674976e-05,
      "loss": 0.3084,
      "step": 749
    },
    {
      "epoch": 0.4704406460718206,
      "grad_norm": 1.0360727310180664,
      "learning_rate": 3.574463358015607e-05,
      "loss": 0.2986,
      "step": 750
    },
    {
      "epoch": 0.471067900266583,
      "grad_norm": 1.0242793560028076,
      "learning_rate": 3.573209449689765e-05,
      "loss": 0.3232,
      "step": 751
    },
    {
      "epoch": 0.47169515446134547,
      "grad_norm": 1.1782797574996948,
      "learning_rate": 3.5719539172844245e-05,
      "loss": 0.3084,
      "step": 752
    },
    {
      "epoch": 0.47232240865610786,
      "grad_norm": 1.0539721250534058,
      "learning_rate": 3.570696762095716e-05,
      "loss": 0.3313,
      "step": 753
    },
    {
      "epoch": 0.4729496628508703,
      "grad_norm": 1.0124305486679077,
      "learning_rate": 3.569437985421447e-05,
      "loss": 0.2994,
      "step": 754
    },
    {
      "epoch": 0.47357691704563276,
      "grad_norm": 1.290825366973877,
      "learning_rate": 3.568177588561095e-05,
      "loss": 0.3425,
      "step": 755
    },
    {
      "epoch": 0.47420417124039516,
      "grad_norm": 1.0773361921310425,
      "learning_rate": 3.566915572815812e-05,
      "loss": 0.2549,
      "step": 756
    },
    {
      "epoch": 0.4748314254351576,
      "grad_norm": 1.2297642230987549,
      "learning_rate": 3.565651939488422e-05,
      "loss": 0.2747,
      "step": 757
    },
    {
      "epoch": 0.47545867962992,
      "grad_norm": 1.120935082435608,
      "learning_rate": 3.5643866898834175e-05,
      "loss": 0.296,
      "step": 758
    },
    {
      "epoch": 0.47608593382468245,
      "grad_norm": 1.1990840435028076,
      "learning_rate": 3.563119825306961e-05,
      "loss": 0.2879,
      "step": 759
    },
    {
      "epoch": 0.4767131880194449,
      "grad_norm": 1.2168325185775757,
      "learning_rate": 3.561851347066881e-05,
      "loss": 0.3373,
      "step": 760
    },
    {
      "epoch": 0.4773404422142073,
      "grad_norm": 1.1412773132324219,
      "learning_rate": 3.560581256472672e-05,
      "loss": 0.3045,
      "step": 761
    },
    {
      "epoch": 0.47796769640896974,
      "grad_norm": 1.0286948680877686,
      "learning_rate": 3.559309554835494e-05,
      "loss": 0.3177,
      "step": 762
    },
    {
      "epoch": 0.47859495060373214,
      "grad_norm": 1.2216492891311646,
      "learning_rate": 3.558036243468169e-05,
      "loss": 0.2898,
      "step": 763
    },
    {
      "epoch": 0.4792222047984946,
      "grad_norm": 1.2539747953414917,
      "learning_rate": 3.556761323685181e-05,
      "loss": 0.3046,
      "step": 764
    },
    {
      "epoch": 0.47984945899325704,
      "grad_norm": 1.0702846050262451,
      "learning_rate": 3.555484796802674e-05,
      "loss": 0.3107,
      "step": 765
    },
    {
      "epoch": 0.48047671318801943,
      "grad_norm": 0.9739949107170105,
      "learning_rate": 3.554206664138453e-05,
      "loss": 0.3133,
      "step": 766
    },
    {
      "epoch": 0.4811039673827819,
      "grad_norm": 1.2004176378250122,
      "learning_rate": 3.5529269270119777e-05,
      "loss": 0.3141,
      "step": 767
    },
    {
      "epoch": 0.4817312215775443,
      "grad_norm": 0.9493788480758667,
      "learning_rate": 3.551645586744368e-05,
      "loss": 0.2525,
      "step": 768
    },
    {
      "epoch": 0.4823584757723067,
      "grad_norm": 1.0642143487930298,
      "learning_rate": 3.550362644658394e-05,
      "loss": 0.3068,
      "step": 769
    },
    {
      "epoch": 0.4829857299670692,
      "grad_norm": 1.0598788261413574,
      "learning_rate": 3.549078102078484e-05,
      "loss": 0.3209,
      "step": 770
    },
    {
      "epoch": 0.48361298416183157,
      "grad_norm": 1.1478248834609985,
      "learning_rate": 3.547791960330716e-05,
      "loss": 0.3304,
      "step": 771
    },
    {
      "epoch": 0.484240238356594,
      "grad_norm": 1.1824041604995728,
      "learning_rate": 3.546504220742818e-05,
      "loss": 0.2554,
      "step": 772
    },
    {
      "epoch": 0.4848674925513564,
      "grad_norm": 0.9910576939582825,
      "learning_rate": 3.5452148846441715e-05,
      "loss": 0.2553,
      "step": 773
    },
    {
      "epoch": 0.48549474674611887,
      "grad_norm": 1.1317683458328247,
      "learning_rate": 3.543923953365801e-05,
      "loss": 0.2876,
      "step": 774
    },
    {
      "epoch": 0.4861220009408813,
      "grad_norm": 1.0811703205108643,
      "learning_rate": 3.5426314282403815e-05,
      "loss": 0.2653,
      "step": 775
    },
    {
      "epoch": 0.4867492551356437,
      "grad_norm": 1.1303654909133911,
      "learning_rate": 3.541337310602233e-05,
      "loss": 0.2883,
      "step": 776
    },
    {
      "epoch": 0.48737650933040616,
      "grad_norm": 1.4702017307281494,
      "learning_rate": 3.5400416017873165e-05,
      "loss": 0.3021,
      "step": 777
    },
    {
      "epoch": 0.48800376352516855,
      "grad_norm": 0.9657360911369324,
      "learning_rate": 3.53874430313324e-05,
      "loss": 0.2672,
      "step": 778
    },
    {
      "epoch": 0.488631017719931,
      "grad_norm": 0.9891339540481567,
      "learning_rate": 3.5374454159792493e-05,
      "loss": 0.2653,
      "step": 779
    },
    {
      "epoch": 0.48925827191469345,
      "grad_norm": 0.9610922932624817,
      "learning_rate": 3.536144941666233e-05,
      "loss": 0.2724,
      "step": 780
    },
    {
      "epoch": 0.48988552610945585,
      "grad_norm": 0.9398078918457031,
      "learning_rate": 3.5348428815367134e-05,
      "loss": 0.2575,
      "step": 781
    },
    {
      "epoch": 0.4905127803042183,
      "grad_norm": 0.9750205278396606,
      "learning_rate": 3.533539236934856e-05,
      "loss": 0.2673,
      "step": 782
    },
    {
      "epoch": 0.4911400344989807,
      "grad_norm": 1.1842604875564575,
      "learning_rate": 3.5322340092064575e-05,
      "loss": 0.2975,
      "step": 783
    },
    {
      "epoch": 0.49176728869374314,
      "grad_norm": 1.1964186429977417,
      "learning_rate": 3.530927199698951e-05,
      "loss": 0.2884,
      "step": 784
    },
    {
      "epoch": 0.4923945428885056,
      "grad_norm": 1.0833983421325684,
      "learning_rate": 3.529618809761402e-05,
      "loss": 0.2802,
      "step": 785
    },
    {
      "epoch": 0.493021797083268,
      "grad_norm": 1.0984426736831665,
      "learning_rate": 3.528308840744506e-05,
      "loss": 0.2913,
      "step": 786
    },
    {
      "epoch": 0.49364905127803044,
      "grad_norm": 1.1128063201904297,
      "learning_rate": 3.526997294000593e-05,
      "loss": 0.311,
      "step": 787
    },
    {
      "epoch": 0.49427630547279283,
      "grad_norm": 1.005001187324524,
      "learning_rate": 3.5256841708836164e-05,
      "loss": 0.276,
      "step": 788
    },
    {
      "epoch": 0.4949035596675553,
      "grad_norm": 1.0180959701538086,
      "learning_rate": 3.524369472749161e-05,
      "loss": 0.3069,
      "step": 789
    },
    {
      "epoch": 0.49553081386231773,
      "grad_norm": 0.9974920153617859,
      "learning_rate": 3.523053200954435e-05,
      "loss": 0.2875,
      "step": 790
    },
    {
      "epoch": 0.4961580680570801,
      "grad_norm": 1.0186494588851929,
      "learning_rate": 3.5217353568582715e-05,
      "loss": 0.2798,
      "step": 791
    },
    {
      "epoch": 0.4967853222518426,
      "grad_norm": 1.0674657821655273,
      "learning_rate": 3.520415941821129e-05,
      "loss": 0.3156,
      "step": 792
    },
    {
      "epoch": 0.49741257644660497,
      "grad_norm": 1.0009441375732422,
      "learning_rate": 3.519094957205085e-05,
      "loss": 0.2827,
      "step": 793
    },
    {
      "epoch": 0.4980398306413674,
      "grad_norm": 0.9863251447677612,
      "learning_rate": 3.517772404373839e-05,
      "loss": 0.2943,
      "step": 794
    },
    {
      "epoch": 0.4986670848361298,
      "grad_norm": 1.0082939863204956,
      "learning_rate": 3.5164482846927084e-05,
      "loss": 0.2979,
      "step": 795
    },
    {
      "epoch": 0.49929433903089226,
      "grad_norm": 1.0588964223861694,
      "learning_rate": 3.515122599528629e-05,
      "loss": 0.2701,
      "step": 796
    },
    {
      "epoch": 0.4999215932256547,
      "grad_norm": 1.0246416330337524,
      "learning_rate": 3.5137953502501516e-05,
      "loss": 0.3135,
      "step": 797
    },
    {
      "epoch": 0.5005488474204172,
      "grad_norm": 0.9969562888145447,
      "learning_rate": 3.512466538227443e-05,
      "loss": 0.315,
      "step": 798
    },
    {
      "epoch": 0.5011761016151796,
      "grad_norm": 0.9564127326011658,
      "learning_rate": 3.511136164832282e-05,
      "loss": 0.2659,
      "step": 799
    },
    {
      "epoch": 0.501803355809942,
      "grad_norm": 0.97264564037323,
      "learning_rate": 3.509804231438061e-05,
      "loss": 0.2534,
      "step": 800
    },
    {
      "epoch": 0.5024306100047045,
      "grad_norm": 1.067979335784912,
      "learning_rate": 3.508470739419781e-05,
      "loss": 0.3077,
      "step": 801
    },
    {
      "epoch": 0.5030578641994669,
      "grad_norm": 1.0767468214035034,
      "learning_rate": 3.5071356901540524e-05,
      "loss": 0.2903,
      "step": 802
    },
    {
      "epoch": 0.5036851183942292,
      "grad_norm": 1.055936336517334,
      "learning_rate": 3.505799085019094e-05,
      "loss": 0.2412,
      "step": 803
    },
    {
      "epoch": 0.5043123725889916,
      "grad_norm": 1.1777431964874268,
      "learning_rate": 3.50446092539473e-05,
      "loss": 0.3098,
      "step": 804
    },
    {
      "epoch": 0.5049396267837541,
      "grad_norm": 1.3164564371109009,
      "learning_rate": 3.50312121266239e-05,
      "loss": 0.3483,
      "step": 805
    },
    {
      "epoch": 0.5055668809785165,
      "grad_norm": 1.227178692817688,
      "learning_rate": 3.5017799482051064e-05,
      "loss": 0.274,
      "step": 806
    },
    {
      "epoch": 0.5061941351732789,
      "grad_norm": 1.074154019355774,
      "learning_rate": 3.500437133407514e-05,
      "loss": 0.2481,
      "step": 807
    },
    {
      "epoch": 0.5068213893680414,
      "grad_norm": 0.9530861973762512,
      "learning_rate": 3.499092769655849e-05,
      "loss": 0.2857,
      "step": 808
    },
    {
      "epoch": 0.5074486435628038,
      "grad_norm": 1.2343348264694214,
      "learning_rate": 3.497746858337943e-05,
      "loss": 0.2962,
      "step": 809
    },
    {
      "epoch": 0.5080758977575662,
      "grad_norm": 1.5502104759216309,
      "learning_rate": 3.4963994008432284e-05,
      "loss": 0.3189,
      "step": 810
    },
    {
      "epoch": 0.5087031519523287,
      "grad_norm": 1.1012052297592163,
      "learning_rate": 3.4950503985627336e-05,
      "loss": 0.3039,
      "step": 811
    },
    {
      "epoch": 0.5093304061470911,
      "grad_norm": 0.9534533023834229,
      "learning_rate": 3.493699852889081e-05,
      "loss": 0.2334,
      "step": 812
    },
    {
      "epoch": 0.5099576603418535,
      "grad_norm": 0.9859058260917664,
      "learning_rate": 3.4923477652164855e-05,
      "loss": 0.2561,
      "step": 813
    },
    {
      "epoch": 0.5105849145366159,
      "grad_norm": 1.1005891561508179,
      "learning_rate": 3.490994136940757e-05,
      "loss": 0.2924,
      "step": 814
    },
    {
      "epoch": 0.5112121687313784,
      "grad_norm": 1.0567169189453125,
      "learning_rate": 3.489638969459292e-05,
      "loss": 0.3075,
      "step": 815
    },
    {
      "epoch": 0.5118394229261408,
      "grad_norm": 0.9554606676101685,
      "learning_rate": 3.488282264171077e-05,
      "loss": 0.2678,
      "step": 816
    },
    {
      "epoch": 0.5124666771209032,
      "grad_norm": 1.0339974164962769,
      "learning_rate": 3.486924022476688e-05,
      "loss": 0.2809,
      "step": 817
    },
    {
      "epoch": 0.5130939313156657,
      "grad_norm": 1.0772579908370972,
      "learning_rate": 3.485564245778285e-05,
      "loss": 0.2799,
      "step": 818
    },
    {
      "epoch": 0.5137211855104281,
      "grad_norm": 0.9478000402450562,
      "learning_rate": 3.484202935479614e-05,
      "loss": 0.2386,
      "step": 819
    },
    {
      "epoch": 0.5143484397051905,
      "grad_norm": 1.0336586236953735,
      "learning_rate": 3.4828400929860023e-05,
      "loss": 0.3335,
      "step": 820
    },
    {
      "epoch": 0.5149756938999529,
      "grad_norm": 1.0443904399871826,
      "learning_rate": 3.481475719704362e-05,
      "loss": 0.304,
      "step": 821
    },
    {
      "epoch": 0.5156029480947154,
      "grad_norm": 1.0557197332382202,
      "learning_rate": 3.480109817043182e-05,
      "loss": 0.2804,
      "step": 822
    },
    {
      "epoch": 0.5162302022894778,
      "grad_norm": 1.3195269107818604,
      "learning_rate": 3.478742386412533e-05,
      "loss": 0.3085,
      "step": 823
    },
    {
      "epoch": 0.5168574564842402,
      "grad_norm": 1.0132356882095337,
      "learning_rate": 3.47737342922406e-05,
      "loss": 0.2515,
      "step": 824
    },
    {
      "epoch": 0.5174847106790027,
      "grad_norm": 0.9900258183479309,
      "learning_rate": 3.4760029468909876e-05,
      "loss": 0.2801,
      "step": 825
    },
    {
      "epoch": 0.5181119648737651,
      "grad_norm": 1.2215389013290405,
      "learning_rate": 3.4746309408281123e-05,
      "loss": 0.295,
      "step": 826
    },
    {
      "epoch": 0.5187392190685275,
      "grad_norm": 1.0417739152908325,
      "learning_rate": 3.473257412451803e-05,
      "loss": 0.2856,
      "step": 827
    },
    {
      "epoch": 0.51936647326329,
      "grad_norm": 1.002961277961731,
      "learning_rate": 3.471882363180004e-05,
      "loss": 0.2985,
      "step": 828
    },
    {
      "epoch": 0.5199937274580524,
      "grad_norm": 1.025743007659912,
      "learning_rate": 3.4705057944322254e-05,
      "loss": 0.3011,
      "step": 829
    },
    {
      "epoch": 0.5206209816528148,
      "grad_norm": 0.9813662767410278,
      "learning_rate": 3.4691277076295464e-05,
      "loss": 0.2533,
      "step": 830
    },
    {
      "epoch": 0.5212482358475772,
      "grad_norm": 0.9720662236213684,
      "learning_rate": 3.467748104194617e-05,
      "loss": 0.2474,
      "step": 831
    },
    {
      "epoch": 0.5218754900423397,
      "grad_norm": 0.9965443015098572,
      "learning_rate": 3.46636698555165e-05,
      "loss": 0.3097,
      "step": 832
    },
    {
      "epoch": 0.5225027442371021,
      "grad_norm": 0.9662636518478394,
      "learning_rate": 3.4649843531264214e-05,
      "loss": 0.297,
      "step": 833
    },
    {
      "epoch": 0.5231299984318645,
      "grad_norm": 1.0954393148422241,
      "learning_rate": 3.4636002083462714e-05,
      "loss": 0.272,
      "step": 834
    },
    {
      "epoch": 0.523757252626627,
      "grad_norm": 1.066875696182251,
      "learning_rate": 3.462214552640103e-05,
      "loss": 0.3458,
      "step": 835
    },
    {
      "epoch": 0.5243845068213894,
      "grad_norm": 1.1262314319610596,
      "learning_rate": 3.460827387438376e-05,
      "loss": 0.2879,
      "step": 836
    },
    {
      "epoch": 0.5250117610161518,
      "grad_norm": 1.11569344997406,
      "learning_rate": 3.4594387141731095e-05,
      "loss": 0.2554,
      "step": 837
    },
    {
      "epoch": 0.5256390152109143,
      "grad_norm": 1.0312837362289429,
      "learning_rate": 3.4580485342778815e-05,
      "loss": 0.2899,
      "step": 838
    },
    {
      "epoch": 0.5262662694056767,
      "grad_norm": 1.1218152046203613,
      "learning_rate": 3.456656849187822e-05,
      "loss": 0.2901,
      "step": 839
    },
    {
      "epoch": 0.5268935236004391,
      "grad_norm": 1.0698727369308472,
      "learning_rate": 3.455263660339618e-05,
      "loss": 0.3009,
      "step": 840
    },
    {
      "epoch": 0.5275207777952015,
      "grad_norm": 1.0733284950256348,
      "learning_rate": 3.453868969171507e-05,
      "loss": 0.2664,
      "step": 841
    },
    {
      "epoch": 0.528148031989964,
      "grad_norm": 1.1404551267623901,
      "learning_rate": 3.452472777123277e-05,
      "loss": 0.2939,
      "step": 842
    },
    {
      "epoch": 0.5287752861847264,
      "grad_norm": 1.1484930515289307,
      "learning_rate": 3.451075085636265e-05,
      "loss": 0.294,
      "step": 843
    },
    {
      "epoch": 0.5294025403794888,
      "grad_norm": 1.1404722929000854,
      "learning_rate": 3.44967589615336e-05,
      "loss": 0.2949,
      "step": 844
    },
    {
      "epoch": 0.5300297945742513,
      "grad_norm": 1.1326576471328735,
      "learning_rate": 3.4482752101189934e-05,
      "loss": 0.2869,
      "step": 845
    },
    {
      "epoch": 0.5306570487690137,
      "grad_norm": 1.152813196182251,
      "learning_rate": 3.4468730289791424e-05,
      "loss": 0.2791,
      "step": 846
    },
    {
      "epoch": 0.531284302963776,
      "grad_norm": 1.0741047859191895,
      "learning_rate": 3.445469354181328e-05,
      "loss": 0.2799,
      "step": 847
    },
    {
      "epoch": 0.5319115571585386,
      "grad_norm": 1.0504900217056274,
      "learning_rate": 3.444064187174613e-05,
      "loss": 0.3542,
      "step": 848
    },
    {
      "epoch": 0.532538811353301,
      "grad_norm": 0.9990710020065308,
      "learning_rate": 3.442657529409601e-05,
      "loss": 0.2624,
      "step": 849
    },
    {
      "epoch": 0.5331660655480633,
      "grad_norm": 1.1457931995391846,
      "learning_rate": 3.4412493823384354e-05,
      "loss": 0.2787,
      "step": 850
    },
    {
      "epoch": 0.5337933197428257,
      "grad_norm": 1.0615814924240112,
      "learning_rate": 3.439839747414795e-05,
      "loss": 0.2402,
      "step": 851
    },
    {
      "epoch": 0.5344205739375882,
      "grad_norm": 1.1380528211593628,
      "learning_rate": 3.438428626093893e-05,
      "loss": 0.2785,
      "step": 852
    },
    {
      "epoch": 0.5350478281323506,
      "grad_norm": 1.1444776058197021,
      "learning_rate": 3.437016019832485e-05,
      "loss": 0.3303,
      "step": 853
    },
    {
      "epoch": 0.535675082327113,
      "grad_norm": 1.1833291053771973,
      "learning_rate": 3.43560193008885e-05,
      "loss": 0.3053,
      "step": 854
    },
    {
      "epoch": 0.5363023365218755,
      "grad_norm": 1.057030200958252,
      "learning_rate": 3.4341863583228035e-05,
      "loss": 0.2852,
      "step": 855
    },
    {
      "epoch": 0.5369295907166379,
      "grad_norm": 1.0322626829147339,
      "learning_rate": 3.432769305995691e-05,
      "loss": 0.2795,
      "step": 856
    },
    {
      "epoch": 0.5375568449114003,
      "grad_norm": 1.1600500345230103,
      "learning_rate": 3.4313507745703855e-05,
      "loss": 0.3004,
      "step": 857
    },
    {
      "epoch": 0.5381840991061628,
      "grad_norm": 1.047410488128662,
      "learning_rate": 3.429930765511286e-05,
      "loss": 0.2798,
      "step": 858
    },
    {
      "epoch": 0.5388113533009252,
      "grad_norm": 1.1107338666915894,
      "learning_rate": 3.428509280284318e-05,
      "loss": 0.2828,
      "step": 859
    },
    {
      "epoch": 0.5394386074956876,
      "grad_norm": 1.0454515218734741,
      "learning_rate": 3.427086320356931e-05,
      "loss": 0.2958,
      "step": 860
    },
    {
      "epoch": 0.54006586169045,
      "grad_norm": 1.0651618242263794,
      "learning_rate": 3.4256618871980964e-05,
      "loss": 0.2995,
      "step": 861
    },
    {
      "epoch": 0.5406931158852125,
      "grad_norm": 1.0665225982666016,
      "learning_rate": 3.4242359822783075e-05,
      "loss": 0.2952,
      "step": 862
    },
    {
      "epoch": 0.5413203700799749,
      "grad_norm": 0.9790911078453064,
      "learning_rate": 3.422808607069575e-05,
      "loss": 0.299,
      "step": 863
    },
    {
      "epoch": 0.5419476242747373,
      "grad_norm": 0.9757672548294067,
      "learning_rate": 3.4213797630454286e-05,
      "loss": 0.2996,
      "step": 864
    },
    {
      "epoch": 0.5425748784694998,
      "grad_norm": 1.08747398853302,
      "learning_rate": 3.419949451680916e-05,
      "loss": 0.3108,
      "step": 865
    },
    {
      "epoch": 0.5432021326642622,
      "grad_norm": 1.0883815288543701,
      "learning_rate": 3.418517674452597e-05,
      "loss": 0.2838,
      "step": 866
    },
    {
      "epoch": 0.5438293868590246,
      "grad_norm": 0.935845136642456,
      "learning_rate": 3.4170844328385446e-05,
      "loss": 0.282,
      "step": 867
    },
    {
      "epoch": 0.544456641053787,
      "grad_norm": 1.0941154956817627,
      "learning_rate": 3.415649728318345e-05,
      "loss": 0.295,
      "step": 868
    },
    {
      "epoch": 0.5450838952485495,
      "grad_norm": 1.1535578966140747,
      "learning_rate": 3.4142135623730954e-05,
      "loss": 0.2653,
      "step": 869
    },
    {
      "epoch": 0.5457111494433119,
      "grad_norm": 0.768397867679596,
      "learning_rate": 3.4127759364854e-05,
      "loss": 0.2877,
      "step": 870
    },
    {
      "epoch": 0.5463384036380743,
      "grad_norm": 0.944739580154419,
      "learning_rate": 3.411336852139369e-05,
      "loss": 0.2566,
      "step": 871
    },
    {
      "epoch": 0.5469656578328368,
      "grad_norm": 1.490142822265625,
      "learning_rate": 3.4098963108206215e-05,
      "loss": 0.262,
      "step": 872
    },
    {
      "epoch": 0.5475929120275992,
      "grad_norm": 1.1872611045837402,
      "learning_rate": 3.408454314016279e-05,
      "loss": 0.3117,
      "step": 873
    },
    {
      "epoch": 0.5482201662223616,
      "grad_norm": 1.2164311408996582,
      "learning_rate": 3.407010863214965e-05,
      "loss": 0.3071,
      "step": 874
    },
    {
      "epoch": 0.5488474204171241,
      "grad_norm": 0.8649755716323853,
      "learning_rate": 3.4055659599068044e-05,
      "loss": 0.256,
      "step": 875
    },
    {
      "epoch": 0.5494746746118865,
      "grad_norm": 1.3158848285675049,
      "learning_rate": 3.4041196055834234e-05,
      "loss": 0.2494,
      "step": 876
    },
    {
      "epoch": 0.5501019288066489,
      "grad_norm": 1.0252341032028198,
      "learning_rate": 3.402671801737942e-05,
      "loss": 0.3016,
      "step": 877
    },
    {
      "epoch": 0.5507291830014113,
      "grad_norm": 1.2007358074188232,
      "learning_rate": 3.40122254986498e-05,
      "loss": 0.3121,
      "step": 878
    },
    {
      "epoch": 0.5513564371961738,
      "grad_norm": 0.943353533744812,
      "learning_rate": 3.399771851460653e-05,
      "loss": 0.2685,
      "step": 879
    },
    {
      "epoch": 0.5519836913909362,
      "grad_norm": 1.036067008972168,
      "learning_rate": 3.398319708022565e-05,
      "loss": 0.3,
      "step": 880
    },
    {
      "epoch": 0.5526109455856986,
      "grad_norm": 1.1122220754623413,
      "learning_rate": 3.396866121049816e-05,
      "loss": 0.277,
      "step": 881
    },
    {
      "epoch": 0.5532381997804611,
      "grad_norm": 0.9816188812255859,
      "learning_rate": 3.395411092042996e-05,
      "loss": 0.2745,
      "step": 882
    },
    {
      "epoch": 0.5538654539752235,
      "grad_norm": 1.0659526586532593,
      "learning_rate": 3.3939546225041816e-05,
      "loss": 0.3018,
      "step": 883
    },
    {
      "epoch": 0.5544927081699859,
      "grad_norm": 1.1326079368591309,
      "learning_rate": 3.392496713936938e-05,
      "loss": 0.312,
      "step": 884
    },
    {
      "epoch": 0.5551199623647484,
      "grad_norm": 1.0870763063430786,
      "learning_rate": 3.391037367846315e-05,
      "loss": 0.2741,
      "step": 885
    },
    {
      "epoch": 0.5557472165595108,
      "grad_norm": 1.2274421453475952,
      "learning_rate": 3.389576585738848e-05,
      "loss": 0.2911,
      "step": 886
    },
    {
      "epoch": 0.5563744707542732,
      "grad_norm": 1.053361177444458,
      "learning_rate": 3.3881143691225534e-05,
      "loss": 0.2496,
      "step": 887
    },
    {
      "epoch": 0.5570017249490355,
      "grad_norm": 0.9817186594009399,
      "learning_rate": 3.386650719506927e-05,
      "loss": 0.2351,
      "step": 888
    },
    {
      "epoch": 0.557628979143798,
      "grad_norm": 1.211648941040039,
      "learning_rate": 3.385185638402949e-05,
      "loss": 0.284,
      "step": 889
    },
    {
      "epoch": 0.5582562333385604,
      "grad_norm": 1.075768232345581,
      "learning_rate": 3.383719127323072e-05,
      "loss": 0.2909,
      "step": 890
    },
    {
      "epoch": 0.5588834875333228,
      "grad_norm": 1.1833033561706543,
      "learning_rate": 3.382251187781228e-05,
      "loss": 0.2985,
      "step": 891
    },
    {
      "epoch": 0.5595107417280853,
      "grad_norm": 1.2154804468154907,
      "learning_rate": 3.380781821292822e-05,
      "loss": 0.2503,
      "step": 892
    },
    {
      "epoch": 0.5601379959228477,
      "grad_norm": 0.6897361278533936,
      "learning_rate": 3.3793110293747334e-05,
      "loss": 0.2373,
      "step": 893
    },
    {
      "epoch": 0.5607652501176101,
      "grad_norm": 1.0964821577072144,
      "learning_rate": 3.377838813545313e-05,
      "loss": 0.2711,
      "step": 894
    },
    {
      "epoch": 0.5613925043123726,
      "grad_norm": 1.0501883029937744,
      "learning_rate": 3.3763651753243783e-05,
      "loss": 0.3124,
      "step": 895
    },
    {
      "epoch": 0.562019758507135,
      "grad_norm": 0.9332281947135925,
      "learning_rate": 3.3748901162332205e-05,
      "loss": 0.291,
      "step": 896
    },
    {
      "epoch": 0.5626470127018974,
      "grad_norm": 1.2695751190185547,
      "learning_rate": 3.373413637794595e-05,
      "loss": 0.295,
      "step": 897
    },
    {
      "epoch": 0.5632742668966598,
      "grad_norm": 1.011654257774353,
      "learning_rate": 3.3719357415327214e-05,
      "loss": 0.3049,
      "step": 898
    },
    {
      "epoch": 0.5639015210914223,
      "grad_norm": 1.0462952852249146,
      "learning_rate": 3.370456428973285e-05,
      "loss": 0.2726,
      "step": 899
    },
    {
      "epoch": 0.5645287752861847,
      "grad_norm": 1.046999454498291,
      "learning_rate": 3.368975701643431e-05,
      "loss": 0.2477,
      "step": 900
    },
    {
      "epoch": 0.5651560294809471,
      "grad_norm": 1.1205613613128662,
      "learning_rate": 3.367493561071767e-05,
      "loss": 0.2988,
      "step": 901
    },
    {
      "epoch": 0.5657832836757096,
      "grad_norm": 1.0621720552444458,
      "learning_rate": 3.366010008788359e-05,
      "loss": 0.3005,
      "step": 902
    },
    {
      "epoch": 0.566410537870472,
      "grad_norm": 1.545395016670227,
      "learning_rate": 3.36452504632473e-05,
      "loss": 0.2953,
      "step": 903
    },
    {
      "epoch": 0.5670377920652344,
      "grad_norm": 0.95975661277771,
      "learning_rate": 3.363038675213859e-05,
      "loss": 0.2769,
      "step": 904
    },
    {
      "epoch": 0.5676650462599968,
      "grad_norm": 1.022896647453308,
      "learning_rate": 3.36155089699018e-05,
      "loss": 0.2779,
      "step": 905
    },
    {
      "epoch": 0.5682923004547593,
      "grad_norm": 1.1125088930130005,
      "learning_rate": 3.360061713189578e-05,
      "loss": 0.2888,
      "step": 906
    },
    {
      "epoch": 0.5689195546495217,
      "grad_norm": 1.1785695552825928,
      "learning_rate": 3.3585711253493904e-05,
      "loss": 0.2538,
      "step": 907
    },
    {
      "epoch": 0.5695468088442841,
      "grad_norm": 1.1789590120315552,
      "learning_rate": 3.3570791350084026e-05,
      "loss": 0.2806,
      "step": 908
    },
    {
      "epoch": 0.5701740630390466,
      "grad_norm": 1.133196234703064,
      "learning_rate": 3.35558574370685e-05,
      "loss": 0.2784,
      "step": 909
    },
    {
      "epoch": 0.570801317233809,
      "grad_norm": 0.9845195412635803,
      "learning_rate": 3.354090952986412e-05,
      "loss": 0.2394,
      "step": 910
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.0305912494659424,
      "learning_rate": 3.352594764390214e-05,
      "loss": 0.2733,
      "step": 911
    },
    {
      "epoch": 0.5720558256233339,
      "grad_norm": 1.159439206123352,
      "learning_rate": 3.3510971794628244e-05,
      "loss": 0.2828,
      "step": 912
    },
    {
      "epoch": 0.5726830798180963,
      "grad_norm": 1.128682255744934,
      "learning_rate": 3.349598199750252e-05,
      "loss": 0.247,
      "step": 913
    },
    {
      "epoch": 0.5733103340128587,
      "grad_norm": 0.911435067653656,
      "learning_rate": 3.348097826799948e-05,
      "loss": 0.2775,
      "step": 914
    },
    {
      "epoch": 0.5739375882076211,
      "grad_norm": 1.0393047332763672,
      "learning_rate": 3.3465960621607995e-05,
      "loss": 0.2769,
      "step": 915
    },
    {
      "epoch": 0.5745648424023836,
      "grad_norm": 0.9499742984771729,
      "learning_rate": 3.3450929073831296e-05,
      "loss": 0.2639,
      "step": 916
    },
    {
      "epoch": 0.575192096597146,
      "grad_norm": 1.0762925148010254,
      "learning_rate": 3.3435883640186996e-05,
      "loss": 0.2847,
      "step": 917
    },
    {
      "epoch": 0.5758193507919084,
      "grad_norm": 1.2723565101623535,
      "learning_rate": 3.342082433620702e-05,
      "loss": 0.3105,
      "step": 918
    },
    {
      "epoch": 0.5764466049866709,
      "grad_norm": 1.0838451385498047,
      "learning_rate": 3.340575117743761e-05,
      "loss": 0.2527,
      "step": 919
    },
    {
      "epoch": 0.5770738591814333,
      "grad_norm": 0.9201546311378479,
      "learning_rate": 3.3390664179439336e-05,
      "loss": 0.2539,
      "step": 920
    },
    {
      "epoch": 0.5777011133761957,
      "grad_norm": 0.951526939868927,
      "learning_rate": 3.337556335778702e-05,
      "loss": 0.2623,
      "step": 921
    },
    {
      "epoch": 0.5783283675709582,
      "grad_norm": 0.9981276988983154,
      "learning_rate": 3.336044872806978e-05,
      "loss": 0.2454,
      "step": 922
    },
    {
      "epoch": 0.5789556217657206,
      "grad_norm": 0.9986087083816528,
      "learning_rate": 3.334532030589098e-05,
      "loss": 0.2502,
      "step": 923
    },
    {
      "epoch": 0.579582875960483,
      "grad_norm": 1.0848922729492188,
      "learning_rate": 3.333017810686823e-05,
      "loss": 0.2748,
      "step": 924
    },
    {
      "epoch": 0.5802101301552454,
      "grad_norm": 0.9143623113632202,
      "learning_rate": 3.331502214663333e-05,
      "loss": 0.2559,
      "step": 925
    },
    {
      "epoch": 0.5808373843500079,
      "grad_norm": 1.2553857564926147,
      "learning_rate": 3.329985244083234e-05,
      "loss": 0.2646,
      "step": 926
    },
    {
      "epoch": 0.5814646385447703,
      "grad_norm": 1.1474798917770386,
      "learning_rate": 3.328466900512547e-05,
      "loss": 0.2959,
      "step": 927
    },
    {
      "epoch": 0.5820918927395327,
      "grad_norm": 0.9720746278762817,
      "learning_rate": 3.3269471855187114e-05,
      "loss": 0.3125,
      "step": 928
    },
    {
      "epoch": 0.5827191469342952,
      "grad_norm": 0.8761600852012634,
      "learning_rate": 3.325426100670583e-05,
      "loss": 0.2655,
      "step": 929
    },
    {
      "epoch": 0.5833464011290576,
      "grad_norm": 1.0103437900543213,
      "learning_rate": 3.3239036475384295e-05,
      "loss": 0.2791,
      "step": 930
    },
    {
      "epoch": 0.58397365532382,
      "grad_norm": 1.0584696531295776,
      "learning_rate": 3.322379827693935e-05,
      "loss": 0.2663,
      "step": 931
    },
    {
      "epoch": 0.5846009095185825,
      "grad_norm": 1.164749264717102,
      "learning_rate": 3.3208546427101915e-05,
      "loss": 0.2835,
      "step": 932
    },
    {
      "epoch": 0.5852281637133449,
      "grad_norm": 1.2362803220748901,
      "learning_rate": 3.3193280941617015e-05,
      "loss": 0.2661,
      "step": 933
    },
    {
      "epoch": 0.5858554179081072,
      "grad_norm": 1.0678917169570923,
      "learning_rate": 3.317800183624374e-05,
      "loss": 0.2567,
      "step": 934
    },
    {
      "epoch": 0.5864826721028696,
      "grad_norm": 1.1115533113479614,
      "learning_rate": 3.316270912675526e-05,
      "loss": 0.2901,
      "step": 935
    },
    {
      "epoch": 0.5871099262976321,
      "grad_norm": 0.9827170372009277,
      "learning_rate": 3.3147402828938764e-05,
      "loss": 0.2706,
      "step": 936
    },
    {
      "epoch": 0.5877371804923945,
      "grad_norm": 1.029221534729004,
      "learning_rate": 3.313208295859549e-05,
      "loss": 0.2538,
      "step": 937
    },
    {
      "epoch": 0.5883644346871569,
      "grad_norm": 1.0566000938415527,
      "learning_rate": 3.311674953154068e-05,
      "loss": 0.2468,
      "step": 938
    },
    {
      "epoch": 0.5889916888819194,
      "grad_norm": 1.0722020864486694,
      "learning_rate": 3.310140256360357e-05,
      "loss": 0.2823,
      "step": 939
    },
    {
      "epoch": 0.5896189430766818,
      "grad_norm": 1.1388366222381592,
      "learning_rate": 3.308604207062738e-05,
      "loss": 0.2806,
      "step": 940
    },
    {
      "epoch": 0.5902461972714442,
      "grad_norm": 0.9928994178771973,
      "learning_rate": 3.307066806846927e-05,
      "loss": 0.2534,
      "step": 941
    },
    {
      "epoch": 0.5908734514662067,
      "grad_norm": 1.0917091369628906,
      "learning_rate": 3.305528057300039e-05,
      "loss": 0.3106,
      "step": 942
    },
    {
      "epoch": 0.5915007056609691,
      "grad_norm": 1.0011969804763794,
      "learning_rate": 3.3039879600105765e-05,
      "loss": 0.2558,
      "step": 943
    },
    {
      "epoch": 0.5921279598557315,
      "grad_norm": 1.084661602973938,
      "learning_rate": 3.302446516568438e-05,
      "loss": 0.2477,
      "step": 944
    },
    {
      "epoch": 0.5927552140504939,
      "grad_norm": 1.0037258863449097,
      "learning_rate": 3.300903728564909e-05,
      "loss": 0.2581,
      "step": 945
    },
    {
      "epoch": 0.5933824682452564,
      "grad_norm": 0.993988573551178,
      "learning_rate": 3.2993595975926644e-05,
      "loss": 0.2829,
      "step": 946
    },
    {
      "epoch": 0.5940097224400188,
      "grad_norm": 1.0701007843017578,
      "learning_rate": 3.297814125245765e-05,
      "loss": 0.258,
      "step": 947
    },
    {
      "epoch": 0.5946369766347812,
      "grad_norm": 1.08833909034729,
      "learning_rate": 3.296267313119656e-05,
      "loss": 0.3067,
      "step": 948
    },
    {
      "epoch": 0.5952642308295437,
      "grad_norm": 1.0432220697402954,
      "learning_rate": 3.294719162811166e-05,
      "loss": 0.2657,
      "step": 949
    },
    {
      "epoch": 0.5958914850243061,
      "grad_norm": 1.281551480293274,
      "learning_rate": 3.293169675918506e-05,
      "loss": 0.3095,
      "step": 950
    },
    {
      "epoch": 0.5965187392190685,
      "grad_norm": 0.9822801351547241,
      "learning_rate": 3.291618854041265e-05,
      "loss": 0.2619,
      "step": 951
    },
    {
      "epoch": 0.5971459934138309,
      "grad_norm": 1.0419542789459229,
      "learning_rate": 3.290066698780409e-05,
      "loss": 0.253,
      "step": 952
    },
    {
      "epoch": 0.5977732476085934,
      "grad_norm": 0.9738790392875671,
      "learning_rate": 3.2885132117382865e-05,
      "loss": 0.2536,
      "step": 953
    },
    {
      "epoch": 0.5984005018033558,
      "grad_norm": 0.9893375039100647,
      "learning_rate": 3.286958394518615e-05,
      "loss": 0.2432,
      "step": 954
    },
    {
      "epoch": 0.5990277559981182,
      "grad_norm": 1.023341178894043,
      "learning_rate": 3.2854022487264865e-05,
      "loss": 0.2955,
      "step": 955
    },
    {
      "epoch": 0.5996550101928807,
      "grad_norm": 1.0082974433898926,
      "learning_rate": 3.2838447759683646e-05,
      "loss": 0.2479,
      "step": 956
    },
    {
      "epoch": 0.6002822643876431,
      "grad_norm": 0.9385503530502319,
      "learning_rate": 3.282285977852086e-05,
      "loss": 0.235,
      "step": 957
    },
    {
      "epoch": 0.6009095185824055,
      "grad_norm": 0.960944652557373,
      "learning_rate": 3.28072585598685e-05,
      "loss": 0.2548,
      "step": 958
    },
    {
      "epoch": 0.601536772777168,
      "grad_norm": 1.061529278755188,
      "learning_rate": 3.2791644119832265e-05,
      "loss": 0.2594,
      "step": 959
    },
    {
      "epoch": 0.6021640269719304,
      "grad_norm": 0.9216713309288025,
      "learning_rate": 3.277601647453149e-05,
      "loss": 0.2867,
      "step": 960
    },
    {
      "epoch": 0.6027912811666928,
      "grad_norm": 1.1350970268249512,
      "learning_rate": 3.276037564009915e-05,
      "loss": 0.2707,
      "step": 961
    },
    {
      "epoch": 0.6034185353614552,
      "grad_norm": 1.0315324068069458,
      "learning_rate": 3.2744721632681816e-05,
      "loss": 0.2968,
      "step": 962
    },
    {
      "epoch": 0.6040457895562177,
      "grad_norm": 1.111251950263977,
      "learning_rate": 3.272905446843968e-05,
      "loss": 0.2876,
      "step": 963
    },
    {
      "epoch": 0.6046730437509801,
      "grad_norm": 0.9278132915496826,
      "learning_rate": 3.2713374163546504e-05,
      "loss": 0.2883,
      "step": 964
    },
    {
      "epoch": 0.6053002979457425,
      "grad_norm": 1.0058590173721313,
      "learning_rate": 3.269768073418962e-05,
      "loss": 0.2652,
      "step": 965
    },
    {
      "epoch": 0.605927552140505,
      "grad_norm": 1.0049445629119873,
      "learning_rate": 3.268197419656991e-05,
      "loss": 0.2581,
      "step": 966
    },
    {
      "epoch": 0.6065548063352674,
      "grad_norm": 0.8966968059539795,
      "learning_rate": 3.266625456690178e-05,
      "loss": 0.2511,
      "step": 967
    },
    {
      "epoch": 0.6071820605300298,
      "grad_norm": 0.9507226347923279,
      "learning_rate": 3.265052186141317e-05,
      "loss": 0.257,
      "step": 968
    },
    {
      "epoch": 0.6078093147247923,
      "grad_norm": 0.849480152130127,
      "learning_rate": 3.263477609634549e-05,
      "loss": 0.2645,
      "step": 969
    },
    {
      "epoch": 0.6084365689195547,
      "grad_norm": 1.1491882801055908,
      "learning_rate": 3.261901728795364e-05,
      "loss": 0.2859,
      "step": 970
    },
    {
      "epoch": 0.6090638231143171,
      "grad_norm": 1.0749951601028442,
      "learning_rate": 3.260324545250602e-05,
      "loss": 0.2804,
      "step": 971
    },
    {
      "epoch": 0.6096910773090795,
      "grad_norm": 0.9273468255996704,
      "learning_rate": 3.2587460606284435e-05,
      "loss": 0.2458,
      "step": 972
    },
    {
      "epoch": 0.610318331503842,
      "grad_norm": 1.2743276357650757,
      "learning_rate": 3.257166276558414e-05,
      "loss": 0.2651,
      "step": 973
    },
    {
      "epoch": 0.6109455856986044,
      "grad_norm": 0.9892858862876892,
      "learning_rate": 3.2555851946713794e-05,
      "loss": 0.2674,
      "step": 974
    },
    {
      "epoch": 0.6115728398933667,
      "grad_norm": 0.8826314210891724,
      "learning_rate": 3.254002816599548e-05,
      "loss": 0.2755,
      "step": 975
    },
    {
      "epoch": 0.6122000940881293,
      "grad_norm": 1.0605132579803467,
      "learning_rate": 3.2524191439764624e-05,
      "loss": 0.2944,
      "step": 976
    },
    {
      "epoch": 0.6128273482828916,
      "grad_norm": 0.9647715091705322,
      "learning_rate": 3.250834178437005e-05,
      "loss": 0.259,
      "step": 977
    },
    {
      "epoch": 0.613454602477654,
      "grad_norm": 0.9203553199768066,
      "learning_rate": 3.249247921617391e-05,
      "loss": 0.2446,
      "step": 978
    },
    {
      "epoch": 0.6140818566724165,
      "grad_norm": 0.9496483206748962,
      "learning_rate": 3.2476603751551695e-05,
      "loss": 0.2573,
      "step": 979
    },
    {
      "epoch": 0.6147091108671789,
      "grad_norm": 0.9901736974716187,
      "learning_rate": 3.24607154068922e-05,
      "loss": 0.2896,
      "step": 980
    },
    {
      "epoch": 0.6153363650619413,
      "grad_norm": 1.043483018875122,
      "learning_rate": 3.244481419859752e-05,
      "loss": 0.2976,
      "step": 981
    },
    {
      "epoch": 0.6159636192567037,
      "grad_norm": 0.9606313109397888,
      "learning_rate": 3.242890014308305e-05,
      "loss": 0.2781,
      "step": 982
    },
    {
      "epoch": 0.6165908734514662,
      "grad_norm": 1.1747736930847168,
      "learning_rate": 3.241297325677742e-05,
      "loss": 0.295,
      "step": 983
    },
    {
      "epoch": 0.6172181276462286,
      "grad_norm": 1.166608214378357,
      "learning_rate": 3.2397033556122496e-05,
      "loss": 0.2732,
      "step": 984
    },
    {
      "epoch": 0.617845381840991,
      "grad_norm": 0.9529145359992981,
      "learning_rate": 3.238108105757342e-05,
      "loss": 0.2793,
      "step": 985
    },
    {
      "epoch": 0.6184726360357535,
      "grad_norm": 0.9561068415641785,
      "learning_rate": 3.23651157775985e-05,
      "loss": 0.24,
      "step": 986
    },
    {
      "epoch": 0.6190998902305159,
      "grad_norm": 1.0726927518844604,
      "learning_rate": 3.234913773267927e-05,
      "loss": 0.2791,
      "step": 987
    },
    {
      "epoch": 0.6197271444252783,
      "grad_norm": 1.0708917379379272,
      "learning_rate": 3.233314693931042e-05,
      "loss": 0.2682,
      "step": 988
    },
    {
      "epoch": 0.6203543986200408,
      "grad_norm": 1.0202605724334717,
      "learning_rate": 3.2317143413999815e-05,
      "loss": 0.2736,
      "step": 989
    },
    {
      "epoch": 0.6209816528148032,
      "grad_norm": 1.00557541847229,
      "learning_rate": 3.2301127173268445e-05,
      "loss": 0.2661,
      "step": 990
    },
    {
      "epoch": 0.6216089070095656,
      "grad_norm": 1.2042478322982788,
      "learning_rate": 3.2285098233650455e-05,
      "loss": 0.3188,
      "step": 991
    },
    {
      "epoch": 0.622236161204328,
      "grad_norm": 1.0422487258911133,
      "learning_rate": 3.2269056611693075e-05,
      "loss": 0.2754,
      "step": 992
    },
    {
      "epoch": 0.6228634153990905,
      "grad_norm": 0.9788997173309326,
      "learning_rate": 3.225300232395664e-05,
      "loss": 0.2528,
      "step": 993
    },
    {
      "epoch": 0.6234906695938529,
      "grad_norm": 1.1364682912826538,
      "learning_rate": 3.223693538701454e-05,
      "loss": 0.2894,
      "step": 994
    },
    {
      "epoch": 0.6241179237886153,
      "grad_norm": 1.0093938112258911,
      "learning_rate": 3.222085581745326e-05,
      "loss": 0.2838,
      "step": 995
    },
    {
      "epoch": 0.6247451779833778,
      "grad_norm": 1.216389775276184,
      "learning_rate": 3.220476363187231e-05,
      "loss": 0.272,
      "step": 996
    },
    {
      "epoch": 0.6253724321781402,
      "grad_norm": 1.0009663105010986,
      "learning_rate": 3.218865884688419e-05,
      "loss": 0.3109,
      "step": 997
    },
    {
      "epoch": 0.6259996863729026,
      "grad_norm": 0.9530799984931946,
      "learning_rate": 3.2172541479114466e-05,
      "loss": 0.2968,
      "step": 998
    },
    {
      "epoch": 0.626626940567665,
      "grad_norm": 0.8794124722480774,
      "learning_rate": 3.215641154520165e-05,
      "loss": 0.2284,
      "step": 999
    },
    {
      "epoch": 0.6272541947624275,
      "grad_norm": 1.0209484100341797,
      "learning_rate": 3.214026906179725e-05,
      "loss": 0.2554,
      "step": 1000
    },
    {
      "epoch": 0.6278814489571899,
      "grad_norm": 0.8757167458534241,
      "learning_rate": 3.21241140455657e-05,
      "loss": 0.2195,
      "step": 1001
    },
    {
      "epoch": 0.6285087031519523,
      "grad_norm": 0.9012476801872253,
      "learning_rate": 3.210794651318442e-05,
      "loss": 0.243,
      "step": 1002
    },
    {
      "epoch": 0.6291359573467148,
      "grad_norm": 0.9644747972488403,
      "learning_rate": 3.20917664813437e-05,
      "loss": 0.2588,
      "step": 1003
    },
    {
      "epoch": 0.6297632115414772,
      "grad_norm": 1.0049502849578857,
      "learning_rate": 3.207557396674677e-05,
      "loss": 0.2262,
      "step": 1004
    },
    {
      "epoch": 0.6303904657362396,
      "grad_norm": 0.924791693687439,
      "learning_rate": 3.205936898610973e-05,
      "loss": 0.2547,
      "step": 1005
    },
    {
      "epoch": 0.6310177199310021,
      "grad_norm": 1.0177044868469238,
      "learning_rate": 3.204315155616154e-05,
      "loss": 0.2557,
      "step": 1006
    },
    {
      "epoch": 0.6316449741257645,
      "grad_norm": 0.9623924493789673,
      "learning_rate": 3.202692169364404e-05,
      "loss": 0.2381,
      "step": 1007
    },
    {
      "epoch": 0.6322722283205269,
      "grad_norm": 1.1295835971832275,
      "learning_rate": 3.2010679415311886e-05,
      "loss": 0.2704,
      "step": 1008
    },
    {
      "epoch": 0.6328994825152893,
      "grad_norm": 1.0597741603851318,
      "learning_rate": 3.199442473793254e-05,
      "loss": 0.2674,
      "step": 1009
    },
    {
      "epoch": 0.6335267367100518,
      "grad_norm": 1.0375666618347168,
      "learning_rate": 3.197815767828629e-05,
      "loss": 0.2909,
      "step": 1010
    },
    {
      "epoch": 0.6341539909048142,
      "grad_norm": 0.9671528339385986,
      "learning_rate": 3.19618782531662e-05,
      "loss": 0.2621,
      "step": 1011
    },
    {
      "epoch": 0.6347812450995766,
      "grad_norm": 1.0041483640670776,
      "learning_rate": 3.1945586479378076e-05,
      "loss": 0.2358,
      "step": 1012
    },
    {
      "epoch": 0.6354084992943391,
      "grad_norm": 1.0127429962158203,
      "learning_rate": 3.1929282373740505e-05,
      "loss": 0.2986,
      "step": 1013
    },
    {
      "epoch": 0.6360357534891015,
      "grad_norm": 1.0777404308319092,
      "learning_rate": 3.191296595308478e-05,
      "loss": 0.2539,
      "step": 1014
    },
    {
      "epoch": 0.6366630076838639,
      "grad_norm": 0.8917059898376465,
      "learning_rate": 3.189663723425492e-05,
      "loss": 0.2578,
      "step": 1015
    },
    {
      "epoch": 0.6372902618786264,
      "grad_norm": 0.9886910915374756,
      "learning_rate": 3.188029623410764e-05,
      "loss": 0.2777,
      "step": 1016
    },
    {
      "epoch": 0.6379175160733888,
      "grad_norm": 0.9957547187805176,
      "learning_rate": 3.186394296951232e-05,
      "loss": 0.2935,
      "step": 1017
    },
    {
      "epoch": 0.6385447702681512,
      "grad_norm": 1.0073680877685547,
      "learning_rate": 3.184757745735102e-05,
      "loss": 0.2772,
      "step": 1018
    },
    {
      "epoch": 0.6391720244629135,
      "grad_norm": 1.0120540857315063,
      "learning_rate": 3.1831199714518436e-05,
      "loss": 0.2751,
      "step": 1019
    },
    {
      "epoch": 0.639799278657676,
      "grad_norm": 1.0629538297653198,
      "learning_rate": 3.181480975792187e-05,
      "loss": 0.2723,
      "step": 1020
    },
    {
      "epoch": 0.6404265328524384,
      "grad_norm": 1.1036429405212402,
      "learning_rate": 3.179840760448127e-05,
      "loss": 0.2914,
      "step": 1021
    },
    {
      "epoch": 0.6410537870472008,
      "grad_norm": 0.9888352751731873,
      "learning_rate": 3.178199327112916e-05,
      "loss": 0.237,
      "step": 1022
    },
    {
      "epoch": 0.6416810412419633,
      "grad_norm": 1.2374451160430908,
      "learning_rate": 3.1765566774810615e-05,
      "loss": 0.2701,
      "step": 1023
    },
    {
      "epoch": 0.6423082954367257,
      "grad_norm": 1.161687970161438,
      "learning_rate": 3.17491281324833e-05,
      "loss": 0.2812,
      "step": 1024
    },
    {
      "epoch": 0.6429355496314881,
      "grad_norm": 1.0261205434799194,
      "learning_rate": 3.17326773611174e-05,
      "loss": 0.2707,
      "step": 1025
    },
    {
      "epoch": 0.6435628038262506,
      "grad_norm": 1.058673620223999,
      "learning_rate": 3.171621447769562e-05,
      "loss": 0.2663,
      "step": 1026
    },
    {
      "epoch": 0.644190058021013,
      "grad_norm": 0.968376100063324,
      "learning_rate": 3.169973949921318e-05,
      "loss": 0.2366,
      "step": 1027
    },
    {
      "epoch": 0.6448173122157754,
      "grad_norm": 0.8703855872154236,
      "learning_rate": 3.168325244267778e-05,
      "loss": 0.2519,
      "step": 1028
    },
    {
      "epoch": 0.6454445664105378,
      "grad_norm": 1.0061933994293213,
      "learning_rate": 3.166675332510959e-05,
      "loss": 0.2435,
      "step": 1029
    },
    {
      "epoch": 0.6460718206053003,
      "grad_norm": 0.9254626631736755,
      "learning_rate": 3.165024216354123e-05,
      "loss": 0.2422,
      "step": 1030
    },
    {
      "epoch": 0.6466990748000627,
      "grad_norm": 1.0174202919006348,
      "learning_rate": 3.163371897501775e-05,
      "loss": 0.2597,
      "step": 1031
    },
    {
      "epoch": 0.6473263289948251,
      "grad_norm": 1.0172829627990723,
      "learning_rate": 3.161718377659663e-05,
      "loss": 0.2645,
      "step": 1032
    },
    {
      "epoch": 0.6479535831895876,
      "grad_norm": 1.0958163738250732,
      "learning_rate": 3.160063658534771e-05,
      "loss": 0.2797,
      "step": 1033
    },
    {
      "epoch": 0.64858083738435,
      "grad_norm": 0.9608843922615051,
      "learning_rate": 3.1584077418353264e-05,
      "loss": 0.253,
      "step": 1034
    },
    {
      "epoch": 0.6492080915791124,
      "grad_norm": 1.0445313453674316,
      "learning_rate": 3.1567506292707904e-05,
      "loss": 0.3066,
      "step": 1035
    },
    {
      "epoch": 0.6498353457738749,
      "grad_norm": 0.8894146680831909,
      "learning_rate": 3.1550923225518564e-05,
      "loss": 0.2694,
      "step": 1036
    },
    {
      "epoch": 0.6504625999686373,
      "grad_norm": 1.1175638437271118,
      "learning_rate": 3.1534328233904554e-05,
      "loss": 0.2864,
      "step": 1037
    },
    {
      "epoch": 0.6510898541633997,
      "grad_norm": 0.9953340291976929,
      "learning_rate": 3.151772133499744e-05,
      "loss": 0.2807,
      "step": 1038
    },
    {
      "epoch": 0.6517171083581621,
      "grad_norm": 0.9881662130355835,
      "learning_rate": 3.1501102545941135e-05,
      "loss": 0.2885,
      "step": 1039
    },
    {
      "epoch": 0.6523443625529246,
      "grad_norm": 1.1190524101257324,
      "learning_rate": 3.148447188389179e-05,
      "loss": 0.2456,
      "step": 1040
    },
    {
      "epoch": 0.652971616747687,
      "grad_norm": 1.0610312223434448,
      "learning_rate": 3.146782936601782e-05,
      "loss": 0.2853,
      "step": 1041
    },
    {
      "epoch": 0.6535988709424494,
      "grad_norm": 1.0105746984481812,
      "learning_rate": 3.145117500949988e-05,
      "loss": 0.2751,
      "step": 1042
    },
    {
      "epoch": 0.6542261251372119,
      "grad_norm": 1.307153344154358,
      "learning_rate": 3.143450883153086e-05,
      "loss": 0.2934,
      "step": 1043
    },
    {
      "epoch": 0.6548533793319743,
      "grad_norm": 1.1009116172790527,
      "learning_rate": 3.1417830849315835e-05,
      "loss": 0.2827,
      "step": 1044
    },
    {
      "epoch": 0.6554806335267367,
      "grad_norm": 0.9631924629211426,
      "learning_rate": 3.140114108007206e-05,
      "loss": 0.284,
      "step": 1045
    },
    {
      "epoch": 0.6561078877214991,
      "grad_norm": 1.0630031824111938,
      "learning_rate": 3.1384439541029004e-05,
      "loss": 0.2785,
      "step": 1046
    },
    {
      "epoch": 0.6567351419162616,
      "grad_norm": 0.908596396446228,
      "learning_rate": 3.1367726249428224e-05,
      "loss": 0.2668,
      "step": 1047
    },
    {
      "epoch": 0.657362396111024,
      "grad_norm": 0.9869924783706665,
      "learning_rate": 3.135100122252345e-05,
      "loss": 0.2626,
      "step": 1048
    },
    {
      "epoch": 0.6579896503057864,
      "grad_norm": 0.807806670665741,
      "learning_rate": 3.133426447758052e-05,
      "loss": 0.2776,
      "step": 1049
    },
    {
      "epoch": 0.6586169045005489,
      "grad_norm": 0.930079996585846,
      "learning_rate": 3.131751603187734e-05,
      "loss": 0.2356,
      "step": 1050
    },
    {
      "epoch": 0.6592441586953113,
      "grad_norm": 0.9175042510032654,
      "learning_rate": 3.130075590270396e-05,
      "loss": 0.2663,
      "step": 1051
    },
    {
      "epoch": 0.6598714128900737,
      "grad_norm": 0.9869060516357422,
      "learning_rate": 3.1283984107362426e-05,
      "loss": 0.254,
      "step": 1052
    },
    {
      "epoch": 0.6604986670848362,
      "grad_norm": 1.0021368265151978,
      "learning_rate": 3.1267200663166866e-05,
      "loss": 0.2679,
      "step": 1053
    },
    {
      "epoch": 0.6611259212795986,
      "grad_norm": 0.9547368884086609,
      "learning_rate": 3.12504055874434e-05,
      "loss": 0.2427,
      "step": 1054
    },
    {
      "epoch": 0.661753175474361,
      "grad_norm": 0.8755058646202087,
      "learning_rate": 3.12335988975302e-05,
      "loss": 0.2045,
      "step": 1055
    },
    {
      "epoch": 0.6623804296691234,
      "grad_norm": 1.010550618171692,
      "learning_rate": 3.121678061077739e-05,
      "loss": 0.2848,
      "step": 1056
    },
    {
      "epoch": 0.6630076838638859,
      "grad_norm": 0.9710196256637573,
      "learning_rate": 3.119995074454708e-05,
      "loss": 0.2818,
      "step": 1057
    },
    {
      "epoch": 0.6636349380586483,
      "grad_norm": 0.9580225944519043,
      "learning_rate": 3.118310931621333e-05,
      "loss": 0.2438,
      "step": 1058
    },
    {
      "epoch": 0.6642621922534107,
      "grad_norm": 0.9465218186378479,
      "learning_rate": 3.116625634316216e-05,
      "loss": 0.2275,
      "step": 1059
    },
    {
      "epoch": 0.6648894464481732,
      "grad_norm": 1.0197880268096924,
      "learning_rate": 3.114939184279146e-05,
      "loss": 0.2715,
      "step": 1060
    },
    {
      "epoch": 0.6655167006429356,
      "grad_norm": 1.1116091012954712,
      "learning_rate": 3.1132515832511064e-05,
      "loss": 0.2522,
      "step": 1061
    },
    {
      "epoch": 0.666143954837698,
      "grad_norm": 0.9236192107200623,
      "learning_rate": 3.111562832974266e-05,
      "loss": 0.266,
      "step": 1062
    },
    {
      "epoch": 0.6667712090324605,
      "grad_norm": 0.8119783401489258,
      "learning_rate": 3.109872935191983e-05,
      "loss": 0.2518,
      "step": 1063
    },
    {
      "epoch": 0.6673984632272228,
      "grad_norm": 1.0820034742355347,
      "learning_rate": 3.108181891648796e-05,
      "loss": 0.2845,
      "step": 1064
    },
    {
      "epoch": 0.6680257174219852,
      "grad_norm": 1.0736851692199707,
      "learning_rate": 3.106489704090431e-05,
      "loss": 0.2814,
      "step": 1065
    },
    {
      "epoch": 0.6686529716167476,
      "grad_norm": 1.084129810333252,
      "learning_rate": 3.104796374263791e-05,
      "loss": 0.2776,
      "step": 1066
    },
    {
      "epoch": 0.6692802258115101,
      "grad_norm": 1.1576014757156372,
      "learning_rate": 3.1031019039169616e-05,
      "loss": 0.3009,
      "step": 1067
    },
    {
      "epoch": 0.6699074800062725,
      "grad_norm": 0.9448037147521973,
      "learning_rate": 3.101406294799203e-05,
      "loss": 0.2486,
      "step": 1068
    },
    {
      "epoch": 0.6705347342010349,
      "grad_norm": 0.9121865034103394,
      "learning_rate": 3.0997095486609527e-05,
      "loss": 0.2472,
      "step": 1069
    },
    {
      "epoch": 0.6711619883957974,
      "grad_norm": 0.9719688892364502,
      "learning_rate": 3.098011667253821e-05,
      "loss": 0.2833,
      "step": 1070
    },
    {
      "epoch": 0.6717892425905598,
      "grad_norm": 1.2267473936080933,
      "learning_rate": 3.096312652330592e-05,
      "loss": 0.2813,
      "step": 1071
    },
    {
      "epoch": 0.6724164967853222,
      "grad_norm": 1.098172903060913,
      "learning_rate": 3.0946125056452164e-05,
      "loss": 0.2527,
      "step": 1072
    },
    {
      "epoch": 0.6730437509800847,
      "grad_norm": 1.091026782989502,
      "learning_rate": 3.092911228952817e-05,
      "loss": 0.2897,
      "step": 1073
    },
    {
      "epoch": 0.6736710051748471,
      "grad_norm": 1.0833760499954224,
      "learning_rate": 3.0912088240096814e-05,
      "loss": 0.2932,
      "step": 1074
    },
    {
      "epoch": 0.6742982593696095,
      "grad_norm": 0.9569733142852783,
      "learning_rate": 3.0895052925732625e-05,
      "loss": 0.238,
      "step": 1075
    },
    {
      "epoch": 0.6749255135643719,
      "grad_norm": 0.8709613084793091,
      "learning_rate": 3.087800636402174e-05,
      "loss": 0.2215,
      "step": 1076
    },
    {
      "epoch": 0.6755527677591344,
      "grad_norm": 1.1714290380477905,
      "learning_rate": 3.0860948572561956e-05,
      "loss": 0.265,
      "step": 1077
    },
    {
      "epoch": 0.6761800219538968,
      "grad_norm": 1.476657509803772,
      "learning_rate": 3.08438795689626e-05,
      "loss": 0.2577,
      "step": 1078
    },
    {
      "epoch": 0.6768072761486592,
      "grad_norm": 1.0154451131820679,
      "learning_rate": 3.082679937084462e-05,
      "loss": 0.2584,
      "step": 1079
    },
    {
      "epoch": 0.6774345303434217,
      "grad_norm": 1.0655109882354736,
      "learning_rate": 3.080970799584051e-05,
      "loss": 0.2903,
      "step": 1080
    },
    {
      "epoch": 0.6780617845381841,
      "grad_norm": 0.8489439487457275,
      "learning_rate": 3.07926054615943e-05,
      "loss": 0.2663,
      "step": 1081
    },
    {
      "epoch": 0.6786890387329465,
      "grad_norm": 1.0146068334579468,
      "learning_rate": 3.0775491785761524e-05,
      "loss": 0.2731,
      "step": 1082
    },
    {
      "epoch": 0.679316292927709,
      "grad_norm": 1.1330286264419556,
      "learning_rate": 3.075836698600924e-05,
      "loss": 0.2706,
      "step": 1083
    },
    {
      "epoch": 0.6799435471224714,
      "grad_norm": 0.9808182716369629,
      "learning_rate": 3.074123108001597e-05,
      "loss": 0.2444,
      "step": 1084
    },
    {
      "epoch": 0.6805708013172338,
      "grad_norm": 0.9529983401298523,
      "learning_rate": 3.072408408547173e-05,
      "loss": 0.2646,
      "step": 1085
    },
    {
      "epoch": 0.6811980555119962,
      "grad_norm": 1.1290792226791382,
      "learning_rate": 3.070692602007797e-05,
      "loss": 0.2799,
      "step": 1086
    },
    {
      "epoch": 0.6818253097067587,
      "grad_norm": 0.9760487675666809,
      "learning_rate": 3.0689756901547546e-05,
      "loss": 0.2323,
      "step": 1087
    },
    {
      "epoch": 0.6824525639015211,
      "grad_norm": 1.0980072021484375,
      "learning_rate": 3.067257674760475e-05,
      "loss": 0.2705,
      "step": 1088
    },
    {
      "epoch": 0.6830798180962835,
      "grad_norm": 0.9709072709083557,
      "learning_rate": 3.065538557598527e-05,
      "loss": 0.2588,
      "step": 1089
    },
    {
      "epoch": 0.683707072291046,
      "grad_norm": 0.9089717864990234,
      "learning_rate": 3.063818340443614e-05,
      "loss": 0.199,
      "step": 1090
    },
    {
      "epoch": 0.6843343264858084,
      "grad_norm": 0.8791358470916748,
      "learning_rate": 3.0620970250715784e-05,
      "loss": 0.2476,
      "step": 1091
    },
    {
      "epoch": 0.6849615806805708,
      "grad_norm": 1.0623133182525635,
      "learning_rate": 3.060374613259394e-05,
      "loss": 0.2524,
      "step": 1092
    },
    {
      "epoch": 0.6855888348753332,
      "grad_norm": 1.2946134805679321,
      "learning_rate": 3.058651106785167e-05,
      "loss": 0.3077,
      "step": 1093
    },
    {
      "epoch": 0.6862160890700957,
      "grad_norm": 1.0471159219741821,
      "learning_rate": 3.056926507428134e-05,
      "loss": 0.2481,
      "step": 1094
    },
    {
      "epoch": 0.6868433432648581,
      "grad_norm": 0.9360001683235168,
      "learning_rate": 3.0552008169686595e-05,
      "loss": 0.2384,
      "step": 1095
    },
    {
      "epoch": 0.6874705974596205,
      "grad_norm": 1.0947415828704834,
      "learning_rate": 3.0534740371882335e-05,
      "loss": 0.2579,
      "step": 1096
    },
    {
      "epoch": 0.688097851654383,
      "grad_norm": 1.0543622970581055,
      "learning_rate": 3.0517461698694737e-05,
      "loss": 0.2634,
      "step": 1097
    },
    {
      "epoch": 0.6887251058491454,
      "grad_norm": 0.9454314708709717,
      "learning_rate": 3.0500172167961168e-05,
      "loss": 0.2363,
      "step": 1098
    },
    {
      "epoch": 0.6893523600439078,
      "grad_norm": 0.9035005569458008,
      "learning_rate": 3.048287179753023e-05,
      "loss": 0.2573,
      "step": 1099
    },
    {
      "epoch": 0.6899796142386703,
      "grad_norm": 0.9288584589958191,
      "learning_rate": 3.046556060526169e-05,
      "loss": 0.2871,
      "step": 1100
    },
    {
      "epoch": 0.6906068684334327,
      "grad_norm": 0.9902932047843933,
      "learning_rate": 3.0448238609026516e-05,
      "loss": 0.2828,
      "step": 1101
    },
    {
      "epoch": 0.6912341226281951,
      "grad_norm": 0.7389153838157654,
      "learning_rate": 3.0430905826706807e-05,
      "loss": 0.222,
      "step": 1102
    },
    {
      "epoch": 0.6918613768229575,
      "grad_norm": 0.9360471963882446,
      "learning_rate": 3.0413562276195803e-05,
      "loss": 0.2641,
      "step": 1103
    },
    {
      "epoch": 0.69248863101772,
      "grad_norm": 0.9741581082344055,
      "learning_rate": 3.0396207975397868e-05,
      "loss": 0.2155,
      "step": 1104
    },
    {
      "epoch": 0.6931158852124824,
      "grad_norm": 0.8907262086868286,
      "learning_rate": 3.037884294222845e-05,
      "loss": 0.2204,
      "step": 1105
    },
    {
      "epoch": 0.6937431394072447,
      "grad_norm": 0.9536752700805664,
      "learning_rate": 3.0361467194614084e-05,
      "loss": 0.2402,
      "step": 1106
    },
    {
      "epoch": 0.6943703936020073,
      "grad_norm": 0.9764878153800964,
      "learning_rate": 3.034408075049238e-05,
      "loss": 0.2383,
      "step": 1107
    },
    {
      "epoch": 0.6949976477967696,
      "grad_norm": 1.0302314758300781,
      "learning_rate": 3.0326683627811966e-05,
      "loss": 0.2781,
      "step": 1108
    },
    {
      "epoch": 0.695624901991532,
      "grad_norm": 0.9398666620254517,
      "learning_rate": 3.03092758445325e-05,
      "loss": 0.2886,
      "step": 1109
    },
    {
      "epoch": 0.6962521561862945,
      "grad_norm": 1.0185917615890503,
      "learning_rate": 3.0291857418624648e-05,
      "loss": 0.3049,
      "step": 1110
    },
    {
      "epoch": 0.6968794103810569,
      "grad_norm": 1.0808707475662231,
      "learning_rate": 3.0274428368070076e-05,
      "loss": 0.2757,
      "step": 1111
    },
    {
      "epoch": 0.6975066645758193,
      "grad_norm": 0.9350001811981201,
      "learning_rate": 3.0256988710861395e-05,
      "loss": 0.2507,
      "step": 1112
    },
    {
      "epoch": 0.6981339187705817,
      "grad_norm": 0.9316309690475464,
      "learning_rate": 3.023953846500219e-05,
      "loss": 0.2604,
      "step": 1113
    },
    {
      "epoch": 0.6987611729653442,
      "grad_norm": 1.0132993459701538,
      "learning_rate": 3.0222077648506958e-05,
      "loss": 0.2496,
      "step": 1114
    },
    {
      "epoch": 0.6993884271601066,
      "grad_norm": 1.0161550045013428,
      "learning_rate": 3.0204606279401107e-05,
      "loss": 0.2851,
      "step": 1115
    },
    {
      "epoch": 0.700015681354869,
      "grad_norm": 1.0576071739196777,
      "learning_rate": 3.0187124375720952e-05,
      "loss": 0.2286,
      "step": 1116
    },
    {
      "epoch": 0.7006429355496315,
      "grad_norm": 1.0544531345367432,
      "learning_rate": 3.016963195551368e-05,
      "loss": 0.3107,
      "step": 1117
    },
    {
      "epoch": 0.7012701897443939,
      "grad_norm": 0.9492296576499939,
      "learning_rate": 3.0152129036837336e-05,
      "loss": 0.2332,
      "step": 1118
    },
    {
      "epoch": 0.7018974439391563,
      "grad_norm": 0.9895444512367249,
      "learning_rate": 3.0134615637760785e-05,
      "loss": 0.2095,
      "step": 1119
    },
    {
      "epoch": 0.7025246981339188,
      "grad_norm": 0.8719656467437744,
      "learning_rate": 3.011709177636374e-05,
      "loss": 0.2219,
      "step": 1120
    },
    {
      "epoch": 0.7031519523286812,
      "grad_norm": 0.9544978141784668,
      "learning_rate": 3.0099557470736695e-05,
      "loss": 0.2206,
      "step": 1121
    },
    {
      "epoch": 0.7037792065234436,
      "grad_norm": 1.007846474647522,
      "learning_rate": 3.0082012738980926e-05,
      "loss": 0.2551,
      "step": 1122
    },
    {
      "epoch": 0.704406460718206,
      "grad_norm": 0.8393685817718506,
      "learning_rate": 3.0064457599208487e-05,
      "loss": 0.2535,
      "step": 1123
    },
    {
      "epoch": 0.7050337149129685,
      "grad_norm": 0.8983904719352722,
      "learning_rate": 3.0046892069542154e-05,
      "loss": 0.2584,
      "step": 1124
    },
    {
      "epoch": 0.7056609691077309,
      "grad_norm": 0.8997087478637695,
      "learning_rate": 3.0029316168115445e-05,
      "loss": 0.2537,
      "step": 1125
    },
    {
      "epoch": 0.7062882233024933,
      "grad_norm": 1.1286370754241943,
      "learning_rate": 3.00117299130726e-05,
      "loss": 0.2693,
      "step": 1126
    },
    {
      "epoch": 0.7069154774972558,
      "grad_norm": 0.8487894535064697,
      "learning_rate": 2.9994133322568505e-05,
      "loss": 0.2597,
      "step": 1127
    },
    {
      "epoch": 0.7075427316920182,
      "grad_norm": 0.999539315700531,
      "learning_rate": 2.9976526414768745e-05,
      "loss": 0.2686,
      "step": 1128
    },
    {
      "epoch": 0.7081699858867806,
      "grad_norm": 0.8149359226226807,
      "learning_rate": 2.9958909207849563e-05,
      "loss": 0.2248,
      "step": 1129
    },
    {
      "epoch": 0.7087972400815431,
      "grad_norm": 1.0735280513763428,
      "learning_rate": 2.994128171999781e-05,
      "loss": 0.2786,
      "step": 1130
    },
    {
      "epoch": 0.7094244942763055,
      "grad_norm": 0.9906747341156006,
      "learning_rate": 2.9923643969410973e-05,
      "loss": 0.2675,
      "step": 1131
    },
    {
      "epoch": 0.7100517484710679,
      "grad_norm": 0.9118907451629639,
      "learning_rate": 2.9905995974297107e-05,
      "loss": 0.2361,
      "step": 1132
    },
    {
      "epoch": 0.7106790026658303,
      "grad_norm": 1.1021521091461182,
      "learning_rate": 2.9888337752874875e-05,
      "loss": 0.2695,
      "step": 1133
    },
    {
      "epoch": 0.7113062568605928,
      "grad_norm": 0.9408007264137268,
      "learning_rate": 2.9870669323373473e-05,
      "loss": 0.2801,
      "step": 1134
    },
    {
      "epoch": 0.7119335110553552,
      "grad_norm": 1.0125051736831665,
      "learning_rate": 2.9852990704032643e-05,
      "loss": 0.2851,
      "step": 1135
    },
    {
      "epoch": 0.7125607652501176,
      "grad_norm": 0.8995490074157715,
      "learning_rate": 2.9835301913102644e-05,
      "loss": 0.2445,
      "step": 1136
    },
    {
      "epoch": 0.7131880194448801,
      "grad_norm": 1.1181654930114746,
      "learning_rate": 2.9817602968844236e-05,
      "loss": 0.2898,
      "step": 1137
    },
    {
      "epoch": 0.7138152736396425,
      "grad_norm": 0.8702493906021118,
      "learning_rate": 2.979989388952867e-05,
      "loss": 0.2305,
      "step": 1138
    },
    {
      "epoch": 0.7144425278344049,
      "grad_norm": 0.9106255769729614,
      "learning_rate": 2.9782174693437654e-05,
      "loss": 0.2199,
      "step": 1139
    },
    {
      "epoch": 0.7150697820291673,
      "grad_norm": 1.2732893228530884,
      "learning_rate": 2.9764445398863328e-05,
      "loss": 0.3188,
      "step": 1140
    },
    {
      "epoch": 0.7156970362239298,
      "grad_norm": 0.9374165534973145,
      "learning_rate": 2.974670602410827e-05,
      "loss": 0.2242,
      "step": 1141
    },
    {
      "epoch": 0.7163242904186922,
      "grad_norm": 0.8125789761543274,
      "learning_rate": 2.972895658748546e-05,
      "loss": 0.2217,
      "step": 1142
    },
    {
      "epoch": 0.7169515446134546,
      "grad_norm": 0.9294368624687195,
      "learning_rate": 2.971119710731827e-05,
      "loss": 0.2253,
      "step": 1143
    },
    {
      "epoch": 0.7175787988082171,
      "grad_norm": 1.0166178941726685,
      "learning_rate": 2.9693427601940433e-05,
      "loss": 0.2191,
      "step": 1144
    },
    {
      "epoch": 0.7182060530029795,
      "grad_norm": 1.0041168928146362,
      "learning_rate": 2.9675648089696038e-05,
      "loss": 0.2146,
      "step": 1145
    },
    {
      "epoch": 0.7188333071977419,
      "grad_norm": 1.0499262809753418,
      "learning_rate": 2.9657858588939502e-05,
      "loss": 0.2383,
      "step": 1146
    },
    {
      "epoch": 0.7194605613925044,
      "grad_norm": 1.1398004293441772,
      "learning_rate": 2.9640059118035555e-05,
      "loss": 0.2968,
      "step": 1147
    },
    {
      "epoch": 0.7200878155872668,
      "grad_norm": 0.99320387840271,
      "learning_rate": 2.9622249695359206e-05,
      "loss": 0.2769,
      "step": 1148
    },
    {
      "epoch": 0.7207150697820291,
      "grad_norm": 1.0669426918029785,
      "learning_rate": 2.9604430339295763e-05,
      "loss": 0.2642,
      "step": 1149
    },
    {
      "epoch": 0.7213423239767915,
      "grad_norm": 1.0032626390457153,
      "learning_rate": 2.958660106824076e-05,
      "loss": 0.2761,
      "step": 1150
    },
    {
      "epoch": 0.721969578171554,
      "grad_norm": 0.8565042018890381,
      "learning_rate": 2.9568761900599992e-05,
      "loss": 0.2337,
      "step": 1151
    },
    {
      "epoch": 0.7225968323663164,
      "grad_norm": 0.980816125869751,
      "learning_rate": 2.9550912854789463e-05,
      "loss": 0.2272,
      "step": 1152
    },
    {
      "epoch": 0.7232240865610788,
      "grad_norm": 0.967366635799408,
      "learning_rate": 2.953305394923536e-05,
      "loss": 0.2384,
      "step": 1153
    },
    {
      "epoch": 0.7238513407558413,
      "grad_norm": 0.904085099697113,
      "learning_rate": 2.9515185202374068e-05,
      "loss": 0.2469,
      "step": 1154
    },
    {
      "epoch": 0.7244785949506037,
      "grad_norm": 0.9692016243934631,
      "learning_rate": 2.949730663265212e-05,
      "loss": 0.2576,
      "step": 1155
    },
    {
      "epoch": 0.7251058491453661,
      "grad_norm": 0.9735581278800964,
      "learning_rate": 2.9479418258526196e-05,
      "loss": 0.2501,
      "step": 1156
    },
    {
      "epoch": 0.7257331033401286,
      "grad_norm": 0.8929852247238159,
      "learning_rate": 2.9461520098463085e-05,
      "loss": 0.2195,
      "step": 1157
    },
    {
      "epoch": 0.726360357534891,
      "grad_norm": 1.1679868698120117,
      "learning_rate": 2.9443612170939705e-05,
      "loss": 0.2593,
      "step": 1158
    },
    {
      "epoch": 0.7269876117296534,
      "grad_norm": 0.9278621077537537,
      "learning_rate": 2.9425694494443024e-05,
      "loss": 0.2605,
      "step": 1159
    },
    {
      "epoch": 0.7276148659244158,
      "grad_norm": 0.9248387813568115,
      "learning_rate": 2.9407767087470094e-05,
      "loss": 0.2295,
      "step": 1160
    },
    {
      "epoch": 0.7282421201191783,
      "grad_norm": 0.9732686281204224,
      "learning_rate": 2.9389829968528013e-05,
      "loss": 0.233,
      "step": 1161
    },
    {
      "epoch": 0.7288693743139407,
      "grad_norm": 0.8317962288856506,
      "learning_rate": 2.9371883156133894e-05,
      "loss": 0.2278,
      "step": 1162
    },
    {
      "epoch": 0.7294966285087031,
      "grad_norm": 0.8876512050628662,
      "learning_rate": 2.935392666881487e-05,
      "loss": 0.2412,
      "step": 1163
    },
    {
      "epoch": 0.7301238827034656,
      "grad_norm": 0.89940345287323,
      "learning_rate": 2.9335960525108052e-05,
      "loss": 0.2202,
      "step": 1164
    },
    {
      "epoch": 0.730751136898228,
      "grad_norm": 0.8470626473426819,
      "learning_rate": 2.9317984743560523e-05,
      "loss": 0.22,
      "step": 1165
    },
    {
      "epoch": 0.7313783910929904,
      "grad_norm": 1.0272356271743774,
      "learning_rate": 2.9299999342729308e-05,
      "loss": 0.3156,
      "step": 1166
    },
    {
      "epoch": 0.7320056452877529,
      "grad_norm": 0.9586045742034912,
      "learning_rate": 2.9282004341181384e-05,
      "loss": 0.2302,
      "step": 1167
    },
    {
      "epoch": 0.7326328994825153,
      "grad_norm": 1.1000994443893433,
      "learning_rate": 2.9263999757493616e-05,
      "loss": 0.2551,
      "step": 1168
    },
    {
      "epoch": 0.7332601536772777,
      "grad_norm": 1.0782554149627686,
      "learning_rate": 2.9245985610252767e-05,
      "loss": 0.2966,
      "step": 1169
    },
    {
      "epoch": 0.7338874078720401,
      "grad_norm": 1.1621781587600708,
      "learning_rate": 2.922796191805548e-05,
      "loss": 0.2416,
      "step": 1170
    },
    {
      "epoch": 0.7345146620668026,
      "grad_norm": 0.9000127911567688,
      "learning_rate": 2.9209928699508244e-05,
      "loss": 0.2071,
      "step": 1171
    },
    {
      "epoch": 0.735141916261565,
      "grad_norm": 1.0584723949432373,
      "learning_rate": 2.9191885973227392e-05,
      "loss": 0.247,
      "step": 1172
    },
    {
      "epoch": 0.7357691704563274,
      "grad_norm": 1.134102702140808,
      "learning_rate": 2.9173833757839055e-05,
      "loss": 0.264,
      "step": 1173
    },
    {
      "epoch": 0.7363964246510899,
      "grad_norm": 0.8952019214630127,
      "learning_rate": 2.9155772071979176e-05,
      "loss": 0.2123,
      "step": 1174
    },
    {
      "epoch": 0.7370236788458523,
      "grad_norm": 0.9572092294692993,
      "learning_rate": 2.9137700934293472e-05,
      "loss": 0.2525,
      "step": 1175
    },
    {
      "epoch": 0.7376509330406147,
      "grad_norm": 1.1004743576049805,
      "learning_rate": 2.9119620363437412e-05,
      "loss": 0.2701,
      "step": 1176
    },
    {
      "epoch": 0.7382781872353771,
      "grad_norm": 0.9321638345718384,
      "learning_rate": 2.9101530378076203e-05,
      "loss": 0.2122,
      "step": 1177
    },
    {
      "epoch": 0.7389054414301396,
      "grad_norm": 0.930284321308136,
      "learning_rate": 2.9083430996884777e-05,
      "loss": 0.2247,
      "step": 1178
    },
    {
      "epoch": 0.739532695624902,
      "grad_norm": 0.9212769865989685,
      "learning_rate": 2.906532223854776e-05,
      "loss": 0.2319,
      "step": 1179
    },
    {
      "epoch": 0.7401599498196644,
      "grad_norm": 0.9486585259437561,
      "learning_rate": 2.9047204121759458e-05,
      "loss": 0.2576,
      "step": 1180
    },
    {
      "epoch": 0.7407872040144269,
      "grad_norm": 0.960632860660553,
      "learning_rate": 2.9029076665223855e-05,
      "loss": 0.2629,
      "step": 1181
    },
    {
      "epoch": 0.7414144582091893,
      "grad_norm": 0.950925350189209,
      "learning_rate": 2.9010939887654547e-05,
      "loss": 0.2804,
      "step": 1182
    },
    {
      "epoch": 0.7420417124039517,
      "grad_norm": 0.9938424825668335,
      "learning_rate": 2.8992793807774768e-05,
      "loss": 0.2598,
      "step": 1183
    },
    {
      "epoch": 0.7426689665987142,
      "grad_norm": 0.8266270160675049,
      "learning_rate": 2.8974638444317373e-05,
      "loss": 0.2562,
      "step": 1184
    },
    {
      "epoch": 0.7432962207934766,
      "grad_norm": 0.9874655604362488,
      "learning_rate": 2.8956473816024764e-05,
      "loss": 0.2668,
      "step": 1185
    },
    {
      "epoch": 0.743923474988239,
      "grad_norm": 1.0183908939361572,
      "learning_rate": 2.893829994164893e-05,
      "loss": 0.2724,
      "step": 1186
    },
    {
      "epoch": 0.7445507291830014,
      "grad_norm": 0.9691697955131531,
      "learning_rate": 2.8920116839951408e-05,
      "loss": 0.2325,
      "step": 1187
    },
    {
      "epoch": 0.7451779833777639,
      "grad_norm": 1.0251258611679077,
      "learning_rate": 2.890192452970325e-05,
      "loss": 0.2714,
      "step": 1188
    },
    {
      "epoch": 0.7458052375725263,
      "grad_norm": 1.0905992984771729,
      "learning_rate": 2.8883723029685017e-05,
      "loss": 0.2714,
      "step": 1189
    },
    {
      "epoch": 0.7464324917672887,
      "grad_norm": 1.0011322498321533,
      "learning_rate": 2.8865512358686757e-05,
      "loss": 0.2719,
      "step": 1190
    },
    {
      "epoch": 0.7470597459620512,
      "grad_norm": 1.1232945919036865,
      "learning_rate": 2.8847292535507994e-05,
      "loss": 0.2959,
      "step": 1191
    },
    {
      "epoch": 0.7476870001568136,
      "grad_norm": 1.031429648399353,
      "learning_rate": 2.8829063578957682e-05,
      "loss": 0.2335,
      "step": 1192
    },
    {
      "epoch": 0.748314254351576,
      "grad_norm": 0.9283905625343323,
      "learning_rate": 2.8810825507854227e-05,
      "loss": 0.2574,
      "step": 1193
    },
    {
      "epoch": 0.7489415085463385,
      "grad_norm": 0.9369612336158752,
      "learning_rate": 2.8792578341025418e-05,
      "loss": 0.2666,
      "step": 1194
    },
    {
      "epoch": 0.7495687627411008,
      "grad_norm": 1.0052746534347534,
      "learning_rate": 2.877432209730846e-05,
      "loss": 0.251,
      "step": 1195
    },
    {
      "epoch": 0.7501960169358632,
      "grad_norm": 1.0359718799591064,
      "learning_rate": 2.8756056795549912e-05,
      "loss": 0.2499,
      "step": 1196
    },
    {
      "epoch": 0.7508232711306256,
      "grad_norm": 0.9635895490646362,
      "learning_rate": 2.873778245460569e-05,
      "loss": 0.2453,
      "step": 1197
    },
    {
      "epoch": 0.7514505253253881,
      "grad_norm": 0.9521752595901489,
      "learning_rate": 2.871949909334103e-05,
      "loss": 0.2607,
      "step": 1198
    },
    {
      "epoch": 0.7520777795201505,
      "grad_norm": 1.0811841487884521,
      "learning_rate": 2.8701206730630507e-05,
      "loss": 0.3175,
      "step": 1199
    },
    {
      "epoch": 0.7527050337149129,
      "grad_norm": 0.9252716302871704,
      "learning_rate": 2.8682905385357954e-05,
      "loss": 0.2498,
      "step": 1200
    },
    {
      "epoch": 0.7533322879096754,
      "grad_norm": 0.9174233078956604,
      "learning_rate": 2.8664595076416497e-05,
      "loss": 0.2515,
      "step": 1201
    },
    {
      "epoch": 0.7539595421044378,
      "grad_norm": 0.865044355392456,
      "learning_rate": 2.8646275822708518e-05,
      "loss": 0.2067,
      "step": 1202
    },
    {
      "epoch": 0.7545867962992002,
      "grad_norm": 0.9602584838867188,
      "learning_rate": 2.8627947643145625e-05,
      "loss": 0.2564,
      "step": 1203
    },
    {
      "epoch": 0.7552140504939627,
      "grad_norm": 0.9477353692054749,
      "learning_rate": 2.8609610556648642e-05,
      "loss": 0.2554,
      "step": 1204
    },
    {
      "epoch": 0.7558413046887251,
      "grad_norm": 0.9231157898902893,
      "learning_rate": 2.8591264582147594e-05,
      "loss": 0.2355,
      "step": 1205
    },
    {
      "epoch": 0.7564685588834875,
      "grad_norm": 0.924030065536499,
      "learning_rate": 2.857290973858167e-05,
      "loss": 0.2244,
      "step": 1206
    },
    {
      "epoch": 0.7570958130782499,
      "grad_norm": 0.9852567911148071,
      "learning_rate": 2.8554546044899223e-05,
      "loss": 0.2543,
      "step": 1207
    },
    {
      "epoch": 0.7577230672730124,
      "grad_norm": 1.033445954322815,
      "learning_rate": 2.8536173520057744e-05,
      "loss": 0.3014,
      "step": 1208
    },
    {
      "epoch": 0.7583503214677748,
      "grad_norm": 0.8468286991119385,
      "learning_rate": 2.851779218302383e-05,
      "loss": 0.2266,
      "step": 1209
    },
    {
      "epoch": 0.7589775756625372,
      "grad_norm": 1.0191102027893066,
      "learning_rate": 2.849940205277319e-05,
      "loss": 0.2342,
      "step": 1210
    },
    {
      "epoch": 0.7596048298572997,
      "grad_norm": 0.8493043780326843,
      "learning_rate": 2.8481003148290606e-05,
      "loss": 0.2553,
      "step": 1211
    },
    {
      "epoch": 0.7602320840520621,
      "grad_norm": 0.855071485042572,
      "learning_rate": 2.84625954885699e-05,
      "loss": 0.2524,
      "step": 1212
    },
    {
      "epoch": 0.7608593382468245,
      "grad_norm": 1.0303425788879395,
      "learning_rate": 2.8444179092613953e-05,
      "loss": 0.2379,
      "step": 1213
    },
    {
      "epoch": 0.761486592441587,
      "grad_norm": 0.9178991913795471,
      "learning_rate": 2.842575397943467e-05,
      "loss": 0.2295,
      "step": 1214
    },
    {
      "epoch": 0.7621138466363494,
      "grad_norm": 0.869573712348938,
      "learning_rate": 2.8407320168052925e-05,
      "loss": 0.2125,
      "step": 1215
    },
    {
      "epoch": 0.7627411008311118,
      "grad_norm": 0.8731804490089417,
      "learning_rate": 2.838887767749861e-05,
      "loss": 0.2302,
      "step": 1216
    },
    {
      "epoch": 0.7633683550258742,
      "grad_norm": 0.7152383923530579,
      "learning_rate": 2.8370426526810545e-05,
      "loss": 0.1908,
      "step": 1217
    },
    {
      "epoch": 0.7639956092206367,
      "grad_norm": 0.9450547099113464,
      "learning_rate": 2.8351966735036502e-05,
      "loss": 0.2948,
      "step": 1218
    },
    {
      "epoch": 0.7646228634153991,
      "grad_norm": 0.875137209892273,
      "learning_rate": 2.8333498321233183e-05,
      "loss": 0.2154,
      "step": 1219
    },
    {
      "epoch": 0.7652501176101615,
      "grad_norm": 0.8878747820854187,
      "learning_rate": 2.831502130446617e-05,
      "loss": 0.236,
      "step": 1220
    },
    {
      "epoch": 0.765877371804924,
      "grad_norm": 0.898963987827301,
      "learning_rate": 2.8296535703809947e-05,
      "loss": 0.271,
      "step": 1221
    },
    {
      "epoch": 0.7665046259996864,
      "grad_norm": 0.9544633626937866,
      "learning_rate": 2.8278041538347847e-05,
      "loss": 0.2678,
      "step": 1222
    },
    {
      "epoch": 0.7671318801944488,
      "grad_norm": 0.8638365268707275,
      "learning_rate": 2.8259538827172047e-05,
      "loss": 0.2071,
      "step": 1223
    },
    {
      "epoch": 0.7677591343892112,
      "grad_norm": 0.8723821640014648,
      "learning_rate": 2.8241027589383553e-05,
      "loss": 0.233,
      "step": 1224
    },
    {
      "epoch": 0.7683863885839737,
      "grad_norm": 0.843165397644043,
      "learning_rate": 2.8222507844092163e-05,
      "loss": 0.2341,
      "step": 1225
    },
    {
      "epoch": 0.7690136427787361,
      "grad_norm": 0.994552731513977,
      "learning_rate": 2.8203979610416457e-05,
      "loss": 0.2601,
      "step": 1226
    },
    {
      "epoch": 0.7696408969734985,
      "grad_norm": 0.9148455858230591,
      "learning_rate": 2.8185442907483786e-05,
      "loss": 0.2302,
      "step": 1227
    },
    {
      "epoch": 0.770268151168261,
      "grad_norm": 0.8745561242103577,
      "learning_rate": 2.816689775443025e-05,
      "loss": 0.2478,
      "step": 1228
    },
    {
      "epoch": 0.7708954053630234,
      "grad_norm": 0.9031297564506531,
      "learning_rate": 2.8148344170400654e-05,
      "loss": 0.2429,
      "step": 1229
    },
    {
      "epoch": 0.7715226595577858,
      "grad_norm": 1.0213173627853394,
      "learning_rate": 2.812978217454851e-05,
      "loss": 0.2483,
      "step": 1230
    },
    {
      "epoch": 0.7721499137525483,
      "grad_norm": 0.9234693646430969,
      "learning_rate": 2.811121178603604e-05,
      "loss": 0.2613,
      "step": 1231
    },
    {
      "epoch": 0.7727771679473107,
      "grad_norm": 0.8913220167160034,
      "learning_rate": 2.809263302403409e-05,
      "loss": 0.2203,
      "step": 1232
    },
    {
      "epoch": 0.7734044221420731,
      "grad_norm": 0.9702144861221313,
      "learning_rate": 2.8074045907722176e-05,
      "loss": 0.2684,
      "step": 1233
    },
    {
      "epoch": 0.7740316763368355,
      "grad_norm": 0.8771833777427673,
      "learning_rate": 2.805545045628844e-05,
      "loss": 0.2265,
      "step": 1234
    },
    {
      "epoch": 0.774658930531598,
      "grad_norm": 0.8456547260284424,
      "learning_rate": 2.8036846688929614e-05,
      "loss": 0.2088,
      "step": 1235
    },
    {
      "epoch": 0.7752861847263604,
      "grad_norm": 0.9730178713798523,
      "learning_rate": 2.8018234624851016e-05,
      "loss": 0.2324,
      "step": 1236
    },
    {
      "epoch": 0.7759134389211227,
      "grad_norm": 1.0039517879486084,
      "learning_rate": 2.7999614283266558e-05,
      "loss": 0.2594,
      "step": 1237
    },
    {
      "epoch": 0.7765406931158852,
      "grad_norm": 0.9516322612762451,
      "learning_rate": 2.7980985683398648e-05,
      "loss": 0.21,
      "step": 1238
    },
    {
      "epoch": 0.7771679473106476,
      "grad_norm": 0.9338817596435547,
      "learning_rate": 2.7962348844478255e-05,
      "loss": 0.229,
      "step": 1239
    },
    {
      "epoch": 0.77779520150541,
      "grad_norm": 0.8492531776428223,
      "learning_rate": 2.794370378574485e-05,
      "loss": 0.2305,
      "step": 1240
    },
    {
      "epoch": 0.7784224557001725,
      "grad_norm": 1.0080124139785767,
      "learning_rate": 2.7925050526446377e-05,
      "loss": 0.2237,
      "step": 1241
    },
    {
      "epoch": 0.7790497098949349,
      "grad_norm": 1.0470832586288452,
      "learning_rate": 2.7906389085839248e-05,
      "loss": 0.2628,
      "step": 1242
    },
    {
      "epoch": 0.7796769640896973,
      "grad_norm": 1.0095301866531372,
      "learning_rate": 2.7887719483188335e-05,
      "loss": 0.2941,
      "step": 1243
    },
    {
      "epoch": 0.7803042182844597,
      "grad_norm": 0.9931348562240601,
      "learning_rate": 2.7869041737766916e-05,
      "loss": 0.2444,
      "step": 1244
    },
    {
      "epoch": 0.7809314724792222,
      "grad_norm": 1.0151265859603882,
      "learning_rate": 2.7850355868856697e-05,
      "loss": 0.2643,
      "step": 1245
    },
    {
      "epoch": 0.7815587266739846,
      "grad_norm": 0.9596518278121948,
      "learning_rate": 2.7831661895747753e-05,
      "loss": 0.2752,
      "step": 1246
    },
    {
      "epoch": 0.782185980868747,
      "grad_norm": 0.9194288849830627,
      "learning_rate": 2.781295983773853e-05,
      "loss": 0.234,
      "step": 1247
    },
    {
      "epoch": 0.7828132350635095,
      "grad_norm": 0.9633300304412842,
      "learning_rate": 2.779424971413582e-05,
      "loss": 0.2282,
      "step": 1248
    },
    {
      "epoch": 0.7834404892582719,
      "grad_norm": 0.8626919388771057,
      "learning_rate": 2.7775531544254735e-05,
      "loss": 0.2708,
      "step": 1249
    },
    {
      "epoch": 0.7840677434530343,
      "grad_norm": 0.9322224855422974,
      "learning_rate": 2.7756805347418717e-05,
      "loss": 0.2398,
      "step": 1250
    },
    {
      "epoch": 0.7846949976477968,
      "grad_norm": 0.949925422668457,
      "learning_rate": 2.7738071142959468e-05,
      "loss": 0.2425,
      "step": 1251
    },
    {
      "epoch": 0.7853222518425592,
      "grad_norm": 0.9091950058937073,
      "learning_rate": 2.7719328950216967e-05,
      "loss": 0.2028,
      "step": 1252
    },
    {
      "epoch": 0.7859495060373216,
      "grad_norm": 0.9050923585891724,
      "learning_rate": 2.770057878853944e-05,
      "loss": 0.2661,
      "step": 1253
    },
    {
      "epoch": 0.786576760232084,
      "grad_norm": 0.8346881866455078,
      "learning_rate": 2.7681820677283338e-05,
      "loss": 0.1991,
      "step": 1254
    },
    {
      "epoch": 0.7872040144268465,
      "grad_norm": 0.9741007685661316,
      "learning_rate": 2.7663054635813325e-05,
      "loss": 0.262,
      "step": 1255
    },
    {
      "epoch": 0.7878312686216089,
      "grad_norm": 0.8977965116500854,
      "learning_rate": 2.7644280683502243e-05,
      "loss": 0.2438,
      "step": 1256
    },
    {
      "epoch": 0.7884585228163713,
      "grad_norm": 1.0643701553344727,
      "learning_rate": 2.762549883973111e-05,
      "loss": 0.2365,
      "step": 1257
    },
    {
      "epoch": 0.7890857770111338,
      "grad_norm": 1.0908955335617065,
      "learning_rate": 2.7606709123889067e-05,
      "loss": 0.2488,
      "step": 1258
    },
    {
      "epoch": 0.7897130312058962,
      "grad_norm": 0.9831537008285522,
      "learning_rate": 2.758791155537341e-05,
      "loss": 0.2613,
      "step": 1259
    },
    {
      "epoch": 0.7903402854006586,
      "grad_norm": 1.0671886205673218,
      "learning_rate": 2.7569106153589536e-05,
      "loss": 0.2367,
      "step": 1260
    },
    {
      "epoch": 0.7909675395954211,
      "grad_norm": 1.1107702255249023,
      "learning_rate": 2.755029293795092e-05,
      "loss": 0.2667,
      "step": 1261
    },
    {
      "epoch": 0.7915947937901835,
      "grad_norm": 1.015174150466919,
      "learning_rate": 2.7531471927879104e-05,
      "loss": 0.2239,
      "step": 1262
    },
    {
      "epoch": 0.7922220479849459,
      "grad_norm": 0.8444801568984985,
      "learning_rate": 2.7512643142803676e-05,
      "loss": 0.2036,
      "step": 1263
    },
    {
      "epoch": 0.7928493021797083,
      "grad_norm": 0.9486952424049377,
      "learning_rate": 2.7493806602162262e-05,
      "loss": 0.2306,
      "step": 1264
    },
    {
      "epoch": 0.7934765563744708,
      "grad_norm": 0.963703989982605,
      "learning_rate": 2.747496232540048e-05,
      "loss": 0.2228,
      "step": 1265
    },
    {
      "epoch": 0.7941038105692332,
      "grad_norm": 1.0590280294418335,
      "learning_rate": 2.7456110331971945e-05,
      "loss": 0.2403,
      "step": 1266
    },
    {
      "epoch": 0.7947310647639956,
      "grad_norm": 1.037078619003296,
      "learning_rate": 2.7437250641338225e-05,
      "loss": 0.2553,
      "step": 1267
    },
    {
      "epoch": 0.7953583189587581,
      "grad_norm": 0.9628890752792358,
      "learning_rate": 2.7418383272968854e-05,
      "loss": 0.2267,
      "step": 1268
    },
    {
      "epoch": 0.7959855731535205,
      "grad_norm": 1.0467262268066406,
      "learning_rate": 2.7399508246341273e-05,
      "loss": 0.2447,
      "step": 1269
    },
    {
      "epoch": 0.7966128273482829,
      "grad_norm": 0.9027785658836365,
      "learning_rate": 2.7380625580940835e-05,
      "loss": 0.2427,
      "step": 1270
    },
    {
      "epoch": 0.7972400815430453,
      "grad_norm": 1.0007820129394531,
      "learning_rate": 2.7361735296260786e-05,
      "loss": 0.2593,
      "step": 1271
    },
    {
      "epoch": 0.7978673357378078,
      "grad_norm": 1.091827392578125,
      "learning_rate": 2.734283741180223e-05,
      "loss": 0.2685,
      "step": 1272
    },
    {
      "epoch": 0.7984945899325702,
      "grad_norm": 1.0432686805725098,
      "learning_rate": 2.7323931947074117e-05,
      "loss": 0.2617,
      "step": 1273
    },
    {
      "epoch": 0.7991218441273326,
      "grad_norm": 0.9992380738258362,
      "learning_rate": 2.7305018921593225e-05,
      "loss": 0.2396,
      "step": 1274
    },
    {
      "epoch": 0.7997490983220951,
      "grad_norm": 0.9518929123878479,
      "learning_rate": 2.7286098354884142e-05,
      "loss": 0.2642,
      "step": 1275
    },
    {
      "epoch": 0.8003763525168575,
      "grad_norm": 1.0347591638565063,
      "learning_rate": 2.726717026647922e-05,
      "loss": 0.2499,
      "step": 1276
    },
    {
      "epoch": 0.8010036067116199,
      "grad_norm": 0.9323564171791077,
      "learning_rate": 2.7248234675918605e-05,
      "loss": 0.265,
      "step": 1277
    },
    {
      "epoch": 0.8016308609063824,
      "grad_norm": 0.8762786388397217,
      "learning_rate": 2.722929160275017e-05,
      "loss": 0.2168,
      "step": 1278
    },
    {
      "epoch": 0.8022581151011448,
      "grad_norm": 0.8518358469009399,
      "learning_rate": 2.721034106652951e-05,
      "loss": 0.243,
      "step": 1279
    },
    {
      "epoch": 0.8028853692959071,
      "grad_norm": 0.8623441457748413,
      "learning_rate": 2.719138308681993e-05,
      "loss": 0.2518,
      "step": 1280
    },
    {
      "epoch": 0.8035126234906695,
      "grad_norm": 0.8846951723098755,
      "learning_rate": 2.7172417683192436e-05,
      "loss": 0.1975,
      "step": 1281
    },
    {
      "epoch": 0.804139877685432,
      "grad_norm": 0.8244737386703491,
      "learning_rate": 2.7153444875225667e-05,
      "loss": 0.2026,
      "step": 1282
    },
    {
      "epoch": 0.8047671318801944,
      "grad_norm": 1.1555349826812744,
      "learning_rate": 2.7134464682505926e-05,
      "loss": 0.2711,
      "step": 1283
    },
    {
      "epoch": 0.8053943860749568,
      "grad_norm": 0.9521437883377075,
      "learning_rate": 2.7115477124627134e-05,
      "loss": 0.2383,
      "step": 1284
    },
    {
      "epoch": 0.8060216402697193,
      "grad_norm": 0.9691430926322937,
      "learning_rate": 2.7096482221190807e-05,
      "loss": 0.2627,
      "step": 1285
    },
    {
      "epoch": 0.8066488944644817,
      "grad_norm": 0.9191070795059204,
      "learning_rate": 2.7077479991806062e-05,
      "loss": 0.1817,
      "step": 1286
    },
    {
      "epoch": 0.8072761486592441,
      "grad_norm": 1.0996882915496826,
      "learning_rate": 2.7058470456089576e-05,
      "loss": 0.2402,
      "step": 1287
    },
    {
      "epoch": 0.8079034028540066,
      "grad_norm": 0.908535361289978,
      "learning_rate": 2.703945363366555e-05,
      "loss": 0.2901,
      "step": 1288
    },
    {
      "epoch": 0.808530657048769,
      "grad_norm": 0.8575511574745178,
      "learning_rate": 2.7020429544165718e-05,
      "loss": 0.2502,
      "step": 1289
    },
    {
      "epoch": 0.8091579112435314,
      "grad_norm": 0.9514098167419434,
      "learning_rate": 2.7001398207229324e-05,
      "loss": 0.2376,
      "step": 1290
    },
    {
      "epoch": 0.8097851654382938,
      "grad_norm": 0.956186056137085,
      "learning_rate": 2.6982359642503082e-05,
      "loss": 0.2388,
      "step": 1291
    },
    {
      "epoch": 0.8104124196330563,
      "grad_norm": 1.0436739921569824,
      "learning_rate": 2.6963313869641177e-05,
      "loss": 0.2925,
      "step": 1292
    },
    {
      "epoch": 0.8110396738278187,
      "grad_norm": 0.7947832345962524,
      "learning_rate": 2.6944260908305233e-05,
      "loss": 0.2262,
      "step": 1293
    },
    {
      "epoch": 0.8116669280225811,
      "grad_norm": 0.7879627346992493,
      "learning_rate": 2.692520077816428e-05,
      "loss": 0.2211,
      "step": 1294
    },
    {
      "epoch": 0.8122941822173436,
      "grad_norm": 0.8735244870185852,
      "learning_rate": 2.690613349889478e-05,
      "loss": 0.2588,
      "step": 1295
    },
    {
      "epoch": 0.812921436412106,
      "grad_norm": 0.880125105381012,
      "learning_rate": 2.6887059090180542e-05,
      "loss": 0.2136,
      "step": 1296
    },
    {
      "epoch": 0.8135486906068684,
      "grad_norm": 0.7848929166793823,
      "learning_rate": 2.686797757171275e-05,
      "loss": 0.2106,
      "step": 1297
    },
    {
      "epoch": 0.8141759448016309,
      "grad_norm": 0.8269034624099731,
      "learning_rate": 2.6848888963189943e-05,
      "loss": 0.2001,
      "step": 1298
    },
    {
      "epoch": 0.8148031989963933,
      "grad_norm": 0.900639533996582,
      "learning_rate": 2.6829793284317945e-05,
      "loss": 0.2377,
      "step": 1299
    },
    {
      "epoch": 0.8154304531911557,
      "grad_norm": 0.8968613743782043,
      "learning_rate": 2.68106905548099e-05,
      "loss": 0.2387,
      "step": 1300
    },
    {
      "epoch": 0.8160577073859181,
      "grad_norm": 1.0977859497070312,
      "learning_rate": 2.679158079438624e-05,
      "loss": 0.2524,
      "step": 1301
    },
    {
      "epoch": 0.8166849615806806,
      "grad_norm": 1.0224528312683105,
      "learning_rate": 2.677246402277464e-05,
      "loss": 0.252,
      "step": 1302
    },
    {
      "epoch": 0.817312215775443,
      "grad_norm": 0.9429143667221069,
      "learning_rate": 2.6753340259710005e-05,
      "loss": 0.2295,
      "step": 1303
    },
    {
      "epoch": 0.8179394699702054,
      "grad_norm": 0.9691178798675537,
      "learning_rate": 2.673420952493448e-05,
      "loss": 0.2387,
      "step": 1304
    },
    {
      "epoch": 0.8185667241649679,
      "grad_norm": 0.8709486722946167,
      "learning_rate": 2.6715071838197393e-05,
      "loss": 0.1901,
      "step": 1305
    },
    {
      "epoch": 0.8191939783597303,
      "grad_norm": 0.9485816955566406,
      "learning_rate": 2.6695927219255248e-05,
      "loss": 0.26,
      "step": 1306
    },
    {
      "epoch": 0.8198212325544927,
      "grad_norm": 0.95992112159729,
      "learning_rate": 2.6676775687871716e-05,
      "loss": 0.2627,
      "step": 1307
    },
    {
      "epoch": 0.8204484867492552,
      "grad_norm": 1.0174671411514282,
      "learning_rate": 2.6657617263817587e-05,
      "loss": 0.2474,
      "step": 1308
    },
    {
      "epoch": 0.8210757409440176,
      "grad_norm": 1.0580779314041138,
      "learning_rate": 2.6638451966870788e-05,
      "loss": 0.2257,
      "step": 1309
    },
    {
      "epoch": 0.82170299513878,
      "grad_norm": 1.1434040069580078,
      "learning_rate": 2.6619279816816324e-05,
      "loss": 0.2668,
      "step": 1310
    },
    {
      "epoch": 0.8223302493335424,
      "grad_norm": 0.9524603486061096,
      "learning_rate": 2.660010083344628e-05,
      "loss": 0.222,
      "step": 1311
    },
    {
      "epoch": 0.8229575035283049,
      "grad_norm": 1.0641026496887207,
      "learning_rate": 2.65809150365598e-05,
      "loss": 0.2243,
      "step": 1312
    },
    {
      "epoch": 0.8235847577230673,
      "grad_norm": 0.9155853390693665,
      "learning_rate": 2.6561722445963054e-05,
      "loss": 0.2534,
      "step": 1313
    },
    {
      "epoch": 0.8242120119178297,
      "grad_norm": 0.8461087942123413,
      "learning_rate": 2.654252308146923e-05,
      "loss": 0.2423,
      "step": 1314
    },
    {
      "epoch": 0.8248392661125922,
      "grad_norm": 1.0586851835250854,
      "learning_rate": 2.6523316962898507e-05,
      "loss": 0.2554,
      "step": 1315
    },
    {
      "epoch": 0.8254665203073546,
      "grad_norm": 0.9544086456298828,
      "learning_rate": 2.650410411007804e-05,
      "loss": 0.2614,
      "step": 1316
    },
    {
      "epoch": 0.826093774502117,
      "grad_norm": 1.1493490934371948,
      "learning_rate": 2.6484884542841922e-05,
      "loss": 0.2487,
      "step": 1317
    },
    {
      "epoch": 0.8267210286968794,
      "grad_norm": 1.180282473564148,
      "learning_rate": 2.6465658281031205e-05,
      "loss": 0.2465,
      "step": 1318
    },
    {
      "epoch": 0.8273482828916419,
      "grad_norm": 0.8955004811286926,
      "learning_rate": 2.6446425344493825e-05,
      "loss": 0.2502,
      "step": 1319
    },
    {
      "epoch": 0.8279755370864043,
      "grad_norm": 0.9922782778739929,
      "learning_rate": 2.6427185753084622e-05,
      "loss": 0.1815,
      "step": 1320
    },
    {
      "epoch": 0.8286027912811667,
      "grad_norm": 1.016087293624878,
      "learning_rate": 2.640793952666531e-05,
      "loss": 0.2208,
      "step": 1321
    },
    {
      "epoch": 0.8292300454759292,
      "grad_norm": 1.1681716442108154,
      "learning_rate": 2.6388686685104438e-05,
      "loss": 0.2762,
      "step": 1322
    },
    {
      "epoch": 0.8298572996706916,
      "grad_norm": 1.0408824682235718,
      "learning_rate": 2.6369427248277387e-05,
      "loss": 0.251,
      "step": 1323
    },
    {
      "epoch": 0.830484553865454,
      "grad_norm": 0.9965060353279114,
      "learning_rate": 2.635016123606637e-05,
      "loss": 0.249,
      "step": 1324
    },
    {
      "epoch": 0.8311118080602165,
      "grad_norm": 0.824463427066803,
      "learning_rate": 2.633088866836036e-05,
      "loss": 0.1927,
      "step": 1325
    },
    {
      "epoch": 0.8317390622549788,
      "grad_norm": 1.0201879739761353,
      "learning_rate": 2.63116095650551e-05,
      "loss": 0.2386,
      "step": 1326
    },
    {
      "epoch": 0.8323663164497412,
      "grad_norm": 0.9999285340309143,
      "learning_rate": 2.6292323946053096e-05,
      "loss": 0.2616,
      "step": 1327
    },
    {
      "epoch": 0.8329935706445036,
      "grad_norm": 1.0046961307525635,
      "learning_rate": 2.6273031831263574e-05,
      "loss": 0.249,
      "step": 1328
    },
    {
      "epoch": 0.8336208248392661,
      "grad_norm": 0.978674054145813,
      "learning_rate": 2.625373324060245e-05,
      "loss": 0.2559,
      "step": 1329
    },
    {
      "epoch": 0.8342480790340285,
      "grad_norm": 0.9145632386207581,
      "learning_rate": 2.623442819399234e-05,
      "loss": 0.2227,
      "step": 1330
    },
    {
      "epoch": 0.8348753332287909,
      "grad_norm": 0.8744785189628601,
      "learning_rate": 2.621511671136254e-05,
      "loss": 0.2334,
      "step": 1331
    },
    {
      "epoch": 0.8355025874235534,
      "grad_norm": 1.0927536487579346,
      "learning_rate": 2.619579881264896e-05,
      "loss": 0.2629,
      "step": 1332
    },
    {
      "epoch": 0.8361298416183158,
      "grad_norm": 0.8395379781723022,
      "learning_rate": 2.6176474517794152e-05,
      "loss": 0.2319,
      "step": 1333
    },
    {
      "epoch": 0.8367570958130782,
      "grad_norm": 1.0170912742614746,
      "learning_rate": 2.6157143846747265e-05,
      "loss": 0.2574,
      "step": 1334
    },
    {
      "epoch": 0.8373843500078407,
      "grad_norm": 0.8509354591369629,
      "learning_rate": 2.613780681946403e-05,
      "loss": 0.1948,
      "step": 1335
    },
    {
      "epoch": 0.8380116042026031,
      "grad_norm": 0.8112233281135559,
      "learning_rate": 2.611846345590674e-05,
      "loss": 0.1838,
      "step": 1336
    },
    {
      "epoch": 0.8386388583973655,
      "grad_norm": 1.1284143924713135,
      "learning_rate": 2.609911377604423e-05,
      "loss": 0.2349,
      "step": 1337
    },
    {
      "epoch": 0.8392661125921279,
      "grad_norm": 0.8029350638389587,
      "learning_rate": 2.6079757799851853e-05,
      "loss": 0.249,
      "step": 1338
    },
    {
      "epoch": 0.8398933667868904,
      "grad_norm": 1.106877326965332,
      "learning_rate": 2.6060395547311476e-05,
      "loss": 0.2542,
      "step": 1339
    },
    {
      "epoch": 0.8405206209816528,
      "grad_norm": 1.000625729560852,
      "learning_rate": 2.6041027038411427e-05,
      "loss": 0.2461,
      "step": 1340
    },
    {
      "epoch": 0.8411478751764152,
      "grad_norm": 1.044447660446167,
      "learning_rate": 2.6021652293146493e-05,
      "loss": 0.2177,
      "step": 1341
    },
    {
      "epoch": 0.8417751293711777,
      "grad_norm": 1.2332682609558105,
      "learning_rate": 2.600227133151791e-05,
      "loss": 0.2558,
      "step": 1342
    },
    {
      "epoch": 0.8424023835659401,
      "grad_norm": 1.0727463960647583,
      "learning_rate": 2.5982884173533335e-05,
      "loss": 0.2329,
      "step": 1343
    },
    {
      "epoch": 0.8430296377607025,
      "grad_norm": 0.9513474106788635,
      "learning_rate": 2.5963490839206806e-05,
      "loss": 0.2508,
      "step": 1344
    },
    {
      "epoch": 0.843656891955465,
      "grad_norm": 0.9870790839195251,
      "learning_rate": 2.5944091348558743e-05,
      "loss": 0.2565,
      "step": 1345
    },
    {
      "epoch": 0.8442841461502274,
      "grad_norm": 0.8659030795097351,
      "learning_rate": 2.5924685721615928e-05,
      "loss": 0.2309,
      "step": 1346
    },
    {
      "epoch": 0.8449114003449898,
      "grad_norm": 1.0346086025238037,
      "learning_rate": 2.5905273978411468e-05,
      "loss": 0.2429,
      "step": 1347
    },
    {
      "epoch": 0.8455386545397522,
      "grad_norm": 1.0064115524291992,
      "learning_rate": 2.58858561389848e-05,
      "loss": 0.2251,
      "step": 1348
    },
    {
      "epoch": 0.8461659087345147,
      "grad_norm": 0.9384128451347351,
      "learning_rate": 2.5866432223381627e-05,
      "loss": 0.2579,
      "step": 1349
    },
    {
      "epoch": 0.8467931629292771,
      "grad_norm": 0.9889576435089111,
      "learning_rate": 2.5847002251653953e-05,
      "loss": 0.1964,
      "step": 1350
    },
    {
      "epoch": 0.8474204171240395,
      "grad_norm": 0.9557651877403259,
      "learning_rate": 2.5827566243860017e-05,
      "loss": 0.2596,
      "step": 1351
    },
    {
      "epoch": 0.848047671318802,
      "grad_norm": 0.9052563309669495,
      "learning_rate": 2.580812422006429e-05,
      "loss": 0.2287,
      "step": 1352
    },
    {
      "epoch": 0.8486749255135644,
      "grad_norm": 0.8695682287216187,
      "learning_rate": 2.578867620033747e-05,
      "loss": 0.215,
      "step": 1353
    },
    {
      "epoch": 0.8493021797083268,
      "grad_norm": 0.9117851853370667,
      "learning_rate": 2.5769222204756426e-05,
      "loss": 0.2254,
      "step": 1354
    },
    {
      "epoch": 0.8499294339030893,
      "grad_norm": 0.98007732629776,
      "learning_rate": 2.57497622534042e-05,
      "loss": 0.2862,
      "step": 1355
    },
    {
      "epoch": 0.8505566880978517,
      "grad_norm": 1.0553463697433472,
      "learning_rate": 2.573029636636998e-05,
      "loss": 0.2336,
      "step": 1356
    },
    {
      "epoch": 0.8511839422926141,
      "grad_norm": 0.931348443031311,
      "learning_rate": 2.5710824563749097e-05,
      "loss": 0.2471,
      "step": 1357
    },
    {
      "epoch": 0.8518111964873765,
      "grad_norm": 0.9068348407745361,
      "learning_rate": 2.569134686564297e-05,
      "loss": 0.2348,
      "step": 1358
    },
    {
      "epoch": 0.852438450682139,
      "grad_norm": 0.9847040176391602,
      "learning_rate": 2.5671863292159122e-05,
      "loss": 0.2385,
      "step": 1359
    },
    {
      "epoch": 0.8530657048769014,
      "grad_norm": 0.7689850926399231,
      "learning_rate": 2.5652373863411124e-05,
      "loss": 0.2093,
      "step": 1360
    },
    {
      "epoch": 0.8536929590716638,
      "grad_norm": 0.9209250211715698,
      "learning_rate": 2.56328785995186e-05,
      "loss": 0.2745,
      "step": 1361
    },
    {
      "epoch": 0.8543202132664263,
      "grad_norm": 1.0016462802886963,
      "learning_rate": 2.56133775206072e-05,
      "loss": 0.217,
      "step": 1362
    },
    {
      "epoch": 0.8549474674611887,
      "grad_norm": 0.8809705972671509,
      "learning_rate": 2.5593870646808575e-05,
      "loss": 0.226,
      "step": 1363
    },
    {
      "epoch": 0.855574721655951,
      "grad_norm": 0.8996021151542664,
      "learning_rate": 2.5574357998260352e-05,
      "loss": 0.2319,
      "step": 1364
    },
    {
      "epoch": 0.8562019758507134,
      "grad_norm": 0.8426946401596069,
      "learning_rate": 2.5554839595106136e-05,
      "loss": 0.2236,
      "step": 1365
    },
    {
      "epoch": 0.856829230045476,
      "grad_norm": 0.8030539751052856,
      "learning_rate": 2.5535315457495455e-05,
      "loss": 0.221,
      "step": 1366
    },
    {
      "epoch": 0.8574564842402383,
      "grad_norm": 0.9328519701957703,
      "learning_rate": 2.5515785605583752e-05,
      "loss": 0.2569,
      "step": 1367
    },
    {
      "epoch": 0.8580837384350007,
      "grad_norm": 1.004563570022583,
      "learning_rate": 2.5496250059532398e-05,
      "loss": 0.2587,
      "step": 1368
    },
    {
      "epoch": 0.8587109926297632,
      "grad_norm": 0.8493061661720276,
      "learning_rate": 2.5476708839508616e-05,
      "loss": 0.2118,
      "step": 1369
    },
    {
      "epoch": 0.8593382468245256,
      "grad_norm": 0.9640793204307556,
      "learning_rate": 2.5457161965685503e-05,
      "loss": 0.2513,
      "step": 1370
    },
    {
      "epoch": 0.859965501019288,
      "grad_norm": 0.9193623065948486,
      "learning_rate": 2.543760945824197e-05,
      "loss": 0.2017,
      "step": 1371
    },
    {
      "epoch": 0.8605927552140505,
      "grad_norm": 0.9694982171058655,
      "learning_rate": 2.541805133736278e-05,
      "loss": 0.2349,
      "step": 1372
    },
    {
      "epoch": 0.8612200094088129,
      "grad_norm": 0.9842789769172668,
      "learning_rate": 2.5398487623238452e-05,
      "loss": 0.2396,
      "step": 1373
    },
    {
      "epoch": 0.8618472636035753,
      "grad_norm": 1.1239285469055176,
      "learning_rate": 2.5378918336065307e-05,
      "loss": 0.2265,
      "step": 1374
    },
    {
      "epoch": 0.8624745177983377,
      "grad_norm": 0.8942157626152039,
      "learning_rate": 2.53593434960454e-05,
      "loss": 0.2217,
      "step": 1375
    },
    {
      "epoch": 0.8631017719931002,
      "grad_norm": 0.8285519480705261,
      "learning_rate": 2.5339763123386537e-05,
      "loss": 0.2228,
      "step": 1376
    },
    {
      "epoch": 0.8637290261878626,
      "grad_norm": 0.9027263522148132,
      "learning_rate": 2.5320177238302222e-05,
      "loss": 0.2288,
      "step": 1377
    },
    {
      "epoch": 0.864356280382625,
      "grad_norm": 0.8536057472229004,
      "learning_rate": 2.5300585861011656e-05,
      "loss": 0.2015,
      "step": 1378
    },
    {
      "epoch": 0.8649835345773875,
      "grad_norm": 1.072025179862976,
      "learning_rate": 2.5280989011739698e-05,
      "loss": 0.2264,
      "step": 1379
    },
    {
      "epoch": 0.8656107887721499,
      "grad_norm": 0.979217529296875,
      "learning_rate": 2.5261386710716872e-05,
      "loss": 0.2375,
      "step": 1380
    },
    {
      "epoch": 0.8662380429669123,
      "grad_norm": 0.9342636466026306,
      "learning_rate": 2.524177897817933e-05,
      "loss": 0.2489,
      "step": 1381
    },
    {
      "epoch": 0.8668652971616748,
      "grad_norm": 0.9185689091682434,
      "learning_rate": 2.522216583436881e-05,
      "loss": 0.2125,
      "step": 1382
    },
    {
      "epoch": 0.8674925513564372,
      "grad_norm": 0.8563762903213501,
      "learning_rate": 2.520254729953267e-05,
      "loss": 0.2628,
      "step": 1383
    },
    {
      "epoch": 0.8681198055511996,
      "grad_norm": 0.897680938243866,
      "learning_rate": 2.518292339392379e-05,
      "loss": 0.1953,
      "step": 1384
    },
    {
      "epoch": 0.868747059745962,
      "grad_norm": 0.8742828965187073,
      "learning_rate": 2.516329413780064e-05,
      "loss": 0.2075,
      "step": 1385
    },
    {
      "epoch": 0.8693743139407245,
      "grad_norm": 0.8942375779151917,
      "learning_rate": 2.5143659551427186e-05,
      "loss": 0.2214,
      "step": 1386
    },
    {
      "epoch": 0.8700015681354869,
      "grad_norm": 1.0169836282730103,
      "learning_rate": 2.5124019655072897e-05,
      "loss": 0.2487,
      "step": 1387
    },
    {
      "epoch": 0.8706288223302493,
      "grad_norm": 0.8696025013923645,
      "learning_rate": 2.5104374469012726e-05,
      "loss": 0.2398,
      "step": 1388
    },
    {
      "epoch": 0.8712560765250118,
      "grad_norm": 0.8479281067848206,
      "learning_rate": 2.5084724013527106e-05,
      "loss": 0.2114,
      "step": 1389
    },
    {
      "epoch": 0.8718833307197742,
      "grad_norm": 0.8906635046005249,
      "learning_rate": 2.506506830890188e-05,
      "loss": 0.1894,
      "step": 1390
    },
    {
      "epoch": 0.8725105849145366,
      "grad_norm": 0.8798310160636902,
      "learning_rate": 2.5045407375428338e-05,
      "loss": 0.237,
      "step": 1391
    },
    {
      "epoch": 0.8731378391092991,
      "grad_norm": 1.0019210577011108,
      "learning_rate": 2.5025741233403143e-05,
      "loss": 0.2334,
      "step": 1392
    },
    {
      "epoch": 0.8737650933040615,
      "grad_norm": 0.9461895227432251,
      "learning_rate": 2.500606990312834e-05,
      "loss": 0.2196,
      "step": 1393
    },
    {
      "epoch": 0.8743923474988239,
      "grad_norm": 0.9007999300956726,
      "learning_rate": 2.498639340491135e-05,
      "loss": 0.216,
      "step": 1394
    },
    {
      "epoch": 0.8750196016935863,
      "grad_norm": 0.9154345393180847,
      "learning_rate": 2.4966711759064906e-05,
      "loss": 0.2569,
      "step": 1395
    },
    {
      "epoch": 0.8756468558883488,
      "grad_norm": 0.8156167268753052,
      "learning_rate": 2.494702498590706e-05,
      "loss": 0.2341,
      "step": 1396
    },
    {
      "epoch": 0.8762741100831112,
      "grad_norm": 0.8945419192314148,
      "learning_rate": 2.4927333105761172e-05,
      "loss": 0.2173,
      "step": 1397
    },
    {
      "epoch": 0.8769013642778736,
      "grad_norm": 0.8813133835792542,
      "learning_rate": 2.490763613895586e-05,
      "loss": 0.2243,
      "step": 1398
    },
    {
      "epoch": 0.8775286184726361,
      "grad_norm": 0.9496026635169983,
      "learning_rate": 2.4887934105824992e-05,
      "loss": 0.2178,
      "step": 1399
    },
    {
      "epoch": 0.8781558726673985,
      "grad_norm": 1.0122556686401367,
      "learning_rate": 2.4868227026707668e-05,
      "loss": 0.2393,
      "step": 1400
    },
    {
      "epoch": 0.8787831268621609,
      "grad_norm": 0.9015643000602722,
      "learning_rate": 2.484851492194821e-05,
      "loss": 0.2274,
      "step": 1401
    },
    {
      "epoch": 0.8794103810569234,
      "grad_norm": 0.8066772222518921,
      "learning_rate": 2.482879781189611e-05,
      "loss": 0.1984,
      "step": 1402
    },
    {
      "epoch": 0.8800376352516858,
      "grad_norm": 1.0243488550186157,
      "learning_rate": 2.4809075716906038e-05,
      "loss": 0.2265,
      "step": 1403
    },
    {
      "epoch": 0.8806648894464482,
      "grad_norm": 0.9530049562454224,
      "learning_rate": 2.478934865733781e-05,
      "loss": 0.2383,
      "step": 1404
    },
    {
      "epoch": 0.8812921436412106,
      "grad_norm": 0.8689175844192505,
      "learning_rate": 2.476961665355635e-05,
      "loss": 0.2527,
      "step": 1405
    },
    {
      "epoch": 0.8819193978359731,
      "grad_norm": 0.8891976475715637,
      "learning_rate": 2.4749879725931716e-05,
      "loss": 0.2017,
      "step": 1406
    },
    {
      "epoch": 0.8825466520307355,
      "grad_norm": 0.9147886633872986,
      "learning_rate": 2.4730137894839023e-05,
      "loss": 0.2672,
      "step": 1407
    },
    {
      "epoch": 0.8831739062254979,
      "grad_norm": 1.0061230659484863,
      "learning_rate": 2.4710391180658465e-05,
      "loss": 0.2183,
      "step": 1408
    },
    {
      "epoch": 0.8838011604202604,
      "grad_norm": 0.9156420230865479,
      "learning_rate": 2.4690639603775263e-05,
      "loss": 0.2306,
      "step": 1409
    },
    {
      "epoch": 0.8844284146150228,
      "grad_norm": 0.9554071426391602,
      "learning_rate": 2.467088318457968e-05,
      "loss": 0.221,
      "step": 1410
    },
    {
      "epoch": 0.8850556688097851,
      "grad_norm": 1.0293577909469604,
      "learning_rate": 2.4651121943466946e-05,
      "loss": 0.2522,
      "step": 1411
    },
    {
      "epoch": 0.8856829230045475,
      "grad_norm": 0.8481536507606506,
      "learning_rate": 2.463135590083729e-05,
      "loss": 0.1971,
      "step": 1412
    },
    {
      "epoch": 0.88631017719931,
      "grad_norm": 1.4141706228256226,
      "learning_rate": 2.4611585077095898e-05,
      "loss": 0.2494,
      "step": 1413
    },
    {
      "epoch": 0.8869374313940724,
      "grad_norm": 0.9040833711624146,
      "learning_rate": 2.4591809492652887e-05,
      "loss": 0.2355,
      "step": 1414
    },
    {
      "epoch": 0.8875646855888348,
      "grad_norm": 0.8359858989715576,
      "learning_rate": 2.457202916792329e-05,
      "loss": 0.2351,
      "step": 1415
    },
    {
      "epoch": 0.8881919397835973,
      "grad_norm": 0.9249632358551025,
      "learning_rate": 2.4552244123327024e-05,
      "loss": 0.2177,
      "step": 1416
    },
    {
      "epoch": 0.8888191939783597,
      "grad_norm": 0.8339369297027588,
      "learning_rate": 2.4532454379288896e-05,
      "loss": 0.2255,
      "step": 1417
    },
    {
      "epoch": 0.8894464481731221,
      "grad_norm": 0.9239262938499451,
      "learning_rate": 2.4512659956238555e-05,
      "loss": 0.2406,
      "step": 1418
    },
    {
      "epoch": 0.8900737023678846,
      "grad_norm": 0.8235772848129272,
      "learning_rate": 2.4492860874610478e-05,
      "loss": 0.2202,
      "step": 1419
    },
    {
      "epoch": 0.890700956562647,
      "grad_norm": 0.7412683367729187,
      "learning_rate": 2.4473057154843952e-05,
      "loss": 0.2467,
      "step": 1420
    },
    {
      "epoch": 0.8913282107574094,
      "grad_norm": 0.8566434383392334,
      "learning_rate": 2.445324881738305e-05,
      "loss": 0.2481,
      "step": 1421
    },
    {
      "epoch": 0.8919554649521718,
      "grad_norm": 0.9057849645614624,
      "learning_rate": 2.4433435882676634e-05,
      "loss": 0.2426,
      "step": 1422
    },
    {
      "epoch": 0.8925827191469343,
      "grad_norm": 0.9396740794181824,
      "learning_rate": 2.4413618371178273e-05,
      "loss": 0.2753,
      "step": 1423
    },
    {
      "epoch": 0.8932099733416967,
      "grad_norm": 0.8927632570266724,
      "learning_rate": 2.4393796303346303e-05,
      "loss": 0.2286,
      "step": 1424
    },
    {
      "epoch": 0.8938372275364591,
      "grad_norm": 1.01885986328125,
      "learning_rate": 2.4373969699643718e-05,
      "loss": 0.2325,
      "step": 1425
    },
    {
      "epoch": 0.8944644817312216,
      "grad_norm": 0.86164790391922,
      "learning_rate": 2.4354138580538237e-05,
      "loss": 0.2242,
      "step": 1426
    },
    {
      "epoch": 0.895091735925984,
      "grad_norm": 0.8852266669273376,
      "learning_rate": 2.433430296650222e-05,
      "loss": 0.2094,
      "step": 1427
    },
    {
      "epoch": 0.8957189901207464,
      "grad_norm": 0.9729646444320679,
      "learning_rate": 2.4314462878012664e-05,
      "loss": 0.2435,
      "step": 1428
    },
    {
      "epoch": 0.8963462443155089,
      "grad_norm": 0.9526697993278503,
      "learning_rate": 2.429461833555119e-05,
      "loss": 0.2385,
      "step": 1429
    },
    {
      "epoch": 0.8969734985102713,
      "grad_norm": 0.9755507707595825,
      "learning_rate": 2.4274769359604028e-05,
      "loss": 0.2457,
      "step": 1430
    },
    {
      "epoch": 0.8976007527050337,
      "grad_norm": 0.9439208507537842,
      "learning_rate": 2.4254915970661968e-05,
      "loss": 0.2438,
      "step": 1431
    },
    {
      "epoch": 0.8982280068997961,
      "grad_norm": 0.9157652854919434,
      "learning_rate": 2.4235058189220353e-05,
      "loss": 0.2378,
      "step": 1432
    },
    {
      "epoch": 0.8988552610945586,
      "grad_norm": 0.8972637057304382,
      "learning_rate": 2.4215196035779086e-05,
      "loss": 0.1998,
      "step": 1433
    },
    {
      "epoch": 0.899482515289321,
      "grad_norm": 0.7806968092918396,
      "learning_rate": 2.419532953084256e-05,
      "loss": 0.2411,
      "step": 1434
    },
    {
      "epoch": 0.9001097694840834,
      "grad_norm": 0.9046010971069336,
      "learning_rate": 2.417545869491966e-05,
      "loss": 0.2189,
      "step": 1435
    },
    {
      "epoch": 0.9007370236788459,
      "grad_norm": 0.9269365072250366,
      "learning_rate": 2.4155583548523768e-05,
      "loss": 0.2323,
      "step": 1436
    },
    {
      "epoch": 0.9013642778736083,
      "grad_norm": 0.9623416662216187,
      "learning_rate": 2.4135704112172683e-05,
      "loss": 0.2403,
      "step": 1437
    },
    {
      "epoch": 0.9019915320683707,
      "grad_norm": 0.9461166858673096,
      "learning_rate": 2.4115820406388646e-05,
      "loss": 0.2322,
      "step": 1438
    },
    {
      "epoch": 0.9026187862631332,
      "grad_norm": 0.8908488750457764,
      "learning_rate": 2.409593245169832e-05,
      "loss": 0.2394,
      "step": 1439
    },
    {
      "epoch": 0.9032460404578956,
      "grad_norm": 1.0029852390289307,
      "learning_rate": 2.4076040268632722e-05,
      "loss": 0.2839,
      "step": 1440
    },
    {
      "epoch": 0.903873294652658,
      "grad_norm": 0.8697084188461304,
      "learning_rate": 2.4056143877727266e-05,
      "loss": 0.2166,
      "step": 1441
    },
    {
      "epoch": 0.9045005488474204,
      "grad_norm": 0.8713654279708862,
      "learning_rate": 2.4036243299521695e-05,
      "loss": 0.1948,
      "step": 1442
    },
    {
      "epoch": 0.9051278030421829,
      "grad_norm": 0.8880418539047241,
      "learning_rate": 2.401633855456008e-05,
      "loss": 0.2412,
      "step": 1443
    },
    {
      "epoch": 0.9057550572369453,
      "grad_norm": 0.810232400894165,
      "learning_rate": 2.3996429663390775e-05,
      "loss": 0.2468,
      "step": 1444
    },
    {
      "epoch": 0.9063823114317077,
      "grad_norm": 0.9070370197296143,
      "learning_rate": 2.3976516646566443e-05,
      "loss": 0.215,
      "step": 1445
    },
    {
      "epoch": 0.9070095656264702,
      "grad_norm": 1.0040373802185059,
      "learning_rate": 2.3956599524643985e-05,
      "loss": 0.2207,
      "step": 1446
    },
    {
      "epoch": 0.9076368198212326,
      "grad_norm": 0.7925657033920288,
      "learning_rate": 2.393667831818455e-05,
      "loss": 0.2389,
      "step": 1447
    },
    {
      "epoch": 0.908264074015995,
      "grad_norm": 0.9810481667518616,
      "learning_rate": 2.3916753047753508e-05,
      "loss": 0.2502,
      "step": 1448
    },
    {
      "epoch": 0.9088913282107574,
      "grad_norm": 0.8342772722244263,
      "learning_rate": 2.3896823733920406e-05,
      "loss": 0.2314,
      "step": 1449
    },
    {
      "epoch": 0.9095185824055199,
      "grad_norm": 0.9573843479156494,
      "learning_rate": 2.387689039725898e-05,
      "loss": 0.2215,
      "step": 1450
    },
    {
      "epoch": 0.9101458366002823,
      "grad_norm": 1.0381935834884644,
      "learning_rate": 2.3856953058347116e-05,
      "loss": 0.2319,
      "step": 1451
    },
    {
      "epoch": 0.9107730907950446,
      "grad_norm": 0.8801733255386353,
      "learning_rate": 2.3837011737766828e-05,
      "loss": 0.18,
      "step": 1452
    },
    {
      "epoch": 0.9114003449898072,
      "grad_norm": 0.8234224915504456,
      "learning_rate": 2.3817066456104253e-05,
      "loss": 0.2063,
      "step": 1453
    },
    {
      "epoch": 0.9120275991845695,
      "grad_norm": 0.828215479850769,
      "learning_rate": 2.379711723394959e-05,
      "loss": 0.1751,
      "step": 1454
    },
    {
      "epoch": 0.9126548533793319,
      "grad_norm": 1.0685749053955078,
      "learning_rate": 2.3777164091897137e-05,
      "loss": 0.2159,
      "step": 1455
    },
    {
      "epoch": 0.9132821075740944,
      "grad_norm": 0.9756271243095398,
      "learning_rate": 2.3757207050545227e-05,
      "loss": 0.2366,
      "step": 1456
    },
    {
      "epoch": 0.9139093617688568,
      "grad_norm": 0.844516396522522,
      "learning_rate": 2.3737246130496206e-05,
      "loss": 0.2433,
      "step": 1457
    },
    {
      "epoch": 0.9145366159636192,
      "grad_norm": 0.8922362327575684,
      "learning_rate": 2.3717281352356438e-05,
      "loss": 0.2433,
      "step": 1458
    },
    {
      "epoch": 0.9151638701583816,
      "grad_norm": 1.1780900955200195,
      "learning_rate": 2.3697312736736268e-05,
      "loss": 0.2332,
      "step": 1459
    },
    {
      "epoch": 0.9157911243531441,
      "grad_norm": 0.9307076334953308,
      "learning_rate": 2.3677340304250003e-05,
      "loss": 0.2548,
      "step": 1460
    },
    {
      "epoch": 0.9164183785479065,
      "grad_norm": 1.1400376558303833,
      "learning_rate": 2.3657364075515884e-05,
      "loss": 0.2158,
      "step": 1461
    },
    {
      "epoch": 0.9170456327426689,
      "grad_norm": 1.0676137208938599,
      "learning_rate": 2.3637384071156078e-05,
      "loss": 0.2487,
      "step": 1462
    },
    {
      "epoch": 0.9176728869374314,
      "grad_norm": 0.9103914499282837,
      "learning_rate": 2.361740031179664e-05,
      "loss": 0.2311,
      "step": 1463
    },
    {
      "epoch": 0.9183001411321938,
      "grad_norm": 0.9785495400428772,
      "learning_rate": 2.359741281806752e-05,
      "loss": 0.2337,
      "step": 1464
    },
    {
      "epoch": 0.9189273953269562,
      "grad_norm": 0.9982824921607971,
      "learning_rate": 2.3577421610602507e-05,
      "loss": 0.2136,
      "step": 1465
    },
    {
      "epoch": 0.9195546495217187,
      "grad_norm": 0.9072036743164062,
      "learning_rate": 2.355742671003922e-05,
      "loss": 0.2117,
      "step": 1466
    },
    {
      "epoch": 0.9201819037164811,
      "grad_norm": 0.8125027418136597,
      "learning_rate": 2.3537428137019105e-05,
      "loss": 0.2219,
      "step": 1467
    },
    {
      "epoch": 0.9208091579112435,
      "grad_norm": 0.9591137170791626,
      "learning_rate": 2.3517425912187397e-05,
      "loss": 0.2372,
      "step": 1468
    },
    {
      "epoch": 0.9214364121060059,
      "grad_norm": 0.9125718474388123,
      "learning_rate": 2.3497420056193095e-05,
      "loss": 0.2641,
      "step": 1469
    },
    {
      "epoch": 0.9220636663007684,
      "grad_norm": 0.8828108906745911,
      "learning_rate": 2.347741058968894e-05,
      "loss": 0.2189,
      "step": 1470
    },
    {
      "epoch": 0.9226909204955308,
      "grad_norm": 0.8246212601661682,
      "learning_rate": 2.3457397533331424e-05,
      "loss": 0.2316,
      "step": 1471
    },
    {
      "epoch": 0.9233181746902932,
      "grad_norm": 0.8619717955589294,
      "learning_rate": 2.3437380907780713e-05,
      "loss": 0.2064,
      "step": 1472
    },
    {
      "epoch": 0.9239454288850557,
      "grad_norm": 0.9750192165374756,
      "learning_rate": 2.3417360733700685e-05,
      "loss": 0.2424,
      "step": 1473
    },
    {
      "epoch": 0.9245726830798181,
      "grad_norm": 0.9998435974121094,
      "learning_rate": 2.339733703175887e-05,
      "loss": 0.2377,
      "step": 1474
    },
    {
      "epoch": 0.9251999372745805,
      "grad_norm": 0.8929082751274109,
      "learning_rate": 2.337730982262643e-05,
      "loss": 0.2397,
      "step": 1475
    },
    {
      "epoch": 0.925827191469343,
      "grad_norm": 0.7877259254455566,
      "learning_rate": 2.335727912697817e-05,
      "loss": 0.2208,
      "step": 1476
    },
    {
      "epoch": 0.9264544456641054,
      "grad_norm": 0.8633270263671875,
      "learning_rate": 2.333724496549248e-05,
      "loss": 0.1959,
      "step": 1477
    },
    {
      "epoch": 0.9270816998588678,
      "grad_norm": 0.9239407777786255,
      "learning_rate": 2.3317207358851322e-05,
      "loss": 0.2076,
      "step": 1478
    },
    {
      "epoch": 0.9277089540536302,
      "grad_norm": 0.8067747950553894,
      "learning_rate": 2.3297166327740234e-05,
      "loss": 0.195,
      "step": 1479
    },
    {
      "epoch": 0.9283362082483927,
      "grad_norm": 0.9614524841308594,
      "learning_rate": 2.327712189284828e-05,
      "loss": 0.267,
      "step": 1480
    },
    {
      "epoch": 0.9289634624431551,
      "grad_norm": 0.6122102737426758,
      "learning_rate": 2.3257074074868023e-05,
      "loss": 0.2062,
      "step": 1481
    },
    {
      "epoch": 0.9295907166379175,
      "grad_norm": 0.8468174338340759,
      "learning_rate": 2.3237022894495532e-05,
      "loss": 0.2433,
      "step": 1482
    },
    {
      "epoch": 0.93021797083268,
      "grad_norm": 0.9263379573822021,
      "learning_rate": 2.3216968372430362e-05,
      "loss": 0.192,
      "step": 1483
    },
    {
      "epoch": 0.9308452250274424,
      "grad_norm": 0.8937593102455139,
      "learning_rate": 2.319691052937548e-05,
      "loss": 0.2115,
      "step": 1484
    },
    {
      "epoch": 0.9314724792222048,
      "grad_norm": 0.8346992135047913,
      "learning_rate": 2.317684938603732e-05,
      "loss": 0.1919,
      "step": 1485
    },
    {
      "epoch": 0.9320997334169673,
      "grad_norm": 1.161743402481079,
      "learning_rate": 2.3156784963125705e-05,
      "loss": 0.2597,
      "step": 1486
    },
    {
      "epoch": 0.9327269876117297,
      "grad_norm": 0.8451173305511475,
      "learning_rate": 2.3136717281353836e-05,
      "loss": 0.2224,
      "step": 1487
    },
    {
      "epoch": 0.9333542418064921,
      "grad_norm": 0.9135767221450806,
      "learning_rate": 2.3116646361438294e-05,
      "loss": 0.2344,
      "step": 1488
    },
    {
      "epoch": 0.9339814960012545,
      "grad_norm": 0.9136664271354675,
      "learning_rate": 2.3096572224099e-05,
      "loss": 0.2214,
      "step": 1489
    },
    {
      "epoch": 0.934608750196017,
      "grad_norm": 0.8622660636901855,
      "learning_rate": 2.3076494890059184e-05,
      "loss": 0.2412,
      "step": 1490
    },
    {
      "epoch": 0.9352360043907794,
      "grad_norm": 1.0650670528411865,
      "learning_rate": 2.3056414380045388e-05,
      "loss": 0.2563,
      "step": 1491
    },
    {
      "epoch": 0.9358632585855418,
      "grad_norm": 1.0359560251235962,
      "learning_rate": 2.3036330714787432e-05,
      "loss": 0.2574,
      "step": 1492
    },
    {
      "epoch": 0.9364905127803043,
      "grad_norm": 0.8969426155090332,
      "learning_rate": 2.3016243915018386e-05,
      "loss": 0.2285,
      "step": 1493
    },
    {
      "epoch": 0.9371177669750667,
      "grad_norm": 0.7989252805709839,
      "learning_rate": 2.299615400147457e-05,
      "loss": 0.245,
      "step": 1494
    },
    {
      "epoch": 0.937745021169829,
      "grad_norm": 1.1010055541992188,
      "learning_rate": 2.2976060994895497e-05,
      "loss": 0.2686,
      "step": 1495
    },
    {
      "epoch": 0.9383722753645914,
      "grad_norm": 0.8471121788024902,
      "learning_rate": 2.295596491602389e-05,
      "loss": 0.2069,
      "step": 1496
    },
    {
      "epoch": 0.938999529559354,
      "grad_norm": 0.9189316034317017,
      "learning_rate": 2.2935865785605636e-05,
      "loss": 0.235,
      "step": 1497
    },
    {
      "epoch": 0.9396267837541163,
      "grad_norm": 0.96001797914505,
      "learning_rate": 2.2915763624389785e-05,
      "loss": 0.2087,
      "step": 1498
    },
    {
      "epoch": 0.9402540379488787,
      "grad_norm": 0.7247317433357239,
      "learning_rate": 2.289565845312849e-05,
      "loss": 0.1773,
      "step": 1499
    },
    {
      "epoch": 0.9408812921436412,
      "grad_norm": 0.7679764628410339,
      "learning_rate": 2.2875550292577043e-05,
      "loss": 0.2105,
      "step": 1500
    },
    {
      "epoch": 0.9415085463384036,
      "grad_norm": 0.8632218241691589,
      "learning_rate": 2.2855439163493796e-05,
      "loss": 0.1966,
      "step": 1501
    },
    {
      "epoch": 0.942135800533166,
      "grad_norm": 0.8614953756332397,
      "learning_rate": 2.2835325086640178e-05,
      "loss": 0.2215,
      "step": 1502
    },
    {
      "epoch": 0.9427630547279285,
      "grad_norm": 1.0454775094985962,
      "learning_rate": 2.2815208082780665e-05,
      "loss": 0.232,
      "step": 1503
    },
    {
      "epoch": 0.9433903089226909,
      "grad_norm": 0.8676419258117676,
      "learning_rate": 2.2795088172682732e-05,
      "loss": 0.2233,
      "step": 1504
    },
    {
      "epoch": 0.9440175631174533,
      "grad_norm": 1.0755561590194702,
      "learning_rate": 2.2774965377116885e-05,
      "loss": 0.2582,
      "step": 1505
    },
    {
      "epoch": 0.9446448173122157,
      "grad_norm": 0.8055601716041565,
      "learning_rate": 2.275483971685659e-05,
      "loss": 0.1712,
      "step": 1506
    },
    {
      "epoch": 0.9452720715069782,
      "grad_norm": 0.9284416437149048,
      "learning_rate": 2.273471121267828e-05,
      "loss": 0.2087,
      "step": 1507
    },
    {
      "epoch": 0.9458993257017406,
      "grad_norm": 1.0512845516204834,
      "learning_rate": 2.2714579885361304e-05,
      "loss": 0.1975,
      "step": 1508
    },
    {
      "epoch": 0.946526579896503,
      "grad_norm": 0.83321613073349,
      "learning_rate": 2.2694445755687956e-05,
      "loss": 0.2174,
      "step": 1509
    },
    {
      "epoch": 0.9471538340912655,
      "grad_norm": 0.9688567519187927,
      "learning_rate": 2.2674308844443394e-05,
      "loss": 0.2499,
      "step": 1510
    },
    {
      "epoch": 0.9477810882860279,
      "grad_norm": 0.7848716974258423,
      "learning_rate": 2.265416917241566e-05,
      "loss": 0.1968,
      "step": 1511
    },
    {
      "epoch": 0.9484083424807903,
      "grad_norm": 0.9451969861984253,
      "learning_rate": 2.2634026760395664e-05,
      "loss": 0.2262,
      "step": 1512
    },
    {
      "epoch": 0.9490355966755528,
      "grad_norm": 1.0070687532424927,
      "learning_rate": 2.2613881629177112e-05,
      "loss": 0.2543,
      "step": 1513
    },
    {
      "epoch": 0.9496628508703152,
      "grad_norm": 1.049132227897644,
      "learning_rate": 2.259373379955653e-05,
      "loss": 0.2521,
      "step": 1514
    },
    {
      "epoch": 0.9502901050650776,
      "grad_norm": 0.9367115497589111,
      "learning_rate": 2.257358329233325e-05,
      "loss": 0.1957,
      "step": 1515
    },
    {
      "epoch": 0.95091735925984,
      "grad_norm": 1.0050894021987915,
      "learning_rate": 2.2553430128309325e-05,
      "loss": 0.2179,
      "step": 1516
    },
    {
      "epoch": 0.9515446134546025,
      "grad_norm": 0.890467643737793,
      "learning_rate": 2.2533274328289593e-05,
      "loss": 0.2328,
      "step": 1517
    },
    {
      "epoch": 0.9521718676493649,
      "grad_norm": 0.8494214415550232,
      "learning_rate": 2.251311591308159e-05,
      "loss": 0.2164,
      "step": 1518
    },
    {
      "epoch": 0.9527991218441273,
      "grad_norm": 0.9842292666435242,
      "learning_rate": 2.2492954903495558e-05,
      "loss": 0.2441,
      "step": 1519
    },
    {
      "epoch": 0.9534263760388898,
      "grad_norm": 0.9736828804016113,
      "learning_rate": 2.247279132034442e-05,
      "loss": 0.2571,
      "step": 1520
    },
    {
      "epoch": 0.9540536302336522,
      "grad_norm": 0.9789125919342041,
      "learning_rate": 2.2452625184443756e-05,
      "loss": 0.2305,
      "step": 1521
    },
    {
      "epoch": 0.9546808844284146,
      "grad_norm": 0.8945363163948059,
      "learning_rate": 2.2432456516611763e-05,
      "loss": 0.2327,
      "step": 1522
    },
    {
      "epoch": 0.9553081386231771,
      "grad_norm": 1.0674930810928345,
      "learning_rate": 2.2412285337669272e-05,
      "loss": 0.2552,
      "step": 1523
    },
    {
      "epoch": 0.9559353928179395,
      "grad_norm": 0.9890167117118835,
      "learning_rate": 2.239211166843972e-05,
      "loss": 0.2399,
      "step": 1524
    },
    {
      "epoch": 0.9565626470127019,
      "grad_norm": 0.8804594278335571,
      "learning_rate": 2.237193552974907e-05,
      "loss": 0.2687,
      "step": 1525
    },
    {
      "epoch": 0.9571899012074643,
      "grad_norm": 0.939110279083252,
      "learning_rate": 2.235175694242587e-05,
      "loss": 0.1726,
      "step": 1526
    },
    {
      "epoch": 0.9578171554022268,
      "grad_norm": 0.8611268401145935,
      "learning_rate": 2.2331575927301198e-05,
      "loss": 0.2008,
      "step": 1527
    },
    {
      "epoch": 0.9584444095969892,
      "grad_norm": 0.9728996753692627,
      "learning_rate": 2.231139250520861e-05,
      "loss": 0.2279,
      "step": 1528
    },
    {
      "epoch": 0.9590716637917516,
      "grad_norm": 0.9554197788238525,
      "learning_rate": 2.2291206696984165e-05,
      "loss": 0.2452,
      "step": 1529
    },
    {
      "epoch": 0.9596989179865141,
      "grad_norm": 0.8852481245994568,
      "learning_rate": 2.2271018523466395e-05,
      "loss": 0.221,
      "step": 1530
    },
    {
      "epoch": 0.9603261721812765,
      "grad_norm": 0.8912515044212341,
      "learning_rate": 2.2250828005496245e-05,
      "loss": 0.244,
      "step": 1531
    },
    {
      "epoch": 0.9609534263760389,
      "grad_norm": 0.7839436531066895,
      "learning_rate": 2.223063516391712e-05,
      "loss": 0.2125,
      "step": 1532
    },
    {
      "epoch": 0.9615806805708014,
      "grad_norm": 0.8890122175216675,
      "learning_rate": 2.2210440019574785e-05,
      "loss": 0.2429,
      "step": 1533
    },
    {
      "epoch": 0.9622079347655638,
      "grad_norm": 0.9256410002708435,
      "learning_rate": 2.2190242593317396e-05,
      "loss": 0.2341,
      "step": 1534
    },
    {
      "epoch": 0.9628351889603262,
      "grad_norm": 0.9652786254882812,
      "learning_rate": 2.2170042905995474e-05,
      "loss": 0.2379,
      "step": 1535
    },
    {
      "epoch": 0.9634624431550886,
      "grad_norm": 1.1375311613082886,
      "learning_rate": 2.2149840978461868e-05,
      "loss": 0.2318,
      "step": 1536
    },
    {
      "epoch": 0.9640896973498511,
      "grad_norm": 0.9609313607215881,
      "learning_rate": 2.212963683157174e-05,
      "loss": 0.2056,
      "step": 1537
    },
    {
      "epoch": 0.9647169515446135,
      "grad_norm": 0.7776309847831726,
      "learning_rate": 2.210943048618253e-05,
      "loss": 0.1841,
      "step": 1538
    },
    {
      "epoch": 0.9653442057393758,
      "grad_norm": 0.8857976198196411,
      "learning_rate": 2.208922196315398e-05,
      "loss": 0.1961,
      "step": 1539
    },
    {
      "epoch": 0.9659714599341384,
      "grad_norm": 0.9588800668716431,
      "learning_rate": 2.2069011283348044e-05,
      "loss": 0.2297,
      "step": 1540
    },
    {
      "epoch": 0.9665987141289007,
      "grad_norm": 0.9649481177330017,
      "learning_rate": 2.2048798467628926e-05,
      "loss": 0.2109,
      "step": 1541
    },
    {
      "epoch": 0.9672259683236631,
      "grad_norm": 1.0013532638549805,
      "learning_rate": 2.2028583536863022e-05,
      "loss": 0.2119,
      "step": 1542
    },
    {
      "epoch": 0.9678532225184255,
      "grad_norm": 1.0082463026046753,
      "learning_rate": 2.2008366511918915e-05,
      "loss": 0.2574,
      "step": 1543
    },
    {
      "epoch": 0.968480476713188,
      "grad_norm": 0.9393874406814575,
      "learning_rate": 2.1988147413667363e-05,
      "loss": 0.2669,
      "step": 1544
    },
    {
      "epoch": 0.9691077309079504,
      "grad_norm": 0.8781605362892151,
      "learning_rate": 2.196792626298124e-05,
      "loss": 0.2055,
      "step": 1545
    },
    {
      "epoch": 0.9697349851027128,
      "grad_norm": 0.9629719853401184,
      "learning_rate": 2.1947703080735564e-05,
      "loss": 0.19,
      "step": 1546
    },
    {
      "epoch": 0.9703622392974753,
      "grad_norm": 0.9251545667648315,
      "learning_rate": 2.1927477887807426e-05,
      "loss": 0.2356,
      "step": 1547
    },
    {
      "epoch": 0.9709894934922377,
      "grad_norm": 0.9196063876152039,
      "learning_rate": 2.190725070507601e-05,
      "loss": 0.2106,
      "step": 1548
    },
    {
      "epoch": 0.9716167476870001,
      "grad_norm": 0.9102229475975037,
      "learning_rate": 2.188702155342255e-05,
      "loss": 0.2273,
      "step": 1549
    },
    {
      "epoch": 0.9722440018817626,
      "grad_norm": 0.8885928392410278,
      "learning_rate": 2.186679045373031e-05,
      "loss": 0.2346,
      "step": 1550
    },
    {
      "epoch": 0.972871256076525,
      "grad_norm": 0.8653280138969421,
      "learning_rate": 2.1846557426884566e-05,
      "loss": 0.199,
      "step": 1551
    },
    {
      "epoch": 0.9734985102712874,
      "grad_norm": 0.9529988169670105,
      "learning_rate": 2.182632249377258e-05,
      "loss": 0.2339,
      "step": 1552
    },
    {
      "epoch": 0.9741257644660498,
      "grad_norm": 0.9183457493782043,
      "learning_rate": 2.18060856752836e-05,
      "loss": 0.2289,
      "step": 1553
    },
    {
      "epoch": 0.9747530186608123,
      "grad_norm": 0.7829636335372925,
      "learning_rate": 2.1785846992308784e-05,
      "loss": 0.2589,
      "step": 1554
    },
    {
      "epoch": 0.9753802728555747,
      "grad_norm": 0.893418550491333,
      "learning_rate": 2.1765606465741253e-05,
      "loss": 0.2363,
      "step": 1555
    },
    {
      "epoch": 0.9760075270503371,
      "grad_norm": 0.7944826483726501,
      "learning_rate": 2.1745364116476017e-05,
      "loss": 0.1994,
      "step": 1556
    },
    {
      "epoch": 0.9766347812450996,
      "grad_norm": 0.9511778354644775,
      "learning_rate": 2.1725119965409957e-05,
      "loss": 0.2518,
      "step": 1557
    },
    {
      "epoch": 0.977262035439862,
      "grad_norm": 0.6436446905136108,
      "learning_rate": 2.170487403344182e-05,
      "loss": 0.243,
      "step": 1558
    },
    {
      "epoch": 0.9778892896346244,
      "grad_norm": 0.8381584882736206,
      "learning_rate": 2.168462634147222e-05,
      "loss": 0.2275,
      "step": 1559
    },
    {
      "epoch": 0.9785165438293869,
      "grad_norm": 0.852128803730011,
      "learning_rate": 2.1664376910403524e-05,
      "loss": 0.2099,
      "step": 1560
    },
    {
      "epoch": 0.9791437980241493,
      "grad_norm": 0.9113161563873291,
      "learning_rate": 2.1644125761139954e-05,
      "loss": 0.2269,
      "step": 1561
    },
    {
      "epoch": 0.9797710522189117,
      "grad_norm": 0.8905954957008362,
      "learning_rate": 2.1623872914587486e-05,
      "loss": 0.2655,
      "step": 1562
    },
    {
      "epoch": 0.9803983064136741,
      "grad_norm": 0.9145949482917786,
      "learning_rate": 2.1603618391653835e-05,
      "loss": 0.1956,
      "step": 1563
    },
    {
      "epoch": 0.9810255606084366,
      "grad_norm": 0.813274621963501,
      "learning_rate": 2.158336221324847e-05,
      "loss": 0.2186,
      "step": 1564
    },
    {
      "epoch": 0.981652814803199,
      "grad_norm": 1.031652569770813,
      "learning_rate": 2.1563104400282554e-05,
      "loss": 0.2218,
      "step": 1565
    },
    {
      "epoch": 0.9822800689979614,
      "grad_norm": 0.8783631324768066,
      "learning_rate": 2.1542844973668927e-05,
      "loss": 0.2237,
      "step": 1566
    },
    {
      "epoch": 0.9829073231927239,
      "grad_norm": 0.9718444347381592,
      "learning_rate": 2.1522583954322128e-05,
      "loss": 0.2356,
      "step": 1567
    },
    {
      "epoch": 0.9835345773874863,
      "grad_norm": 0.7708685398101807,
      "learning_rate": 2.1502321363158315e-05,
      "loss": 0.1831,
      "step": 1568
    },
    {
      "epoch": 0.9841618315822487,
      "grad_norm": 0.8163751363754272,
      "learning_rate": 2.1482057221095265e-05,
      "loss": 0.1993,
      "step": 1569
    },
    {
      "epoch": 0.9847890857770112,
      "grad_norm": 0.8872279524803162,
      "learning_rate": 2.1461791549052374e-05,
      "loss": 0.239,
      "step": 1570
    },
    {
      "epoch": 0.9854163399717736,
      "grad_norm": 0.8377917408943176,
      "learning_rate": 2.1441524367950608e-05,
      "loss": 0.2046,
      "step": 1571
    },
    {
      "epoch": 0.986043594166536,
      "grad_norm": 0.8231236338615417,
      "learning_rate": 2.1421255698712492e-05,
      "loss": 0.2061,
      "step": 1572
    },
    {
      "epoch": 0.9866708483612984,
      "grad_norm": 0.7960736751556396,
      "learning_rate": 2.140098556226208e-05,
      "loss": 0.1915,
      "step": 1573
    },
    {
      "epoch": 0.9872981025560609,
      "grad_norm": 0.932736337184906,
      "learning_rate": 2.1380713979524956e-05,
      "loss": 0.2073,
      "step": 1574
    },
    {
      "epoch": 0.9879253567508233,
      "grad_norm": 0.8400558233261108,
      "learning_rate": 2.136044097142818e-05,
      "loss": 0.1567,
      "step": 1575
    },
    {
      "epoch": 0.9885526109455857,
      "grad_norm": 0.912680983543396,
      "learning_rate": 2.1340166558900294e-05,
      "loss": 0.2281,
      "step": 1576
    },
    {
      "epoch": 0.9891798651403482,
      "grad_norm": 0.9152987003326416,
      "learning_rate": 2.1319890762871294e-05,
      "loss": 0.1829,
      "step": 1577
    },
    {
      "epoch": 0.9898071193351106,
      "grad_norm": 1.1444475650787354,
      "learning_rate": 2.1299613604272594e-05,
      "loss": 0.2924,
      "step": 1578
    },
    {
      "epoch": 0.990434373529873,
      "grad_norm": 0.9289518594741821,
      "learning_rate": 2.1279335104037025e-05,
      "loss": 0.2289,
      "step": 1579
    },
    {
      "epoch": 0.9910616277246355,
      "grad_norm": 0.9576685428619385,
      "learning_rate": 2.1259055283098782e-05,
      "loss": 0.2195,
      "step": 1580
    },
    {
      "epoch": 0.9916888819193979,
      "grad_norm": 0.893746554851532,
      "learning_rate": 2.123877416239345e-05,
      "loss": 0.2407,
      "step": 1581
    },
    {
      "epoch": 0.9923161361141603,
      "grad_norm": 0.8622097969055176,
      "learning_rate": 2.1218491762857943e-05,
      "loss": 0.2014,
      "step": 1582
    },
    {
      "epoch": 0.9929433903089226,
      "grad_norm": 0.8856543898582458,
      "learning_rate": 2.119820810543049e-05,
      "loss": 0.1957,
      "step": 1583
    },
    {
      "epoch": 0.9935706445036852,
      "grad_norm": 1.0095499753952026,
      "learning_rate": 2.1177923211050628e-05,
      "loss": 0.2179,
      "step": 1584
    },
    {
      "epoch": 0.9941978986984475,
      "grad_norm": 0.9774998426437378,
      "learning_rate": 2.1157637100659173e-05,
      "loss": 0.194,
      "step": 1585
    },
    {
      "epoch": 0.9948251528932099,
      "grad_norm": 0.815413773059845,
      "learning_rate": 2.1137349795198182e-05,
      "loss": 0.202,
      "step": 1586
    },
    {
      "epoch": 0.9954524070879724,
      "grad_norm": 0.8930078148841858,
      "learning_rate": 2.111706131561096e-05,
      "loss": 0.2171,
      "step": 1587
    },
    {
      "epoch": 0.9960796612827348,
      "grad_norm": 1.0153467655181885,
      "learning_rate": 2.1096771682842016e-05,
      "loss": 0.2223,
      "step": 1588
    },
    {
      "epoch": 0.9967069154774972,
      "grad_norm": 0.9196330904960632,
      "learning_rate": 2.107648091783705e-05,
      "loss": 0.2069,
      "step": 1589
    },
    {
      "epoch": 0.9973341696722596,
      "grad_norm": 0.8963049650192261,
      "learning_rate": 2.1056189041542938e-05,
      "loss": 0.224,
      "step": 1590
    },
    {
      "epoch": 0.9979614238670221,
      "grad_norm": 0.8232291340827942,
      "learning_rate": 2.10358960749077e-05,
      "loss": 0.2092,
      "step": 1591
    },
    {
      "epoch": 0.9985886780617845,
      "grad_norm": 1.0254426002502441,
      "learning_rate": 2.101560203888047e-05,
      "loss": 0.2107,
      "step": 1592
    },
    {
      "epoch": 0.9992159322565469,
      "grad_norm": 0.9355930685997009,
      "learning_rate": 2.099530695441151e-05,
      "loss": 0.2155,
      "step": 1593
    },
    {
      "epoch": 0.9998431864513094,
      "grad_norm": 0.9351606965065002,
      "learning_rate": 2.0975010842452138e-05,
      "loss": 0.2061,
      "step": 1594
    },
    {
      "epoch": 1.0004704406460718,
      "grad_norm": 0.7437397837638855,
      "learning_rate": 2.095471372395475e-05,
      "loss": 0.1526,
      "step": 1595
    },
    {
      "epoch": 1.0010976948408343,
      "grad_norm": 0.9794983863830566,
      "learning_rate": 2.0934415619872773e-05,
      "loss": 0.1749,
      "step": 1596
    },
    {
      "epoch": 1.0017249490355966,
      "grad_norm": 0.8396321535110474,
      "learning_rate": 2.0914116551160663e-05,
      "loss": 0.1517,
      "step": 1597
    },
    {
      "epoch": 1.0023522032303591,
      "grad_norm": 0.7245031595230103,
      "learning_rate": 2.0893816538773853e-05,
      "loss": 0.1462,
      "step": 1598
    },
    {
      "epoch": 1.0029794574251216,
      "grad_norm": 0.8666381239891052,
      "learning_rate": 2.087351560366876e-05,
      "loss": 0.1772,
      "step": 1599
    },
    {
      "epoch": 1.003606711619884,
      "grad_norm": 0.8373695015907288,
      "learning_rate": 2.0853213766802763e-05,
      "loss": 0.1766,
      "step": 1600
    },
    {
      "epoch": 1.0042339658146464,
      "grad_norm": 0.811313271522522,
      "learning_rate": 2.0832911049134154e-05,
      "loss": 0.1562,
      "step": 1601
    },
    {
      "epoch": 1.004861220009409,
      "grad_norm": 0.6714876890182495,
      "learning_rate": 2.0812607471622152e-05,
      "loss": 0.1355,
      "step": 1602
    },
    {
      "epoch": 1.0054884742041712,
      "grad_norm": 0.7351215481758118,
      "learning_rate": 2.079230305522685e-05,
      "loss": 0.1478,
      "step": 1603
    },
    {
      "epoch": 1.0061157283989337,
      "grad_norm": 0.5395284295082092,
      "learning_rate": 2.077199782090921e-05,
      "loss": 0.1277,
      "step": 1604
    },
    {
      "epoch": 1.006742982593696,
      "grad_norm": 1.017524242401123,
      "learning_rate": 2.0751691789631037e-05,
      "loss": 0.176,
      "step": 1605
    },
    {
      "epoch": 1.0073702367884585,
      "grad_norm": 0.8757364749908447,
      "learning_rate": 2.0731384982354977e-05,
      "loss": 0.1736,
      "step": 1606
    },
    {
      "epoch": 1.007997490983221,
      "grad_norm": 1.1049363613128662,
      "learning_rate": 2.0711077420044448e-05,
      "loss": 0.2043,
      "step": 1607
    },
    {
      "epoch": 1.0086247451779833,
      "grad_norm": 0.8929004669189453,
      "learning_rate": 2.069076912366366e-05,
      "loss": 0.1678,
      "step": 1608
    },
    {
      "epoch": 1.0092519993727458,
      "grad_norm": 0.9069939255714417,
      "learning_rate": 2.0670460114177597e-05,
      "loss": 0.1718,
      "step": 1609
    },
    {
      "epoch": 1.0098792535675083,
      "grad_norm": 0.9183425903320312,
      "learning_rate": 2.0650150412551958e-05,
      "loss": 0.14,
      "step": 1610
    },
    {
      "epoch": 1.0105065077622706,
      "grad_norm": 0.84815913438797,
      "learning_rate": 2.062984003975315e-05,
      "loss": 0.1478,
      "step": 1611
    },
    {
      "epoch": 1.011133761957033,
      "grad_norm": 1.0737385749816895,
      "learning_rate": 2.0609529016748304e-05,
      "loss": 0.218,
      "step": 1612
    },
    {
      "epoch": 1.0117610161517956,
      "grad_norm": 0.8041346073150635,
      "learning_rate": 2.058921736450519e-05,
      "loss": 0.1647,
      "step": 1613
    },
    {
      "epoch": 1.0123882703465579,
      "grad_norm": 0.8681744933128357,
      "learning_rate": 2.056890510399225e-05,
      "loss": 0.1537,
      "step": 1614
    },
    {
      "epoch": 1.0130155245413204,
      "grad_norm": 0.7892470359802246,
      "learning_rate": 2.054859225617855e-05,
      "loss": 0.1578,
      "step": 1615
    },
    {
      "epoch": 1.0136427787360829,
      "grad_norm": 0.8767917156219482,
      "learning_rate": 2.052827884203374e-05,
      "loss": 0.1697,
      "step": 1616
    },
    {
      "epoch": 1.0142700329308452,
      "grad_norm": 1.0207929611206055,
      "learning_rate": 2.0507964882528092e-05,
      "loss": 0.1829,
      "step": 1617
    },
    {
      "epoch": 1.0148972871256077,
      "grad_norm": 0.777074933052063,
      "learning_rate": 2.0487650398632408e-05,
      "loss": 0.1492,
      "step": 1618
    },
    {
      "epoch": 1.0155245413203702,
      "grad_norm": 0.8585703372955322,
      "learning_rate": 2.0467335411318056e-05,
      "loss": 0.1783,
      "step": 1619
    },
    {
      "epoch": 1.0161517955151325,
      "grad_norm": 0.7692691683769226,
      "learning_rate": 2.0447019941556902e-05,
      "loss": 0.1432,
      "step": 1620
    },
    {
      "epoch": 1.016779049709895,
      "grad_norm": 0.8173703551292419,
      "learning_rate": 2.0426704010321327e-05,
      "loss": 0.1565,
      "step": 1621
    },
    {
      "epoch": 1.0174063039046572,
      "grad_norm": 0.8283787965774536,
      "learning_rate": 2.040638763858418e-05,
      "loss": 0.1531,
      "step": 1622
    },
    {
      "epoch": 1.0180335580994198,
      "grad_norm": 0.9298859238624573,
      "learning_rate": 2.038607084731877e-05,
      "loss": 0.1648,
      "step": 1623
    },
    {
      "epoch": 1.0186608122941823,
      "grad_norm": 0.8722273707389832,
      "learning_rate": 2.0365753657498838e-05,
      "loss": 0.1807,
      "step": 1624
    },
    {
      "epoch": 1.0192880664889445,
      "grad_norm": 0.7768132090568542,
      "learning_rate": 2.0345436090098517e-05,
      "loss": 0.1519,
      "step": 1625
    },
    {
      "epoch": 1.019915320683707,
      "grad_norm": 0.8653201460838318,
      "learning_rate": 2.032511816609237e-05,
      "loss": 0.1723,
      "step": 1626
    },
    {
      "epoch": 1.0205425748784696,
      "grad_norm": 0.8122580051422119,
      "learning_rate": 2.030479990645529e-05,
      "loss": 0.1677,
      "step": 1627
    },
    {
      "epoch": 1.0211698290732318,
      "grad_norm": 0.8771058917045593,
      "learning_rate": 2.0284481332162534e-05,
      "loss": 0.1628,
      "step": 1628
    },
    {
      "epoch": 1.0217970832679943,
      "grad_norm": 0.7717471122741699,
      "learning_rate": 2.0264162464189692e-05,
      "loss": 0.154,
      "step": 1629
    },
    {
      "epoch": 1.0224243374627568,
      "grad_norm": 0.8961104154586792,
      "learning_rate": 2.0243843323512632e-05,
      "loss": 0.1597,
      "step": 1630
    },
    {
      "epoch": 1.0230515916575191,
      "grad_norm": 0.8729629516601562,
      "learning_rate": 2.0223523931107523e-05,
      "loss": 0.1504,
      "step": 1631
    },
    {
      "epoch": 1.0236788458522816,
      "grad_norm": 0.9216261506080627,
      "learning_rate": 2.0203204307950796e-05,
      "loss": 0.1669,
      "step": 1632
    },
    {
      "epoch": 1.0243061000470441,
      "grad_norm": 0.965466320514679,
      "learning_rate": 2.0182884475019108e-05,
      "loss": 0.1608,
      "step": 1633
    },
    {
      "epoch": 1.0249333542418064,
      "grad_norm": 0.8710114359855652,
      "learning_rate": 2.0162564453289333e-05,
      "loss": 0.1709,
      "step": 1634
    },
    {
      "epoch": 1.025560608436569,
      "grad_norm": 0.8559769988059998,
      "learning_rate": 2.014224426373856e-05,
      "loss": 0.161,
      "step": 1635
    },
    {
      "epoch": 1.0261878626313314,
      "grad_norm": 0.8614868521690369,
      "learning_rate": 2.0121923927344018e-05,
      "loss": 0.1535,
      "step": 1636
    },
    {
      "epoch": 1.0268151168260937,
      "grad_norm": 0.7848617434501648,
      "learning_rate": 2.0101603465083114e-05,
      "loss": 0.1281,
      "step": 1637
    },
    {
      "epoch": 1.0274423710208562,
      "grad_norm": 0.8579975366592407,
      "learning_rate": 2.008128289793338e-05,
      "loss": 0.151,
      "step": 1638
    },
    {
      "epoch": 1.0280696252156187,
      "grad_norm": 1.0866503715515137,
      "learning_rate": 2.0060962246872448e-05,
      "loss": 0.177,
      "step": 1639
    },
    {
      "epoch": 1.028696879410381,
      "grad_norm": 0.9099308252334595,
      "learning_rate": 2.004064153287804e-05,
      "loss": 0.1956,
      "step": 1640
    },
    {
      "epoch": 1.0293241336051435,
      "grad_norm": 0.8447299003601074,
      "learning_rate": 2.0020320776927947e-05,
      "loss": 0.1667,
      "step": 1641
    },
    {
      "epoch": 1.0299513877999058,
      "grad_norm": 0.9115908741950989,
      "learning_rate": 2e-05,
      "loss": 0.1908,
      "step": 1642
    },
    {
      "epoch": 1.0305786419946683,
      "grad_norm": 0.9186663627624512,
      "learning_rate": 1.9979679223072056e-05,
      "loss": 0.1649,
      "step": 1643
    },
    {
      "epoch": 1.0312058961894308,
      "grad_norm": 0.9531528353691101,
      "learning_rate": 1.9959358467121966e-05,
      "loss": 0.1584,
      "step": 1644
    },
    {
      "epoch": 1.031833150384193,
      "grad_norm": 0.8582137227058411,
      "learning_rate": 1.9939037753127556e-05,
      "loss": 0.1472,
      "step": 1645
    },
    {
      "epoch": 1.0324604045789556,
      "grad_norm": 0.8075041770935059,
      "learning_rate": 1.9918717102066622e-05,
      "loss": 0.162,
      "step": 1646
    },
    {
      "epoch": 1.033087658773718,
      "grad_norm": 0.9009033441543579,
      "learning_rate": 1.989839653491689e-05,
      "loss": 0.1703,
      "step": 1647
    },
    {
      "epoch": 1.0337149129684804,
      "grad_norm": 0.826914370059967,
      "learning_rate": 1.9878076072655985e-05,
      "loss": 0.1507,
      "step": 1648
    },
    {
      "epoch": 1.034342167163243,
      "grad_norm": 0.8428979516029358,
      "learning_rate": 1.985775573626145e-05,
      "loss": 0.1562,
      "step": 1649
    },
    {
      "epoch": 1.0349694213580054,
      "grad_norm": 0.7274463772773743,
      "learning_rate": 1.983743554671067e-05,
      "loss": 0.1406,
      "step": 1650
    },
    {
      "epoch": 1.0355966755527677,
      "grad_norm": 0.8123782277107239,
      "learning_rate": 1.9817115524980895e-05,
      "loss": 0.1669,
      "step": 1651
    },
    {
      "epoch": 1.0362239297475302,
      "grad_norm": 0.8698951601982117,
      "learning_rate": 1.9796795692049207e-05,
      "loss": 0.1598,
      "step": 1652
    },
    {
      "epoch": 1.0368511839422927,
      "grad_norm": 0.7227709293365479,
      "learning_rate": 1.977647606889248e-05,
      "loss": 0.1469,
      "step": 1653
    },
    {
      "epoch": 1.037478438137055,
      "grad_norm": 0.7994180917739868,
      "learning_rate": 1.975615667648737e-05,
      "loss": 0.1385,
      "step": 1654
    },
    {
      "epoch": 1.0381056923318175,
      "grad_norm": 0.7999747395515442,
      "learning_rate": 1.9735837535810318e-05,
      "loss": 0.1601,
      "step": 1655
    },
    {
      "epoch": 1.03873294652658,
      "grad_norm": 0.8406004905700684,
      "learning_rate": 1.9715518667837473e-05,
      "loss": 0.1591,
      "step": 1656
    },
    {
      "epoch": 1.0393602007213423,
      "grad_norm": 0.8649516105651855,
      "learning_rate": 1.969520009354472e-05,
      "loss": 0.156,
      "step": 1657
    },
    {
      "epoch": 1.0399874549161048,
      "grad_norm": 0.8329012393951416,
      "learning_rate": 1.9674881833907638e-05,
      "loss": 0.1434,
      "step": 1658
    },
    {
      "epoch": 1.0406147091108673,
      "grad_norm": 0.8117157816886902,
      "learning_rate": 1.965456390990149e-05,
      "loss": 0.157,
      "step": 1659
    },
    {
      "epoch": 1.0412419633056296,
      "grad_norm": 0.8369453549385071,
      "learning_rate": 1.9634246342501172e-05,
      "loss": 0.1673,
      "step": 1660
    },
    {
      "epoch": 1.041869217500392,
      "grad_norm": 0.9822908043861389,
      "learning_rate": 1.9613929152681234e-05,
      "loss": 0.18,
      "step": 1661
    },
    {
      "epoch": 1.0424964716951544,
      "grad_norm": 1.0761197805404663,
      "learning_rate": 1.9593612361415824e-05,
      "loss": 0.1992,
      "step": 1662
    },
    {
      "epoch": 1.0431237258899169,
      "grad_norm": 0.8977817893028259,
      "learning_rate": 1.9573295989678677e-05,
      "loss": 0.1493,
      "step": 1663
    },
    {
      "epoch": 1.0437509800846794,
      "grad_norm": 0.7583467364311218,
      "learning_rate": 1.95529800584431e-05,
      "loss": 0.1324,
      "step": 1664
    },
    {
      "epoch": 1.0443782342794417,
      "grad_norm": 0.8257169127464294,
      "learning_rate": 1.953266458868195e-05,
      "loss": 0.1681,
      "step": 1665
    },
    {
      "epoch": 1.0450054884742042,
      "grad_norm": 0.8764519691467285,
      "learning_rate": 1.9512349601367596e-05,
      "loss": 0.1469,
      "step": 1666
    },
    {
      "epoch": 1.0456327426689667,
      "grad_norm": 0.8687977194786072,
      "learning_rate": 1.9492035117471918e-05,
      "loss": 0.1456,
      "step": 1667
    },
    {
      "epoch": 1.046259996863729,
      "grad_norm": 0.84560227394104,
      "learning_rate": 1.9471721157966267e-05,
      "loss": 0.1521,
      "step": 1668
    },
    {
      "epoch": 1.0468872510584915,
      "grad_norm": 0.81574946641922,
      "learning_rate": 1.9451407743821458e-05,
      "loss": 0.1692,
      "step": 1669
    },
    {
      "epoch": 1.047514505253254,
      "grad_norm": 0.7992091178894043,
      "learning_rate": 1.9431094896007757e-05,
      "loss": 0.1605,
      "step": 1670
    },
    {
      "epoch": 1.0481417594480162,
      "grad_norm": 0.9301627278327942,
      "learning_rate": 1.9410782635494815e-05,
      "loss": 0.1776,
      "step": 1671
    },
    {
      "epoch": 1.0487690136427787,
      "grad_norm": 0.8068720698356628,
      "learning_rate": 1.93904709832517e-05,
      "loss": 0.1646,
      "step": 1672
    },
    {
      "epoch": 1.0493962678375413,
      "grad_norm": 0.775219202041626,
      "learning_rate": 1.9370159960246853e-05,
      "loss": 0.1407,
      "step": 1673
    },
    {
      "epoch": 1.0500235220323035,
      "grad_norm": 0.88523268699646,
      "learning_rate": 1.934984958744805e-05,
      "loss": 0.1696,
      "step": 1674
    },
    {
      "epoch": 1.050650776227066,
      "grad_norm": 0.8547267913818359,
      "learning_rate": 1.9329539885822403e-05,
      "loss": 0.1464,
      "step": 1675
    },
    {
      "epoch": 1.0512780304218285,
      "grad_norm": 0.8653497695922852,
      "learning_rate": 1.9309230876336337e-05,
      "loss": 0.1675,
      "step": 1676
    },
    {
      "epoch": 1.0519052846165908,
      "grad_norm": 1.0238136053085327,
      "learning_rate": 1.9288922579955552e-05,
      "loss": 0.1653,
      "step": 1677
    },
    {
      "epoch": 1.0525325388113533,
      "grad_norm": 0.7988049387931824,
      "learning_rate": 1.9268615017645023e-05,
      "loss": 0.1587,
      "step": 1678
    },
    {
      "epoch": 1.0531597930061158,
      "grad_norm": 0.8365591764450073,
      "learning_rate": 1.9248308210368962e-05,
      "loss": 0.1705,
      "step": 1679
    },
    {
      "epoch": 1.0537870472008781,
      "grad_norm": 0.8894667029380798,
      "learning_rate": 1.9228002179090793e-05,
      "loss": 0.1811,
      "step": 1680
    },
    {
      "epoch": 1.0544143013956406,
      "grad_norm": 0.7917067408561707,
      "learning_rate": 1.920769694477316e-05,
      "loss": 0.1506,
      "step": 1681
    },
    {
      "epoch": 1.055041555590403,
      "grad_norm": 0.8458444476127625,
      "learning_rate": 1.9187392528377855e-05,
      "loss": 0.1511,
      "step": 1682
    },
    {
      "epoch": 1.0556688097851654,
      "grad_norm": 0.870349109172821,
      "learning_rate": 1.9167088950865853e-05,
      "loss": 0.1583,
      "step": 1683
    },
    {
      "epoch": 1.056296063979928,
      "grad_norm": 0.7882643342018127,
      "learning_rate": 1.9146786233197247e-05,
      "loss": 0.1429,
      "step": 1684
    },
    {
      "epoch": 1.0569233181746902,
      "grad_norm": 0.7835047245025635,
      "learning_rate": 1.912648439633125e-05,
      "loss": 0.1432,
      "step": 1685
    },
    {
      "epoch": 1.0575505723694527,
      "grad_norm": 0.8960422873497009,
      "learning_rate": 1.910618346122616e-05,
      "loss": 0.179,
      "step": 1686
    },
    {
      "epoch": 1.0581778265642152,
      "grad_norm": 0.8102217316627502,
      "learning_rate": 1.9085883448839347e-05,
      "loss": 0.157,
      "step": 1687
    },
    {
      "epoch": 1.0588050807589775,
      "grad_norm": 0.8699423670768738,
      "learning_rate": 1.9065584380127237e-05,
      "loss": 0.1655,
      "step": 1688
    },
    {
      "epoch": 1.05943233495374,
      "grad_norm": 0.8518077731132507,
      "learning_rate": 1.904528627604526e-05,
      "loss": 0.1653,
      "step": 1689
    },
    {
      "epoch": 1.0600595891485025,
      "grad_norm": 0.7880319952964783,
      "learning_rate": 1.902498915754787e-05,
      "loss": 0.1574,
      "step": 1690
    },
    {
      "epoch": 1.0606868433432648,
      "grad_norm": 0.8024574518203735,
      "learning_rate": 1.90046930455885e-05,
      "loss": 0.1671,
      "step": 1691
    },
    {
      "epoch": 1.0613140975380273,
      "grad_norm": 0.7825784683227539,
      "learning_rate": 1.8984397961119533e-05,
      "loss": 0.1459,
      "step": 1692
    },
    {
      "epoch": 1.0619413517327898,
      "grad_norm": 0.7848218679428101,
      "learning_rate": 1.8964103925092304e-05,
      "loss": 0.1638,
      "step": 1693
    },
    {
      "epoch": 1.062568605927552,
      "grad_norm": 0.7753551602363586,
      "learning_rate": 1.8943810958457066e-05,
      "loss": 0.1304,
      "step": 1694
    },
    {
      "epoch": 1.0631958601223146,
      "grad_norm": 0.7678571343421936,
      "learning_rate": 1.8923519082162956e-05,
      "loss": 0.1498,
      "step": 1695
    },
    {
      "epoch": 1.0638231143170769,
      "grad_norm": 0.8414244055747986,
      "learning_rate": 1.890322831715799e-05,
      "loss": 0.1661,
      "step": 1696
    },
    {
      "epoch": 1.0644503685118394,
      "grad_norm": 0.8053593635559082,
      "learning_rate": 1.888293868438905e-05,
      "loss": 0.1605,
      "step": 1697
    },
    {
      "epoch": 1.065077622706602,
      "grad_norm": 0.5466974377632141,
      "learning_rate": 1.886265020480182e-05,
      "loss": 0.1214,
      "step": 1698
    },
    {
      "epoch": 1.0657048769013642,
      "grad_norm": 0.8131811618804932,
      "learning_rate": 1.8842362899340834e-05,
      "loss": 0.1454,
      "step": 1699
    },
    {
      "epoch": 1.0663321310961267,
      "grad_norm": 1.0224452018737793,
      "learning_rate": 1.8822076788949376e-05,
      "loss": 0.1797,
      "step": 1700
    },
    {
      "epoch": 1.0669593852908892,
      "grad_norm": 0.7746246457099915,
      "learning_rate": 1.8801791894569513e-05,
      "loss": 0.1576,
      "step": 1701
    },
    {
      "epoch": 1.0675866394856515,
      "grad_norm": 0.8536249995231628,
      "learning_rate": 1.8781508237142064e-05,
      "loss": 0.1587,
      "step": 1702
    },
    {
      "epoch": 1.068213893680414,
      "grad_norm": 0.9272940158843994,
      "learning_rate": 1.8761225837606552e-05,
      "loss": 0.1656,
      "step": 1703
    },
    {
      "epoch": 1.0688411478751765,
      "grad_norm": 0.9104244709014893,
      "learning_rate": 1.8740944716901218e-05,
      "loss": 0.1744,
      "step": 1704
    },
    {
      "epoch": 1.0694684020699388,
      "grad_norm": 0.8467017412185669,
      "learning_rate": 1.872066489596298e-05,
      "loss": 0.1486,
      "step": 1705
    },
    {
      "epoch": 1.0700956562647013,
      "grad_norm": 0.768990695476532,
      "learning_rate": 1.8700386395727403e-05,
      "loss": 0.1367,
      "step": 1706
    },
    {
      "epoch": 1.0707229104594638,
      "grad_norm": 0.9271062016487122,
      "learning_rate": 1.8680109237128712e-05,
      "loss": 0.1706,
      "step": 1707
    },
    {
      "epoch": 1.071350164654226,
      "grad_norm": 0.791062593460083,
      "learning_rate": 1.865983344109971e-05,
      "loss": 0.1624,
      "step": 1708
    },
    {
      "epoch": 1.0719774188489886,
      "grad_norm": 0.8225182294845581,
      "learning_rate": 1.8639559028571832e-05,
      "loss": 0.1357,
      "step": 1709
    },
    {
      "epoch": 1.072604673043751,
      "grad_norm": 0.8716280460357666,
      "learning_rate": 1.8619286020475054e-05,
      "loss": 0.1525,
      "step": 1710
    },
    {
      "epoch": 1.0732319272385134,
      "grad_norm": 1.0095592737197876,
      "learning_rate": 1.8599014437737924e-05,
      "loss": 0.1718,
      "step": 1711
    },
    {
      "epoch": 1.0738591814332759,
      "grad_norm": 0.8246644139289856,
      "learning_rate": 1.8578744301287518e-05,
      "loss": 0.156,
      "step": 1712
    },
    {
      "epoch": 1.0744864356280384,
      "grad_norm": 0.939094603061676,
      "learning_rate": 1.85584756320494e-05,
      "loss": 0.1817,
      "step": 1713
    },
    {
      "epoch": 1.0751136898228006,
      "grad_norm": 0.8982072472572327,
      "learning_rate": 1.8538208450947632e-05,
      "loss": 0.1633,
      "step": 1714
    },
    {
      "epoch": 1.0757409440175631,
      "grad_norm": 1.0006636381149292,
      "learning_rate": 1.851794277890474e-05,
      "loss": 0.157,
      "step": 1715
    },
    {
      "epoch": 1.0763681982123257,
      "grad_norm": 0.7728843688964844,
      "learning_rate": 1.849767863684169e-05,
      "loss": 0.1486,
      "step": 1716
    },
    {
      "epoch": 1.076995452407088,
      "grad_norm": 0.7604233622550964,
      "learning_rate": 1.8477416045677875e-05,
      "loss": 0.1469,
      "step": 1717
    },
    {
      "epoch": 1.0776227066018504,
      "grad_norm": 0.8282923698425293,
      "learning_rate": 1.8457155026331077e-05,
      "loss": 0.1576,
      "step": 1718
    },
    {
      "epoch": 1.0782499607966127,
      "grad_norm": 0.7721243500709534,
      "learning_rate": 1.8436895599717456e-05,
      "loss": 0.1365,
      "step": 1719
    },
    {
      "epoch": 1.0788772149913752,
      "grad_norm": 1.1395034790039062,
      "learning_rate": 1.8416637786751537e-05,
      "loss": 0.1945,
      "step": 1720
    },
    {
      "epoch": 1.0795044691861377,
      "grad_norm": 0.7299290895462036,
      "learning_rate": 1.8396381608346168e-05,
      "loss": 0.1142,
      "step": 1721
    },
    {
      "epoch": 1.0801317233809,
      "grad_norm": 0.8539410829544067,
      "learning_rate": 1.837612708541252e-05,
      "loss": 0.1388,
      "step": 1722
    },
    {
      "epoch": 1.0807589775756625,
      "grad_norm": 0.9020755887031555,
      "learning_rate": 1.8355874238860053e-05,
      "loss": 0.1621,
      "step": 1723
    },
    {
      "epoch": 1.081386231770425,
      "grad_norm": 0.7987878322601318,
      "learning_rate": 1.8335623089596483e-05,
      "loss": 0.1441,
      "step": 1724
    },
    {
      "epoch": 1.0820134859651873,
      "grad_norm": 0.778041660785675,
      "learning_rate": 1.8315373658527788e-05,
      "loss": 0.1513,
      "step": 1725
    },
    {
      "epoch": 1.0826407401599498,
      "grad_norm": 0.9035307168960571,
      "learning_rate": 1.8295125966558175e-05,
      "loss": 0.1855,
      "step": 1726
    },
    {
      "epoch": 1.0832679943547123,
      "grad_norm": 0.9106894135475159,
      "learning_rate": 1.8274880034590043e-05,
      "loss": 0.1713,
      "step": 1727
    },
    {
      "epoch": 1.0838952485494746,
      "grad_norm": 0.8942307829856873,
      "learning_rate": 1.8254635883523986e-05,
      "loss": 0.1654,
      "step": 1728
    },
    {
      "epoch": 1.0845225027442371,
      "grad_norm": 0.868350625038147,
      "learning_rate": 1.8234393534258744e-05,
      "loss": 0.1444,
      "step": 1729
    },
    {
      "epoch": 1.0851497569389996,
      "grad_norm": 0.7952873110771179,
      "learning_rate": 1.8214153007691216e-05,
      "loss": 0.1429,
      "step": 1730
    },
    {
      "epoch": 1.085777011133762,
      "grad_norm": 0.9205557107925415,
      "learning_rate": 1.819391432471641e-05,
      "loss": 0.1652,
      "step": 1731
    },
    {
      "epoch": 1.0864042653285244,
      "grad_norm": 0.8573682904243469,
      "learning_rate": 1.817367750622742e-05,
      "loss": 0.1892,
      "step": 1732
    },
    {
      "epoch": 1.087031519523287,
      "grad_norm": 0.8403246402740479,
      "learning_rate": 1.8153442573115445e-05,
      "loss": 0.1425,
      "step": 1733
    },
    {
      "epoch": 1.0876587737180492,
      "grad_norm": 0.9676584601402283,
      "learning_rate": 1.81332095462697e-05,
      "loss": 0.1943,
      "step": 1734
    },
    {
      "epoch": 1.0882860279128117,
      "grad_norm": 0.7942099571228027,
      "learning_rate": 1.811297844657746e-05,
      "loss": 0.153,
      "step": 1735
    },
    {
      "epoch": 1.088913282107574,
      "grad_norm": 0.8771933913230896,
      "learning_rate": 1.8092749294924e-05,
      "loss": 0.1371,
      "step": 1736
    },
    {
      "epoch": 1.0895405363023365,
      "grad_norm": 0.86127769947052,
      "learning_rate": 1.807252211219258e-05,
      "loss": 0.1902,
      "step": 1737
    },
    {
      "epoch": 1.090167790497099,
      "grad_norm": 0.957034170627594,
      "learning_rate": 1.8052296919264447e-05,
      "loss": 0.1623,
      "step": 1738
    },
    {
      "epoch": 1.0907950446918613,
      "grad_norm": 0.8910097479820251,
      "learning_rate": 1.8032073737018766e-05,
      "loss": 0.1546,
      "step": 1739
    },
    {
      "epoch": 1.0914222988866238,
      "grad_norm": 0.9799900650978088,
      "learning_rate": 1.8011852586332643e-05,
      "loss": 0.1501,
      "step": 1740
    },
    {
      "epoch": 1.0920495530813863,
      "grad_norm": 0.8283793330192566,
      "learning_rate": 1.7991633488081092e-05,
      "loss": 0.1539,
      "step": 1741
    },
    {
      "epoch": 1.0926768072761486,
      "grad_norm": 0.7672228813171387,
      "learning_rate": 1.7971416463136984e-05,
      "loss": 0.1345,
      "step": 1742
    },
    {
      "epoch": 1.093304061470911,
      "grad_norm": 0.8319811224937439,
      "learning_rate": 1.7951201532371078e-05,
      "loss": 0.1458,
      "step": 1743
    },
    {
      "epoch": 1.0939313156656736,
      "grad_norm": 0.7316347360610962,
      "learning_rate": 1.7930988716651963e-05,
      "loss": 0.1445,
      "step": 1744
    },
    {
      "epoch": 1.0945585698604359,
      "grad_norm": 0.9188635349273682,
      "learning_rate": 1.7910778036846026e-05,
      "loss": 0.1725,
      "step": 1745
    },
    {
      "epoch": 1.0951858240551984,
      "grad_norm": 0.8345009684562683,
      "learning_rate": 1.7890569513817473e-05,
      "loss": 0.144,
      "step": 1746
    },
    {
      "epoch": 1.0958130782499609,
      "grad_norm": 0.8347561955451965,
      "learning_rate": 1.787036316842827e-05,
      "loss": 0.1568,
      "step": 1747
    },
    {
      "epoch": 1.0964403324447232,
      "grad_norm": 0.7968394160270691,
      "learning_rate": 1.7850159021538135e-05,
      "loss": 0.1358,
      "step": 1748
    },
    {
      "epoch": 1.0970675866394857,
      "grad_norm": 0.7191606163978577,
      "learning_rate": 1.7829957094004532e-05,
      "loss": 0.1455,
      "step": 1749
    },
    {
      "epoch": 1.0976948408342482,
      "grad_norm": 0.7700591087341309,
      "learning_rate": 1.780975740668261e-05,
      "loss": 0.1414,
      "step": 1750
    },
    {
      "epoch": 1.0983220950290105,
      "grad_norm": 0.8214758038520813,
      "learning_rate": 1.778955998042522e-05,
      "loss": 0.1571,
      "step": 1751
    },
    {
      "epoch": 1.098949349223773,
      "grad_norm": 0.8999004364013672,
      "learning_rate": 1.7769364836082886e-05,
      "loss": 0.1753,
      "step": 1752
    },
    {
      "epoch": 1.0995766034185355,
      "grad_norm": 0.8949674367904663,
      "learning_rate": 1.7749171994503752e-05,
      "loss": 0.1605,
      "step": 1753
    },
    {
      "epoch": 1.1002038576132978,
      "grad_norm": 0.8810386061668396,
      "learning_rate": 1.7728981476533608e-05,
      "loss": 0.1645,
      "step": 1754
    },
    {
      "epoch": 1.1008311118080603,
      "grad_norm": 0.8616423606872559,
      "learning_rate": 1.7708793303015835e-05,
      "loss": 0.1594,
      "step": 1755
    },
    {
      "epoch": 1.1014583660028225,
      "grad_norm": 0.8571405410766602,
      "learning_rate": 1.7688607494791394e-05,
      "loss": 0.1804,
      "step": 1756
    },
    {
      "epoch": 1.102085620197585,
      "grad_norm": 0.7818455696105957,
      "learning_rate": 1.7668424072698806e-05,
      "loss": 0.1613,
      "step": 1757
    },
    {
      "epoch": 1.1027128743923476,
      "grad_norm": 0.8706740140914917,
      "learning_rate": 1.7648243057574136e-05,
      "loss": 0.1821,
      "step": 1758
    },
    {
      "epoch": 1.1033401285871098,
      "grad_norm": 0.7925953269004822,
      "learning_rate": 1.762806447025094e-05,
      "loss": 0.1467,
      "step": 1759
    },
    {
      "epoch": 1.1039673827818723,
      "grad_norm": 0.9717243909835815,
      "learning_rate": 1.760788833156029e-05,
      "loss": 0.1661,
      "step": 1760
    },
    {
      "epoch": 1.1045946369766348,
      "grad_norm": 1.0377686023712158,
      "learning_rate": 1.758771466233073e-05,
      "loss": 0.1895,
      "step": 1761
    },
    {
      "epoch": 1.1052218911713971,
      "grad_norm": 0.876140832901001,
      "learning_rate": 1.756754348338825e-05,
      "loss": 0.1473,
      "step": 1762
    },
    {
      "epoch": 1.1058491453661596,
      "grad_norm": 0.6880743503570557,
      "learning_rate": 1.7547374815556257e-05,
      "loss": 0.119,
      "step": 1763
    },
    {
      "epoch": 1.1064763995609221,
      "grad_norm": 0.9143416881561279,
      "learning_rate": 1.7527208679655587e-05,
      "loss": 0.1698,
      "step": 1764
    },
    {
      "epoch": 1.1071036537556844,
      "grad_norm": 0.8823429346084595,
      "learning_rate": 1.7507045096504446e-05,
      "loss": 0.1396,
      "step": 1765
    },
    {
      "epoch": 1.107730907950447,
      "grad_norm": 0.9152858853340149,
      "learning_rate": 1.7486884086918414e-05,
      "loss": 0.1541,
      "step": 1766
    },
    {
      "epoch": 1.1083581621452094,
      "grad_norm": 0.7654640078544617,
      "learning_rate": 1.7466725671710417e-05,
      "loss": 0.1334,
      "step": 1767
    },
    {
      "epoch": 1.1089854163399717,
      "grad_norm": 0.7074160575866699,
      "learning_rate": 1.744656987169068e-05,
      "loss": 0.1412,
      "step": 1768
    },
    {
      "epoch": 1.1096126705347342,
      "grad_norm": 0.8951393961906433,
      "learning_rate": 1.742641670766676e-05,
      "loss": 0.1636,
      "step": 1769
    },
    {
      "epoch": 1.1102399247294967,
      "grad_norm": 0.8777140378952026,
      "learning_rate": 1.7406266200443472e-05,
      "loss": 0.1382,
      "step": 1770
    },
    {
      "epoch": 1.110867178924259,
      "grad_norm": 0.8367090821266174,
      "learning_rate": 1.7386118370822895e-05,
      "loss": 0.182,
      "step": 1771
    },
    {
      "epoch": 1.1114944331190215,
      "grad_norm": 0.751369833946228,
      "learning_rate": 1.7365973239604342e-05,
      "loss": 0.146,
      "step": 1772
    },
    {
      "epoch": 1.1121216873137838,
      "grad_norm": 0.9562219977378845,
      "learning_rate": 1.7345830827584344e-05,
      "loss": 0.1548,
      "step": 1773
    },
    {
      "epoch": 1.1127489415085463,
      "grad_norm": 0.804206907749176,
      "learning_rate": 1.7325691155556613e-05,
      "loss": 0.1645,
      "step": 1774
    },
    {
      "epoch": 1.1133761957033088,
      "grad_norm": 0.8179658055305481,
      "learning_rate": 1.7305554244312054e-05,
      "loss": 0.1617,
      "step": 1775
    },
    {
      "epoch": 1.114003449898071,
      "grad_norm": 0.8444033265113831,
      "learning_rate": 1.7285420114638703e-05,
      "loss": 0.158,
      "step": 1776
    },
    {
      "epoch": 1.1146307040928336,
      "grad_norm": 0.8494364023208618,
      "learning_rate": 1.7265288787321728e-05,
      "loss": 0.1606,
      "step": 1777
    },
    {
      "epoch": 1.115257958287596,
      "grad_norm": 0.8242295980453491,
      "learning_rate": 1.7245160283143415e-05,
      "loss": 0.1507,
      "step": 1778
    },
    {
      "epoch": 1.1158852124823584,
      "grad_norm": 0.860016942024231,
      "learning_rate": 1.7225034622883118e-05,
      "loss": 0.1604,
      "step": 1779
    },
    {
      "epoch": 1.116512466677121,
      "grad_norm": 0.7402696013450623,
      "learning_rate": 1.7204911827317268e-05,
      "loss": 0.1454,
      "step": 1780
    },
    {
      "epoch": 1.1171397208718834,
      "grad_norm": 0.804934024810791,
      "learning_rate": 1.7184791917219345e-05,
      "loss": 0.1468,
      "step": 1781
    },
    {
      "epoch": 1.1177669750666457,
      "grad_norm": 0.8034117221832275,
      "learning_rate": 1.7164674913359822e-05,
      "loss": 0.1401,
      "step": 1782
    },
    {
      "epoch": 1.1183942292614082,
      "grad_norm": 0.7933990955352783,
      "learning_rate": 1.7144560836506204e-05,
      "loss": 0.1571,
      "step": 1783
    },
    {
      "epoch": 1.1190214834561707,
      "grad_norm": 0.8252025246620178,
      "learning_rate": 1.7124449707422963e-05,
      "loss": 0.1603,
      "step": 1784
    },
    {
      "epoch": 1.119648737650933,
      "grad_norm": 0.8010521531105042,
      "learning_rate": 1.7104341546871515e-05,
      "loss": 0.1708,
      "step": 1785
    },
    {
      "epoch": 1.1202759918456955,
      "grad_norm": 0.7285284996032715,
      "learning_rate": 1.7084236375610225e-05,
      "loss": 0.1218,
      "step": 1786
    },
    {
      "epoch": 1.120903246040458,
      "grad_norm": 0.7609200477600098,
      "learning_rate": 1.7064134214394367e-05,
      "loss": 0.1528,
      "step": 1787
    },
    {
      "epoch": 1.1215305002352203,
      "grad_norm": 0.9738415479660034,
      "learning_rate": 1.704403508397612e-05,
      "loss": 0.1575,
      "step": 1788
    },
    {
      "epoch": 1.1221577544299828,
      "grad_norm": 0.7736012935638428,
      "learning_rate": 1.7023939005104513e-05,
      "loss": 0.1312,
      "step": 1789
    },
    {
      "epoch": 1.1227850086247453,
      "grad_norm": 0.8661496043205261,
      "learning_rate": 1.7003845998525438e-05,
      "loss": 0.145,
      "step": 1790
    },
    {
      "epoch": 1.1234122628195076,
      "grad_norm": 0.9104440808296204,
      "learning_rate": 1.698375608498162e-05,
      "loss": 0.1496,
      "step": 1791
    },
    {
      "epoch": 1.12403951701427,
      "grad_norm": 0.9806966781616211,
      "learning_rate": 1.6963669285212575e-05,
      "loss": 0.1438,
      "step": 1792
    },
    {
      "epoch": 1.1246667712090324,
      "grad_norm": 0.8026207089424133,
      "learning_rate": 1.6943585619954623e-05,
      "loss": 0.1556,
      "step": 1793
    },
    {
      "epoch": 1.1252940254037949,
      "grad_norm": 0.7851147055625916,
      "learning_rate": 1.6923505109940823e-05,
      "loss": 0.1505,
      "step": 1794
    },
    {
      "epoch": 1.1259212795985574,
      "grad_norm": 0.7137161493301392,
      "learning_rate": 1.6903427775901006e-05,
      "loss": 0.126,
      "step": 1795
    },
    {
      "epoch": 1.1265485337933197,
      "grad_norm": 0.8256273865699768,
      "learning_rate": 1.6883353638561713e-05,
      "loss": 0.1476,
      "step": 1796
    },
    {
      "epoch": 1.1271757879880822,
      "grad_norm": 1.027342438697815,
      "learning_rate": 1.686328271864617e-05,
      "loss": 0.1624,
      "step": 1797
    },
    {
      "epoch": 1.1278030421828447,
      "grad_norm": 0.957568883895874,
      "learning_rate": 1.68432150368743e-05,
      "loss": 0.1767,
      "step": 1798
    },
    {
      "epoch": 1.128430296377607,
      "grad_norm": 0.9146484732627869,
      "learning_rate": 1.6823150613962686e-05,
      "loss": 0.1623,
      "step": 1799
    },
    {
      "epoch": 1.1290575505723695,
      "grad_norm": 0.9292408227920532,
      "learning_rate": 1.6803089470624524e-05,
      "loss": 0.1562,
      "step": 1800
    },
    {
      "epoch": 1.129684804767132,
      "grad_norm": 0.8380172848701477,
      "learning_rate": 1.6783031627569645e-05,
      "loss": 0.163,
      "step": 1801
    },
    {
      "epoch": 1.1303120589618942,
      "grad_norm": 0.798717737197876,
      "learning_rate": 1.676297710550447e-05,
      "loss": 0.1305,
      "step": 1802
    },
    {
      "epoch": 1.1309393131566567,
      "grad_norm": 0.8696675896644592,
      "learning_rate": 1.6742925925131984e-05,
      "loss": 0.1803,
      "step": 1803
    },
    {
      "epoch": 1.1315665673514193,
      "grad_norm": 0.8462779521942139,
      "learning_rate": 1.6722878107151723e-05,
      "loss": 0.18,
      "step": 1804
    },
    {
      "epoch": 1.1321938215461815,
      "grad_norm": 0.8923316597938538,
      "learning_rate": 1.6702833672259763e-05,
      "loss": 0.1615,
      "step": 1805
    },
    {
      "epoch": 1.132821075740944,
      "grad_norm": 0.9322481751441956,
      "learning_rate": 1.6682792641148674e-05,
      "loss": 0.1612,
      "step": 1806
    },
    {
      "epoch": 1.1334483299357063,
      "grad_norm": 0.8081589937210083,
      "learning_rate": 1.6662755034507525e-05,
      "loss": 0.1488,
      "step": 1807
    },
    {
      "epoch": 1.1340755841304688,
      "grad_norm": 0.902973473072052,
      "learning_rate": 1.6642720873021833e-05,
      "loss": 0.1581,
      "step": 1808
    },
    {
      "epoch": 1.1347028383252313,
      "grad_norm": 0.791506826877594,
      "learning_rate": 1.662269017737357e-05,
      "loss": 0.1386,
      "step": 1809
    },
    {
      "epoch": 1.1353300925199936,
      "grad_norm": 0.6856691241264343,
      "learning_rate": 1.6602662968241145e-05,
      "loss": 0.1253,
      "step": 1810
    },
    {
      "epoch": 1.1359573467147561,
      "grad_norm": 0.7945860624313354,
      "learning_rate": 1.6582639266299328e-05,
      "loss": 0.1403,
      "step": 1811
    },
    {
      "epoch": 1.1365846009095186,
      "grad_norm": 0.8668363690376282,
      "learning_rate": 1.6562619092219298e-05,
      "loss": 0.1491,
      "step": 1812
    },
    {
      "epoch": 1.137211855104281,
      "grad_norm": 0.8494446277618408,
      "learning_rate": 1.6542602466668586e-05,
      "loss": 0.1414,
      "step": 1813
    },
    {
      "epoch": 1.1378391092990434,
      "grad_norm": 0.9679626822471619,
      "learning_rate": 1.6522589410311067e-05,
      "loss": 0.1808,
      "step": 1814
    },
    {
      "epoch": 1.138466363493806,
      "grad_norm": 0.8872665762901306,
      "learning_rate": 1.6502579943806915e-05,
      "loss": 0.1499,
      "step": 1815
    },
    {
      "epoch": 1.1390936176885682,
      "grad_norm": 0.7795957326889038,
      "learning_rate": 1.648257408781261e-05,
      "loss": 0.1473,
      "step": 1816
    },
    {
      "epoch": 1.1397208718833307,
      "grad_norm": 0.8384416103363037,
      "learning_rate": 1.6462571862980902e-05,
      "loss": 0.1642,
      "step": 1817
    },
    {
      "epoch": 1.1403481260780932,
      "grad_norm": 0.8306025862693787,
      "learning_rate": 1.6442573289960788e-05,
      "loss": 0.1547,
      "step": 1818
    },
    {
      "epoch": 1.1409753802728555,
      "grad_norm": 0.8154910206794739,
      "learning_rate": 1.64225783893975e-05,
      "loss": 0.1569,
      "step": 1819
    },
    {
      "epoch": 1.141602634467618,
      "grad_norm": 0.9459575414657593,
      "learning_rate": 1.6402587181932488e-05,
      "loss": 0.1817,
      "step": 1820
    },
    {
      "epoch": 1.1422298886623805,
      "grad_norm": 0.7928250432014465,
      "learning_rate": 1.6382599688203362e-05,
      "loss": 0.1214,
      "step": 1821
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 0.8686726689338684,
      "learning_rate": 1.636261592884393e-05,
      "loss": 0.1406,
      "step": 1822
    },
    {
      "epoch": 1.1434843970519053,
      "grad_norm": 0.849258303642273,
      "learning_rate": 1.6342635924484123e-05,
      "loss": 0.1286,
      "step": 1823
    },
    {
      "epoch": 1.1441116512466678,
      "grad_norm": 0.8623886704444885,
      "learning_rate": 1.6322659695750003e-05,
      "loss": 0.1682,
      "step": 1824
    },
    {
      "epoch": 1.14473890544143,
      "grad_norm": 0.8463473916053772,
      "learning_rate": 1.6302687263263735e-05,
      "loss": 0.1682,
      "step": 1825
    },
    {
      "epoch": 1.1453661596361926,
      "grad_norm": 0.779043436050415,
      "learning_rate": 1.628271864764357e-05,
      "loss": 0.1671,
      "step": 1826
    },
    {
      "epoch": 1.145993413830955,
      "grad_norm": 1.2017778158187866,
      "learning_rate": 1.6262753869503797e-05,
      "loss": 0.1809,
      "step": 1827
    },
    {
      "epoch": 1.1466206680257174,
      "grad_norm": 0.8701357245445251,
      "learning_rate": 1.624279294945478e-05,
      "loss": 0.1583,
      "step": 1828
    },
    {
      "epoch": 1.1472479222204799,
      "grad_norm": 0.8049497008323669,
      "learning_rate": 1.6222835908102866e-05,
      "loss": 0.1343,
      "step": 1829
    },
    {
      "epoch": 1.1478751764152424,
      "grad_norm": 0.8608588576316833,
      "learning_rate": 1.6202882766050412e-05,
      "loss": 0.1714,
      "step": 1830
    },
    {
      "epoch": 1.1485024306100047,
      "grad_norm": 0.7564876079559326,
      "learning_rate": 1.6182933543895757e-05,
      "loss": 0.1312,
      "step": 1831
    },
    {
      "epoch": 1.1491296848047672,
      "grad_norm": 0.8293046951293945,
      "learning_rate": 1.6162988262233176e-05,
      "loss": 0.153,
      "step": 1832
    },
    {
      "epoch": 1.1497569389995295,
      "grad_norm": 0.7152065634727478,
      "learning_rate": 1.6143046941652887e-05,
      "loss": 0.1418,
      "step": 1833
    },
    {
      "epoch": 1.150384193194292,
      "grad_norm": 0.7662420868873596,
      "learning_rate": 1.6123109602741025e-05,
      "loss": 0.1358,
      "step": 1834
    },
    {
      "epoch": 1.1510114473890545,
      "grad_norm": 0.6615224480628967,
      "learning_rate": 1.6103176266079598e-05,
      "loss": 0.1275,
      "step": 1835
    },
    {
      "epoch": 1.1516387015838168,
      "grad_norm": 0.9032341241836548,
      "learning_rate": 1.6083246952246502e-05,
      "loss": 0.1534,
      "step": 1836
    },
    {
      "epoch": 1.1522659557785793,
      "grad_norm": 0.8961499929428101,
      "learning_rate": 1.6063321681815454e-05,
      "loss": 0.1574,
      "step": 1837
    },
    {
      "epoch": 1.1528932099733418,
      "grad_norm": 0.8611046671867371,
      "learning_rate": 1.6043400475356025e-05,
      "loss": 0.1766,
      "step": 1838
    },
    {
      "epoch": 1.153520464168104,
      "grad_norm": 0.8358910083770752,
      "learning_rate": 1.6023483353433567e-05,
      "loss": 0.1366,
      "step": 1839
    },
    {
      "epoch": 1.1541477183628666,
      "grad_norm": 0.8706210255622864,
      "learning_rate": 1.6003570336609235e-05,
      "loss": 0.1796,
      "step": 1840
    },
    {
      "epoch": 1.154774972557629,
      "grad_norm": 2.1682252883911133,
      "learning_rate": 1.5983661445439933e-05,
      "loss": 0.186,
      "step": 1841
    },
    {
      "epoch": 1.1554022267523913,
      "grad_norm": 0.9323776364326477,
      "learning_rate": 1.596375670047831e-05,
      "loss": 0.1578,
      "step": 1842
    },
    {
      "epoch": 1.1560294809471539,
      "grad_norm": 0.8486548066139221,
      "learning_rate": 1.594385612227274e-05,
      "loss": 0.1508,
      "step": 1843
    },
    {
      "epoch": 1.1566567351419161,
      "grad_norm": 0.8433166146278381,
      "learning_rate": 1.5923959731367285e-05,
      "loss": 0.1526,
      "step": 1844
    },
    {
      "epoch": 1.1572839893366786,
      "grad_norm": 0.8981841802597046,
      "learning_rate": 1.5904067548301688e-05,
      "loss": 0.1853,
      "step": 1845
    },
    {
      "epoch": 1.1579112435314411,
      "grad_norm": 0.8817017078399658,
      "learning_rate": 1.5884179593611357e-05,
      "loss": 0.1443,
      "step": 1846
    },
    {
      "epoch": 1.1585384977262034,
      "grad_norm": 0.9960587024688721,
      "learning_rate": 1.5864295887827324e-05,
      "loss": 0.1676,
      "step": 1847
    },
    {
      "epoch": 1.159165751920966,
      "grad_norm": 0.8518686890602112,
      "learning_rate": 1.5844416451476236e-05,
      "loss": 0.167,
      "step": 1848
    },
    {
      "epoch": 1.1597930061157284,
      "grad_norm": 0.8934053182601929,
      "learning_rate": 1.5824541305080342e-05,
      "loss": 0.1212,
      "step": 1849
    },
    {
      "epoch": 1.1604202603104907,
      "grad_norm": 0.8080291152000427,
      "learning_rate": 1.5804670469157447e-05,
      "loss": 0.1349,
      "step": 1850
    },
    {
      "epoch": 1.1610475145052532,
      "grad_norm": 0.7122689485549927,
      "learning_rate": 1.5784803964220918e-05,
      "loss": 0.1521,
      "step": 1851
    },
    {
      "epoch": 1.1616747687000157,
      "grad_norm": 0.9208701252937317,
      "learning_rate": 1.5764941810779653e-05,
      "loss": 0.1893,
      "step": 1852
    },
    {
      "epoch": 1.162302022894778,
      "grad_norm": 0.9723432064056396,
      "learning_rate": 1.574508402933804e-05,
      "loss": 0.1564,
      "step": 1853
    },
    {
      "epoch": 1.1629292770895405,
      "grad_norm": 1.0611326694488525,
      "learning_rate": 1.5725230640395982e-05,
      "loss": 0.1711,
      "step": 1854
    },
    {
      "epoch": 1.163556531284303,
      "grad_norm": 1.152522087097168,
      "learning_rate": 1.570538166444881e-05,
      "loss": 0.1553,
      "step": 1855
    },
    {
      "epoch": 1.1641837854790653,
      "grad_norm": 0.9438923597335815,
      "learning_rate": 1.568553712198734e-05,
      "loss": 0.1629,
      "step": 1856
    },
    {
      "epoch": 1.1648110396738278,
      "grad_norm": 0.7039708495140076,
      "learning_rate": 1.5665697033497788e-05,
      "loss": 0.1373,
      "step": 1857
    },
    {
      "epoch": 1.1654382938685903,
      "grad_norm": 0.9011950492858887,
      "learning_rate": 1.5645861419461766e-05,
      "loss": 0.1603,
      "step": 1858
    },
    {
      "epoch": 1.1660655480633526,
      "grad_norm": 0.9298847913742065,
      "learning_rate": 1.5626030300356282e-05,
      "loss": 0.1461,
      "step": 1859
    },
    {
      "epoch": 1.1666928022581151,
      "grad_norm": 1.023101568222046,
      "learning_rate": 1.5606203696653704e-05,
      "loss": 0.1704,
      "step": 1860
    },
    {
      "epoch": 1.1673200564528776,
      "grad_norm": 0.8772397041320801,
      "learning_rate": 1.5586381628821724e-05,
      "loss": 0.1733,
      "step": 1861
    },
    {
      "epoch": 1.16794731064764,
      "grad_norm": 0.8202458620071411,
      "learning_rate": 1.5566564117323376e-05,
      "loss": 0.141,
      "step": 1862
    },
    {
      "epoch": 1.1685745648424024,
      "grad_norm": 0.7463909983634949,
      "learning_rate": 1.5546751182616953e-05,
      "loss": 0.14,
      "step": 1863
    },
    {
      "epoch": 1.169201819037165,
      "grad_norm": 0.9149695634841919,
      "learning_rate": 1.552694284515606e-05,
      "loss": 0.1575,
      "step": 1864
    },
    {
      "epoch": 1.1698290732319272,
      "grad_norm": 0.7667670845985413,
      "learning_rate": 1.5507139125389532e-05,
      "loss": 0.1377,
      "step": 1865
    },
    {
      "epoch": 1.1704563274266897,
      "grad_norm": 0.9685397148132324,
      "learning_rate": 1.5487340043761452e-05,
      "loss": 0.1708,
      "step": 1866
    },
    {
      "epoch": 1.1710835816214522,
      "grad_norm": 0.7025471329689026,
      "learning_rate": 1.546754562071111e-05,
      "loss": 0.1304,
      "step": 1867
    },
    {
      "epoch": 1.1717108358162145,
      "grad_norm": 0.7441644668579102,
      "learning_rate": 1.544775587667298e-05,
      "loss": 0.1225,
      "step": 1868
    },
    {
      "epoch": 1.172338090010977,
      "grad_norm": 0.7965696454048157,
      "learning_rate": 1.5427970832076717e-05,
      "loss": 0.1664,
      "step": 1869
    },
    {
      "epoch": 1.1729653442057393,
      "grad_norm": 0.9268738031387329,
      "learning_rate": 1.540819050734712e-05,
      "loss": 0.1625,
      "step": 1870
    },
    {
      "epoch": 1.1735925984005018,
      "grad_norm": 0.9017819166183472,
      "learning_rate": 1.5388414922904106e-05,
      "loss": 0.1594,
      "step": 1871
    },
    {
      "epoch": 1.1742198525952643,
      "grad_norm": 0.7799805998802185,
      "learning_rate": 1.5368644099162717e-05,
      "loss": 0.1351,
      "step": 1872
    },
    {
      "epoch": 1.1748471067900266,
      "grad_norm": 0.880843997001648,
      "learning_rate": 1.5348878056533064e-05,
      "loss": 0.1697,
      "step": 1873
    },
    {
      "epoch": 1.175474360984789,
      "grad_norm": 0.9333112835884094,
      "learning_rate": 1.5329116815420325e-05,
      "loss": 0.1457,
      "step": 1874
    },
    {
      "epoch": 1.1761016151795516,
      "grad_norm": 0.8713697195053101,
      "learning_rate": 1.530936039622474e-05,
      "loss": 0.1601,
      "step": 1875
    },
    {
      "epoch": 1.1767288693743139,
      "grad_norm": 0.8385375142097473,
      "learning_rate": 1.5289608819341538e-05,
      "loss": 0.1611,
      "step": 1876
    },
    {
      "epoch": 1.1773561235690764,
      "grad_norm": 0.8101088404655457,
      "learning_rate": 1.526986210516098e-05,
      "loss": 0.1552,
      "step": 1877
    },
    {
      "epoch": 1.1779833777638389,
      "grad_norm": 0.7988011240959167,
      "learning_rate": 1.5250120274068292e-05,
      "loss": 0.1547,
      "step": 1878
    },
    {
      "epoch": 1.1786106319586012,
      "grad_norm": 0.8540886640548706,
      "learning_rate": 1.5230383346443654e-05,
      "loss": 0.1571,
      "step": 1879
    },
    {
      "epoch": 1.1792378861533637,
      "grad_norm": 0.7687944173812866,
      "learning_rate": 1.5210651342662197e-05,
      "loss": 0.127,
      "step": 1880
    },
    {
      "epoch": 1.1798651403481262,
      "grad_norm": 0.6662045121192932,
      "learning_rate": 1.5190924283093967e-05,
      "loss": 0.1338,
      "step": 1881
    },
    {
      "epoch": 1.1804923945428885,
      "grad_norm": 0.8382959961891174,
      "learning_rate": 1.517120218810389e-05,
      "loss": 0.1502,
      "step": 1882
    },
    {
      "epoch": 1.181119648737651,
      "grad_norm": 0.7887606024742126,
      "learning_rate": 1.515148507805179e-05,
      "loss": 0.1623,
      "step": 1883
    },
    {
      "epoch": 1.1817469029324132,
      "grad_norm": 1.0467164516448975,
      "learning_rate": 1.5131772973292334e-05,
      "loss": 0.1724,
      "step": 1884
    },
    {
      "epoch": 1.1823741571271758,
      "grad_norm": 0.972026526927948,
      "learning_rate": 1.5112065894175011e-05,
      "loss": 0.1625,
      "step": 1885
    },
    {
      "epoch": 1.1830014113219383,
      "grad_norm": 0.702070415019989,
      "learning_rate": 1.5092363861044146e-05,
      "loss": 0.1297,
      "step": 1886
    },
    {
      "epoch": 1.1836286655167005,
      "grad_norm": 0.9498781561851501,
      "learning_rate": 1.5072666894238835e-05,
      "loss": 0.1802,
      "step": 1887
    },
    {
      "epoch": 1.184255919711463,
      "grad_norm": 0.8403610587120056,
      "learning_rate": 1.5052975014092947e-05,
      "loss": 0.1514,
      "step": 1888
    },
    {
      "epoch": 1.1848831739062256,
      "grad_norm": 0.9342296719551086,
      "learning_rate": 1.5033288240935106e-05,
      "loss": 0.1659,
      "step": 1889
    },
    {
      "epoch": 1.1855104281009878,
      "grad_norm": 0.738856315612793,
      "learning_rate": 1.5013606595088663e-05,
      "loss": 0.1297,
      "step": 1890
    },
    {
      "epoch": 1.1861376822957503,
      "grad_norm": 0.7916891574859619,
      "learning_rate": 1.4993930096871668e-05,
      "loss": 0.1625,
      "step": 1891
    },
    {
      "epoch": 1.1867649364905128,
      "grad_norm": 0.7939260005950928,
      "learning_rate": 1.4974258766596866e-05,
      "loss": 0.1635,
      "step": 1892
    },
    {
      "epoch": 1.1873921906852751,
      "grad_norm": 0.8619648218154907,
      "learning_rate": 1.495459262457167e-05,
      "loss": 0.168,
      "step": 1893
    },
    {
      "epoch": 1.1880194448800376,
      "grad_norm": 0.9105474352836609,
      "learning_rate": 1.493493169109812e-05,
      "loss": 0.1868,
      "step": 1894
    },
    {
      "epoch": 1.1886466990748001,
      "grad_norm": 0.9018944501876831,
      "learning_rate": 1.4915275986472896e-05,
      "loss": 0.1671,
      "step": 1895
    },
    {
      "epoch": 1.1892739532695624,
      "grad_norm": 0.906952440738678,
      "learning_rate": 1.4895625530987277e-05,
      "loss": 0.1768,
      "step": 1896
    },
    {
      "epoch": 1.189901207464325,
      "grad_norm": 0.7171775102615356,
      "learning_rate": 1.4875980344927112e-05,
      "loss": 0.1674,
      "step": 1897
    },
    {
      "epoch": 1.1905284616590874,
      "grad_norm": 0.7971838116645813,
      "learning_rate": 1.4856340448572821e-05,
      "loss": 0.1412,
      "step": 1898
    },
    {
      "epoch": 1.1911557158538497,
      "grad_norm": 0.9042351841926575,
      "learning_rate": 1.4836705862199367e-05,
      "loss": 0.1784,
      "step": 1899
    },
    {
      "epoch": 1.1917829700486122,
      "grad_norm": 0.7793372869491577,
      "learning_rate": 1.4817076606076213e-05,
      "loss": 0.1529,
      "step": 1900
    },
    {
      "epoch": 1.1924102242433747,
      "grad_norm": 1.0335478782653809,
      "learning_rate": 1.4797452700467339e-05,
      "loss": 0.1947,
      "step": 1901
    },
    {
      "epoch": 1.193037478438137,
      "grad_norm": 0.8207297921180725,
      "learning_rate": 1.4777834165631193e-05,
      "loss": 0.1745,
      "step": 1902
    },
    {
      "epoch": 1.1936647326328995,
      "grad_norm": 0.8895414471626282,
      "learning_rate": 1.4758221021820678e-05,
      "loss": 0.1516,
      "step": 1903
    },
    {
      "epoch": 1.194291986827662,
      "grad_norm": 0.7220791578292847,
      "learning_rate": 1.4738613289283134e-05,
      "loss": 0.1381,
      "step": 1904
    },
    {
      "epoch": 1.1949192410224243,
      "grad_norm": 0.8622303009033203,
      "learning_rate": 1.4719010988260309e-05,
      "loss": 0.1603,
      "step": 1905
    },
    {
      "epoch": 1.1955464952171868,
      "grad_norm": 0.875914454460144,
      "learning_rate": 1.4699414138988352e-05,
      "loss": 0.1518,
      "step": 1906
    },
    {
      "epoch": 1.1961737494119493,
      "grad_norm": 0.7252442240715027,
      "learning_rate": 1.4679822761697784e-05,
      "loss": 0.1415,
      "step": 1907
    },
    {
      "epoch": 1.1968010036067116,
      "grad_norm": 0.7394492626190186,
      "learning_rate": 1.4660236876613465e-05,
      "loss": 0.1274,
      "step": 1908
    },
    {
      "epoch": 1.197428257801474,
      "grad_norm": 0.9310888648033142,
      "learning_rate": 1.46406565039546e-05,
      "loss": 0.145,
      "step": 1909
    },
    {
      "epoch": 1.1980555119962364,
      "grad_norm": 0.8693878054618835,
      "learning_rate": 1.46210816639347e-05,
      "loss": 0.1623,
      "step": 1910
    },
    {
      "epoch": 1.198682766190999,
      "grad_norm": 0.6786130666732788,
      "learning_rate": 1.4601512376761551e-05,
      "loss": 0.1269,
      "step": 1911
    },
    {
      "epoch": 1.1993100203857614,
      "grad_norm": 0.8285700678825378,
      "learning_rate": 1.458194866263722e-05,
      "loss": 0.1727,
      "step": 1912
    },
    {
      "epoch": 1.1999372745805237,
      "grad_norm": 0.841225266456604,
      "learning_rate": 1.4562390541758034e-05,
      "loss": 0.1618,
      "step": 1913
    },
    {
      "epoch": 1.2005645287752862,
      "grad_norm": 0.721674382686615,
      "learning_rate": 1.454283803431451e-05,
      "loss": 0.1402,
      "step": 1914
    },
    {
      "epoch": 1.2011917829700487,
      "grad_norm": 0.8150315284729004,
      "learning_rate": 1.452329116049139e-05,
      "loss": 0.1542,
      "step": 1915
    },
    {
      "epoch": 1.201819037164811,
      "grad_norm": 0.9047179222106934,
      "learning_rate": 1.450374994046761e-05,
      "loss": 0.1654,
      "step": 1916
    },
    {
      "epoch": 1.2024462913595735,
      "grad_norm": 0.8281776905059814,
      "learning_rate": 1.4484214394416256e-05,
      "loss": 0.1549,
      "step": 1917
    },
    {
      "epoch": 1.203073545554336,
      "grad_norm": 0.9416970610618591,
      "learning_rate": 1.4464684542504559e-05,
      "loss": 0.1579,
      "step": 1918
    },
    {
      "epoch": 1.2037007997490983,
      "grad_norm": 0.93312007188797,
      "learning_rate": 1.444516040489387e-05,
      "loss": 0.1842,
      "step": 1919
    },
    {
      "epoch": 1.2043280539438608,
      "grad_norm": 0.8868372440338135,
      "learning_rate": 1.4425642001739651e-05,
      "loss": 0.1647,
      "step": 1920
    },
    {
      "epoch": 1.204955308138623,
      "grad_norm": 0.8125755786895752,
      "learning_rate": 1.440612935319143e-05,
      "loss": 0.1755,
      "step": 1921
    },
    {
      "epoch": 1.2055825623333856,
      "grad_norm": 0.823745608329773,
      "learning_rate": 1.4386622479392807e-05,
      "loss": 0.1257,
      "step": 1922
    },
    {
      "epoch": 1.206209816528148,
      "grad_norm": 0.7787184119224548,
      "learning_rate": 1.4367121400481405e-05,
      "loss": 0.138,
      "step": 1923
    },
    {
      "epoch": 1.2068370707229104,
      "grad_norm": 0.810903012752533,
      "learning_rate": 1.4347626136588881e-05,
      "loss": 0.1487,
      "step": 1924
    },
    {
      "epoch": 1.2074643249176729,
      "grad_norm": 0.7609171867370605,
      "learning_rate": 1.4328136707840885e-05,
      "loss": 0.1444,
      "step": 1925
    },
    {
      "epoch": 1.2080915791124354,
      "grad_norm": 0.7199875712394714,
      "learning_rate": 1.4308653134357034e-05,
      "loss": 0.1343,
      "step": 1926
    },
    {
      "epoch": 1.2087188333071976,
      "grad_norm": 0.7784453630447388,
      "learning_rate": 1.4289175436250908e-05,
      "loss": 0.1401,
      "step": 1927
    },
    {
      "epoch": 1.2093460875019602,
      "grad_norm": 0.8262721300125122,
      "learning_rate": 1.4269703633630028e-05,
      "loss": 0.1307,
      "step": 1928
    },
    {
      "epoch": 1.2099733416967227,
      "grad_norm": 0.7545008063316345,
      "learning_rate": 1.425023774659581e-05,
      "loss": 0.1401,
      "step": 1929
    },
    {
      "epoch": 1.210600595891485,
      "grad_norm": 0.7401162385940552,
      "learning_rate": 1.4230777795243577e-05,
      "loss": 0.1198,
      "step": 1930
    },
    {
      "epoch": 1.2112278500862474,
      "grad_norm": 0.8155125975608826,
      "learning_rate": 1.4211323799662533e-05,
      "loss": 0.1568,
      "step": 1931
    },
    {
      "epoch": 1.21185510428101,
      "grad_norm": 0.8564190864562988,
      "learning_rate": 1.4191875779935709e-05,
      "loss": 0.1618,
      "step": 1932
    },
    {
      "epoch": 1.2124823584757722,
      "grad_norm": 0.9466654062271118,
      "learning_rate": 1.417243375613999e-05,
      "loss": 0.163,
      "step": 1933
    },
    {
      "epoch": 1.2131096126705347,
      "grad_norm": 0.7641735076904297,
      "learning_rate": 1.4152997748346052e-05,
      "loss": 0.1456,
      "step": 1934
    },
    {
      "epoch": 1.2137368668652972,
      "grad_norm": 0.8033687472343445,
      "learning_rate": 1.4133567776618376e-05,
      "loss": 0.1373,
      "step": 1935
    },
    {
      "epoch": 1.2143641210600595,
      "grad_norm": 0.812255322933197,
      "learning_rate": 1.4114143861015208e-05,
      "loss": 0.1571,
      "step": 1936
    },
    {
      "epoch": 1.214991375254822,
      "grad_norm": 0.7928274869918823,
      "learning_rate": 1.4094726021588532e-05,
      "loss": 0.1458,
      "step": 1937
    },
    {
      "epoch": 1.2156186294495845,
      "grad_norm": 0.851686418056488,
      "learning_rate": 1.4075314278384074e-05,
      "loss": 0.1336,
      "step": 1938
    },
    {
      "epoch": 1.2162458836443468,
      "grad_norm": 0.8840798139572144,
      "learning_rate": 1.4055908651441263e-05,
      "loss": 0.1512,
      "step": 1939
    },
    {
      "epoch": 1.2168731378391093,
      "grad_norm": 0.7764620780944824,
      "learning_rate": 1.4036509160793206e-05,
      "loss": 0.1341,
      "step": 1940
    },
    {
      "epoch": 1.2175003920338718,
      "grad_norm": 0.8411002159118652,
      "learning_rate": 1.4017115826466673e-05,
      "loss": 0.1472,
      "step": 1941
    },
    {
      "epoch": 1.2181276462286341,
      "grad_norm": 0.673572838306427,
      "learning_rate": 1.3997728668482093e-05,
      "loss": 0.1179,
      "step": 1942
    },
    {
      "epoch": 1.2187549004233966,
      "grad_norm": 0.838390052318573,
      "learning_rate": 1.3978347706853518e-05,
      "loss": 0.1295,
      "step": 1943
    },
    {
      "epoch": 1.2193821546181591,
      "grad_norm": 0.8127596974372864,
      "learning_rate": 1.3958972961588583e-05,
      "loss": 0.1454,
      "step": 1944
    },
    {
      "epoch": 1.2200094088129214,
      "grad_norm": 0.7803665399551392,
      "learning_rate": 1.393960445268853e-05,
      "loss": 0.1448,
      "step": 1945
    },
    {
      "epoch": 1.220636663007684,
      "grad_norm": 0.8282830119132996,
      "learning_rate": 1.392024220014815e-05,
      "loss": 0.1598,
      "step": 1946
    },
    {
      "epoch": 1.2212639172024462,
      "grad_norm": 0.7870911359786987,
      "learning_rate": 1.3900886223955778e-05,
      "loss": 0.1562,
      "step": 1947
    },
    {
      "epoch": 1.2218911713972087,
      "grad_norm": 0.8417776823043823,
      "learning_rate": 1.3881536544093266e-05,
      "loss": 0.16,
      "step": 1948
    },
    {
      "epoch": 1.2225184255919712,
      "grad_norm": 0.9041074514389038,
      "learning_rate": 1.3862193180535979e-05,
      "loss": 0.1648,
      "step": 1949
    },
    {
      "epoch": 1.2231456797867335,
      "grad_norm": 1.0169646739959717,
      "learning_rate": 1.3842856153252737e-05,
      "loss": 0.1943,
      "step": 1950
    },
    {
      "epoch": 1.223772933981496,
      "grad_norm": 0.9628655910491943,
      "learning_rate": 1.382352548220585e-05,
      "loss": 0.2029,
      "step": 1951
    },
    {
      "epoch": 1.2244001881762585,
      "grad_norm": 0.7748029828071594,
      "learning_rate": 1.380420118735104e-05,
      "loss": 0.128,
      "step": 1952
    },
    {
      "epoch": 1.2250274423710208,
      "grad_norm": 0.7870191335678101,
      "learning_rate": 1.378488328863746e-05,
      "loss": 0.1264,
      "step": 1953
    },
    {
      "epoch": 1.2256546965657833,
      "grad_norm": 0.9672967195510864,
      "learning_rate": 1.376557180600766e-05,
      "loss": 0.1577,
      "step": 1954
    },
    {
      "epoch": 1.2262819507605458,
      "grad_norm": 0.8693897128105164,
      "learning_rate": 1.374626675939756e-05,
      "loss": 0.1349,
      "step": 1955
    },
    {
      "epoch": 1.226909204955308,
      "grad_norm": 0.8322724103927612,
      "learning_rate": 1.3726968168736434e-05,
      "loss": 0.141,
      "step": 1956
    },
    {
      "epoch": 1.2275364591500706,
      "grad_norm": 0.8421970009803772,
      "learning_rate": 1.3707676053946909e-05,
      "loss": 0.1399,
      "step": 1957
    },
    {
      "epoch": 1.2281637133448329,
      "grad_norm": 0.8611185550689697,
      "learning_rate": 1.3688390434944903e-05,
      "loss": 0.1438,
      "step": 1958
    },
    {
      "epoch": 1.2287909675395954,
      "grad_norm": 0.8574026823043823,
      "learning_rate": 1.3669111331639644e-05,
      "loss": 0.1369,
      "step": 1959
    },
    {
      "epoch": 1.2294182217343579,
      "grad_norm": 0.9769577383995056,
      "learning_rate": 1.3649838763933633e-05,
      "loss": 0.1591,
      "step": 1960
    },
    {
      "epoch": 1.2300454759291202,
      "grad_norm": 0.8494389057159424,
      "learning_rate": 1.3630572751722611e-05,
      "loss": 0.1361,
      "step": 1961
    },
    {
      "epoch": 1.2306727301238827,
      "grad_norm": 0.8490456938743591,
      "learning_rate": 1.3611313314895567e-05,
      "loss": 0.1638,
      "step": 1962
    },
    {
      "epoch": 1.2312999843186452,
      "grad_norm": 0.9514238834381104,
      "learning_rate": 1.3592060473334697e-05,
      "loss": 0.1557,
      "step": 1963
    },
    {
      "epoch": 1.2319272385134075,
      "grad_norm": 0.8845386505126953,
      "learning_rate": 1.357281424691538e-05,
      "loss": 0.1457,
      "step": 1964
    },
    {
      "epoch": 1.23255449270817,
      "grad_norm": 0.8918384313583374,
      "learning_rate": 1.3553574655506185e-05,
      "loss": 0.1574,
      "step": 1965
    },
    {
      "epoch": 1.2331817469029325,
      "grad_norm": 0.7711067795753479,
      "learning_rate": 1.3534341718968805e-05,
      "loss": 0.1289,
      "step": 1966
    },
    {
      "epoch": 1.2338090010976948,
      "grad_norm": 0.7397522330284119,
      "learning_rate": 1.3515115457158088e-05,
      "loss": 0.1431,
      "step": 1967
    },
    {
      "epoch": 1.2344362552924573,
      "grad_norm": 0.7456501722335815,
      "learning_rate": 1.3495895889921973e-05,
      "loss": 0.1564,
      "step": 1968
    },
    {
      "epoch": 1.2350635094872198,
      "grad_norm": 0.7631916999816895,
      "learning_rate": 1.3476683037101506e-05,
      "loss": 0.1354,
      "step": 1969
    },
    {
      "epoch": 1.235690763681982,
      "grad_norm": 0.9010176658630371,
      "learning_rate": 1.3457476918530779e-05,
      "loss": 0.1641,
      "step": 1970
    },
    {
      "epoch": 1.2363180178767446,
      "grad_norm": 0.7279947400093079,
      "learning_rate": 1.343827755403695e-05,
      "loss": 0.1375,
      "step": 1971
    },
    {
      "epoch": 1.236945272071507,
      "grad_norm": 0.832119882106781,
      "learning_rate": 1.3419084963440207e-05,
      "loss": 0.1682,
      "step": 1972
    },
    {
      "epoch": 1.2375725262662693,
      "grad_norm": 0.7645889520645142,
      "learning_rate": 1.3399899166553722e-05,
      "loss": 0.1405,
      "step": 1973
    },
    {
      "epoch": 1.2381997804610319,
      "grad_norm": 0.7561132907867432,
      "learning_rate": 1.3380720183183677e-05,
      "loss": 0.1493,
      "step": 1974
    },
    {
      "epoch": 1.2388270346557944,
      "grad_norm": 0.7001355886459351,
      "learning_rate": 1.3361548033129217e-05,
      "loss": 0.1265,
      "step": 1975
    },
    {
      "epoch": 1.2394542888505566,
      "grad_norm": 0.7829381823539734,
      "learning_rate": 1.3342382736182415e-05,
      "loss": 0.1576,
      "step": 1976
    },
    {
      "epoch": 1.2400815430453191,
      "grad_norm": 0.9344700574874878,
      "learning_rate": 1.332322431212829e-05,
      "loss": 0.159,
      "step": 1977
    },
    {
      "epoch": 1.2407087972400817,
      "grad_norm": 0.8084545135498047,
      "learning_rate": 1.3304072780744759e-05,
      "loss": 0.1359,
      "step": 1978
    },
    {
      "epoch": 1.241336051434844,
      "grad_norm": 0.9364140629768372,
      "learning_rate": 1.3284928161802614e-05,
      "loss": 0.1706,
      "step": 1979
    },
    {
      "epoch": 1.2419633056296064,
      "grad_norm": 0.7873445153236389,
      "learning_rate": 1.3265790475065523e-05,
      "loss": 0.121,
      "step": 1980
    },
    {
      "epoch": 1.242590559824369,
      "grad_norm": 0.6898486614227295,
      "learning_rate": 1.3246659740290002e-05,
      "loss": 0.1254,
      "step": 1981
    },
    {
      "epoch": 1.2432178140191312,
      "grad_norm": 0.944259762763977,
      "learning_rate": 1.3227535977225364e-05,
      "loss": 0.1484,
      "step": 1982
    },
    {
      "epoch": 1.2438450682138937,
      "grad_norm": 0.8232392072677612,
      "learning_rate": 1.3208419205613762e-05,
      "loss": 0.1266,
      "step": 1983
    },
    {
      "epoch": 1.244472322408656,
      "grad_norm": 0.8024344444274902,
      "learning_rate": 1.31893094451901e-05,
      "loss": 0.146,
      "step": 1984
    },
    {
      "epoch": 1.2450995766034185,
      "grad_norm": 0.7504177093505859,
      "learning_rate": 1.3170206715682058e-05,
      "loss": 0.1398,
      "step": 1985
    },
    {
      "epoch": 1.245726830798181,
      "grad_norm": 0.9191885590553284,
      "learning_rate": 1.3151111036810065e-05,
      "loss": 0.1498,
      "step": 1986
    },
    {
      "epoch": 1.2463540849929433,
      "grad_norm": 0.9054408073425293,
      "learning_rate": 1.313202242828725e-05,
      "loss": 0.1392,
      "step": 1987
    },
    {
      "epoch": 1.2469813391877058,
      "grad_norm": 0.8019949197769165,
      "learning_rate": 1.311294090981946e-05,
      "loss": 0.1446,
      "step": 1988
    },
    {
      "epoch": 1.2476085933824683,
      "grad_norm": 0.8508473634719849,
      "learning_rate": 1.3093866501105226e-05,
      "loss": 0.1444,
      "step": 1989
    },
    {
      "epoch": 1.2482358475772306,
      "grad_norm": 0.9246096611022949,
      "learning_rate": 1.3074799221835728e-05,
      "loss": 0.1505,
      "step": 1990
    },
    {
      "epoch": 1.2488631017719931,
      "grad_norm": 0.8947358727455139,
      "learning_rate": 1.3055739091694779e-05,
      "loss": 0.1621,
      "step": 1991
    },
    {
      "epoch": 1.2494903559667556,
      "grad_norm": 0.7604474425315857,
      "learning_rate": 1.3036686130358828e-05,
      "loss": 0.1221,
      "step": 1992
    },
    {
      "epoch": 1.250117610161518,
      "grad_norm": 0.8415438532829285,
      "learning_rate": 1.3017640357496925e-05,
      "loss": 0.1724,
      "step": 1993
    },
    {
      "epoch": 1.2507448643562804,
      "grad_norm": 0.8152422308921814,
      "learning_rate": 1.2998601792770684e-05,
      "loss": 0.1332,
      "step": 1994
    },
    {
      "epoch": 1.2513721185510427,
      "grad_norm": 0.870513916015625,
      "learning_rate": 1.2979570455834289e-05,
      "loss": 0.1415,
      "step": 1995
    },
    {
      "epoch": 1.2519993727458052,
      "grad_norm": 0.8987338542938232,
      "learning_rate": 1.2960546366334458e-05,
      "loss": 0.1698,
      "step": 1996
    },
    {
      "epoch": 1.2526266269405677,
      "grad_norm": 0.8302509784698486,
      "learning_rate": 1.2941529543910429e-05,
      "loss": 0.1508,
      "step": 1997
    },
    {
      "epoch": 1.25325388113533,
      "grad_norm": 0.9404724836349487,
      "learning_rate": 1.292252000819394e-05,
      "loss": 0.1458,
      "step": 1998
    },
    {
      "epoch": 1.2538811353300925,
      "grad_norm": 0.6932850480079651,
      "learning_rate": 1.2903517778809196e-05,
      "loss": 0.12,
      "step": 1999
    },
    {
      "epoch": 1.254508389524855,
      "grad_norm": 0.6985989212989807,
      "learning_rate": 1.2884522875372874e-05,
      "loss": 0.127,
      "step": 2000
    },
    {
      "epoch": 1.2551356437196173,
      "grad_norm": 0.7476592659950256,
      "learning_rate": 1.2865535317494083e-05,
      "loss": 0.1517,
      "step": 2001
    },
    {
      "epoch": 1.2557628979143798,
      "grad_norm": 0.7512590885162354,
      "learning_rate": 1.2846555124774338e-05,
      "loss": 0.1344,
      "step": 2002
    },
    {
      "epoch": 1.2563901521091423,
      "grad_norm": 0.7579845190048218,
      "learning_rate": 1.2827582316807568e-05,
      "loss": 0.1335,
      "step": 2003
    },
    {
      "epoch": 1.2570174063039046,
      "grad_norm": 0.8196955323219299,
      "learning_rate": 1.2808616913180073e-05,
      "loss": 0.1488,
      "step": 2004
    },
    {
      "epoch": 1.257644660498667,
      "grad_norm": 0.8363924026489258,
      "learning_rate": 1.2789658933470497e-05,
      "loss": 0.1378,
      "step": 2005
    },
    {
      "epoch": 1.2582719146934296,
      "grad_norm": 0.8742525577545166,
      "learning_rate": 1.2770708397249836e-05,
      "loss": 0.1406,
      "step": 2006
    },
    {
      "epoch": 1.2588991688881919,
      "grad_norm": 0.9705551862716675,
      "learning_rate": 1.27517653240814e-05,
      "loss": 0.1775,
      "step": 2007
    },
    {
      "epoch": 1.2595264230829544,
      "grad_norm": 0.8819530606269836,
      "learning_rate": 1.2732829733520782e-05,
      "loss": 0.1589,
      "step": 2008
    },
    {
      "epoch": 1.2601536772777169,
      "grad_norm": 0.7328044772148132,
      "learning_rate": 1.2713901645115861e-05,
      "loss": 0.1168,
      "step": 2009
    },
    {
      "epoch": 1.2607809314724792,
      "grad_norm": 0.8703687787055969,
      "learning_rate": 1.2694981078406775e-05,
      "loss": 0.1561,
      "step": 2010
    },
    {
      "epoch": 1.2614081856672417,
      "grad_norm": 0.7671546339988708,
      "learning_rate": 1.2676068052925883e-05,
      "loss": 0.1458,
      "step": 2011
    },
    {
      "epoch": 1.2620354398620042,
      "grad_norm": 0.9110318422317505,
      "learning_rate": 1.2657162588197774e-05,
      "loss": 0.1748,
      "step": 2012
    },
    {
      "epoch": 1.2626626940567665,
      "grad_norm": 0.919498085975647,
      "learning_rate": 1.2638264703739215e-05,
      "loss": 0.1471,
      "step": 2013
    },
    {
      "epoch": 1.263289948251529,
      "grad_norm": 0.9211678504943848,
      "learning_rate": 1.2619374419059165e-05,
      "loss": 0.18,
      "step": 2014
    },
    {
      "epoch": 1.2639172024462915,
      "grad_norm": 0.9267973303794861,
      "learning_rate": 1.2600491753658735e-05,
      "loss": 0.1491,
      "step": 2015
    },
    {
      "epoch": 1.2645444566410537,
      "grad_norm": 0.8233495354652405,
      "learning_rate": 1.2581616727031159e-05,
      "loss": 0.1558,
      "step": 2016
    },
    {
      "epoch": 1.2651717108358163,
      "grad_norm": 0.8945194482803345,
      "learning_rate": 1.2562749358661783e-05,
      "loss": 0.143,
      "step": 2017
    },
    {
      "epoch": 1.2657989650305788,
      "grad_norm": 0.7861148118972778,
      "learning_rate": 1.2543889668028064e-05,
      "loss": 0.1596,
      "step": 2018
    },
    {
      "epoch": 1.266426219225341,
      "grad_norm": 0.8920468688011169,
      "learning_rate": 1.2525037674599529e-05,
      "loss": 0.1391,
      "step": 2019
    },
    {
      "epoch": 1.2670534734201035,
      "grad_norm": 1.0049667358398438,
      "learning_rate": 1.2506193397837746e-05,
      "loss": 0.1426,
      "step": 2020
    },
    {
      "epoch": 1.267680727614866,
      "grad_norm": 0.9976416230201721,
      "learning_rate": 1.2487356857196331e-05,
      "loss": 0.1685,
      "step": 2021
    },
    {
      "epoch": 1.2683079818096283,
      "grad_norm": 0.7630930542945862,
      "learning_rate": 1.2468528072120908e-05,
      "loss": 0.1473,
      "step": 2022
    },
    {
      "epoch": 1.2689352360043908,
      "grad_norm": 0.8132131695747375,
      "learning_rate": 1.2449707062049088e-05,
      "loss": 0.1658,
      "step": 2023
    },
    {
      "epoch": 1.2695624901991531,
      "grad_norm": 0.8889040946960449,
      "learning_rate": 1.2430893846410465e-05,
      "loss": 0.1531,
      "step": 2024
    },
    {
      "epoch": 1.2701897443939156,
      "grad_norm": 0.7996199727058411,
      "learning_rate": 1.2412088444626593e-05,
      "loss": 0.1516,
      "step": 2025
    },
    {
      "epoch": 1.2708169985886781,
      "grad_norm": 0.8301317095756531,
      "learning_rate": 1.239329087611094e-05,
      "loss": 0.1449,
      "step": 2026
    },
    {
      "epoch": 1.2714442527834404,
      "grad_norm": 0.8246314525604248,
      "learning_rate": 1.2374501160268899e-05,
      "loss": 0.1217,
      "step": 2027
    },
    {
      "epoch": 1.272071506978203,
      "grad_norm": 0.7794330716133118,
      "learning_rate": 1.235571931649776e-05,
      "loss": 0.1351,
      "step": 2028
    },
    {
      "epoch": 1.2726987611729652,
      "grad_norm": 0.9376643300056458,
      "learning_rate": 1.2336945364186676e-05,
      "loss": 0.1797,
      "step": 2029
    },
    {
      "epoch": 1.2733260153677277,
      "grad_norm": 0.8853619694709778,
      "learning_rate": 1.2318179322716665e-05,
      "loss": 0.1446,
      "step": 2030
    },
    {
      "epoch": 1.2739532695624902,
      "grad_norm": 0.8093166947364807,
      "learning_rate": 1.2299421211460564e-05,
      "loss": 0.1457,
      "step": 2031
    },
    {
      "epoch": 1.2745805237572525,
      "grad_norm": 0.852875828742981,
      "learning_rate": 1.2280671049783037e-05,
      "loss": 0.1688,
      "step": 2032
    },
    {
      "epoch": 1.275207777952015,
      "grad_norm": 0.9001780152320862,
      "learning_rate": 1.226192885704054e-05,
      "loss": 0.1567,
      "step": 2033
    },
    {
      "epoch": 1.2758350321467775,
      "grad_norm": 0.8690427541732788,
      "learning_rate": 1.2243194652581288e-05,
      "loss": 0.1473,
      "step": 2034
    },
    {
      "epoch": 1.2764622863415398,
      "grad_norm": 0.8091393709182739,
      "learning_rate": 1.2224468455745266e-05,
      "loss": 0.1603,
      "step": 2035
    },
    {
      "epoch": 1.2770895405363023,
      "grad_norm": 0.8217348456382751,
      "learning_rate": 1.2205750285864188e-05,
      "loss": 0.1657,
      "step": 2036
    },
    {
      "epoch": 1.2777167947310648,
      "grad_norm": 0.8783489465713501,
      "learning_rate": 1.2187040162261475e-05,
      "loss": 0.1197,
      "step": 2037
    },
    {
      "epoch": 1.278344048925827,
      "grad_norm": 0.7721499800682068,
      "learning_rate": 1.2168338104252247e-05,
      "loss": 0.1313,
      "step": 2038
    },
    {
      "epoch": 1.2789713031205896,
      "grad_norm": 0.8439178466796875,
      "learning_rate": 1.2149644131143303e-05,
      "loss": 0.1257,
      "step": 2039
    },
    {
      "epoch": 1.279598557315352,
      "grad_norm": 0.8752883672714233,
      "learning_rate": 1.213095826223308e-05,
      "loss": 0.1364,
      "step": 2040
    },
    {
      "epoch": 1.2802258115101144,
      "grad_norm": 0.8068944811820984,
      "learning_rate": 1.2112280516811667e-05,
      "loss": 0.1578,
      "step": 2041
    },
    {
      "epoch": 1.280853065704877,
      "grad_norm": 0.779299795627594,
      "learning_rate": 1.2093610914160759e-05,
      "loss": 0.1328,
      "step": 2042
    },
    {
      "epoch": 1.2814803198996394,
      "grad_norm": 0.808593213558197,
      "learning_rate": 1.2074949473553636e-05,
      "loss": 0.1445,
      "step": 2043
    },
    {
      "epoch": 1.2821075740944017,
      "grad_norm": 0.7457171082496643,
      "learning_rate": 1.205629621425516e-05,
      "loss": 0.1281,
      "step": 2044
    },
    {
      "epoch": 1.2827348282891642,
      "grad_norm": 0.7942360043525696,
      "learning_rate": 1.203765115552175e-05,
      "loss": 0.1379,
      "step": 2045
    },
    {
      "epoch": 1.2833620824839267,
      "grad_norm": 0.8352336883544922,
      "learning_rate": 1.201901431660136e-05,
      "loss": 0.1461,
      "step": 2046
    },
    {
      "epoch": 1.283989336678689,
      "grad_norm": 0.7953838109970093,
      "learning_rate": 1.2000385716733449e-05,
      "loss": 0.1471,
      "step": 2047
    },
    {
      "epoch": 1.2846165908734515,
      "grad_norm": 0.8717171549797058,
      "learning_rate": 1.1981765375148985e-05,
      "loss": 0.1448,
      "step": 2048
    },
    {
      "epoch": 1.285243845068214,
      "grad_norm": 0.7027206420898438,
      "learning_rate": 1.1963153311070392e-05,
      "loss": 0.1176,
      "step": 2049
    },
    {
      "epoch": 1.2858710992629763,
      "grad_norm": 0.7868241667747498,
      "learning_rate": 1.1944549543711563e-05,
      "loss": 0.146,
      "step": 2050
    },
    {
      "epoch": 1.2864983534577388,
      "grad_norm": 0.9464139938354492,
      "learning_rate": 1.1925954092277827e-05,
      "loss": 0.1678,
      "step": 2051
    },
    {
      "epoch": 1.2871256076525013,
      "grad_norm": 0.7622083425521851,
      "learning_rate": 1.1907366975965916e-05,
      "loss": 0.1458,
      "step": 2052
    },
    {
      "epoch": 1.2877528618472636,
      "grad_norm": 0.7768915891647339,
      "learning_rate": 1.1888788213963966e-05,
      "loss": 0.133,
      "step": 2053
    },
    {
      "epoch": 1.288380116042026,
      "grad_norm": 0.8117609024047852,
      "learning_rate": 1.1870217825451493e-05,
      "loss": 0.1599,
      "step": 2054
    },
    {
      "epoch": 1.2890073702367886,
      "grad_norm": 0.8793721199035645,
      "learning_rate": 1.1851655829599355e-05,
      "loss": 0.1462,
      "step": 2055
    },
    {
      "epoch": 1.2896346244315509,
      "grad_norm": 0.8287612795829773,
      "learning_rate": 1.1833102245569755e-05,
      "loss": 0.1656,
      "step": 2056
    },
    {
      "epoch": 1.2902618786263134,
      "grad_norm": 0.736072301864624,
      "learning_rate": 1.1814557092516218e-05,
      "loss": 0.1303,
      "step": 2057
    },
    {
      "epoch": 1.2908891328210759,
      "grad_norm": 0.802575409412384,
      "learning_rate": 1.179602038958355e-05,
      "loss": 0.1532,
      "step": 2058
    },
    {
      "epoch": 1.2915163870158382,
      "grad_norm": 0.8144501447677612,
      "learning_rate": 1.1777492155907844e-05,
      "loss": 0.1316,
      "step": 2059
    },
    {
      "epoch": 1.2921436412106007,
      "grad_norm": 0.8047012090682983,
      "learning_rate": 1.1758972410616452e-05,
      "loss": 0.1355,
      "step": 2060
    },
    {
      "epoch": 1.292770895405363,
      "grad_norm": 0.8402323722839355,
      "learning_rate": 1.1740461172827953e-05,
      "loss": 0.1408,
      "step": 2061
    },
    {
      "epoch": 1.2933981496001254,
      "grad_norm": 0.781400203704834,
      "learning_rate": 1.1721958461652156e-05,
      "loss": 0.1382,
      "step": 2062
    },
    {
      "epoch": 1.294025403794888,
      "grad_norm": 0.9326108694076538,
      "learning_rate": 1.1703464296190056e-05,
      "loss": 0.1801,
      "step": 2063
    },
    {
      "epoch": 1.2946526579896502,
      "grad_norm": 0.7394958138465881,
      "learning_rate": 1.1684978695533831e-05,
      "loss": 0.1244,
      "step": 2064
    },
    {
      "epoch": 1.2952799121844127,
      "grad_norm": 0.8357262015342712,
      "learning_rate": 1.166650167876682e-05,
      "loss": 0.1501,
      "step": 2065
    },
    {
      "epoch": 1.295907166379175,
      "grad_norm": 0.7975276112556458,
      "learning_rate": 1.1648033264963503e-05,
      "loss": 0.1338,
      "step": 2066
    },
    {
      "epoch": 1.2965344205739375,
      "grad_norm": 0.8282522559165955,
      "learning_rate": 1.1629573473189461e-05,
      "loss": 0.1567,
      "step": 2067
    },
    {
      "epoch": 1.2971616747687,
      "grad_norm": 0.8194451332092285,
      "learning_rate": 1.1611122322501403e-05,
      "loss": 0.1302,
      "step": 2068
    },
    {
      "epoch": 1.2977889289634623,
      "grad_norm": 0.9897724390029907,
      "learning_rate": 1.1592679831947083e-05,
      "loss": 0.1664,
      "step": 2069
    },
    {
      "epoch": 1.2984161831582248,
      "grad_norm": 0.8145666122436523,
      "learning_rate": 1.1574246020565341e-05,
      "loss": 0.1356,
      "step": 2070
    },
    {
      "epoch": 1.2990434373529873,
      "grad_norm": 0.7804147601127625,
      "learning_rate": 1.155582090738605e-05,
      "loss": 0.1444,
      "step": 2071
    },
    {
      "epoch": 1.2996706915477496,
      "grad_norm": 0.7265585064888,
      "learning_rate": 1.1537404511430106e-05,
      "loss": 0.1287,
      "step": 2072
    },
    {
      "epoch": 1.3002979457425121,
      "grad_norm": 0.8573222756385803,
      "learning_rate": 1.1518996851709408e-05,
      "loss": 0.1432,
      "step": 2073
    },
    {
      "epoch": 1.3009251999372746,
      "grad_norm": 0.893895149230957,
      "learning_rate": 1.1500597947226818e-05,
      "loss": 0.1472,
      "step": 2074
    },
    {
      "epoch": 1.301552454132037,
      "grad_norm": 0.8899487853050232,
      "learning_rate": 1.1482207816976174e-05,
      "loss": 0.1469,
      "step": 2075
    },
    {
      "epoch": 1.3021797083267994,
      "grad_norm": 0.8849236369132996,
      "learning_rate": 1.1463826479942261e-05,
      "loss": 0.1775,
      "step": 2076
    },
    {
      "epoch": 1.302806962521562,
      "grad_norm": 0.974793553352356,
      "learning_rate": 1.144545395510078e-05,
      "loss": 0.145,
      "step": 2077
    },
    {
      "epoch": 1.3034342167163242,
      "grad_norm": 0.8298199772834778,
      "learning_rate": 1.1427090261418332e-05,
      "loss": 0.1311,
      "step": 2078
    },
    {
      "epoch": 1.3040614709110867,
      "grad_norm": 0.9482800364494324,
      "learning_rate": 1.1408735417852414e-05,
      "loss": 0.1453,
      "step": 2079
    },
    {
      "epoch": 1.3046887251058492,
      "grad_norm": 0.831490695476532,
      "learning_rate": 1.1390389443351361e-05,
      "loss": 0.1385,
      "step": 2080
    },
    {
      "epoch": 1.3053159793006115,
      "grad_norm": 0.9137451648712158,
      "learning_rate": 1.1372052356854382e-05,
      "loss": 0.1524,
      "step": 2081
    },
    {
      "epoch": 1.305943233495374,
      "grad_norm": 0.7532322406768799,
      "learning_rate": 1.1353724177291485e-05,
      "loss": 0.1246,
      "step": 2082
    },
    {
      "epoch": 1.3065704876901365,
      "grad_norm": 0.8048055768013,
      "learning_rate": 1.1335404923583507e-05,
      "loss": 0.1384,
      "step": 2083
    },
    {
      "epoch": 1.3071977418848988,
      "grad_norm": 0.8215882778167725,
      "learning_rate": 1.131709461464206e-05,
      "loss": 0.165,
      "step": 2084
    },
    {
      "epoch": 1.3078249960796613,
      "grad_norm": 0.7621756196022034,
      "learning_rate": 1.1298793269369503e-05,
      "loss": 0.1272,
      "step": 2085
    },
    {
      "epoch": 1.3084522502744238,
      "grad_norm": 0.7866765260696411,
      "learning_rate": 1.128050090665897e-05,
      "loss": 0.1617,
      "step": 2086
    },
    {
      "epoch": 1.309079504469186,
      "grad_norm": 0.8137136697769165,
      "learning_rate": 1.1262217545394312e-05,
      "loss": 0.1435,
      "step": 2087
    },
    {
      "epoch": 1.3097067586639486,
      "grad_norm": 0.891494870185852,
      "learning_rate": 1.1243943204450086e-05,
      "loss": 0.1516,
      "step": 2088
    },
    {
      "epoch": 1.310334012858711,
      "grad_norm": 0.863868772983551,
      "learning_rate": 1.1225677902691537e-05,
      "loss": 0.1487,
      "step": 2089
    },
    {
      "epoch": 1.3109612670534734,
      "grad_norm": 0.8466315269470215,
      "learning_rate": 1.1207421658974585e-05,
      "loss": 0.1538,
      "step": 2090
    },
    {
      "epoch": 1.3115885212482359,
      "grad_norm": 0.9184805750846863,
      "learning_rate": 1.1189174492145778e-05,
      "loss": 0.1691,
      "step": 2091
    },
    {
      "epoch": 1.3122157754429984,
      "grad_norm": 0.8500133156776428,
      "learning_rate": 1.117093642104232e-05,
      "loss": 0.1371,
      "step": 2092
    },
    {
      "epoch": 1.3128430296377607,
      "grad_norm": 0.8608570694923401,
      "learning_rate": 1.115270746449202e-05,
      "loss": 0.1809,
      "step": 2093
    },
    {
      "epoch": 1.3134702838325232,
      "grad_norm": 0.8924458622932434,
      "learning_rate": 1.113448764131325e-05,
      "loss": 0.1551,
      "step": 2094
    },
    {
      "epoch": 1.3140975380272857,
      "grad_norm": 0.7744616866111755,
      "learning_rate": 1.1116276970314993e-05,
      "loss": 0.1326,
      "step": 2095
    },
    {
      "epoch": 1.314724792222048,
      "grad_norm": 0.8079293370246887,
      "learning_rate": 1.1098075470296756e-05,
      "loss": 0.1497,
      "step": 2096
    },
    {
      "epoch": 1.3153520464168105,
      "grad_norm": 0.9292060136795044,
      "learning_rate": 1.1079883160048604e-05,
      "loss": 0.1667,
      "step": 2097
    },
    {
      "epoch": 1.315979300611573,
      "grad_norm": 0.7225872874259949,
      "learning_rate": 1.1061700058351078e-05,
      "loss": 0.1397,
      "step": 2098
    },
    {
      "epoch": 1.3166065548063353,
      "grad_norm": 0.7413331866264343,
      "learning_rate": 1.1043526183975244e-05,
      "loss": 0.1194,
      "step": 2099
    },
    {
      "epoch": 1.3172338090010978,
      "grad_norm": 0.7993394732475281,
      "learning_rate": 1.1025361555682634e-05,
      "loss": 0.1374,
      "step": 2100
    },
    {
      "epoch": 1.31786106319586,
      "grad_norm": 0.7846838235855103,
      "learning_rate": 1.100720619222523e-05,
      "loss": 0.1577,
      "step": 2101
    },
    {
      "epoch": 1.3184883173906226,
      "grad_norm": 0.8102317452430725,
      "learning_rate": 1.0989060112345466e-05,
      "loss": 0.1337,
      "step": 2102
    },
    {
      "epoch": 1.319115571585385,
      "grad_norm": 0.9342648983001709,
      "learning_rate": 1.0970923334776155e-05,
      "loss": 0.1631,
      "step": 2103
    },
    {
      "epoch": 1.3197428257801473,
      "grad_norm": 0.8400986194610596,
      "learning_rate": 1.0952795878240546e-05,
      "loss": 0.1371,
      "step": 2104
    },
    {
      "epoch": 1.3203700799749098,
      "grad_norm": 0.7187572717666626,
      "learning_rate": 1.0934677761452248e-05,
      "loss": 0.1234,
      "step": 2105
    },
    {
      "epoch": 1.3209973341696721,
      "grad_norm": 0.7044083476066589,
      "learning_rate": 1.091656900311523e-05,
      "loss": 0.1224,
      "step": 2106
    },
    {
      "epoch": 1.3216245883644346,
      "grad_norm": 0.9172436594963074,
      "learning_rate": 1.0898469621923802e-05,
      "loss": 0.1749,
      "step": 2107
    },
    {
      "epoch": 1.3222518425591971,
      "grad_norm": 0.8661630153656006,
      "learning_rate": 1.08803796365626e-05,
      "loss": 0.1642,
      "step": 2108
    },
    {
      "epoch": 1.3228790967539594,
      "grad_norm": 1.0116245746612549,
      "learning_rate": 1.0862299065706536e-05,
      "loss": 0.169,
      "step": 2109
    },
    {
      "epoch": 1.323506350948722,
      "grad_norm": 0.7918195128440857,
      "learning_rate": 1.0844227928020828e-05,
      "loss": 0.1259,
      "step": 2110
    },
    {
      "epoch": 1.3241336051434844,
      "grad_norm": 0.9659956693649292,
      "learning_rate": 1.082616624216095e-05,
      "loss": 0.141,
      "step": 2111
    },
    {
      "epoch": 1.3247608593382467,
      "grad_norm": 0.8624740242958069,
      "learning_rate": 1.0808114026772612e-05,
      "loss": 0.1337,
      "step": 2112
    },
    {
      "epoch": 1.3253881135330092,
      "grad_norm": 1.061497449874878,
      "learning_rate": 1.0790071300491754e-05,
      "loss": 0.1813,
      "step": 2113
    },
    {
      "epoch": 1.3260153677277717,
      "grad_norm": 0.8342097401618958,
      "learning_rate": 1.0772038081944527e-05,
      "loss": 0.153,
      "step": 2114
    },
    {
      "epoch": 1.326642621922534,
      "grad_norm": 0.9330176115036011,
      "learning_rate": 1.0754014389747236e-05,
      "loss": 0.1259,
      "step": 2115
    },
    {
      "epoch": 1.3272698761172965,
      "grad_norm": 0.7098188996315002,
      "learning_rate": 1.0736000242506389e-05,
      "loss": 0.1224,
      "step": 2116
    },
    {
      "epoch": 1.327897130312059,
      "grad_norm": 0.7948756814002991,
      "learning_rate": 1.0717995658818616e-05,
      "loss": 0.1603,
      "step": 2117
    },
    {
      "epoch": 1.3285243845068213,
      "grad_norm": 0.8998217582702637,
      "learning_rate": 1.0700000657270688e-05,
      "loss": 0.1668,
      "step": 2118
    },
    {
      "epoch": 1.3291516387015838,
      "grad_norm": 0.9300773739814758,
      "learning_rate": 1.0682015256439487e-05,
      "loss": 0.1639,
      "step": 2119
    },
    {
      "epoch": 1.3297788928963463,
      "grad_norm": 0.9498723745346069,
      "learning_rate": 1.0664039474891961e-05,
      "loss": 0.1553,
      "step": 2120
    },
    {
      "epoch": 1.3304061470911086,
      "grad_norm": 0.8543471693992615,
      "learning_rate": 1.064607333118514e-05,
      "loss": 0.1323,
      "step": 2121
    },
    {
      "epoch": 1.331033401285871,
      "grad_norm": 0.7313621640205383,
      "learning_rate": 1.0628116843866114e-05,
      "loss": 0.1268,
      "step": 2122
    },
    {
      "epoch": 1.3316606554806336,
      "grad_norm": 0.8691371083259583,
      "learning_rate": 1.0610170031471995e-05,
      "loss": 0.1634,
      "step": 2123
    },
    {
      "epoch": 1.332287909675396,
      "grad_norm": 0.8829277753829956,
      "learning_rate": 1.0592232912529913e-05,
      "loss": 0.1572,
      "step": 2124
    },
    {
      "epoch": 1.3329151638701584,
      "grad_norm": 0.9373019337654114,
      "learning_rate": 1.0574305505556983e-05,
      "loss": 0.1677,
      "step": 2125
    },
    {
      "epoch": 1.333542418064921,
      "grad_norm": 0.6969695091247559,
      "learning_rate": 1.0556387829060308e-05,
      "loss": 0.1268,
      "step": 2126
    },
    {
      "epoch": 1.3341696722596832,
      "grad_norm": 0.804305374622345,
      "learning_rate": 1.0538479901536922e-05,
      "loss": 0.1354,
      "step": 2127
    },
    {
      "epoch": 1.3347969264544457,
      "grad_norm": 0.8717243075370789,
      "learning_rate": 1.0520581741473813e-05,
      "loss": 0.1543,
      "step": 2128
    },
    {
      "epoch": 1.3354241806492082,
      "grad_norm": 0.9425841569900513,
      "learning_rate": 1.0502693367347882e-05,
      "loss": 0.1507,
      "step": 2129
    },
    {
      "epoch": 1.3360514348439705,
      "grad_norm": 0.7681815028190613,
      "learning_rate": 1.0484814797625936e-05,
      "loss": 0.1347,
      "step": 2130
    },
    {
      "epoch": 1.336678689038733,
      "grad_norm": 0.8692004680633545,
      "learning_rate": 1.046694605076464e-05,
      "loss": 0.1577,
      "step": 2131
    },
    {
      "epoch": 1.3373059432334955,
      "grad_norm": 0.780180037021637,
      "learning_rate": 1.0449087145210543e-05,
      "loss": 0.1404,
      "step": 2132
    },
    {
      "epoch": 1.3379331974282578,
      "grad_norm": 0.7524372339248657,
      "learning_rate": 1.043123809940001e-05,
      "loss": 0.1369,
      "step": 2133
    },
    {
      "epoch": 1.3385604516230203,
      "grad_norm": 0.8653045296669006,
      "learning_rate": 1.0413398931759243e-05,
      "loss": 0.1605,
      "step": 2134
    },
    {
      "epoch": 1.3391877058177828,
      "grad_norm": 0.8803753852844238,
      "learning_rate": 1.0395569660704243e-05,
      "loss": 0.1603,
      "step": 2135
    },
    {
      "epoch": 1.339814960012545,
      "grad_norm": 0.8828588724136353,
      "learning_rate": 1.0377750304640794e-05,
      "loss": 0.1433,
      "step": 2136
    },
    {
      "epoch": 1.3404422142073076,
      "grad_norm": 0.9002618193626404,
      "learning_rate": 1.0359940881964454e-05,
      "loss": 0.1584,
      "step": 2137
    },
    {
      "epoch": 1.3410694684020699,
      "grad_norm": 0.7786149978637695,
      "learning_rate": 1.03421414110605e-05,
      "loss": 0.1446,
      "step": 2138
    },
    {
      "epoch": 1.3416967225968324,
      "grad_norm": 0.7834008932113647,
      "learning_rate": 1.0324351910303965e-05,
      "loss": 0.1397,
      "step": 2139
    },
    {
      "epoch": 1.3423239767915949,
      "grad_norm": 0.8710561990737915,
      "learning_rate": 1.0306572398059566e-05,
      "loss": 0.1281,
      "step": 2140
    },
    {
      "epoch": 1.3429512309863572,
      "grad_norm": 0.8711461424827576,
      "learning_rate": 1.028880289268173e-05,
      "loss": 0.1391,
      "step": 2141
    },
    {
      "epoch": 1.3435784851811197,
      "grad_norm": 0.854386031627655,
      "learning_rate": 1.0271043412514539e-05,
      "loss": 0.146,
      "step": 2142
    },
    {
      "epoch": 1.344205739375882,
      "grad_norm": 0.7929691672325134,
      "learning_rate": 1.0253293975891739e-05,
      "loss": 0.1254,
      "step": 2143
    },
    {
      "epoch": 1.3448329935706445,
      "grad_norm": 0.8346929550170898,
      "learning_rate": 1.0235554601136677e-05,
      "loss": 0.1246,
      "step": 2144
    },
    {
      "epoch": 1.345460247765407,
      "grad_norm": 0.7438914179801941,
      "learning_rate": 1.021782530656236e-05,
      "loss": 0.1282,
      "step": 2145
    },
    {
      "epoch": 1.3460875019601692,
      "grad_norm": 0.691235363483429,
      "learning_rate": 1.0200106110471335e-05,
      "loss": 0.1277,
      "step": 2146
    },
    {
      "epoch": 1.3467147561549317,
      "grad_norm": 0.7530035376548767,
      "learning_rate": 1.0182397031155769e-05,
      "loss": 0.1366,
      "step": 2147
    },
    {
      "epoch": 1.3473420103496943,
      "grad_norm": 0.9042640328407288,
      "learning_rate": 1.0164698086897365e-05,
      "loss": 0.1582,
      "step": 2148
    },
    {
      "epoch": 1.3479692645444565,
      "grad_norm": 0.8310264348983765,
      "learning_rate": 1.014700929596737e-05,
      "loss": 0.1189,
      "step": 2149
    },
    {
      "epoch": 1.348596518739219,
      "grad_norm": 0.8199609518051147,
      "learning_rate": 1.0129330676626537e-05,
      "loss": 0.1344,
      "step": 2150
    },
    {
      "epoch": 1.3492237729339815,
      "grad_norm": 0.7133363485336304,
      "learning_rate": 1.0111662247125131e-05,
      "loss": 0.1178,
      "step": 2151
    },
    {
      "epoch": 1.3498510271287438,
      "grad_norm": 0.8384912014007568,
      "learning_rate": 1.0094004025702894e-05,
      "loss": 0.1475,
      "step": 2152
    },
    {
      "epoch": 1.3504782813235063,
      "grad_norm": 0.8178553581237793,
      "learning_rate": 1.0076356030589033e-05,
      "loss": 0.1367,
      "step": 2153
    },
    {
      "epoch": 1.3511055355182688,
      "grad_norm": 0.7055070400238037,
      "learning_rate": 1.005871828000219e-05,
      "loss": 0.1074,
      "step": 2154
    },
    {
      "epoch": 1.3517327897130311,
      "grad_norm": 0.7852804660797119,
      "learning_rate": 1.0041090792150446e-05,
      "loss": 0.1443,
      "step": 2155
    },
    {
      "epoch": 1.3523600439077936,
      "grad_norm": 0.882070004940033,
      "learning_rate": 1.0023473585231258e-05,
      "loss": 0.1518,
      "step": 2156
    },
    {
      "epoch": 1.3529872981025561,
      "grad_norm": 0.8362320065498352,
      "learning_rate": 1.0005866677431503e-05,
      "loss": 0.1446,
      "step": 2157
    },
    {
      "epoch": 1.3536145522973184,
      "grad_norm": 0.976771891117096,
      "learning_rate": 9.988270086927407e-06,
      "loss": 0.1376,
      "step": 2158
    },
    {
      "epoch": 1.354241806492081,
      "grad_norm": 0.8807855248451233,
      "learning_rate": 9.970683831884554e-06,
      "loss": 0.1428,
      "step": 2159
    },
    {
      "epoch": 1.3548690606868434,
      "grad_norm": 0.8212952017784119,
      "learning_rate": 9.953107930457848e-06,
      "loss": 0.1441,
      "step": 2160
    },
    {
      "epoch": 1.3554963148816057,
      "grad_norm": 0.8011760115623474,
      "learning_rate": 9.935542400791525e-06,
      "loss": 0.1242,
      "step": 2161
    },
    {
      "epoch": 1.3561235690763682,
      "grad_norm": 0.7925354242324829,
      "learning_rate": 9.91798726101908e-06,
      "loss": 0.1198,
      "step": 2162
    },
    {
      "epoch": 1.3567508232711307,
      "grad_norm": 0.843718409538269,
      "learning_rate": 9.90044252926331e-06,
      "loss": 0.1294,
      "step": 2163
    },
    {
      "epoch": 1.357378077465893,
      "grad_norm": 0.90981525182724,
      "learning_rate": 9.882908223636261e-06,
      "loss": 0.1601,
      "step": 2164
    },
    {
      "epoch": 1.3580053316606555,
      "grad_norm": 0.7558876276016235,
      "learning_rate": 9.865384362239212e-06,
      "loss": 0.1311,
      "step": 2165
    },
    {
      "epoch": 1.358632585855418,
      "grad_norm": 0.8670477271080017,
      "learning_rate": 9.847870963162671e-06,
      "loss": 0.1396,
      "step": 2166
    },
    {
      "epoch": 1.3592598400501803,
      "grad_norm": 0.9721325039863586,
      "learning_rate": 9.830368044486322e-06,
      "loss": 0.1452,
      "step": 2167
    },
    {
      "epoch": 1.3598870942449428,
      "grad_norm": 0.9157195687294006,
      "learning_rate": 9.812875624279051e-06,
      "loss": 0.172,
      "step": 2168
    },
    {
      "epoch": 1.3605143484397053,
      "grad_norm": 0.7177037000656128,
      "learning_rate": 9.795393720598896e-06,
      "loss": 0.1239,
      "step": 2169
    },
    {
      "epoch": 1.3611416026344676,
      "grad_norm": 0.8860791921615601,
      "learning_rate": 9.777922351493045e-06,
      "loss": 0.1261,
      "step": 2170
    },
    {
      "epoch": 1.36176885682923,
      "grad_norm": 0.8846379518508911,
      "learning_rate": 9.760461534997816e-06,
      "loss": 0.1367,
      "step": 2171
    },
    {
      "epoch": 1.3623961110239926,
      "grad_norm": 0.7757281064987183,
      "learning_rate": 9.743011289138606e-06,
      "loss": 0.1177,
      "step": 2172
    },
    {
      "epoch": 1.363023365218755,
      "grad_norm": 0.8190012574195862,
      "learning_rate": 9.725571631929938e-06,
      "loss": 0.1431,
      "step": 2173
    },
    {
      "epoch": 1.3636506194135174,
      "grad_norm": 0.7819278240203857,
      "learning_rate": 9.708142581375362e-06,
      "loss": 0.1329,
      "step": 2174
    },
    {
      "epoch": 1.3642778736082797,
      "grad_norm": 0.8488762974739075,
      "learning_rate": 9.690724155467515e-06,
      "loss": 0.1272,
      "step": 2175
    },
    {
      "epoch": 1.3649051278030422,
      "grad_norm": 0.7727980613708496,
      "learning_rate": 9.673316372188047e-06,
      "loss": 0.1347,
      "step": 2176
    },
    {
      "epoch": 1.3655323819978047,
      "grad_norm": 0.7058163285255432,
      "learning_rate": 9.655919249507626e-06,
      "loss": 0.1192,
      "step": 2177
    },
    {
      "epoch": 1.366159636192567,
      "grad_norm": 0.9108099341392517,
      "learning_rate": 9.638532805385915e-06,
      "loss": 0.1378,
      "step": 2178
    },
    {
      "epoch": 1.3667868903873295,
      "grad_norm": 0.8259838819503784,
      "learning_rate": 9.62115705777156e-06,
      "loss": 0.1641,
      "step": 2179
    },
    {
      "epoch": 1.3674141445820918,
      "grad_norm": 0.7587583661079407,
      "learning_rate": 9.60379202460214e-06,
      "loss": 0.1308,
      "step": 2180
    },
    {
      "epoch": 1.3680413987768543,
      "grad_norm": 0.7948460578918457,
      "learning_rate": 9.586437723804204e-06,
      "loss": 0.1543,
      "step": 2181
    },
    {
      "epoch": 1.3686686529716168,
      "grad_norm": 0.7781640887260437,
      "learning_rate": 9.569094173293198e-06,
      "loss": 0.1281,
      "step": 2182
    },
    {
      "epoch": 1.369295907166379,
      "grad_norm": 0.8927295207977295,
      "learning_rate": 9.551761390973487e-06,
      "loss": 0.145,
      "step": 2183
    },
    {
      "epoch": 1.3699231613611416,
      "grad_norm": 1.0925737619400024,
      "learning_rate": 9.534439394738316e-06,
      "loss": 0.1768,
      "step": 2184
    },
    {
      "epoch": 1.370550415555904,
      "grad_norm": 0.886824369430542,
      "learning_rate": 9.51712820246978e-06,
      "loss": 0.1516,
      "step": 2185
    },
    {
      "epoch": 1.3711776697506664,
      "grad_norm": 0.8679983615875244,
      "learning_rate": 9.499827832038835e-06,
      "loss": 0.1275,
      "step": 2186
    },
    {
      "epoch": 1.3718049239454289,
      "grad_norm": 0.786601185798645,
      "learning_rate": 9.482538301305266e-06,
      "loss": 0.1204,
      "step": 2187
    },
    {
      "epoch": 1.3724321781401914,
      "grad_norm": 0.871163547039032,
      "learning_rate": 9.465259628117665e-06,
      "loss": 0.1466,
      "step": 2188
    },
    {
      "epoch": 1.3730594323349536,
      "grad_norm": 0.8171364665031433,
      "learning_rate": 9.447991830313412e-06,
      "loss": 0.1484,
      "step": 2189
    },
    {
      "epoch": 1.3736866865297161,
      "grad_norm": 0.709475576877594,
      "learning_rate": 9.43073492571867e-06,
      "loss": 0.1163,
      "step": 2190
    },
    {
      "epoch": 1.3743139407244787,
      "grad_norm": 0.8342685699462891,
      "learning_rate": 9.413488932148336e-06,
      "loss": 0.1522,
      "step": 2191
    },
    {
      "epoch": 1.374941194919241,
      "grad_norm": 0.8246030807495117,
      "learning_rate": 9.396253867406061e-06,
      "loss": 0.1484,
      "step": 2192
    },
    {
      "epoch": 1.3755684491140034,
      "grad_norm": 0.9686318635940552,
      "learning_rate": 9.379029749284216e-06,
      "loss": 0.2097,
      "step": 2193
    },
    {
      "epoch": 1.376195703308766,
      "grad_norm": 0.9869811534881592,
      "learning_rate": 9.361816595563855e-06,
      "loss": 0.1497,
      "step": 2194
    },
    {
      "epoch": 1.3768229575035282,
      "grad_norm": 0.7771713733673096,
      "learning_rate": 9.344614424014738e-06,
      "loss": 0.1242,
      "step": 2195
    },
    {
      "epoch": 1.3774502116982907,
      "grad_norm": 0.7462970614433289,
      "learning_rate": 9.327423252395252e-06,
      "loss": 0.1203,
      "step": 2196
    },
    {
      "epoch": 1.3780774658930532,
      "grad_norm": 0.9352504014968872,
      "learning_rate": 9.310243098452464e-06,
      "loss": 0.1713,
      "step": 2197
    },
    {
      "epoch": 1.3787047200878155,
      "grad_norm": 0.8267216682434082,
      "learning_rate": 9.293073979922041e-06,
      "loss": 0.1593,
      "step": 2198
    },
    {
      "epoch": 1.379331974282578,
      "grad_norm": 0.8669594526290894,
      "learning_rate": 9.275915914528271e-06,
      "loss": 0.1142,
      "step": 2199
    },
    {
      "epoch": 1.3799592284773405,
      "grad_norm": 1.0197635889053345,
      "learning_rate": 9.258768919984032e-06,
      "loss": 0.1358,
      "step": 2200
    },
    {
      "epoch": 1.3805864826721028,
      "grad_norm": 0.746362030506134,
      "learning_rate": 9.24163301399077e-06,
      "loss": 0.1249,
      "step": 2201
    },
    {
      "epoch": 1.3812137368668653,
      "grad_norm": 0.7596211433410645,
      "learning_rate": 9.22450821423849e-06,
      "loss": 0.1354,
      "step": 2202
    },
    {
      "epoch": 1.3818409910616278,
      "grad_norm": 0.7407150268554688,
      "learning_rate": 9.20739453840571e-06,
      "loss": 0.1444,
      "step": 2203
    },
    {
      "epoch": 1.3824682452563901,
      "grad_norm": 0.7925580143928528,
      "learning_rate": 9.190292004159491e-06,
      "loss": 0.1371,
      "step": 2204
    },
    {
      "epoch": 1.3830954994511526,
      "grad_norm": 0.764828085899353,
      "learning_rate": 9.173200629155379e-06,
      "loss": 0.1216,
      "step": 2205
    },
    {
      "epoch": 1.3837227536459151,
      "grad_norm": 0.8288745284080505,
      "learning_rate": 9.156120431037401e-06,
      "loss": 0.1504,
      "step": 2206
    },
    {
      "epoch": 1.3843500078406774,
      "grad_norm": 0.8584113717079163,
      "learning_rate": 9.139051427438048e-06,
      "loss": 0.15,
      "step": 2207
    },
    {
      "epoch": 1.38497726203544,
      "grad_norm": 0.8606441020965576,
      "learning_rate": 9.121993635978262e-06,
      "loss": 0.1503,
      "step": 2208
    },
    {
      "epoch": 1.3856045162302024,
      "grad_norm": 0.8186155557632446,
      "learning_rate": 9.104947074267385e-06,
      "loss": 0.1226,
      "step": 2209
    },
    {
      "epoch": 1.3862317704249647,
      "grad_norm": 0.7461327314376831,
      "learning_rate": 9.08791175990319e-06,
      "loss": 0.1476,
      "step": 2210
    },
    {
      "epoch": 1.3868590246197272,
      "grad_norm": 0.9064809083938599,
      "learning_rate": 9.070887710471832e-06,
      "loss": 0.1608,
      "step": 2211
    },
    {
      "epoch": 1.3874862788144895,
      "grad_norm": 0.9225057363510132,
      "learning_rate": 9.053874943547837e-06,
      "loss": 0.1507,
      "step": 2212
    },
    {
      "epoch": 1.388113533009252,
      "grad_norm": 0.7884825468063354,
      "learning_rate": 9.036873476694092e-06,
      "loss": 0.1394,
      "step": 2213
    },
    {
      "epoch": 1.3887407872040145,
      "grad_norm": 0.7843522429466248,
      "learning_rate": 9.019883327461794e-06,
      "loss": 0.1303,
      "step": 2214
    },
    {
      "epoch": 1.3893680413987768,
      "grad_norm": 0.7671374082565308,
      "learning_rate": 9.002904513390478e-06,
      "loss": 0.1265,
      "step": 2215
    },
    {
      "epoch": 1.3899952955935393,
      "grad_norm": 0.7901222109794617,
      "learning_rate": 8.985937052007973e-06,
      "loss": 0.1425,
      "step": 2216
    },
    {
      "epoch": 1.3906225497883016,
      "grad_norm": 0.8651731610298157,
      "learning_rate": 8.968980960830384e-06,
      "loss": 0.14,
      "step": 2217
    },
    {
      "epoch": 1.391249803983064,
      "grad_norm": 1.0074193477630615,
      "learning_rate": 8.952036257362086e-06,
      "loss": 0.1685,
      "step": 2218
    },
    {
      "epoch": 1.3918770581778266,
      "grad_norm": 0.803781270980835,
      "learning_rate": 8.935102959095697e-06,
      "loss": 0.1368,
      "step": 2219
    },
    {
      "epoch": 1.3925043123725889,
      "grad_norm": 0.8673598766326904,
      "learning_rate": 8.918181083512041e-06,
      "loss": 0.1201,
      "step": 2220
    },
    {
      "epoch": 1.3931315665673514,
      "grad_norm": 0.7912781238555908,
      "learning_rate": 8.901270648080175e-06,
      "loss": 0.1543,
      "step": 2221
    },
    {
      "epoch": 1.3937588207621139,
      "grad_norm": 0.8338043093681335,
      "learning_rate": 8.884371670257345e-06,
      "loss": 0.143,
      "step": 2222
    },
    {
      "epoch": 1.3943860749568762,
      "grad_norm": 0.922680675983429,
      "learning_rate": 8.867484167488946e-06,
      "loss": 0.1425,
      "step": 2223
    },
    {
      "epoch": 1.3950133291516387,
      "grad_norm": 0.7326045632362366,
      "learning_rate": 8.850608157208547e-06,
      "loss": 0.1283,
      "step": 2224
    },
    {
      "epoch": 1.3956405833464012,
      "grad_norm": 0.8759211897850037,
      "learning_rate": 8.833743656837847e-06,
      "loss": 0.1519,
      "step": 2225
    },
    {
      "epoch": 1.3962678375411635,
      "grad_norm": 0.8462759852409363,
      "learning_rate": 8.816890683786674e-06,
      "loss": 0.1214,
      "step": 2226
    },
    {
      "epoch": 1.396895091735926,
      "grad_norm": 0.9513422846794128,
      "learning_rate": 8.800049255452929e-06,
      "loss": 0.1515,
      "step": 2227
    },
    {
      "epoch": 1.3975223459306885,
      "grad_norm": 0.7908011674880981,
      "learning_rate": 8.783219389222616e-06,
      "loss": 0.1326,
      "step": 2228
    },
    {
      "epoch": 1.3981496001254508,
      "grad_norm": 0.7702666521072388,
      "learning_rate": 8.766401102469803e-06,
      "loss": 0.1398,
      "step": 2229
    },
    {
      "epoch": 1.3987768543202133,
      "grad_norm": 0.9492427110671997,
      "learning_rate": 8.7495944125566e-06,
      "loss": 0.16,
      "step": 2230
    },
    {
      "epoch": 1.3994041085149758,
      "grad_norm": 0.8880034685134888,
      "learning_rate": 8.732799336833147e-06,
      "loss": 0.1613,
      "step": 2231
    },
    {
      "epoch": 1.400031362709738,
      "grad_norm": 0.7824158072471619,
      "learning_rate": 8.71601589263758e-06,
      "loss": 0.1329,
      "step": 2232
    },
    {
      "epoch": 1.4006586169045006,
      "grad_norm": 0.8795452117919922,
      "learning_rate": 8.699244097296044e-06,
      "loss": 0.1497,
      "step": 2233
    },
    {
      "epoch": 1.401285871099263,
      "grad_norm": 0.8127107620239258,
      "learning_rate": 8.682483968122659e-06,
      "loss": 0.1474,
      "step": 2234
    },
    {
      "epoch": 1.4019131252940253,
      "grad_norm": 0.9448059797286987,
      "learning_rate": 8.665735522419491e-06,
      "loss": 0.1628,
      "step": 2235
    },
    {
      "epoch": 1.4025403794887878,
      "grad_norm": 0.84309321641922,
      "learning_rate": 8.648998777476556e-06,
      "loss": 0.1658,
      "step": 2236
    },
    {
      "epoch": 1.4031676336835504,
      "grad_norm": 0.814670979976654,
      "learning_rate": 8.632273750571786e-06,
      "loss": 0.1416,
      "step": 2237
    },
    {
      "epoch": 1.4037948878783126,
      "grad_norm": 0.7217344641685486,
      "learning_rate": 8.615560458971004e-06,
      "loss": 0.1204,
      "step": 2238
    },
    {
      "epoch": 1.4044221420730751,
      "grad_norm": 0.8013978600502014,
      "learning_rate": 8.598858919927937e-06,
      "loss": 0.1431,
      "step": 2239
    },
    {
      "epoch": 1.4050493962678376,
      "grad_norm": 0.7642830014228821,
      "learning_rate": 8.58216915068417e-06,
      "loss": 0.1377,
      "step": 2240
    },
    {
      "epoch": 1.4056766504626,
      "grad_norm": 0.9084851741790771,
      "learning_rate": 8.56549116846914e-06,
      "loss": 0.1417,
      "step": 2241
    },
    {
      "epoch": 1.4063039046573624,
      "grad_norm": 0.7790862917900085,
      "learning_rate": 8.548824990500123e-06,
      "loss": 0.146,
      "step": 2242
    },
    {
      "epoch": 1.406931158852125,
      "grad_norm": 0.9662855267524719,
      "learning_rate": 8.532170633982185e-06,
      "loss": 0.1202,
      "step": 2243
    },
    {
      "epoch": 1.4075584130468872,
      "grad_norm": 0.8067671656608582,
      "learning_rate": 8.515528116108214e-06,
      "loss": 0.1252,
      "step": 2244
    },
    {
      "epoch": 1.4081856672416497,
      "grad_norm": 0.8642739653587341,
      "learning_rate": 8.498897454058863e-06,
      "loss": 0.1304,
      "step": 2245
    },
    {
      "epoch": 1.4088129214364122,
      "grad_norm": 0.8284696340560913,
      "learning_rate": 8.482278665002556e-06,
      "loss": 0.1579,
      "step": 2246
    },
    {
      "epoch": 1.4094401756311745,
      "grad_norm": 0.8237106204032898,
      "learning_rate": 8.46567176609545e-06,
      "loss": 0.1441,
      "step": 2247
    },
    {
      "epoch": 1.410067429825937,
      "grad_norm": 0.8551800847053528,
      "learning_rate": 8.449076774481439e-06,
      "loss": 0.1277,
      "step": 2248
    },
    {
      "epoch": 1.4106946840206993,
      "grad_norm": 0.8437981605529785,
      "learning_rate": 8.432493707292111e-06,
      "loss": 0.1382,
      "step": 2249
    },
    {
      "epoch": 1.4113219382154618,
      "grad_norm": 0.7792211771011353,
      "learning_rate": 8.415922581646743e-06,
      "loss": 0.1492,
      "step": 2250
    },
    {
      "epoch": 1.4119491924102243,
      "grad_norm": 0.8435136079788208,
      "learning_rate": 8.399363414652298e-06,
      "loss": 0.1437,
      "step": 2251
    },
    {
      "epoch": 1.4125764466049866,
      "grad_norm": 0.8436769843101501,
      "learning_rate": 8.382816223403386e-06,
      "loss": 0.1361,
      "step": 2252
    },
    {
      "epoch": 1.413203700799749,
      "grad_norm": 0.8829277753829956,
      "learning_rate": 8.366281024982252e-06,
      "loss": 0.1494,
      "step": 2253
    },
    {
      "epoch": 1.4138309549945114,
      "grad_norm": 0.9682825803756714,
      "learning_rate": 8.349757836458771e-06,
      "loss": 0.1632,
      "step": 2254
    },
    {
      "epoch": 1.414458209189274,
      "grad_norm": 1.1217553615570068,
      "learning_rate": 8.333246674890417e-06,
      "loss": 0.1383,
      "step": 2255
    },
    {
      "epoch": 1.4150854633840364,
      "grad_norm": 0.7459167838096619,
      "learning_rate": 8.316747557322224e-06,
      "loss": 0.1099,
      "step": 2256
    },
    {
      "epoch": 1.4157127175787987,
      "grad_norm": 0.880784809589386,
      "learning_rate": 8.300260500786825e-06,
      "loss": 0.14,
      "step": 2257
    },
    {
      "epoch": 1.4163399717735612,
      "grad_norm": 0.9267962574958801,
      "learning_rate": 8.283785522304386e-06,
      "loss": 0.1327,
      "step": 2258
    },
    {
      "epoch": 1.4169672259683237,
      "grad_norm": 1.0290577411651611,
      "learning_rate": 8.267322638882606e-06,
      "loss": 0.1669,
      "step": 2259
    },
    {
      "epoch": 1.417594480163086,
      "grad_norm": 0.8680477738380432,
      "learning_rate": 8.250871867516713e-06,
      "loss": 0.1392,
      "step": 2260
    },
    {
      "epoch": 1.4182217343578485,
      "grad_norm": 1.068716287612915,
      "learning_rate": 8.234433225189393e-06,
      "loss": 0.1546,
      "step": 2261
    },
    {
      "epoch": 1.418848988552611,
      "grad_norm": 1.1250039339065552,
      "learning_rate": 8.218006728870848e-06,
      "loss": 0.1675,
      "step": 2262
    },
    {
      "epoch": 1.4194762427473733,
      "grad_norm": 0.9158503413200378,
      "learning_rate": 8.201592395518732e-06,
      "loss": 0.1378,
      "step": 2263
    },
    {
      "epoch": 1.4201034969421358,
      "grad_norm": 0.7678124308586121,
      "learning_rate": 8.185190242078132e-06,
      "loss": 0.1241,
      "step": 2264
    },
    {
      "epoch": 1.4207307511368983,
      "grad_norm": 1.0234214067459106,
      "learning_rate": 8.168800285481571e-06,
      "loss": 0.1592,
      "step": 2265
    },
    {
      "epoch": 1.4213580053316606,
      "grad_norm": 0.8566369414329529,
      "learning_rate": 8.152422542648988e-06,
      "loss": 0.157,
      "step": 2266
    },
    {
      "epoch": 1.421985259526423,
      "grad_norm": 0.9309266805648804,
      "learning_rate": 8.136057030487684e-06,
      "loss": 0.1463,
      "step": 2267
    },
    {
      "epoch": 1.4226125137211856,
      "grad_norm": 0.9169596433639526,
      "learning_rate": 8.119703765892366e-06,
      "loss": 0.1627,
      "step": 2268
    },
    {
      "epoch": 1.4232397679159479,
      "grad_norm": 0.8501594066619873,
      "learning_rate": 8.103362765745081e-06,
      "loss": 0.1558,
      "step": 2269
    },
    {
      "epoch": 1.4238670221107104,
      "grad_norm": 0.902891993522644,
      "learning_rate": 8.08703404691522e-06,
      "loss": 0.1365,
      "step": 2270
    },
    {
      "epoch": 1.4244942763054729,
      "grad_norm": 0.9229341745376587,
      "learning_rate": 8.070717626259497e-06,
      "loss": 0.1449,
      "step": 2271
    },
    {
      "epoch": 1.4251215305002352,
      "grad_norm": 0.7798023819923401,
      "learning_rate": 8.054413520621929e-06,
      "loss": 0.1418,
      "step": 2272
    },
    {
      "epoch": 1.4257487846949977,
      "grad_norm": 0.9616074562072754,
      "learning_rate": 8.038121746833807e-06,
      "loss": 0.1979,
      "step": 2273
    },
    {
      "epoch": 1.4263760388897602,
      "grad_norm": 0.8707701563835144,
      "learning_rate": 8.021842321713718e-06,
      "loss": 0.1491,
      "step": 2274
    },
    {
      "epoch": 1.4270032930845225,
      "grad_norm": 0.9088772535324097,
      "learning_rate": 8.005575262067466e-06,
      "loss": 0.1491,
      "step": 2275
    },
    {
      "epoch": 1.427630547279285,
      "grad_norm": 0.891663134098053,
      "learning_rate": 7.989320584688125e-06,
      "loss": 0.1414,
      "step": 2276
    },
    {
      "epoch": 1.4282578014740475,
      "grad_norm": 0.9096094965934753,
      "learning_rate": 7.973078306355965e-06,
      "loss": 0.1336,
      "step": 2277
    },
    {
      "epoch": 1.4288850556688097,
      "grad_norm": 0.8948990702629089,
      "learning_rate": 7.956848443838469e-06,
      "loss": 0.1348,
      "step": 2278
    },
    {
      "epoch": 1.4295123098635723,
      "grad_norm": 1.0273152589797974,
      "learning_rate": 7.940631013890283e-06,
      "loss": 0.154,
      "step": 2279
    },
    {
      "epoch": 1.4301395640583348,
      "grad_norm": 0.768093466758728,
      "learning_rate": 7.924426033253239e-06,
      "loss": 0.13,
      "step": 2280
    },
    {
      "epoch": 1.430766818253097,
      "grad_norm": 0.8098708391189575,
      "learning_rate": 7.908233518656304e-06,
      "loss": 0.1386,
      "step": 2281
    },
    {
      "epoch": 1.4313940724478595,
      "grad_norm": 0.990982174873352,
      "learning_rate": 7.892053486815585e-06,
      "loss": 0.1493,
      "step": 2282
    },
    {
      "epoch": 1.432021326642622,
      "grad_norm": 0.8665327429771423,
      "learning_rate": 7.8758859544343e-06,
      "loss": 0.1387,
      "step": 2283
    },
    {
      "epoch": 1.4326485808373843,
      "grad_norm": 0.9492656588554382,
      "learning_rate": 7.859730938202763e-06,
      "loss": 0.1488,
      "step": 2284
    },
    {
      "epoch": 1.4332758350321468,
      "grad_norm": 0.8344292044639587,
      "learning_rate": 7.843588454798357e-06,
      "loss": 0.1373,
      "step": 2285
    },
    {
      "epoch": 1.4339030892269091,
      "grad_norm": 0.7794461250305176,
      "learning_rate": 7.827458520885541e-06,
      "loss": 0.1182,
      "step": 2286
    },
    {
      "epoch": 1.4345303434216716,
      "grad_norm": 0.8400171995162964,
      "learning_rate": 7.811341153115813e-06,
      "loss": 0.121,
      "step": 2287
    },
    {
      "epoch": 1.4351575976164341,
      "grad_norm": 0.9185622930526733,
      "learning_rate": 7.795236368127698e-06,
      "loss": 0.1426,
      "step": 2288
    },
    {
      "epoch": 1.4357848518111964,
      "grad_norm": 0.8727949261665344,
      "learning_rate": 7.779144182546746e-06,
      "loss": 0.1224,
      "step": 2289
    },
    {
      "epoch": 1.436412106005959,
      "grad_norm": 0.7786707878112793,
      "learning_rate": 7.763064612985465e-06,
      "loss": 0.1445,
      "step": 2290
    },
    {
      "epoch": 1.4370393602007212,
      "grad_norm": 0.8271812200546265,
      "learning_rate": 7.746997676043372e-06,
      "loss": 0.1361,
      "step": 2291
    },
    {
      "epoch": 1.4376666143954837,
      "grad_norm": 0.7741566300392151,
      "learning_rate": 7.73094338830693e-06,
      "loss": 0.1596,
      "step": 2292
    },
    {
      "epoch": 1.4382938685902462,
      "grad_norm": 0.8974960446357727,
      "learning_rate": 7.714901766349548e-06,
      "loss": 0.1529,
      "step": 2293
    },
    {
      "epoch": 1.4389211227850085,
      "grad_norm": 0.9158388376235962,
      "learning_rate": 7.698872826731551e-06,
      "loss": 0.1371,
      "step": 2294
    },
    {
      "epoch": 1.439548376979771,
      "grad_norm": 0.9374398589134216,
      "learning_rate": 7.682856586000192e-06,
      "loss": 0.1659,
      "step": 2295
    },
    {
      "epoch": 1.4401756311745335,
      "grad_norm": 0.8173784613609314,
      "learning_rate": 7.666853060689581e-06,
      "loss": 0.1652,
      "step": 2296
    },
    {
      "epoch": 1.4408028853692958,
      "grad_norm": 0.9208851456642151,
      "learning_rate": 7.650862267320731e-06,
      "loss": 0.1512,
      "step": 2297
    },
    {
      "epoch": 1.4414301395640583,
      "grad_norm": 0.8968892097473145,
      "learning_rate": 7.634884222401497e-06,
      "loss": 0.1545,
      "step": 2298
    },
    {
      "epoch": 1.4420573937588208,
      "grad_norm": 0.8849417567253113,
      "learning_rate": 7.6189189424265805e-06,
      "loss": 0.1416,
      "step": 2299
    },
    {
      "epoch": 1.442684647953583,
      "grad_norm": 0.7522055506706238,
      "learning_rate": 7.602966443877509e-06,
      "loss": 0.1145,
      "step": 2300
    },
    {
      "epoch": 1.4433119021483456,
      "grad_norm": 0.8512992858886719,
      "learning_rate": 7.587026743222592e-06,
      "loss": 0.1479,
      "step": 2301
    },
    {
      "epoch": 1.443939156343108,
      "grad_norm": 1.0570125579833984,
      "learning_rate": 7.5710998569169595e-06,
      "loss": 0.1714,
      "step": 2302
    },
    {
      "epoch": 1.4445664105378704,
      "grad_norm": 0.8478622436523438,
      "learning_rate": 7.555185801402483e-06,
      "loss": 0.138,
      "step": 2303
    },
    {
      "epoch": 1.4451936647326329,
      "grad_norm": 0.7958086133003235,
      "learning_rate": 7.539284593107807e-06,
      "loss": 0.1362,
      "step": 2304
    },
    {
      "epoch": 1.4458209189273954,
      "grad_norm": 1.0083904266357422,
      "learning_rate": 7.523396248448314e-06,
      "loss": 0.1692,
      "step": 2305
    },
    {
      "epoch": 1.4464481731221577,
      "grad_norm": 0.5430143475532532,
      "learning_rate": 7.5075207838260945e-06,
      "loss": 0.1385,
      "step": 2306
    },
    {
      "epoch": 1.4470754273169202,
      "grad_norm": 0.7983195781707764,
      "learning_rate": 7.49165821562996e-06,
      "loss": 0.1292,
      "step": 2307
    },
    {
      "epoch": 1.4477026815116827,
      "grad_norm": 0.9012365341186523,
      "learning_rate": 7.475808560235385e-06,
      "loss": 0.1676,
      "step": 2308
    },
    {
      "epoch": 1.448329935706445,
      "grad_norm": 0.8816755414009094,
      "learning_rate": 7.4599718340045315e-06,
      "loss": 0.1437,
      "step": 2309
    },
    {
      "epoch": 1.4489571899012075,
      "grad_norm": 0.8799365758895874,
      "learning_rate": 7.444148053286211e-06,
      "loss": 0.1346,
      "step": 2310
    },
    {
      "epoch": 1.44958444409597,
      "grad_norm": 0.8972303867340088,
      "learning_rate": 7.42833723441587e-06,
      "loss": 0.153,
      "step": 2311
    },
    {
      "epoch": 1.4502116982907323,
      "grad_norm": 0.9287340044975281,
      "learning_rate": 7.41253939371557e-06,
      "loss": 0.1559,
      "step": 2312
    },
    {
      "epoch": 1.4508389524854948,
      "grad_norm": 0.7804572582244873,
      "learning_rate": 7.396754547493989e-06,
      "loss": 0.1198,
      "step": 2313
    },
    {
      "epoch": 1.4514662066802573,
      "grad_norm": 0.8070683479309082,
      "learning_rate": 7.380982712046365e-06,
      "loss": 0.1202,
      "step": 2314
    },
    {
      "epoch": 1.4520934608750196,
      "grad_norm": 0.8352468013763428,
      "learning_rate": 7.365223903654524e-06,
      "loss": 0.1347,
      "step": 2315
    },
    {
      "epoch": 1.452720715069782,
      "grad_norm": 0.7718518376350403,
      "learning_rate": 7.349478138586841e-06,
      "loss": 0.1399,
      "step": 2316
    },
    {
      "epoch": 1.4533479692645446,
      "grad_norm": 0.8753253221511841,
      "learning_rate": 7.333745433098221e-06,
      "loss": 0.1415,
      "step": 2317
    },
    {
      "epoch": 1.4539752234593069,
      "grad_norm": 0.805276095867157,
      "learning_rate": 7.318025803430089e-06,
      "loss": 0.131,
      "step": 2318
    },
    {
      "epoch": 1.4546024776540694,
      "grad_norm": 0.9128892421722412,
      "learning_rate": 7.302319265810383e-06,
      "loss": 0.1329,
      "step": 2319
    },
    {
      "epoch": 1.4552297318488319,
      "grad_norm": 0.8427088856697083,
      "learning_rate": 7.286625836453498e-06,
      "loss": 0.1357,
      "step": 2320
    },
    {
      "epoch": 1.4558569860435941,
      "grad_norm": 0.8445613384246826,
      "learning_rate": 7.270945531560323e-06,
      "loss": 0.1459,
      "step": 2321
    },
    {
      "epoch": 1.4564842402383567,
      "grad_norm": 0.8637903928756714,
      "learning_rate": 7.255278367318186e-06,
      "loss": 0.1341,
      "step": 2322
    },
    {
      "epoch": 1.4571114944331192,
      "grad_norm": 0.8728572726249695,
      "learning_rate": 7.239624359900852e-06,
      "loss": 0.1483,
      "step": 2323
    },
    {
      "epoch": 1.4577387486278814,
      "grad_norm": 1.0005639791488647,
      "learning_rate": 7.223983525468514e-06,
      "loss": 0.1511,
      "step": 2324
    },
    {
      "epoch": 1.458366002822644,
      "grad_norm": 0.823235034942627,
      "learning_rate": 7.208355880167746e-06,
      "loss": 0.1274,
      "step": 2325
    },
    {
      "epoch": 1.4589932570174062,
      "grad_norm": 0.9102848172187805,
      "learning_rate": 7.192741440131512e-06,
      "loss": 0.1474,
      "step": 2326
    },
    {
      "epoch": 1.4596205112121687,
      "grad_norm": 0.808292031288147,
      "learning_rate": 7.177140221479153e-06,
      "loss": 0.1299,
      "step": 2327
    },
    {
      "epoch": 1.4602477654069312,
      "grad_norm": 0.9165015816688538,
      "learning_rate": 7.161552240316356e-06,
      "loss": 0.1385,
      "step": 2328
    },
    {
      "epoch": 1.4608750196016935,
      "grad_norm": 0.8659415245056152,
      "learning_rate": 7.1459775127351446e-06,
      "loss": 0.1426,
      "step": 2329
    },
    {
      "epoch": 1.461502273796456,
      "grad_norm": 0.816756010055542,
      "learning_rate": 7.130416054813858e-06,
      "loss": 0.1402,
      "step": 2330
    },
    {
      "epoch": 1.4621295279912183,
      "grad_norm": 0.9069549441337585,
      "learning_rate": 7.114867882617143e-06,
      "loss": 0.1424,
      "step": 2331
    },
    {
      "epoch": 1.4627567821859808,
      "grad_norm": 0.9110038876533508,
      "learning_rate": 7.099333012195913e-06,
      "loss": 0.1834,
      "step": 2332
    },
    {
      "epoch": 1.4633840363807433,
      "grad_norm": 0.7811709046363831,
      "learning_rate": 7.083811459587364e-06,
      "loss": 0.1143,
      "step": 2333
    },
    {
      "epoch": 1.4640112905755056,
      "grad_norm": 1.0482608079910278,
      "learning_rate": 7.068303240814947e-06,
      "loss": 0.1392,
      "step": 2334
    },
    {
      "epoch": 1.4646385447702681,
      "grad_norm": 0.7268643379211426,
      "learning_rate": 7.05280837188834e-06,
      "loss": 0.124,
      "step": 2335
    },
    {
      "epoch": 1.4652657989650306,
      "grad_norm": 1.0208662748336792,
      "learning_rate": 7.03732686880344e-06,
      "loss": 0.1467,
      "step": 2336
    },
    {
      "epoch": 1.465893053159793,
      "grad_norm": 0.8626693487167358,
      "learning_rate": 7.021858747542356e-06,
      "loss": 0.1407,
      "step": 2337
    },
    {
      "epoch": 1.4665203073545554,
      "grad_norm": 0.7776216268539429,
      "learning_rate": 7.006404024073359e-06,
      "loss": 0.1161,
      "step": 2338
    },
    {
      "epoch": 1.467147561549318,
      "grad_norm": 0.8511480689048767,
      "learning_rate": 6.990962714350913e-06,
      "loss": 0.1597,
      "step": 2339
    },
    {
      "epoch": 1.4677748157440802,
      "grad_norm": 0.9013397693634033,
      "learning_rate": 6.975534834315625e-06,
      "loss": 0.1465,
      "step": 2340
    },
    {
      "epoch": 1.4684020699388427,
      "grad_norm": 1.035351037979126,
      "learning_rate": 6.9601203998942394e-06,
      "loss": 0.1426,
      "step": 2341
    },
    {
      "epoch": 1.4690293241336052,
      "grad_norm": 1.0567203760147095,
      "learning_rate": 6.944719426999624e-06,
      "loss": 0.1542,
      "step": 2342
    },
    {
      "epoch": 1.4696565783283675,
      "grad_norm": 0.9788340926170349,
      "learning_rate": 6.929331931530734e-06,
      "loss": 0.1605,
      "step": 2343
    },
    {
      "epoch": 1.47028383252313,
      "grad_norm": 0.8828837275505066,
      "learning_rate": 6.913957929372628e-06,
      "loss": 0.13,
      "step": 2344
    },
    {
      "epoch": 1.4709110867178925,
      "grad_norm": 0.8553276062011719,
      "learning_rate": 6.898597436396428e-06,
      "loss": 0.1285,
      "step": 2345
    },
    {
      "epoch": 1.4715383409126548,
      "grad_norm": 0.8594926595687866,
      "learning_rate": 6.883250468459317e-06,
      "loss": 0.1283,
      "step": 2346
    },
    {
      "epoch": 1.4721655951074173,
      "grad_norm": 0.8016849160194397,
      "learning_rate": 6.867917041404506e-06,
      "loss": 0.1166,
      "step": 2347
    },
    {
      "epoch": 1.4727928493021798,
      "grad_norm": 0.8469088077545166,
      "learning_rate": 6.852597171061239e-06,
      "loss": 0.1329,
      "step": 2348
    },
    {
      "epoch": 1.473420103496942,
      "grad_norm": 0.7968443632125854,
      "learning_rate": 6.837290873244744e-06,
      "loss": 0.1445,
      "step": 2349
    },
    {
      "epoch": 1.4740473576917046,
      "grad_norm": 0.8820021152496338,
      "learning_rate": 6.821998163756258e-06,
      "loss": 0.1442,
      "step": 2350
    },
    {
      "epoch": 1.474674611886467,
      "grad_norm": 0.8443328738212585,
      "learning_rate": 6.806719058382991e-06,
      "loss": 0.1397,
      "step": 2351
    },
    {
      "epoch": 1.4753018660812294,
      "grad_norm": 0.8024336695671082,
      "learning_rate": 6.791453572898088e-06,
      "loss": 0.1443,
      "step": 2352
    },
    {
      "epoch": 1.4759291202759919,
      "grad_norm": 0.7728897333145142,
      "learning_rate": 6.776201723060651e-06,
      "loss": 0.1235,
      "step": 2353
    },
    {
      "epoch": 1.4765563744707544,
      "grad_norm": 0.8861445188522339,
      "learning_rate": 6.760963524615711e-06,
      "loss": 0.1513,
      "step": 2354
    },
    {
      "epoch": 1.4771836286655167,
      "grad_norm": 0.9280038475990295,
      "learning_rate": 6.745738993294184e-06,
      "loss": 0.1581,
      "step": 2355
    },
    {
      "epoch": 1.4778108828602792,
      "grad_norm": 0.9169256687164307,
      "learning_rate": 6.730528144812893e-06,
      "loss": 0.1441,
      "step": 2356
    },
    {
      "epoch": 1.4784381370550417,
      "grad_norm": 0.9574286937713623,
      "learning_rate": 6.715330994874534e-06,
      "loss": 0.1394,
      "step": 2357
    },
    {
      "epoch": 1.479065391249804,
      "grad_norm": 0.9301524758338928,
      "learning_rate": 6.700147559167662e-06,
      "loss": 0.149,
      "step": 2358
    },
    {
      "epoch": 1.4796926454445665,
      "grad_norm": 0.8063904643058777,
      "learning_rate": 6.684977853366667e-06,
      "loss": 0.1208,
      "step": 2359
    },
    {
      "epoch": 1.480319899639329,
      "grad_norm": 0.927034854888916,
      "learning_rate": 6.669821893131783e-06,
      "loss": 0.1394,
      "step": 2360
    },
    {
      "epoch": 1.4809471538340913,
      "grad_norm": 0.7899417877197266,
      "learning_rate": 6.654679694109025e-06,
      "loss": 0.1121,
      "step": 2361
    },
    {
      "epoch": 1.4815744080288538,
      "grad_norm": 0.8486856818199158,
      "learning_rate": 6.639551271930225e-06,
      "loss": 0.1416,
      "step": 2362
    },
    {
      "epoch": 1.482201662223616,
      "grad_norm": 0.7845723032951355,
      "learning_rate": 6.624436642212983e-06,
      "loss": 0.1247,
      "step": 2363
    },
    {
      "epoch": 1.4828289164183786,
      "grad_norm": 0.8707875609397888,
      "learning_rate": 6.609335820560669e-06,
      "loss": 0.1515,
      "step": 2364
    },
    {
      "epoch": 1.483456170613141,
      "grad_norm": 0.8075608611106873,
      "learning_rate": 6.59424882256239e-06,
      "loss": 0.1463,
      "step": 2365
    },
    {
      "epoch": 1.4840834248079033,
      "grad_norm": 0.9238713979721069,
      "learning_rate": 6.579175663792991e-06,
      "loss": 0.1459,
      "step": 2366
    },
    {
      "epoch": 1.4847106790026658,
      "grad_norm": 0.832962155342102,
      "learning_rate": 6.564116359813011e-06,
      "loss": 0.1452,
      "step": 2367
    },
    {
      "epoch": 1.4853379331974281,
      "grad_norm": 0.9457533955574036,
      "learning_rate": 6.549070926168708e-06,
      "loss": 0.1586,
      "step": 2368
    },
    {
      "epoch": 1.4859651873921906,
      "grad_norm": 0.744782567024231,
      "learning_rate": 6.534039378392012e-06,
      "loss": 0.1159,
      "step": 2369
    },
    {
      "epoch": 1.4865924415869531,
      "grad_norm": 0.8590942621231079,
      "learning_rate": 6.519021732000519e-06,
      "loss": 0.1231,
      "step": 2370
    },
    {
      "epoch": 1.4872196957817154,
      "grad_norm": 0.8148005604743958,
      "learning_rate": 6.504018002497481e-06,
      "loss": 0.1512,
      "step": 2371
    },
    {
      "epoch": 1.487846949976478,
      "grad_norm": 0.8527929782867432,
      "learning_rate": 6.489028205371762e-06,
      "loss": 0.1511,
      "step": 2372
    },
    {
      "epoch": 1.4884742041712404,
      "grad_norm": 0.9304593205451965,
      "learning_rate": 6.4740523560978644e-06,
      "loss": 0.1425,
      "step": 2373
    },
    {
      "epoch": 1.4891014583660027,
      "grad_norm": 0.9466644525527954,
      "learning_rate": 6.4590904701358845e-06,
      "loss": 0.1528,
      "step": 2374
    },
    {
      "epoch": 1.4897287125607652,
      "grad_norm": 0.772213339805603,
      "learning_rate": 6.444142562931503e-06,
      "loss": 0.1157,
      "step": 2375
    },
    {
      "epoch": 1.4903559667555277,
      "grad_norm": 0.8148625493049622,
      "learning_rate": 6.429208649915974e-06,
      "loss": 0.1354,
      "step": 2376
    },
    {
      "epoch": 1.49098322095029,
      "grad_norm": 0.8912999629974365,
      "learning_rate": 6.414288746506106e-06,
      "loss": 0.1302,
      "step": 2377
    },
    {
      "epoch": 1.4916104751450525,
      "grad_norm": 0.9323480129241943,
      "learning_rate": 6.399382868104231e-06,
      "loss": 0.1335,
      "step": 2378
    },
    {
      "epoch": 1.492237729339815,
      "grad_norm": 0.8342500329017639,
      "learning_rate": 6.3844910300982095e-06,
      "loss": 0.1314,
      "step": 2379
    },
    {
      "epoch": 1.4928649835345773,
      "grad_norm": 0.8314619064331055,
      "learning_rate": 6.369613247861417e-06,
      "loss": 0.1294,
      "step": 2380
    },
    {
      "epoch": 1.4934922377293398,
      "grad_norm": 0.8739069700241089,
      "learning_rate": 6.354749536752709e-06,
      "loss": 0.1407,
      "step": 2381
    },
    {
      "epoch": 1.4941194919241023,
      "grad_norm": 0.9383721351623535,
      "learning_rate": 6.3398999121164185e-06,
      "loss": 0.1653,
      "step": 2382
    },
    {
      "epoch": 1.4947467461188646,
      "grad_norm": 1.0760291814804077,
      "learning_rate": 6.325064389282338e-06,
      "loss": 0.159,
      "step": 2383
    },
    {
      "epoch": 1.495374000313627,
      "grad_norm": 0.8461733460426331,
      "learning_rate": 6.310242983565704e-06,
      "loss": 0.1215,
      "step": 2384
    },
    {
      "epoch": 1.4960012545083896,
      "grad_norm": 0.7973076701164246,
      "learning_rate": 6.2954357102671636e-06,
      "loss": 0.1161,
      "step": 2385
    },
    {
      "epoch": 1.496628508703152,
      "grad_norm": 0.8130058646202087,
      "learning_rate": 6.2806425846727914e-06,
      "loss": 0.1276,
      "step": 2386
    },
    {
      "epoch": 1.4972557628979144,
      "grad_norm": 0.7450674176216125,
      "learning_rate": 6.265863622054053e-06,
      "loss": 0.1151,
      "step": 2387
    },
    {
      "epoch": 1.497883017092677,
      "grad_norm": 0.9922374486923218,
      "learning_rate": 6.251098837667792e-06,
      "loss": 0.1458,
      "step": 2388
    },
    {
      "epoch": 1.4985102712874392,
      "grad_norm": 0.8384523391723633,
      "learning_rate": 6.236348246756223e-06,
      "loss": 0.1408,
      "step": 2389
    },
    {
      "epoch": 1.4991375254822017,
      "grad_norm": 0.9825587868690491,
      "learning_rate": 6.221611864546884e-06,
      "loss": 0.1508,
      "step": 2390
    },
    {
      "epoch": 1.4997647796769642,
      "grad_norm": 0.6755114793777466,
      "learning_rate": 6.206889706252668e-06,
      "loss": 0.1055,
      "step": 2391
    },
    {
      "epoch": 1.5003920338717265,
      "grad_norm": 0.788532018661499,
      "learning_rate": 6.192181787071781e-06,
      "loss": 0.1165,
      "step": 2392
    },
    {
      "epoch": 1.501019288066489,
      "grad_norm": 0.9015203714370728,
      "learning_rate": 6.17748812218772e-06,
      "loss": 0.1237,
      "step": 2393
    },
    {
      "epoch": 1.5016465422612515,
      "grad_norm": 0.8451139330863953,
      "learning_rate": 6.162808726769276e-06,
      "loss": 0.1367,
      "step": 2394
    },
    {
      "epoch": 1.5022737964560138,
      "grad_norm": 0.7379468679428101,
      "learning_rate": 6.148143615970514e-06,
      "loss": 0.1031,
      "step": 2395
    },
    {
      "epoch": 1.5029010506507763,
      "grad_norm": 0.932864248752594,
      "learning_rate": 6.133492804930728e-06,
      "loss": 0.1188,
      "step": 2396
    },
    {
      "epoch": 1.5035283048455388,
      "grad_norm": 0.9268395304679871,
      "learning_rate": 6.118856308774472e-06,
      "loss": 0.136,
      "step": 2397
    },
    {
      "epoch": 1.504155559040301,
      "grad_norm": 0.7807483077049255,
      "learning_rate": 6.10423414261152e-06,
      "loss": 0.1183,
      "step": 2398
    },
    {
      "epoch": 1.5047828132350634,
      "grad_norm": 0.9302142858505249,
      "learning_rate": 6.0896263215368454e-06,
      "loss": 0.1417,
      "step": 2399
    },
    {
      "epoch": 1.505410067429826,
      "grad_norm": 0.8320773839950562,
      "learning_rate": 6.075032860630625e-06,
      "loss": 0.1358,
      "step": 2400
    },
    {
      "epoch": 1.5060373216245884,
      "grad_norm": 0.7353841066360474,
      "learning_rate": 6.060453774958186e-06,
      "loss": 0.1291,
      "step": 2401
    },
    {
      "epoch": 1.5066645758193506,
      "grad_norm": 0.7645012736320496,
      "learning_rate": 6.045889079570042e-06,
      "loss": 0.1086,
      "step": 2402
    },
    {
      "epoch": 1.5072918300141134,
      "grad_norm": 0.9877032041549683,
      "learning_rate": 6.0313387895018435e-06,
      "loss": 0.1613,
      "step": 2403
    },
    {
      "epoch": 1.5079190842088757,
      "grad_norm": 0.9352614283561707,
      "learning_rate": 6.016802919774358e-06,
      "loss": 0.1537,
      "step": 2404
    },
    {
      "epoch": 1.508546338403638,
      "grad_norm": 0.8109005093574524,
      "learning_rate": 6.002281485393482e-06,
      "loss": 0.1295,
      "step": 2405
    },
    {
      "epoch": 1.5091735925984004,
      "grad_norm": 0.8273382782936096,
      "learning_rate": 5.987774501350198e-06,
      "loss": 0.1185,
      "step": 2406
    },
    {
      "epoch": 1.509800846793163,
      "grad_norm": 0.8394461274147034,
      "learning_rate": 5.973281982620589e-06,
      "loss": 0.1049,
      "step": 2407
    },
    {
      "epoch": 1.5104281009879252,
      "grad_norm": 0.537996768951416,
      "learning_rate": 5.9588039441657766e-06,
      "loss": 0.1327,
      "step": 2408
    },
    {
      "epoch": 1.5110553551826877,
      "grad_norm": 0.9410959482192993,
      "learning_rate": 5.9443404009319565e-06,
      "loss": 0.1489,
      "step": 2409
    },
    {
      "epoch": 1.5116826093774502,
      "grad_norm": 0.9202421307563782,
      "learning_rate": 5.929891367850354e-06,
      "loss": 0.1627,
      "step": 2410
    },
    {
      "epoch": 1.5123098635722125,
      "grad_norm": 0.8550167083740234,
      "learning_rate": 5.915456859837212e-06,
      "loss": 0.1599,
      "step": 2411
    },
    {
      "epoch": 1.512937117766975,
      "grad_norm": 0.9108312726020813,
      "learning_rate": 5.901036891793785e-06,
      "loss": 0.1359,
      "step": 2412
    },
    {
      "epoch": 1.5135643719617375,
      "grad_norm": 0.8570768237113953,
      "learning_rate": 5.886631478606317e-06,
      "loss": 0.1366,
      "step": 2413
    },
    {
      "epoch": 1.5141916261564998,
      "grad_norm": 1.050724983215332,
      "learning_rate": 5.872240635146011e-06,
      "loss": 0.1494,
      "step": 2414
    },
    {
      "epoch": 1.5148188803512623,
      "grad_norm": 0.91434645652771,
      "learning_rate": 5.857864376269051e-06,
      "loss": 0.1391,
      "step": 2415
    },
    {
      "epoch": 1.5154461345460248,
      "grad_norm": 0.7957209348678589,
      "learning_rate": 5.843502716816552e-06,
      "loss": 0.1419,
      "step": 2416
    },
    {
      "epoch": 1.5160733887407871,
      "grad_norm": 0.881434977054596,
      "learning_rate": 5.8291556716145595e-06,
      "loss": 0.1381,
      "step": 2417
    },
    {
      "epoch": 1.5167006429355496,
      "grad_norm": 0.8157554268836975,
      "learning_rate": 5.814823255474043e-06,
      "loss": 0.1319,
      "step": 2418
    },
    {
      "epoch": 1.5173278971303121,
      "grad_norm": 0.9174538254737854,
      "learning_rate": 5.800505483190846e-06,
      "loss": 0.1293,
      "step": 2419
    },
    {
      "epoch": 1.5179551513250744,
      "grad_norm": 0.9651191830635071,
      "learning_rate": 5.786202369545712e-06,
      "loss": 0.1572,
      "step": 2420
    },
    {
      "epoch": 1.518582405519837,
      "grad_norm": 0.8735745549201965,
      "learning_rate": 5.771913929304256e-06,
      "loss": 0.121,
      "step": 2421
    },
    {
      "epoch": 1.5192096597145994,
      "grad_norm": 0.889881432056427,
      "learning_rate": 5.757640177216932e-06,
      "loss": 0.137,
      "step": 2422
    },
    {
      "epoch": 1.5198369139093617,
      "grad_norm": 0.784126341342926,
      "learning_rate": 5.743381128019036e-06,
      "loss": 0.137,
      "step": 2423
    },
    {
      "epoch": 1.5204641681041242,
      "grad_norm": 0.9778151512145996,
      "learning_rate": 5.7291367964306985e-06,
      "loss": 0.1468,
      "step": 2424
    },
    {
      "epoch": 1.5210914222988867,
      "grad_norm": 0.8209559321403503,
      "learning_rate": 5.71490719715683e-06,
      "loss": 0.1224,
      "step": 2425
    },
    {
      "epoch": 1.521718676493649,
      "grad_norm": 0.9283830523490906,
      "learning_rate": 5.700692344887149e-06,
      "loss": 0.1296,
      "step": 2426
    },
    {
      "epoch": 1.5223459306884115,
      "grad_norm": 0.8546496033668518,
      "learning_rate": 5.6864922542961496e-06,
      "loss": 0.122,
      "step": 2427
    },
    {
      "epoch": 1.522973184883174,
      "grad_norm": 0.856823742389679,
      "learning_rate": 5.672306940043093e-06,
      "loss": 0.1161,
      "step": 2428
    },
    {
      "epoch": 1.5236004390779363,
      "grad_norm": 0.8136732578277588,
      "learning_rate": 5.658136416771966e-06,
      "loss": 0.1287,
      "step": 2429
    },
    {
      "epoch": 1.5242276932726988,
      "grad_norm": 0.9049329161643982,
      "learning_rate": 5.643980699111507e-06,
      "loss": 0.1346,
      "step": 2430
    },
    {
      "epoch": 1.5248549474674613,
      "grad_norm": 0.8317587971687317,
      "learning_rate": 5.629839801675161e-06,
      "loss": 0.1389,
      "step": 2431
    },
    {
      "epoch": 1.5254822016622236,
      "grad_norm": 0.8538785576820374,
      "learning_rate": 5.615713739061068e-06,
      "loss": 0.1425,
      "step": 2432
    },
    {
      "epoch": 1.526109455856986,
      "grad_norm": 1.051171064376831,
      "learning_rate": 5.601602525852063e-06,
      "loss": 0.147,
      "step": 2433
    },
    {
      "epoch": 1.5267367100517486,
      "grad_norm": 0.8037875294685364,
      "learning_rate": 5.587506176615651e-06,
      "loss": 0.1166,
      "step": 2434
    },
    {
      "epoch": 1.5273639642465109,
      "grad_norm": 0.7876270413398743,
      "learning_rate": 5.573424705903987e-06,
      "loss": 0.1363,
      "step": 2435
    },
    {
      "epoch": 1.5279912184412732,
      "grad_norm": 0.8768879771232605,
      "learning_rate": 5.5593581282538755e-06,
      "loss": 0.1276,
      "step": 2436
    },
    {
      "epoch": 1.528618472636036,
      "grad_norm": 0.8771581649780273,
      "learning_rate": 5.545306458186728e-06,
      "loss": 0.1408,
      "step": 2437
    },
    {
      "epoch": 1.5292457268307982,
      "grad_norm": 0.8572736382484436,
      "learning_rate": 5.531269710208582e-06,
      "loss": 0.1241,
      "step": 2438
    },
    {
      "epoch": 1.5298729810255605,
      "grad_norm": 0.8630357980728149,
      "learning_rate": 5.517247898810072e-06,
      "loss": 0.1242,
      "step": 2439
    },
    {
      "epoch": 1.5305002352203232,
      "grad_norm": 0.9446988105773926,
      "learning_rate": 5.503241038466403e-06,
      "loss": 0.1338,
      "step": 2440
    },
    {
      "epoch": 1.5311274894150855,
      "grad_norm": 0.9041091799736023,
      "learning_rate": 5.48924914363735e-06,
      "loss": 0.1302,
      "step": 2441
    },
    {
      "epoch": 1.5317547436098478,
      "grad_norm": 0.9294798970222473,
      "learning_rate": 5.475272228767248e-06,
      "loss": 0.1445,
      "step": 2442
    },
    {
      "epoch": 1.5323819978046103,
      "grad_norm": 0.8078564405441284,
      "learning_rate": 5.461310308284941e-06,
      "loss": 0.1208,
      "step": 2443
    },
    {
      "epoch": 1.5330092519993728,
      "grad_norm": 0.8502971529960632,
      "learning_rate": 5.447363396603822e-06,
      "loss": 0.1247,
      "step": 2444
    },
    {
      "epoch": 1.533636506194135,
      "grad_norm": 0.8516446948051453,
      "learning_rate": 5.433431508121778e-06,
      "loss": 0.1428,
      "step": 2445
    },
    {
      "epoch": 1.5342637603888976,
      "grad_norm": 0.9445379376411438,
      "learning_rate": 5.419514657221185e-06,
      "loss": 0.1303,
      "step": 2446
    },
    {
      "epoch": 1.53489101458366,
      "grad_norm": 1.029537558555603,
      "learning_rate": 5.405612858268907e-06,
      "loss": 0.1331,
      "step": 2447
    },
    {
      "epoch": 1.5355182687784223,
      "grad_norm": 0.8077148795127869,
      "learning_rate": 5.391726125616248e-06,
      "loss": 0.1246,
      "step": 2448
    },
    {
      "epoch": 1.5361455229731849,
      "grad_norm": 0.8353478908538818,
      "learning_rate": 5.377854473598976e-06,
      "loss": 0.1435,
      "step": 2449
    },
    {
      "epoch": 1.5367727771679474,
      "grad_norm": 0.8239552974700928,
      "learning_rate": 5.363997916537287e-06,
      "loss": 0.1339,
      "step": 2450
    },
    {
      "epoch": 1.5374000313627096,
      "grad_norm": 0.7541061639785767,
      "learning_rate": 5.350156468735792e-06,
      "loss": 0.1248,
      "step": 2451
    },
    {
      "epoch": 1.5380272855574721,
      "grad_norm": 0.7956366539001465,
      "learning_rate": 5.3363301444835035e-06,
      "loss": 0.1181,
      "step": 2452
    },
    {
      "epoch": 1.5386545397522347,
      "grad_norm": 0.9982238411903381,
      "learning_rate": 5.3225189580538304e-06,
      "loss": 0.1423,
      "step": 2453
    },
    {
      "epoch": 1.539281793946997,
      "grad_norm": 0.9753595590591431,
      "learning_rate": 5.308722923704542e-06,
      "loss": 0.1282,
      "step": 2454
    },
    {
      "epoch": 1.5399090481417594,
      "grad_norm": 0.9468235969543457,
      "learning_rate": 5.294942055677759e-06,
      "loss": 0.1324,
      "step": 2455
    },
    {
      "epoch": 1.540536302336522,
      "grad_norm": 1.0198054313659668,
      "learning_rate": 5.281176368199967e-06,
      "loss": 0.1355,
      "step": 2456
    },
    {
      "epoch": 1.5411635565312842,
      "grad_norm": 1.0024781227111816,
      "learning_rate": 5.267425875481971e-06,
      "loss": 0.1334,
      "step": 2457
    },
    {
      "epoch": 1.5417908107260467,
      "grad_norm": 0.9851219654083252,
      "learning_rate": 5.253690591718885e-06,
      "loss": 0.1347,
      "step": 2458
    },
    {
      "epoch": 1.5424180649208092,
      "grad_norm": 0.789305567741394,
      "learning_rate": 5.239970531090128e-06,
      "loss": 0.1562,
      "step": 2459
    },
    {
      "epoch": 1.5430453191155715,
      "grad_norm": 0.7955427169799805,
      "learning_rate": 5.226265707759408e-06,
      "loss": 0.1251,
      "step": 2460
    },
    {
      "epoch": 1.543672573310334,
      "grad_norm": 0.8244566917419434,
      "learning_rate": 5.212576135874683e-06,
      "loss": 0.1199,
      "step": 2461
    },
    {
      "epoch": 1.5442998275050965,
      "grad_norm": 0.8037431240081787,
      "learning_rate": 5.1989018295681855e-06,
      "loss": 0.1174,
      "step": 2462
    },
    {
      "epoch": 1.5449270816998588,
      "grad_norm": 0.8412739038467407,
      "learning_rate": 5.185242802956387e-06,
      "loss": 0.1297,
      "step": 2463
    },
    {
      "epoch": 1.5455543358946213,
      "grad_norm": 0.9513403177261353,
      "learning_rate": 5.171599070139976e-06,
      "loss": 0.1319,
      "step": 2464
    },
    {
      "epoch": 1.5461815900893838,
      "grad_norm": 0.806678056716919,
      "learning_rate": 5.1579706452038695e-06,
      "loss": 0.1091,
      "step": 2465
    },
    {
      "epoch": 1.5468088442841461,
      "grad_norm": 0.8771329522132874,
      "learning_rate": 5.144357542217155e-06,
      "loss": 0.1339,
      "step": 2466
    },
    {
      "epoch": 1.5474360984789086,
      "grad_norm": 0.8276905417442322,
      "learning_rate": 5.130759775233125e-06,
      "loss": 0.1428,
      "step": 2467
    },
    {
      "epoch": 1.5480633526736711,
      "grad_norm": 0.8733172416687012,
      "learning_rate": 5.117177358289234e-06,
      "loss": 0.1434,
      "step": 2468
    },
    {
      "epoch": 1.5486906068684334,
      "grad_norm": 0.8888440728187561,
      "learning_rate": 5.1036103054070876e-06,
      "loss": 0.1376,
      "step": 2469
    },
    {
      "epoch": 1.549317861063196,
      "grad_norm": 0.810010552406311,
      "learning_rate": 5.090058630592434e-06,
      "loss": 0.1232,
      "step": 2470
    },
    {
      "epoch": 1.5499451152579584,
      "grad_norm": 0.8017184734344482,
      "learning_rate": 5.076522347835147e-06,
      "loss": 0.1197,
      "step": 2471
    },
    {
      "epoch": 1.5505723694527207,
      "grad_norm": 0.8911049365997314,
      "learning_rate": 5.0630014711091965e-06,
      "loss": 0.14,
      "step": 2472
    },
    {
      "epoch": 1.551199623647483,
      "grad_norm": 0.7320021390914917,
      "learning_rate": 5.04949601437267e-06,
      "loss": 0.1095,
      "step": 2473
    },
    {
      "epoch": 1.5518268778422457,
      "grad_norm": 0.9221493601799011,
      "learning_rate": 5.03600599156772e-06,
      "loss": 0.1368,
      "step": 2474
    },
    {
      "epoch": 1.552454132037008,
      "grad_norm": 0.8870469927787781,
      "learning_rate": 5.022531416620577e-06,
      "loss": 0.1362,
      "step": 2475
    },
    {
      "epoch": 1.5530813862317703,
      "grad_norm": 0.9370519518852234,
      "learning_rate": 5.009072303441516e-06,
      "loss": 0.1277,
      "step": 2476
    },
    {
      "epoch": 1.553708640426533,
      "grad_norm": 0.9179979562759399,
      "learning_rate": 4.9956286659248585e-06,
      "loss": 0.1386,
      "step": 2477
    },
    {
      "epoch": 1.5543358946212953,
      "grad_norm": 0.7572410106658936,
      "learning_rate": 4.9822005179489365e-06,
      "loss": 0.1081,
      "step": 2478
    },
    {
      "epoch": 1.5549631488160576,
      "grad_norm": 0.9240134954452515,
      "learning_rate": 4.968787873376102e-06,
      "loss": 0.1454,
      "step": 2479
    },
    {
      "epoch": 1.5555904030108203,
      "grad_norm": 0.7326767444610596,
      "learning_rate": 4.9553907460527084e-06,
      "loss": 0.1135,
      "step": 2480
    },
    {
      "epoch": 1.5562176572055826,
      "grad_norm": 0.825032114982605,
      "learning_rate": 4.942009149809068e-06,
      "loss": 0.1298,
      "step": 2481
    },
    {
      "epoch": 1.5568449114003449,
      "grad_norm": 0.9178546667098999,
      "learning_rate": 4.928643098459485e-06,
      "loss": 0.1362,
      "step": 2482
    },
    {
      "epoch": 1.5574721655951074,
      "grad_norm": 0.7995597720146179,
      "learning_rate": 4.915292605802202e-06,
      "loss": 0.1216,
      "step": 2483
    },
    {
      "epoch": 1.5580994197898699,
      "grad_norm": 0.8662647008895874,
      "learning_rate": 4.901957685619398e-06,
      "loss": 0.1247,
      "step": 2484
    },
    {
      "epoch": 1.5587266739846322,
      "grad_norm": 0.8464028835296631,
      "learning_rate": 4.888638351677184e-06,
      "loss": 0.15,
      "step": 2485
    },
    {
      "epoch": 1.5593539281793947,
      "grad_norm": 0.8262073993682861,
      "learning_rate": 4.8753346177255755e-06,
      "loss": 0.1082,
      "step": 2486
    },
    {
      "epoch": 1.5599811823741572,
      "grad_norm": 0.9026674032211304,
      "learning_rate": 4.86204649749849e-06,
      "loss": 0.1163,
      "step": 2487
    },
    {
      "epoch": 1.5606084365689195,
      "grad_norm": 0.8266077637672424,
      "learning_rate": 4.848774004713714e-06,
      "loss": 0.122,
      "step": 2488
    },
    {
      "epoch": 1.561235690763682,
      "grad_norm": 0.955435574054718,
      "learning_rate": 4.835517153072924e-06,
      "loss": 0.1325,
      "step": 2489
    },
    {
      "epoch": 1.5618629449584445,
      "grad_norm": 0.8240264654159546,
      "learning_rate": 4.822275956261617e-06,
      "loss": 0.147,
      "step": 2490
    },
    {
      "epoch": 1.5624901991532067,
      "grad_norm": 0.9496490359306335,
      "learning_rate": 4.809050427949156e-06,
      "loss": 0.1397,
      "step": 2491
    },
    {
      "epoch": 1.5631174533479693,
      "grad_norm": 0.9616060256958008,
      "learning_rate": 4.795840581788715e-06,
      "loss": 0.1487,
      "step": 2492
    },
    {
      "epoch": 1.5637447075427318,
      "grad_norm": 0.736541211605072,
      "learning_rate": 4.782646431417288e-06,
      "loss": 0.1207,
      "step": 2493
    },
    {
      "epoch": 1.564371961737494,
      "grad_norm": 0.845628023147583,
      "learning_rate": 4.769467990455663e-06,
      "loss": 0.1271,
      "step": 2494
    },
    {
      "epoch": 1.5649992159322565,
      "grad_norm": 0.9478265047073364,
      "learning_rate": 4.7563052725083995e-06,
      "loss": 0.1478,
      "step": 2495
    },
    {
      "epoch": 1.565626470127019,
      "grad_norm": 0.8801705837249756,
      "learning_rate": 4.743158291163838e-06,
      "loss": 0.1515,
      "step": 2496
    },
    {
      "epoch": 1.5662537243217813,
      "grad_norm": 0.9115733504295349,
      "learning_rate": 4.730027059994073e-06,
      "loss": 0.1133,
      "step": 2497
    },
    {
      "epoch": 1.5668809785165438,
      "grad_norm": 0.8305857181549072,
      "learning_rate": 4.7169115925549356e-06,
      "loss": 0.1234,
      "step": 2498
    },
    {
      "epoch": 1.5675082327113063,
      "grad_norm": 0.8001347184181213,
      "learning_rate": 4.703811902385984e-06,
      "loss": 0.1146,
      "step": 2499
    },
    {
      "epoch": 1.5681354869060686,
      "grad_norm": 0.9758232235908508,
      "learning_rate": 4.690728003010496e-06,
      "loss": 0.1534,
      "step": 2500
    },
    {
      "epoch": 1.5687627411008311,
      "grad_norm": 0.8125451803207397,
      "learning_rate": 4.677659907935428e-06,
      "loss": 0.1211,
      "step": 2501
    },
    {
      "epoch": 1.5693899952955936,
      "grad_norm": 0.7471377849578857,
      "learning_rate": 4.664607630651443e-06,
      "loss": 0.1113,
      "step": 2502
    },
    {
      "epoch": 1.570017249490356,
      "grad_norm": 0.9791470170021057,
      "learning_rate": 4.651571184632866e-06,
      "loss": 0.1512,
      "step": 2503
    },
    {
      "epoch": 1.5706445036851184,
      "grad_norm": 0.7867892980575562,
      "learning_rate": 4.6385505833376755e-06,
      "loss": 0.1191,
      "step": 2504
    },
    {
      "epoch": 1.571271757879881,
      "grad_norm": 0.8695036172866821,
      "learning_rate": 4.625545840207501e-06,
      "loss": 0.122,
      "step": 2505
    },
    {
      "epoch": 1.5718990120746432,
      "grad_norm": 0.8763251304626465,
      "learning_rate": 4.612556968667603e-06,
      "loss": 0.1349,
      "step": 2506
    },
    {
      "epoch": 1.5725262662694057,
      "grad_norm": 0.9683310389518738,
      "learning_rate": 4.59958398212684e-06,
      "loss": 0.153,
      "step": 2507
    },
    {
      "epoch": 1.5731535204641682,
      "grad_norm": 0.777633547782898,
      "learning_rate": 4.586626893977681e-06,
      "loss": 0.1204,
      "step": 2508
    },
    {
      "epoch": 1.5737807746589305,
      "grad_norm": 0.9412129521369934,
      "learning_rate": 4.57368571759619e-06,
      "loss": 0.1661,
      "step": 2509
    },
    {
      "epoch": 1.5744080288536928,
      "grad_norm": 0.9540891647338867,
      "learning_rate": 4.5607604663419955e-06,
      "loss": 0.1441,
      "step": 2510
    },
    {
      "epoch": 1.5750352830484555,
      "grad_norm": 0.8474165201187134,
      "learning_rate": 4.547851153558293e-06,
      "loss": 0.1125,
      "step": 2511
    },
    {
      "epoch": 1.5756625372432178,
      "grad_norm": 0.7620456218719482,
      "learning_rate": 4.534957792571826e-06,
      "loss": 0.1125,
      "step": 2512
    },
    {
      "epoch": 1.57628979143798,
      "grad_norm": 0.7852665185928345,
      "learning_rate": 4.522080396692852e-06,
      "loss": 0.1172,
      "step": 2513
    },
    {
      "epoch": 1.5769170456327428,
      "grad_norm": 0.8036596775054932,
      "learning_rate": 4.509218979215166e-06,
      "loss": 0.1255,
      "step": 2514
    },
    {
      "epoch": 1.577544299827505,
      "grad_norm": 0.7978661060333252,
      "learning_rate": 4.4963735534160625e-06,
      "loss": 0.1309,
      "step": 2515
    },
    {
      "epoch": 1.5781715540222674,
      "grad_norm": 0.7998348474502563,
      "learning_rate": 4.483544132556328e-06,
      "loss": 0.1087,
      "step": 2516
    },
    {
      "epoch": 1.5787988082170301,
      "grad_norm": 0.8894103169441223,
      "learning_rate": 4.470730729880222e-06,
      "loss": 0.1476,
      "step": 2517
    },
    {
      "epoch": 1.5794260624117924,
      "grad_norm": 0.8098848462104797,
      "learning_rate": 4.457933358615478e-06,
      "loss": 0.1247,
      "step": 2518
    },
    {
      "epoch": 1.5800533166065547,
      "grad_norm": 0.7475268840789795,
      "learning_rate": 4.445152031973263e-06,
      "loss": 0.1193,
      "step": 2519
    },
    {
      "epoch": 1.5806805708013172,
      "grad_norm": 0.8451122641563416,
      "learning_rate": 4.432386763148197e-06,
      "loss": 0.1528,
      "step": 2520
    },
    {
      "epoch": 1.5813078249960797,
      "grad_norm": 0.8940575122833252,
      "learning_rate": 4.419637565318313e-06,
      "loss": 0.1447,
      "step": 2521
    },
    {
      "epoch": 1.581935079190842,
      "grad_norm": 1.0932834148406982,
      "learning_rate": 4.40690445164506e-06,
      "loss": 0.1416,
      "step": 2522
    },
    {
      "epoch": 1.5825623333856045,
      "grad_norm": 0.8093680143356323,
      "learning_rate": 4.394187435273278e-06,
      "loss": 0.1442,
      "step": 2523
    },
    {
      "epoch": 1.583189587580367,
      "grad_norm": 0.7737075686454773,
      "learning_rate": 4.381486529331196e-06,
      "loss": 0.112,
      "step": 2524
    },
    {
      "epoch": 1.5838168417751293,
      "grad_norm": 0.8742368221282959,
      "learning_rate": 4.368801746930396e-06,
      "loss": 0.1209,
      "step": 2525
    },
    {
      "epoch": 1.5844440959698918,
      "grad_norm": 0.8883172869682312,
      "learning_rate": 4.356133101165829e-06,
      "loss": 0.1252,
      "step": 2526
    },
    {
      "epoch": 1.5850713501646543,
      "grad_norm": 0.8819630146026611,
      "learning_rate": 4.343480605115787e-06,
      "loss": 0.1195,
      "step": 2527
    },
    {
      "epoch": 1.5856986043594166,
      "grad_norm": 0.8639972805976868,
      "learning_rate": 4.330844271841885e-06,
      "loss": 0.1405,
      "step": 2528
    },
    {
      "epoch": 1.586325858554179,
      "grad_norm": 0.8901621103286743,
      "learning_rate": 4.3182241143890605e-06,
      "loss": 0.1451,
      "step": 2529
    },
    {
      "epoch": 1.5869531127489416,
      "grad_norm": 0.7380594611167908,
      "learning_rate": 4.305620145785536e-06,
      "loss": 0.1122,
      "step": 2530
    },
    {
      "epoch": 1.5875803669437039,
      "grad_norm": 0.8391431570053101,
      "learning_rate": 4.2930323790428404e-06,
      "loss": 0.1464,
      "step": 2531
    },
    {
      "epoch": 1.5882076211384664,
      "grad_norm": 0.7736996412277222,
      "learning_rate": 4.280460827155764e-06,
      "loss": 0.1121,
      "step": 2532
    },
    {
      "epoch": 1.5888348753332289,
      "grad_norm": 0.7368905544281006,
      "learning_rate": 4.267905503102358e-06,
      "loss": 0.1255,
      "step": 2533
    },
    {
      "epoch": 1.5894621295279912,
      "grad_norm": 0.7692631483078003,
      "learning_rate": 4.2553664198439356e-06,
      "loss": 0.1386,
      "step": 2534
    },
    {
      "epoch": 1.5900893837227537,
      "grad_norm": 0.9218795895576477,
      "learning_rate": 4.242843590325028e-06,
      "loss": 0.1337,
      "step": 2535
    },
    {
      "epoch": 1.5907166379175162,
      "grad_norm": 0.8344318866729736,
      "learning_rate": 4.230337027473399e-06,
      "loss": 0.1419,
      "step": 2536
    },
    {
      "epoch": 1.5913438921122784,
      "grad_norm": 0.8635849952697754,
      "learning_rate": 4.217846744200003e-06,
      "loss": 0.1236,
      "step": 2537
    },
    {
      "epoch": 1.591971146307041,
      "grad_norm": 0.7777072787284851,
      "learning_rate": 4.205372753399006e-06,
      "loss": 0.0933,
      "step": 2538
    },
    {
      "epoch": 1.5925984005018035,
      "grad_norm": 0.9210771322250366,
      "learning_rate": 4.192915067947747e-06,
      "loss": 0.1345,
      "step": 2539
    },
    {
      "epoch": 1.5932256546965657,
      "grad_norm": 0.8880153894424438,
      "learning_rate": 4.180473700706733e-06,
      "loss": 0.1363,
      "step": 2540
    },
    {
      "epoch": 1.5938529088913282,
      "grad_norm": 0.7697297930717468,
      "learning_rate": 4.1680486645196235e-06,
      "loss": 0.1174,
      "step": 2541
    },
    {
      "epoch": 1.5944801630860908,
      "grad_norm": 0.7962117195129395,
      "learning_rate": 4.15563997221323e-06,
      "loss": 0.1193,
      "step": 2542
    },
    {
      "epoch": 1.595107417280853,
      "grad_norm": 0.8743627071380615,
      "learning_rate": 4.143247636597469e-06,
      "loss": 0.1508,
      "step": 2543
    },
    {
      "epoch": 1.5957346714756155,
      "grad_norm": 0.8989070653915405,
      "learning_rate": 4.1308716704653926e-06,
      "loss": 0.1421,
      "step": 2544
    },
    {
      "epoch": 1.596361925670378,
      "grad_norm": 0.8142381310462952,
      "learning_rate": 4.118512086593143e-06,
      "loss": 0.1182,
      "step": 2545
    },
    {
      "epoch": 1.5969891798651403,
      "grad_norm": 0.9252510070800781,
      "learning_rate": 4.106168897739955e-06,
      "loss": 0.1669,
      "step": 2546
    },
    {
      "epoch": 1.5976164340599026,
      "grad_norm": 0.8701350688934326,
      "learning_rate": 4.093842116648146e-06,
      "loss": 0.1618,
      "step": 2547
    },
    {
      "epoch": 1.5982436882546653,
      "grad_norm": 0.9177315831184387,
      "learning_rate": 4.081531756043069e-06,
      "loss": 0.1338,
      "step": 2548
    },
    {
      "epoch": 1.5988709424494276,
      "grad_norm": 0.7954108119010925,
      "learning_rate": 4.06923782863315e-06,
      "loss": 0.1295,
      "step": 2549
    },
    {
      "epoch": 1.59949819664419,
      "grad_norm": 0.8146564364433289,
      "learning_rate": 4.056960347109846e-06,
      "loss": 0.1243,
      "step": 2550
    },
    {
      "epoch": 1.6001254508389526,
      "grad_norm": 0.8106422424316406,
      "learning_rate": 4.044699324147632e-06,
      "loss": 0.1272,
      "step": 2551
    },
    {
      "epoch": 1.600752705033715,
      "grad_norm": 0.7793592214584351,
      "learning_rate": 4.032454772403993e-06,
      "loss": 0.1333,
      "step": 2552
    },
    {
      "epoch": 1.6013799592284772,
      "grad_norm": 0.9862551689147949,
      "learning_rate": 4.020226704519416e-06,
      "loss": 0.1681,
      "step": 2553
    },
    {
      "epoch": 1.60200721342324,
      "grad_norm": 0.852508008480072,
      "learning_rate": 4.008015133117356e-06,
      "loss": 0.1148,
      "step": 2554
    },
    {
      "epoch": 1.6026344676180022,
      "grad_norm": 0.9125605821609497,
      "learning_rate": 3.995820070804253e-06,
      "loss": 0.1256,
      "step": 2555
    },
    {
      "epoch": 1.6032617218127645,
      "grad_norm": 0.8080053329467773,
      "learning_rate": 3.983641530169497e-06,
      "loss": 0.1151,
      "step": 2556
    },
    {
      "epoch": 1.603888976007527,
      "grad_norm": 0.8587335348129272,
      "learning_rate": 3.971479523785435e-06,
      "loss": 0.1102,
      "step": 2557
    },
    {
      "epoch": 1.6045162302022895,
      "grad_norm": 0.9160087704658508,
      "learning_rate": 3.959334064207318e-06,
      "loss": 0.1205,
      "step": 2558
    },
    {
      "epoch": 1.6051434843970518,
      "grad_norm": 0.965299129486084,
      "learning_rate": 3.947205163973347e-06,
      "loss": 0.1425,
      "step": 2559
    },
    {
      "epoch": 1.6057707385918143,
      "grad_norm": 0.8616262674331665,
      "learning_rate": 3.935092835604597e-06,
      "loss": 0.1215,
      "step": 2560
    },
    {
      "epoch": 1.6063979927865768,
      "grad_norm": 0.8299267292022705,
      "learning_rate": 3.922997091605059e-06,
      "loss": 0.1329,
      "step": 2561
    },
    {
      "epoch": 1.607025246981339,
      "grad_norm": 0.8477101922035217,
      "learning_rate": 3.9109179444615965e-06,
      "loss": 0.1245,
      "step": 2562
    },
    {
      "epoch": 1.6076525011761016,
      "grad_norm": 0.8761797547340393,
      "learning_rate": 3.898855406643935e-06,
      "loss": 0.1444,
      "step": 2563
    },
    {
      "epoch": 1.608279755370864,
      "grad_norm": 0.7521011233329773,
      "learning_rate": 3.886809490604661e-06,
      "loss": 0.1182,
      "step": 2564
    },
    {
      "epoch": 1.6089070095656264,
      "grad_norm": 0.8995316624641418,
      "learning_rate": 3.874780208779201e-06,
      "loss": 0.1287,
      "step": 2565
    },
    {
      "epoch": 1.6095342637603889,
      "grad_norm": 0.8634257912635803,
      "learning_rate": 3.862767573585795e-06,
      "loss": 0.1505,
      "step": 2566
    },
    {
      "epoch": 1.6101615179551514,
      "grad_norm": 0.8332936763763428,
      "learning_rate": 3.850771597425515e-06,
      "loss": 0.1372,
      "step": 2567
    },
    {
      "epoch": 1.6107887721499137,
      "grad_norm": 1.0028297901153564,
      "learning_rate": 3.83879229268223e-06,
      "loss": 0.1472,
      "step": 2568
    },
    {
      "epoch": 1.6114160263446762,
      "grad_norm": 0.8762323260307312,
      "learning_rate": 3.826829671722596e-06,
      "loss": 0.1237,
      "step": 2569
    },
    {
      "epoch": 1.6120432805394387,
      "grad_norm": 0.8835314512252808,
      "learning_rate": 3.81488374689605e-06,
      "loss": 0.1247,
      "step": 2570
    },
    {
      "epoch": 1.612670534734201,
      "grad_norm": 0.8781018257141113,
      "learning_rate": 3.802954530534795e-06,
      "loss": 0.1283,
      "step": 2571
    },
    {
      "epoch": 1.6132977889289635,
      "grad_norm": 0.706692636013031,
      "learning_rate": 3.7910420349537714e-06,
      "loss": 0.0883,
      "step": 2572
    },
    {
      "epoch": 1.613925043123726,
      "grad_norm": 0.8565866351127625,
      "learning_rate": 3.779146272450671e-06,
      "loss": 0.1276,
      "step": 2573
    },
    {
      "epoch": 1.6145522973184883,
      "grad_norm": 0.8782655596733093,
      "learning_rate": 3.767267255305911e-06,
      "loss": 0.1428,
      "step": 2574
    },
    {
      "epoch": 1.6151795515132508,
      "grad_norm": 0.8743089437484741,
      "learning_rate": 3.7554049957826166e-06,
      "loss": 0.1405,
      "step": 2575
    },
    {
      "epoch": 1.6158068057080133,
      "grad_norm": 0.9464399218559265,
      "learning_rate": 3.7435595061266216e-06,
      "loss": 0.1462,
      "step": 2576
    },
    {
      "epoch": 1.6164340599027756,
      "grad_norm": 0.9497423768043518,
      "learning_rate": 3.731730798566433e-06,
      "loss": 0.1555,
      "step": 2577
    },
    {
      "epoch": 1.617061314097538,
      "grad_norm": 0.877797544002533,
      "learning_rate": 3.719918885313247e-06,
      "loss": 0.1341,
      "step": 2578
    },
    {
      "epoch": 1.6176885682923006,
      "grad_norm": 0.8285330533981323,
      "learning_rate": 3.7081237785609147e-06,
      "loss": 0.1228,
      "step": 2579
    },
    {
      "epoch": 1.6183158224870628,
      "grad_norm": 0.9165721535682678,
      "learning_rate": 3.696345490485944e-06,
      "loss": 0.1289,
      "step": 2580
    },
    {
      "epoch": 1.6189430766818254,
      "grad_norm": 0.7632472515106201,
      "learning_rate": 3.6845840332474715e-06,
      "loss": 0.127,
      "step": 2581
    },
    {
      "epoch": 1.6195703308765879,
      "grad_norm": 0.8485530018806458,
      "learning_rate": 3.672839418987275e-06,
      "loss": 0.1282,
      "step": 2582
    },
    {
      "epoch": 1.6201975850713501,
      "grad_norm": 0.8329845070838928,
      "learning_rate": 3.6611116598297213e-06,
      "loss": 0.1356,
      "step": 2583
    },
    {
      "epoch": 1.6208248392661124,
      "grad_norm": 0.84341961145401,
      "learning_rate": 3.64940076788179e-06,
      "loss": 0.1126,
      "step": 2584
    },
    {
      "epoch": 1.6214520934608752,
      "grad_norm": 0.9287056922912598,
      "learning_rate": 3.637706755233048e-06,
      "loss": 0.1452,
      "step": 2585
    },
    {
      "epoch": 1.6220793476556374,
      "grad_norm": 0.7329952716827393,
      "learning_rate": 3.626029633955639e-06,
      "loss": 0.1081,
      "step": 2586
    },
    {
      "epoch": 1.6227066018503997,
      "grad_norm": 0.8448726534843445,
      "learning_rate": 3.6143694161042643e-06,
      "loss": 0.1279,
      "step": 2587
    },
    {
      "epoch": 1.6233338560451624,
      "grad_norm": 0.9451857805252075,
      "learning_rate": 3.6027261137161774e-06,
      "loss": 0.15,
      "step": 2588
    },
    {
      "epoch": 1.6239611102399247,
      "grad_norm": 0.9256661534309387,
      "learning_rate": 3.5910997388111767e-06,
      "loss": 0.1312,
      "step": 2589
    },
    {
      "epoch": 1.624588364434687,
      "grad_norm": 0.8317173719406128,
      "learning_rate": 3.579490303391564e-06,
      "loss": 0.1181,
      "step": 2590
    },
    {
      "epoch": 1.6252156186294497,
      "grad_norm": 0.8810442686080933,
      "learning_rate": 3.5678978194421787e-06,
      "loss": 0.1285,
      "step": 2591
    },
    {
      "epoch": 1.625842872824212,
      "grad_norm": 0.8828345537185669,
      "learning_rate": 3.5563222989303458e-06,
      "loss": 0.1187,
      "step": 2592
    },
    {
      "epoch": 1.6264701270189743,
      "grad_norm": 0.8922680616378784,
      "learning_rate": 3.544763753805882e-06,
      "loss": 0.142,
      "step": 2593
    },
    {
      "epoch": 1.6270973812137368,
      "grad_norm": 0.8479432463645935,
      "learning_rate": 3.5332221960010892e-06,
      "loss": 0.1437,
      "step": 2594
    },
    {
      "epoch": 1.6277246354084993,
      "grad_norm": 0.7755201458930969,
      "learning_rate": 3.5216976374307123e-06,
      "loss": 0.129,
      "step": 2595
    },
    {
      "epoch": 1.6283518896032616,
      "grad_norm": 0.7489088773727417,
      "learning_rate": 3.5101900899919626e-06,
      "loss": 0.1231,
      "step": 2596
    },
    {
      "epoch": 1.628979143798024,
      "grad_norm": 0.8229769468307495,
      "learning_rate": 3.4986995655644874e-06,
      "loss": 0.1143,
      "step": 2597
    },
    {
      "epoch": 1.6296063979927866,
      "grad_norm": 0.9471097588539124,
      "learning_rate": 3.487226076010359e-06,
      "loss": 0.1612,
      "step": 2598
    },
    {
      "epoch": 1.630233652187549,
      "grad_norm": 0.8467300534248352,
      "learning_rate": 3.4757696331740666e-06,
      "loss": 0.137,
      "step": 2599
    },
    {
      "epoch": 1.6308609063823114,
      "grad_norm": 0.8977757692337036,
      "learning_rate": 3.4643302488825017e-06,
      "loss": 0.1168,
      "step": 2600
    },
    {
      "epoch": 1.631488160577074,
      "grad_norm": 0.8392802476882935,
      "learning_rate": 3.4529079349449358e-06,
      "loss": 0.1483,
      "step": 2601
    },
    {
      "epoch": 1.6321154147718362,
      "grad_norm": 0.9977639317512512,
      "learning_rate": 3.441502703153028e-06,
      "loss": 0.1422,
      "step": 2602
    },
    {
      "epoch": 1.6327426689665987,
      "grad_norm": 1.0911308526992798,
      "learning_rate": 3.4301145652808022e-06,
      "loss": 0.1435,
      "step": 2603
    },
    {
      "epoch": 1.6333699231613612,
      "grad_norm": 0.6966939568519592,
      "learning_rate": 3.418743533084634e-06,
      "loss": 0.1079,
      "step": 2604
    },
    {
      "epoch": 1.6339971773561235,
      "grad_norm": 0.870735228061676,
      "learning_rate": 3.407389618303245e-06,
      "loss": 0.137,
      "step": 2605
    },
    {
      "epoch": 1.634624431550886,
      "grad_norm": 0.8636801838874817,
      "learning_rate": 3.3960528326576724e-06,
      "loss": 0.1127,
      "step": 2606
    },
    {
      "epoch": 1.6352516857456485,
      "grad_norm": 0.9069063067436218,
      "learning_rate": 3.3847331878512814e-06,
      "loss": 0.1503,
      "step": 2607
    },
    {
      "epoch": 1.6358789399404108,
      "grad_norm": 0.8030480742454529,
      "learning_rate": 3.3734306955697417e-06,
      "loss": 0.1182,
      "step": 2608
    },
    {
      "epoch": 1.6365061941351733,
      "grad_norm": 0.7923745512962341,
      "learning_rate": 3.36214536748102e-06,
      "loss": 0.1091,
      "step": 2609
    },
    {
      "epoch": 1.6371334483299358,
      "grad_norm": 0.843481719493866,
      "learning_rate": 3.3508772152353464e-06,
      "loss": 0.1318,
      "step": 2610
    },
    {
      "epoch": 1.637760702524698,
      "grad_norm": 0.7991217970848083,
      "learning_rate": 3.3396262504652353e-06,
      "loss": 0.1253,
      "step": 2611
    },
    {
      "epoch": 1.6383879567194606,
      "grad_norm": 0.8465505242347717,
      "learning_rate": 3.3283924847854588e-06,
      "loss": 0.1455,
      "step": 2612
    },
    {
      "epoch": 1.639015210914223,
      "grad_norm": 0.7803559303283691,
      "learning_rate": 3.3171759297930194e-06,
      "loss": 0.0915,
      "step": 2613
    },
    {
      "epoch": 1.6396424651089854,
      "grad_norm": 1.0076162815093994,
      "learning_rate": 3.3059765970671688e-06,
      "loss": 0.1335,
      "step": 2614
    },
    {
      "epoch": 1.6402697193037479,
      "grad_norm": 0.9411731362342834,
      "learning_rate": 3.2947944981693693e-06,
      "loss": 0.1426,
      "step": 2615
    },
    {
      "epoch": 1.6408969734985104,
      "grad_norm": 0.8761032223701477,
      "learning_rate": 3.283629644643296e-06,
      "loss": 0.1259,
      "step": 2616
    },
    {
      "epoch": 1.6415242276932727,
      "grad_norm": 0.7741644382476807,
      "learning_rate": 3.2724820480148202e-06,
      "loss": 0.1206,
      "step": 2617
    },
    {
      "epoch": 1.6421514818880352,
      "grad_norm": 0.8049193620681763,
      "learning_rate": 3.261351719792005e-06,
      "loss": 0.1224,
      "step": 2618
    },
    {
      "epoch": 1.6427787360827977,
      "grad_norm": 0.8451152443885803,
      "learning_rate": 3.250238671465067e-06,
      "loss": 0.1262,
      "step": 2619
    },
    {
      "epoch": 1.64340599027756,
      "grad_norm": 0.8271937370300293,
      "learning_rate": 3.239142914506406e-06,
      "loss": 0.126,
      "step": 2620
    },
    {
      "epoch": 1.6440332444723225,
      "grad_norm": 0.7851383090019226,
      "learning_rate": 3.228064460370559e-06,
      "loss": 0.1078,
      "step": 2621
    },
    {
      "epoch": 1.644660498667085,
      "grad_norm": 0.8023667931556702,
      "learning_rate": 3.2170033204942073e-06,
      "loss": 0.1078,
      "step": 2622
    },
    {
      "epoch": 1.6452877528618473,
      "grad_norm": 0.7937674522399902,
      "learning_rate": 3.2059595062961568e-06,
      "loss": 0.1263,
      "step": 2623
    },
    {
      "epoch": 1.6459150070566095,
      "grad_norm": 0.889525830745697,
      "learning_rate": 3.1949330291773183e-06,
      "loss": 0.1452,
      "step": 2624
    },
    {
      "epoch": 1.6465422612513723,
      "grad_norm": 1.006691336631775,
      "learning_rate": 3.1839239005207156e-06,
      "loss": 0.157,
      "step": 2625
    },
    {
      "epoch": 1.6471695154461345,
      "grad_norm": 0.8261495232582092,
      "learning_rate": 3.1729321316914617e-06,
      "loss": 0.1197,
      "step": 2626
    },
    {
      "epoch": 1.6477967696408968,
      "grad_norm": 0.816116988658905,
      "learning_rate": 3.1619577340367445e-06,
      "loss": 0.1204,
      "step": 2627
    },
    {
      "epoch": 1.6484240238356596,
      "grad_norm": 0.8447760343551636,
      "learning_rate": 3.1510007188858214e-06,
      "loss": 0.1361,
      "step": 2628
    },
    {
      "epoch": 1.6490512780304218,
      "grad_norm": 0.8570554256439209,
      "learning_rate": 3.140061097550011e-06,
      "loss": 0.1388,
      "step": 2629
    },
    {
      "epoch": 1.6496785322251841,
      "grad_norm": 0.8310171961784363,
      "learning_rate": 3.1291388813226574e-06,
      "loss": 0.1351,
      "step": 2630
    },
    {
      "epoch": 1.6503057864199466,
      "grad_norm": 0.9378737807273865,
      "learning_rate": 3.1182340814791567e-06,
      "loss": 0.1333,
      "step": 2631
    },
    {
      "epoch": 1.6509330406147091,
      "grad_norm": 0.6849513649940491,
      "learning_rate": 3.1073467092769148e-06,
      "loss": 0.0923,
      "step": 2632
    },
    {
      "epoch": 1.6515602948094714,
      "grad_norm": 0.8584526181221008,
      "learning_rate": 3.096476775955353e-06,
      "loss": 0.1311,
      "step": 2633
    },
    {
      "epoch": 1.652187549004234,
      "grad_norm": 0.8407584428787231,
      "learning_rate": 3.085624292735887e-06,
      "loss": 0.1438,
      "step": 2634
    },
    {
      "epoch": 1.6528148031989964,
      "grad_norm": 0.9150035381317139,
      "learning_rate": 3.0747892708219095e-06,
      "loss": 0.1515,
      "step": 2635
    },
    {
      "epoch": 1.6534420573937587,
      "grad_norm": 0.9339888095855713,
      "learning_rate": 3.0639717213988064e-06,
      "loss": 0.1405,
      "step": 2636
    },
    {
      "epoch": 1.6540693115885212,
      "grad_norm": 0.8367180824279785,
      "learning_rate": 3.053171655633902e-06,
      "loss": 0.1204,
      "step": 2637
    },
    {
      "epoch": 1.6546965657832837,
      "grad_norm": 0.8894864320755005,
      "learning_rate": 3.0423890846764914e-06,
      "loss": 0.1461,
      "step": 2638
    },
    {
      "epoch": 1.655323819978046,
      "grad_norm": 0.7762343287467957,
      "learning_rate": 3.0316240196577993e-06,
      "loss": 0.1161,
      "step": 2639
    },
    {
      "epoch": 1.6559510741728085,
      "grad_norm": 0.910273551940918,
      "learning_rate": 3.0208764716909854e-06,
      "loss": 0.1278,
      "step": 2640
    },
    {
      "epoch": 1.656578328367571,
      "grad_norm": 0.6781446933746338,
      "learning_rate": 3.0101464518711232e-06,
      "loss": 0.143,
      "step": 2641
    },
    {
      "epoch": 1.6572055825623333,
      "grad_norm": 0.8713527917861938,
      "learning_rate": 2.999433971275181e-06,
      "loss": 0.1454,
      "step": 2642
    },
    {
      "epoch": 1.6578328367570958,
      "grad_norm": 0.8391584753990173,
      "learning_rate": 2.9887390409620363e-06,
      "loss": 0.1311,
      "step": 2643
    },
    {
      "epoch": 1.6584600909518583,
      "grad_norm": 0.9159807562828064,
      "learning_rate": 2.9780616719724407e-06,
      "loss": 0.1355,
      "step": 2644
    },
    {
      "epoch": 1.6590873451466206,
      "grad_norm": 0.8163064122200012,
      "learning_rate": 2.967401875329019e-06,
      "loss": 0.1267,
      "step": 2645
    },
    {
      "epoch": 1.659714599341383,
      "grad_norm": 0.8416815996170044,
      "learning_rate": 2.956759662036253e-06,
      "loss": 0.1349,
      "step": 2646
    },
    {
      "epoch": 1.6603418535361456,
      "grad_norm": 0.858839213848114,
      "learning_rate": 2.94613504308048e-06,
      "loss": 0.153,
      "step": 2647
    },
    {
      "epoch": 1.660969107730908,
      "grad_norm": 1.034128189086914,
      "learning_rate": 2.9355280294298595e-06,
      "loss": 0.1144,
      "step": 2648
    },
    {
      "epoch": 1.6615963619256704,
      "grad_norm": 0.9208030700683594,
      "learning_rate": 2.9249386320343863e-06,
      "loss": 0.1586,
      "step": 2649
    },
    {
      "epoch": 1.662223616120433,
      "grad_norm": 0.8803120255470276,
      "learning_rate": 2.9143668618258723e-06,
      "loss": 0.1444,
      "step": 2650
    },
    {
      "epoch": 1.6628508703151952,
      "grad_norm": 0.8457797169685364,
      "learning_rate": 2.9038127297179273e-06,
      "loss": 0.1185,
      "step": 2651
    },
    {
      "epoch": 1.6634781245099577,
      "grad_norm": 0.8965927362442017,
      "learning_rate": 2.8932762466059563e-06,
      "loss": 0.1329,
      "step": 2652
    },
    {
      "epoch": 1.6641053787047202,
      "grad_norm": 0.8928632736206055,
      "learning_rate": 2.882757423367133e-06,
      "loss": 0.1345,
      "step": 2653
    },
    {
      "epoch": 1.6647326328994825,
      "grad_norm": 1.097901463508606,
      "learning_rate": 2.872256270860414e-06,
      "loss": 0.1435,
      "step": 2654
    },
    {
      "epoch": 1.665359887094245,
      "grad_norm": 0.7610005140304565,
      "learning_rate": 2.861772799926508e-06,
      "loss": 0.1171,
      "step": 2655
    },
    {
      "epoch": 1.6659871412890075,
      "grad_norm": 0.8609413504600525,
      "learning_rate": 2.85130702138787e-06,
      "loss": 0.1315,
      "step": 2656
    },
    {
      "epoch": 1.6666143954837698,
      "grad_norm": 0.8173717856407166,
      "learning_rate": 2.840858946048697e-06,
      "loss": 0.1396,
      "step": 2657
    },
    {
      "epoch": 1.6672416496785323,
      "grad_norm": 0.8212631344795227,
      "learning_rate": 2.830428584694904e-06,
      "loss": 0.1376,
      "step": 2658
    },
    {
      "epoch": 1.6678689038732948,
      "grad_norm": 0.8188292384147644,
      "learning_rate": 2.8200159480941125e-06,
      "loss": 0.1194,
      "step": 2659
    },
    {
      "epoch": 1.668496158068057,
      "grad_norm": 0.7810712456703186,
      "learning_rate": 2.809621046995663e-06,
      "loss": 0.1218,
      "step": 2660
    },
    {
      "epoch": 1.6691234122628194,
      "grad_norm": 0.8900648951530457,
      "learning_rate": 2.79924389213057e-06,
      "loss": 0.1168,
      "step": 2661
    },
    {
      "epoch": 1.669750666457582,
      "grad_norm": 0.8871285319328308,
      "learning_rate": 2.788884494211539e-06,
      "loss": 0.1426,
      "step": 2662
    },
    {
      "epoch": 1.6703779206523444,
      "grad_norm": 0.9678382873535156,
      "learning_rate": 2.7785428639329427e-06,
      "loss": 0.1472,
      "step": 2663
    },
    {
      "epoch": 1.6710051748471066,
      "grad_norm": 0.796698808670044,
      "learning_rate": 2.7682190119708074e-06,
      "loss": 0.1071,
      "step": 2664
    },
    {
      "epoch": 1.6716324290418694,
      "grad_norm": 0.9197902083396912,
      "learning_rate": 2.7579129489828173e-06,
      "loss": 0.1468,
      "step": 2665
    },
    {
      "epoch": 1.6722596832366317,
      "grad_norm": 1.0166407823562622,
      "learning_rate": 2.74762468560827e-06,
      "loss": 0.1384,
      "step": 2666
    },
    {
      "epoch": 1.672886937431394,
      "grad_norm": 0.8844588994979858,
      "learning_rate": 2.7373542324681126e-06,
      "loss": 0.143,
      "step": 2667
    },
    {
      "epoch": 1.6735141916261567,
      "grad_norm": 0.7596662044525146,
      "learning_rate": 2.7271016001648877e-06,
      "loss": 0.0926,
      "step": 2668
    },
    {
      "epoch": 1.674141445820919,
      "grad_norm": 0.9629493951797485,
      "learning_rate": 2.716866799282756e-06,
      "loss": 0.1499,
      "step": 2669
    },
    {
      "epoch": 1.6747687000156812,
      "grad_norm": 0.8590734004974365,
      "learning_rate": 2.706649840387463e-06,
      "loss": 0.1347,
      "step": 2670
    },
    {
      "epoch": 1.6753959542104437,
      "grad_norm": 0.7214009165763855,
      "learning_rate": 2.696450734026328e-06,
      "loss": 0.115,
      "step": 2671
    },
    {
      "epoch": 1.6760232084052062,
      "grad_norm": 0.8959335684776306,
      "learning_rate": 2.6862694907282506e-06,
      "loss": 0.1403,
      "step": 2672
    },
    {
      "epoch": 1.6766504625999685,
      "grad_norm": 0.8892949819564819,
      "learning_rate": 2.6761061210036877e-06,
      "loss": 0.13,
      "step": 2673
    },
    {
      "epoch": 1.677277716794731,
      "grad_norm": 0.8894079327583313,
      "learning_rate": 2.665960635344642e-06,
      "loss": 0.1321,
      "step": 2674
    },
    {
      "epoch": 1.6779049709894935,
      "grad_norm": 1.1011821031570435,
      "learning_rate": 2.655833044224658e-06,
      "loss": 0.1542,
      "step": 2675
    },
    {
      "epoch": 1.6785322251842558,
      "grad_norm": 0.945216715335846,
      "learning_rate": 2.6457233580988083e-06,
      "loss": 0.1472,
      "step": 2676
    },
    {
      "epoch": 1.6791594793790183,
      "grad_norm": 0.953877329826355,
      "learning_rate": 2.6356315874036685e-06,
      "loss": 0.1247,
      "step": 2677
    },
    {
      "epoch": 1.6797867335737808,
      "grad_norm": 0.8800573945045471,
      "learning_rate": 2.625557742557332e-06,
      "loss": 0.1112,
      "step": 2678
    },
    {
      "epoch": 1.6804139877685431,
      "grad_norm": 0.7551929950714111,
      "learning_rate": 2.615501833959384e-06,
      "loss": 0.1242,
      "step": 2679
    },
    {
      "epoch": 1.6810412419633056,
      "grad_norm": 0.9266740679740906,
      "learning_rate": 2.605463871990894e-06,
      "loss": 0.1424,
      "step": 2680
    },
    {
      "epoch": 1.6816684961580681,
      "grad_norm": 0.9326827526092529,
      "learning_rate": 2.5954438670144044e-06,
      "loss": 0.1296,
      "step": 2681
    },
    {
      "epoch": 1.6822957503528304,
      "grad_norm": 0.9425510168075562,
      "learning_rate": 2.5854418293739204e-06,
      "loss": 0.1474,
      "step": 2682
    },
    {
      "epoch": 1.682923004547593,
      "grad_norm": 0.9114566445350647,
      "learning_rate": 2.575457769394891e-06,
      "loss": 0.1332,
      "step": 2683
    },
    {
      "epoch": 1.6835502587423554,
      "grad_norm": 0.9489275217056274,
      "learning_rate": 2.5654916973842168e-06,
      "loss": 0.1237,
      "step": 2684
    },
    {
      "epoch": 1.6841775129371177,
      "grad_norm": 0.8633596897125244,
      "learning_rate": 2.5555436236302254e-06,
      "loss": 0.1306,
      "step": 2685
    },
    {
      "epoch": 1.6848047671318802,
      "grad_norm": 0.9675174355506897,
      "learning_rate": 2.5456135584026664e-06,
      "loss": 0.1303,
      "step": 2686
    },
    {
      "epoch": 1.6854320213266427,
      "grad_norm": 0.9195914268493652,
      "learning_rate": 2.5357015119526882e-06,
      "loss": 0.1418,
      "step": 2687
    },
    {
      "epoch": 1.686059275521405,
      "grad_norm": 0.9421305060386658,
      "learning_rate": 2.525807494512853e-06,
      "loss": 0.1124,
      "step": 2688
    },
    {
      "epoch": 1.6866865297161675,
      "grad_norm": 0.8732928037643433,
      "learning_rate": 2.515931516297094e-06,
      "loss": 0.1361,
      "step": 2689
    },
    {
      "epoch": 1.68731378391093,
      "grad_norm": 0.7002175450325012,
      "learning_rate": 2.5060735875007392e-06,
      "loss": 0.1036,
      "step": 2690
    },
    {
      "epoch": 1.6879410381056923,
      "grad_norm": 0.8896955251693726,
      "learning_rate": 2.4962337183004714e-06,
      "loss": 0.1205,
      "step": 2691
    },
    {
      "epoch": 1.6885682923004548,
      "grad_norm": 0.8062140941619873,
      "learning_rate": 2.4864119188543368e-06,
      "loss": 0.1283,
      "step": 2692
    },
    {
      "epoch": 1.6891955464952173,
      "grad_norm": 0.9641050696372986,
      "learning_rate": 2.476608199301722e-06,
      "loss": 0.1402,
      "step": 2693
    },
    {
      "epoch": 1.6898228006899796,
      "grad_norm": 0.9702235460281372,
      "learning_rate": 2.4668225697633586e-06,
      "loss": 0.1447,
      "step": 2694
    },
    {
      "epoch": 1.690450054884742,
      "grad_norm": 0.7521775364875793,
      "learning_rate": 2.4570550403412854e-06,
      "loss": 0.1153,
      "step": 2695
    },
    {
      "epoch": 1.6910773090795046,
      "grad_norm": 0.8183186054229736,
      "learning_rate": 2.447305621118874e-06,
      "loss": 0.1217,
      "step": 2696
    },
    {
      "epoch": 1.6917045632742669,
      "grad_norm": 0.7753691673278809,
      "learning_rate": 2.437574322160792e-06,
      "loss": 0.1029,
      "step": 2697
    },
    {
      "epoch": 1.6923318174690292,
      "grad_norm": 0.8064447641372681,
      "learning_rate": 2.4278611535130005e-06,
      "loss": 0.1164,
      "step": 2698
    },
    {
      "epoch": 1.692959071663792,
      "grad_norm": 0.930521547794342,
      "learning_rate": 2.418166125202752e-06,
      "loss": 0.1381,
      "step": 2699
    },
    {
      "epoch": 1.6935863258585542,
      "grad_norm": 0.869580864906311,
      "learning_rate": 2.4084892472385545e-06,
      "loss": 0.1373,
      "step": 2700
    },
    {
      "epoch": 1.6942135800533165,
      "grad_norm": 0.9124370217323303,
      "learning_rate": 2.3988305296101965e-06,
      "loss": 0.171,
      "step": 2701
    },
    {
      "epoch": 1.6948408342480792,
      "grad_norm": 1.0721293687820435,
      "learning_rate": 2.389189982288711e-06,
      "loss": 0.118,
      "step": 2702
    },
    {
      "epoch": 1.6954680884428415,
      "grad_norm": 0.8098970651626587,
      "learning_rate": 2.3795676152263746e-06,
      "loss": 0.119,
      "step": 2703
    },
    {
      "epoch": 1.6960953426376038,
      "grad_norm": 0.9644606113433838,
      "learning_rate": 2.3699634383566973e-06,
      "loss": 0.1485,
      "step": 2704
    },
    {
      "epoch": 1.6967225968323665,
      "grad_norm": 0.9537535309791565,
      "learning_rate": 2.3603774615944098e-06,
      "loss": 0.1513,
      "step": 2705
    },
    {
      "epoch": 1.6973498510271288,
      "grad_norm": 0.9182994961738586,
      "learning_rate": 2.3508096948354453e-06,
      "loss": 0.1406,
      "step": 2706
    },
    {
      "epoch": 1.697977105221891,
      "grad_norm": 0.9129054546356201,
      "learning_rate": 2.341260147956952e-06,
      "loss": 0.129,
      "step": 2707
    },
    {
      "epoch": 1.6986043594166536,
      "grad_norm": 0.824851930141449,
      "learning_rate": 2.331728830817264e-06,
      "loss": 0.1117,
      "step": 2708
    },
    {
      "epoch": 1.699231613611416,
      "grad_norm": 0.7824007272720337,
      "learning_rate": 2.3222157532558944e-06,
      "loss": 0.1115,
      "step": 2709
    },
    {
      "epoch": 1.6998588678061783,
      "grad_norm": 0.7941312193870544,
      "learning_rate": 2.312720925093528e-06,
      "loss": 0.1452,
      "step": 2710
    },
    {
      "epoch": 1.7004861220009408,
      "grad_norm": 0.9578248858451843,
      "learning_rate": 2.303244356132015e-06,
      "loss": 0.1447,
      "step": 2711
    },
    {
      "epoch": 1.7011133761957034,
      "grad_norm": 0.839382529258728,
      "learning_rate": 2.2937860561543457e-06,
      "loss": 0.1208,
      "step": 2712
    },
    {
      "epoch": 1.7017406303904656,
      "grad_norm": 0.8220772743225098,
      "learning_rate": 2.284346034924654e-06,
      "loss": 0.1278,
      "step": 2713
    },
    {
      "epoch": 1.7023678845852281,
      "grad_norm": 0.7147632241249084,
      "learning_rate": 2.2749243021882085e-06,
      "loss": 0.1155,
      "step": 2714
    },
    {
      "epoch": 1.7029951387799906,
      "grad_norm": 0.8697649836540222,
      "learning_rate": 2.2655208676713957e-06,
      "loss": 0.1293,
      "step": 2715
    },
    {
      "epoch": 1.703622392974753,
      "grad_norm": 0.7797321081161499,
      "learning_rate": 2.256135741081711e-06,
      "loss": 0.1135,
      "step": 2716
    },
    {
      "epoch": 1.7042496471695154,
      "grad_norm": 1.2459222078323364,
      "learning_rate": 2.246768932107759e-06,
      "loss": 0.1219,
      "step": 2717
    },
    {
      "epoch": 1.704876901364278,
      "grad_norm": 0.8825497627258301,
      "learning_rate": 2.2374204504192163e-06,
      "loss": 0.134,
      "step": 2718
    },
    {
      "epoch": 1.7055041555590402,
      "grad_norm": 0.9286799430847168,
      "learning_rate": 2.2280903056668523e-06,
      "loss": 0.1337,
      "step": 2719
    },
    {
      "epoch": 1.7061314097538027,
      "grad_norm": 0.8434159159660339,
      "learning_rate": 2.218778507482504e-06,
      "loss": 0.1069,
      "step": 2720
    },
    {
      "epoch": 1.7067586639485652,
      "grad_norm": 1.0079706907272339,
      "learning_rate": 2.209485065479069e-06,
      "loss": 0.1464,
      "step": 2721
    },
    {
      "epoch": 1.7073859181433275,
      "grad_norm": 0.8292884230613708,
      "learning_rate": 2.2002099892504967e-06,
      "loss": 0.1369,
      "step": 2722
    },
    {
      "epoch": 1.70801317233809,
      "grad_norm": 0.9433373212814331,
      "learning_rate": 2.1909532883717753e-06,
      "loss": 0.1464,
      "step": 2723
    },
    {
      "epoch": 1.7086404265328525,
      "grad_norm": 0.8257498741149902,
      "learning_rate": 2.1817149723989185e-06,
      "loss": 0.1345,
      "step": 2724
    },
    {
      "epoch": 1.7092676807276148,
      "grad_norm": 0.8046967387199402,
      "learning_rate": 2.1724950508689656e-06,
      "loss": 0.1067,
      "step": 2725
    },
    {
      "epoch": 1.7098949349223773,
      "grad_norm": 0.8804117441177368,
      "learning_rate": 2.1632935332999706e-06,
      "loss": 0.15,
      "step": 2726
    },
    {
      "epoch": 1.7105221891171398,
      "grad_norm": 0.9799605011940002,
      "learning_rate": 2.1541104291909787e-06,
      "loss": 0.1411,
      "step": 2727
    },
    {
      "epoch": 1.711149443311902,
      "grad_norm": 0.8501812219619751,
      "learning_rate": 2.144945748022036e-06,
      "loss": 0.1242,
      "step": 2728
    },
    {
      "epoch": 1.7117766975066646,
      "grad_norm": 0.9060745239257812,
      "learning_rate": 2.1357994992541653e-06,
      "loss": 0.1176,
      "step": 2729
    },
    {
      "epoch": 1.7124039517014271,
      "grad_norm": 0.8523778319358826,
      "learning_rate": 2.1266716923293583e-06,
      "loss": 0.1443,
      "step": 2730
    },
    {
      "epoch": 1.7130312058961894,
      "grad_norm": 0.8337844610214233,
      "learning_rate": 2.117562336670571e-06,
      "loss": 0.1128,
      "step": 2731
    },
    {
      "epoch": 1.713658460090952,
      "grad_norm": 0.6823314428329468,
      "learning_rate": 2.1084714416817144e-06,
      "loss": 0.1215,
      "step": 2732
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 0.9481781125068665,
      "learning_rate": 2.0993990167476387e-06,
      "loss": 0.1317,
      "step": 2733
    },
    {
      "epoch": 1.7149129684804767,
      "grad_norm": 0.973198652267456,
      "learning_rate": 2.09034507123413e-06,
      "loss": 0.1483,
      "step": 2734
    },
    {
      "epoch": 1.715540222675239,
      "grad_norm": 0.8245511054992676,
      "learning_rate": 2.0813096144878895e-06,
      "loss": 0.1278,
      "step": 2735
    },
    {
      "epoch": 1.7161674768700017,
      "grad_norm": 0.8046514391899109,
      "learning_rate": 2.072292655836541e-06,
      "loss": 0.1166,
      "step": 2736
    },
    {
      "epoch": 1.716794731064764,
      "grad_norm": 0.9386309385299683,
      "learning_rate": 2.063294204588606e-06,
      "loss": 0.1192,
      "step": 2737
    },
    {
      "epoch": 1.7174219852595263,
      "grad_norm": 0.9409434199333191,
      "learning_rate": 2.0543142700335106e-06,
      "loss": 0.1485,
      "step": 2738
    },
    {
      "epoch": 1.718049239454289,
      "grad_norm": 0.811586320400238,
      "learning_rate": 2.045352861441545e-06,
      "loss": 0.114,
      "step": 2739
    },
    {
      "epoch": 1.7186764936490513,
      "grad_norm": 0.8804102540016174,
      "learning_rate": 2.036409988063892e-06,
      "loss": 0.1384,
      "step": 2740
    },
    {
      "epoch": 1.7193037478438136,
      "grad_norm": 0.8443262577056885,
      "learning_rate": 2.027485659132602e-06,
      "loss": 0.1226,
      "step": 2741
    },
    {
      "epoch": 1.7199310020385763,
      "grad_norm": 0.9686700701713562,
      "learning_rate": 2.018579883860561e-06,
      "loss": 0.147,
      "step": 2742
    },
    {
      "epoch": 1.7205582562333386,
      "grad_norm": 0.915995180606842,
      "learning_rate": 2.009692671441521e-06,
      "loss": 0.1287,
      "step": 2743
    },
    {
      "epoch": 1.7211855104281009,
      "grad_norm": 0.906000018119812,
      "learning_rate": 2.000824031050064e-06,
      "loss": 0.1383,
      "step": 2744
    },
    {
      "epoch": 1.7218127646228634,
      "grad_norm": 0.9242513179779053,
      "learning_rate": 1.9919739718415987e-06,
      "loss": 0.1238,
      "step": 2745
    },
    {
      "epoch": 1.7224400188176259,
      "grad_norm": 0.8009757399559021,
      "learning_rate": 1.9831425029523534e-06,
      "loss": 0.1066,
      "step": 2746
    },
    {
      "epoch": 1.7230672730123882,
      "grad_norm": 0.9236252307891846,
      "learning_rate": 1.97432963349937e-06,
      "loss": 0.1482,
      "step": 2747
    },
    {
      "epoch": 1.7236945272071507,
      "grad_norm": 0.854402482509613,
      "learning_rate": 1.965535372580474e-06,
      "loss": 0.1272,
      "step": 2748
    },
    {
      "epoch": 1.7243217814019132,
      "grad_norm": 0.9159762859344482,
      "learning_rate": 1.9567597292742934e-06,
      "loss": 0.145,
      "step": 2749
    },
    {
      "epoch": 1.7249490355966755,
      "grad_norm": 0.8271155953407288,
      "learning_rate": 1.9480027126402357e-06,
      "loss": 0.1165,
      "step": 2750
    },
    {
      "epoch": 1.725576289791438,
      "grad_norm": 0.7804858088493347,
      "learning_rate": 1.939264331718478e-06,
      "loss": 0.1198,
      "step": 2751
    },
    {
      "epoch": 1.7262035439862005,
      "grad_norm": 0.8453112840652466,
      "learning_rate": 1.9305445955299594e-06,
      "loss": 0.1066,
      "step": 2752
    },
    {
      "epoch": 1.7268307981809627,
      "grad_norm": 0.8296924233436584,
      "learning_rate": 1.921843513076367e-06,
      "loss": 0.118,
      "step": 2753
    },
    {
      "epoch": 1.7274580523757253,
      "grad_norm": 0.9044693112373352,
      "learning_rate": 1.9131610933401366e-06,
      "loss": 0.141,
      "step": 2754
    },
    {
      "epoch": 1.7280853065704878,
      "grad_norm": 0.791060209274292,
      "learning_rate": 1.904497345284433e-06,
      "loss": 0.1169,
      "step": 2755
    },
    {
      "epoch": 1.72871256076525,
      "grad_norm": 0.8272935152053833,
      "learning_rate": 1.8958522778531518e-06,
      "loss": 0.1336,
      "step": 2756
    },
    {
      "epoch": 1.7293398149600125,
      "grad_norm": 0.8372897505760193,
      "learning_rate": 1.8872258999708992e-06,
      "loss": 0.125,
      "step": 2757
    },
    {
      "epoch": 1.729967069154775,
      "grad_norm": 0.9450139403343201,
      "learning_rate": 1.8786182205429938e-06,
      "loss": 0.1385,
      "step": 2758
    },
    {
      "epoch": 1.7305943233495373,
      "grad_norm": 0.8529419898986816,
      "learning_rate": 1.8700292484554362e-06,
      "loss": 0.1308,
      "step": 2759
    },
    {
      "epoch": 1.7312215775442998,
      "grad_norm": 0.8922459483146667,
      "learning_rate": 1.8614589925749293e-06,
      "loss": 0.1282,
      "step": 2760
    },
    {
      "epoch": 1.7318488317390623,
      "grad_norm": 0.7917664051055908,
      "learning_rate": 1.8529074617488497e-06,
      "loss": 0.122,
      "step": 2761
    },
    {
      "epoch": 1.7324760859338246,
      "grad_norm": 1.03391432762146,
      "learning_rate": 1.844374664805244e-06,
      "loss": 0.1622,
      "step": 2762
    },
    {
      "epoch": 1.7331033401285871,
      "grad_norm": 0.8683478832244873,
      "learning_rate": 1.8358606105528242e-06,
      "loss": 0.1234,
      "step": 2763
    },
    {
      "epoch": 1.7337305943233496,
      "grad_norm": 0.8627936840057373,
      "learning_rate": 1.8273653077809395e-06,
      "loss": 0.1297,
      "step": 2764
    },
    {
      "epoch": 1.734357848518112,
      "grad_norm": 0.8438630700111389,
      "learning_rate": 1.8188887652595966e-06,
      "loss": 0.1156,
      "step": 2765
    },
    {
      "epoch": 1.7349851027128744,
      "grad_norm": 0.7723935842514038,
      "learning_rate": 1.8104309917394248e-06,
      "loss": 0.1155,
      "step": 2766
    },
    {
      "epoch": 1.735612356907637,
      "grad_norm": 0.8524870872497559,
      "learning_rate": 1.8019919959516796e-06,
      "loss": 0.1332,
      "step": 2767
    },
    {
      "epoch": 1.7362396111023992,
      "grad_norm": 0.8028182983398438,
      "learning_rate": 1.793571786608237e-06,
      "loss": 0.1027,
      "step": 2768
    },
    {
      "epoch": 1.7368668652971617,
      "grad_norm": 0.9504607915878296,
      "learning_rate": 1.7851703724015745e-06,
      "loss": 0.1234,
      "step": 2769
    },
    {
      "epoch": 1.7374941194919242,
      "grad_norm": 0.7929568886756897,
      "learning_rate": 1.7767877620047723e-06,
      "loss": 0.1258,
      "step": 2770
    },
    {
      "epoch": 1.7381213736866865,
      "grad_norm": 0.9246236085891724,
      "learning_rate": 1.7684239640714862e-06,
      "loss": 0.1306,
      "step": 2771
    },
    {
      "epoch": 1.7387486278814488,
      "grad_norm": 0.8959844708442688,
      "learning_rate": 1.7600789872359624e-06,
      "loss": 0.1223,
      "step": 2772
    },
    {
      "epoch": 1.7393758820762115,
      "grad_norm": 0.8668091893196106,
      "learning_rate": 1.7517528401130168e-06,
      "loss": 0.1261,
      "step": 2773
    },
    {
      "epoch": 1.7400031362709738,
      "grad_norm": 0.8081668615341187,
      "learning_rate": 1.7434455312980203e-06,
      "loss": 0.1152,
      "step": 2774
    },
    {
      "epoch": 1.740630390465736,
      "grad_norm": 0.9705616235733032,
      "learning_rate": 1.735157069366904e-06,
      "loss": 0.136,
      "step": 2775
    },
    {
      "epoch": 1.7412576446604988,
      "grad_norm": 0.8360944390296936,
      "learning_rate": 1.7268874628761389e-06,
      "loss": 0.1317,
      "step": 2776
    },
    {
      "epoch": 1.741884898855261,
      "grad_norm": 0.7722758054733276,
      "learning_rate": 1.7186367203627274e-06,
      "loss": 0.1298,
      "step": 2777
    },
    {
      "epoch": 1.7425121530500234,
      "grad_norm": 1.0844967365264893,
      "learning_rate": 1.7104048503442028e-06,
      "loss": 0.1545,
      "step": 2778
    },
    {
      "epoch": 1.743139407244786,
      "grad_norm": 0.7706978917121887,
      "learning_rate": 1.702191861318614e-06,
      "loss": 0.1155,
      "step": 2779
    },
    {
      "epoch": 1.7437666614395484,
      "grad_norm": 0.8536537885665894,
      "learning_rate": 1.6939977617645186e-06,
      "loss": 0.1042,
      "step": 2780
    },
    {
      "epoch": 1.7443939156343107,
      "grad_norm": 0.8721825480461121,
      "learning_rate": 1.6858225601409794e-06,
      "loss": 0.1234,
      "step": 2781
    },
    {
      "epoch": 1.7450211698290732,
      "grad_norm": 0.8884703516960144,
      "learning_rate": 1.6776662648875386e-06,
      "loss": 0.1409,
      "step": 2782
    },
    {
      "epoch": 1.7456484240238357,
      "grad_norm": 0.8294079899787903,
      "learning_rate": 1.6695288844242274e-06,
      "loss": 0.1196,
      "step": 2783
    },
    {
      "epoch": 1.746275678218598,
      "grad_norm": 1.0719852447509766,
      "learning_rate": 1.6614104271515508e-06,
      "loss": 0.1359,
      "step": 2784
    },
    {
      "epoch": 1.7469029324133605,
      "grad_norm": 0.9199581146240234,
      "learning_rate": 1.6533109014504777e-06,
      "loss": 0.1251,
      "step": 2785
    },
    {
      "epoch": 1.747530186608123,
      "grad_norm": 0.8534532785415649,
      "learning_rate": 1.6452303156824357e-06,
      "loss": 0.1343,
      "step": 2786
    },
    {
      "epoch": 1.7481574408028853,
      "grad_norm": 0.9075412154197693,
      "learning_rate": 1.6371686781893003e-06,
      "loss": 0.1252,
      "step": 2787
    },
    {
      "epoch": 1.7487846949976478,
      "grad_norm": 0.8409426808357239,
      "learning_rate": 1.629125997293377e-06,
      "loss": 0.1122,
      "step": 2788
    },
    {
      "epoch": 1.7494119491924103,
      "grad_norm": 0.8862391114234924,
      "learning_rate": 1.6211022812974152e-06,
      "loss": 0.1331,
      "step": 2789
    },
    {
      "epoch": 1.7500392033871726,
      "grad_norm": 0.9557252526283264,
      "learning_rate": 1.6130975384845715e-06,
      "loss": 0.1566,
      "step": 2790
    },
    {
      "epoch": 1.750666457581935,
      "grad_norm": 0.7308382391929626,
      "learning_rate": 1.605111777118429e-06,
      "loss": 0.1056,
      "step": 2791
    },
    {
      "epoch": 1.7512937117766976,
      "grad_norm": 0.9058966040611267,
      "learning_rate": 1.5971450054429682e-06,
      "loss": 0.1309,
      "step": 2792
    },
    {
      "epoch": 1.7519209659714599,
      "grad_norm": 0.9257788062095642,
      "learning_rate": 1.5891972316825688e-06,
      "loss": 0.1473,
      "step": 2793
    },
    {
      "epoch": 1.7525482201662224,
      "grad_norm": 1.0357989072799683,
      "learning_rate": 1.5812684640420028e-06,
      "loss": 0.1596,
      "step": 2794
    },
    {
      "epoch": 1.7531754743609849,
      "grad_norm": 0.8932484984397888,
      "learning_rate": 1.5733587107064051e-06,
      "loss": 0.1342,
      "step": 2795
    },
    {
      "epoch": 1.7538027285557471,
      "grad_norm": 0.908905029296875,
      "learning_rate": 1.5654679798412997e-06,
      "loss": 0.137,
      "step": 2796
    },
    {
      "epoch": 1.7544299827505097,
      "grad_norm": 0.84003084897995,
      "learning_rate": 1.557596279592566e-06,
      "loss": 0.1196,
      "step": 2797
    },
    {
      "epoch": 1.7550572369452722,
      "grad_norm": 0.8938060402870178,
      "learning_rate": 1.5497436180864344e-06,
      "loss": 0.134,
      "step": 2798
    },
    {
      "epoch": 1.7556844911400344,
      "grad_norm": 0.9037652611732483,
      "learning_rate": 1.5419100034294876e-06,
      "loss": 0.1254,
      "step": 2799
    },
    {
      "epoch": 1.756311745334797,
      "grad_norm": 0.9804080128669739,
      "learning_rate": 1.5340954437086387e-06,
      "loss": 0.1376,
      "step": 2800
    },
    {
      "epoch": 1.7569389995295595,
      "grad_norm": 0.8713802695274353,
      "learning_rate": 1.5262999469911322e-06,
      "loss": 0.1367,
      "step": 2801
    },
    {
      "epoch": 1.7575662537243217,
      "grad_norm": 0.9325335025787354,
      "learning_rate": 1.5185235213245352e-06,
      "loss": 0.1314,
      "step": 2802
    },
    {
      "epoch": 1.7581935079190842,
      "grad_norm": 0.7857791185379028,
      "learning_rate": 1.5107661747367263e-06,
      "loss": 0.119,
      "step": 2803
    },
    {
      "epoch": 1.7588207621138467,
      "grad_norm": 0.8473135232925415,
      "learning_rate": 1.5030279152358862e-06,
      "loss": 0.1118,
      "step": 2804
    },
    {
      "epoch": 1.759448016308609,
      "grad_norm": 0.9458979368209839,
      "learning_rate": 1.4953087508104958e-06,
      "loss": 0.1407,
      "step": 2805
    },
    {
      "epoch": 1.7600752705033715,
      "grad_norm": 1.1570056676864624,
      "learning_rate": 1.4876086894293163e-06,
      "loss": 0.1667,
      "step": 2806
    },
    {
      "epoch": 1.760702524698134,
      "grad_norm": 0.7316725254058838,
      "learning_rate": 1.4799277390413913e-06,
      "loss": 0.1002,
      "step": 2807
    },
    {
      "epoch": 1.7613297788928963,
      "grad_norm": 0.969429075717926,
      "learning_rate": 1.4722659075760404e-06,
      "loss": 0.1343,
      "step": 2808
    },
    {
      "epoch": 1.7619570330876586,
      "grad_norm": 0.8668022751808167,
      "learning_rate": 1.4646232029428387e-06,
      "loss": 0.1336,
      "step": 2809
    },
    {
      "epoch": 1.7625842872824213,
      "grad_norm": 0.8704190850257874,
      "learning_rate": 1.4569996330316238e-06,
      "loss": 0.129,
      "step": 2810
    },
    {
      "epoch": 1.7632115414771836,
      "grad_norm": 0.829531729221344,
      "learning_rate": 1.4493952057124715e-06,
      "loss": 0.1453,
      "step": 2811
    },
    {
      "epoch": 1.763838795671946,
      "grad_norm": 0.8531040549278259,
      "learning_rate": 1.4418099288356979e-06,
      "loss": 0.1488,
      "step": 2812
    },
    {
      "epoch": 1.7644660498667086,
      "grad_norm": 0.7893531918525696,
      "learning_rate": 1.4342438102318546e-06,
      "loss": 0.1131,
      "step": 2813
    },
    {
      "epoch": 1.765093304061471,
      "grad_norm": 0.788480818271637,
      "learning_rate": 1.4266968577117114e-06,
      "loss": 0.1316,
      "step": 2814
    },
    {
      "epoch": 1.7657205582562332,
      "grad_norm": 0.8491449356079102,
      "learning_rate": 1.419169079066256e-06,
      "loss": 0.1291,
      "step": 2815
    },
    {
      "epoch": 1.766347812450996,
      "grad_norm": 0.9434386491775513,
      "learning_rate": 1.4116604820666746e-06,
      "loss": 0.151,
      "step": 2816
    },
    {
      "epoch": 1.7669750666457582,
      "grad_norm": 0.8945111036300659,
      "learning_rate": 1.40417107446436e-06,
      "loss": 0.1329,
      "step": 2817
    },
    {
      "epoch": 1.7676023208405205,
      "grad_norm": 0.8145248889923096,
      "learning_rate": 1.3967008639908851e-06,
      "loss": 0.1332,
      "step": 2818
    },
    {
      "epoch": 1.768229575035283,
      "grad_norm": 0.8825253844261169,
      "learning_rate": 1.389249858358017e-06,
      "loss": 0.134,
      "step": 2819
    },
    {
      "epoch": 1.7688568292300455,
      "grad_norm": 0.9387741684913635,
      "learning_rate": 1.3818180652576895e-06,
      "loss": 0.1431,
      "step": 2820
    },
    {
      "epoch": 1.7694840834248078,
      "grad_norm": 0.8380280137062073,
      "learning_rate": 1.3744054923620053e-06,
      "loss": 0.1157,
      "step": 2821
    },
    {
      "epoch": 1.7701113376195703,
      "grad_norm": 0.781251847743988,
      "learning_rate": 1.3670121473232211e-06,
      "loss": 0.1094,
      "step": 2822
    },
    {
      "epoch": 1.7707385918143328,
      "grad_norm": 0.965782880783081,
      "learning_rate": 1.3596380377737539e-06,
      "loss": 0.1408,
      "step": 2823
    },
    {
      "epoch": 1.771365846009095,
      "grad_norm": 0.8496629595756531,
      "learning_rate": 1.3522831713261498e-06,
      "loss": 0.1261,
      "step": 2824
    },
    {
      "epoch": 1.7719931002038576,
      "grad_norm": 0.9037855267524719,
      "learning_rate": 1.3449475555730973e-06,
      "loss": 0.127,
      "step": 2825
    },
    {
      "epoch": 1.77262035439862,
      "grad_norm": 0.8562514781951904,
      "learning_rate": 1.3376311980874146e-06,
      "loss": 0.1256,
      "step": 2826
    },
    {
      "epoch": 1.7732476085933824,
      "grad_norm": 0.5597630143165588,
      "learning_rate": 1.330334106422031e-06,
      "loss": 0.1119,
      "step": 2827
    },
    {
      "epoch": 1.7738748627881449,
      "grad_norm": 0.8352013826370239,
      "learning_rate": 1.3230562881099963e-06,
      "loss": 0.1163,
      "step": 2828
    },
    {
      "epoch": 1.7745021169829074,
      "grad_norm": 0.9308125972747803,
      "learning_rate": 1.3157977506644515e-06,
      "loss": 0.1497,
      "step": 2829
    },
    {
      "epoch": 1.7751293711776697,
      "grad_norm": 0.8582794070243835,
      "learning_rate": 1.308558501578643e-06,
      "loss": 0.1329,
      "step": 2830
    },
    {
      "epoch": 1.7757566253724322,
      "grad_norm": 0.8619057536125183,
      "learning_rate": 1.3013385483259055e-06,
      "loss": 0.1218,
      "step": 2831
    },
    {
      "epoch": 1.7763838795671947,
      "grad_norm": 0.8505441546440125,
      "learning_rate": 1.2941378983596464e-06,
      "loss": 0.1488,
      "step": 2832
    },
    {
      "epoch": 1.777011133761957,
      "grad_norm": 0.714715838432312,
      "learning_rate": 1.2869565591133548e-06,
      "loss": 0.0959,
      "step": 2833
    },
    {
      "epoch": 1.7776383879567195,
      "grad_norm": 0.8384492993354797,
      "learning_rate": 1.2797945380005805e-06,
      "loss": 0.1251,
      "step": 2834
    },
    {
      "epoch": 1.778265642151482,
      "grad_norm": 0.9021497368812561,
      "learning_rate": 1.2726518424149247e-06,
      "loss": 0.1364,
      "step": 2835
    },
    {
      "epoch": 1.7788928963462443,
      "grad_norm": 1.215956687927246,
      "learning_rate": 1.2655284797300449e-06,
      "loss": 0.1395,
      "step": 2836
    },
    {
      "epoch": 1.7795201505410068,
      "grad_norm": 0.9060066342353821,
      "learning_rate": 1.2584244572996406e-06,
      "loss": 0.1421,
      "step": 2837
    },
    {
      "epoch": 1.7801474047357693,
      "grad_norm": 0.8602017760276794,
      "learning_rate": 1.2513397824574436e-06,
      "loss": 0.1158,
      "step": 2838
    },
    {
      "epoch": 1.7807746589305316,
      "grad_norm": 0.9450860023498535,
      "learning_rate": 1.2442744625172165e-06,
      "loss": 0.1341,
      "step": 2839
    },
    {
      "epoch": 1.781401913125294,
      "grad_norm": 0.9497483372688293,
      "learning_rate": 1.237228504772734e-06,
      "loss": 0.1579,
      "step": 2840
    },
    {
      "epoch": 1.7820291673200566,
      "grad_norm": 1.0387485027313232,
      "learning_rate": 1.2302019164977886e-06,
      "loss": 0.1303,
      "step": 2841
    },
    {
      "epoch": 1.7826564215148188,
      "grad_norm": 0.9795166850090027,
      "learning_rate": 1.2231947049461712e-06,
      "loss": 0.1569,
      "step": 2842
    },
    {
      "epoch": 1.7832836757095814,
      "grad_norm": 0.908534824848175,
      "learning_rate": 1.2162068773516732e-06,
      "loss": 0.1184,
      "step": 2843
    },
    {
      "epoch": 1.7839109299043439,
      "grad_norm": 0.8440238237380981,
      "learning_rate": 1.2092384409280778e-06,
      "loss": 0.1279,
      "step": 2844
    },
    {
      "epoch": 1.7845381840991061,
      "grad_norm": 0.9186972379684448,
      "learning_rate": 1.202289402869148e-06,
      "loss": 0.1268,
      "step": 2845
    },
    {
      "epoch": 1.7851654382938686,
      "grad_norm": 0.8728666305541992,
      "learning_rate": 1.1953597703486208e-06,
      "loss": 0.1273,
      "step": 2846
    },
    {
      "epoch": 1.7857926924886312,
      "grad_norm": 0.7897111177444458,
      "learning_rate": 1.1884495505201944e-06,
      "loss": 0.1269,
      "step": 2847
    },
    {
      "epoch": 1.7864199466833934,
      "grad_norm": 0.8890313506126404,
      "learning_rate": 1.1815587505175351e-06,
      "loss": 0.1314,
      "step": 2848
    },
    {
      "epoch": 1.7870472008781557,
      "grad_norm": 0.9216718077659607,
      "learning_rate": 1.1746873774542578e-06,
      "loss": 0.1202,
      "step": 2849
    },
    {
      "epoch": 1.7876744550729184,
      "grad_norm": 1.0509883165359497,
      "learning_rate": 1.1678354384239255e-06,
      "loss": 0.1229,
      "step": 2850
    },
    {
      "epoch": 1.7883017092676807,
      "grad_norm": 0.8760930299758911,
      "learning_rate": 1.1610029405000335e-06,
      "loss": 0.1305,
      "step": 2851
    },
    {
      "epoch": 1.788928963462443,
      "grad_norm": 0.9807936549186707,
      "learning_rate": 1.1541898907360128e-06,
      "loss": 0.1388,
      "step": 2852
    },
    {
      "epoch": 1.7895562176572057,
      "grad_norm": 0.8667734265327454,
      "learning_rate": 1.1473962961652107e-06,
      "loss": 0.1146,
      "step": 2853
    },
    {
      "epoch": 1.790183471851968,
      "grad_norm": 1.0074526071548462,
      "learning_rate": 1.1406221638008952e-06,
      "loss": 0.1526,
      "step": 2854
    },
    {
      "epoch": 1.7908107260467303,
      "grad_norm": 0.736807107925415,
      "learning_rate": 1.1338675006362432e-06,
      "loss": 0.1202,
      "step": 2855
    },
    {
      "epoch": 1.7914379802414928,
      "grad_norm": 0.8689996004104614,
      "learning_rate": 1.127132313644328e-06,
      "loss": 0.1405,
      "step": 2856
    },
    {
      "epoch": 1.7920652344362553,
      "grad_norm": 0.9014766812324524,
      "learning_rate": 1.1204166097781278e-06,
      "loss": 0.1191,
      "step": 2857
    },
    {
      "epoch": 1.7926924886310176,
      "grad_norm": 0.9162267446517944,
      "learning_rate": 1.1137203959704901e-06,
      "loss": 0.113,
      "step": 2858
    },
    {
      "epoch": 1.79331974282578,
      "grad_norm": 0.8885189890861511,
      "learning_rate": 1.1070436791341566e-06,
      "loss": 0.115,
      "step": 2859
    },
    {
      "epoch": 1.7939469970205426,
      "grad_norm": 1.0271906852722168,
      "learning_rate": 1.100386466161738e-06,
      "loss": 0.1321,
      "step": 2860
    },
    {
      "epoch": 1.794574251215305,
      "grad_norm": 0.8138547539710999,
      "learning_rate": 1.0937487639257083e-06,
      "loss": 0.1115,
      "step": 2861
    },
    {
      "epoch": 1.7952015054100674,
      "grad_norm": 0.8766812086105347,
      "learning_rate": 1.0871305792784015e-06,
      "loss": 0.1207,
      "step": 2862
    },
    {
      "epoch": 1.79582875960483,
      "grad_norm": 0.7150856256484985,
      "learning_rate": 1.080531919052008e-06,
      "loss": 0.1062,
      "step": 2863
    },
    {
      "epoch": 1.7964560137995922,
      "grad_norm": 0.9094497561454773,
      "learning_rate": 1.0739527900585478e-06,
      "loss": 0.1241,
      "step": 2864
    },
    {
      "epoch": 1.7970832679943547,
      "grad_norm": 0.8997982740402222,
      "learning_rate": 1.06739319908989e-06,
      "loss": 0.1271,
      "step": 2865
    },
    {
      "epoch": 1.7977105221891172,
      "grad_norm": 0.8818061351776123,
      "learning_rate": 1.060853152917738e-06,
      "loss": 0.1137,
      "step": 2866
    },
    {
      "epoch": 1.7983377763838795,
      "grad_norm": 0.8932895660400391,
      "learning_rate": 1.054332658293602e-06,
      "loss": 0.1255,
      "step": 2867
    },
    {
      "epoch": 1.798965030578642,
      "grad_norm": 0.7263736128807068,
      "learning_rate": 1.0478317219488244e-06,
      "loss": 0.1058,
      "step": 2868
    },
    {
      "epoch": 1.7995922847734045,
      "grad_norm": 0.8277838230133057,
      "learning_rate": 1.0413503505945477e-06,
      "loss": 0.1432,
      "step": 2869
    },
    {
      "epoch": 1.8002195389681668,
      "grad_norm": 0.8268225193023682,
      "learning_rate": 1.0348885509217222e-06,
      "loss": 0.1216,
      "step": 2870
    },
    {
      "epoch": 1.8008467931629293,
      "grad_norm": 0.8564212322235107,
      "learning_rate": 1.0284463296010893e-06,
      "loss": 0.1059,
      "step": 2871
    },
    {
      "epoch": 1.8014740473576918,
      "grad_norm": 0.9777878522872925,
      "learning_rate": 1.0220236932831807e-06,
      "loss": 0.1444,
      "step": 2872
    },
    {
      "epoch": 1.802101301552454,
      "grad_norm": 0.9540963172912598,
      "learning_rate": 1.0156206485983123e-06,
      "loss": 0.1415,
      "step": 2873
    },
    {
      "epoch": 1.8027285557472166,
      "grad_norm": 0.9117012619972229,
      "learning_rate": 1.0092372021565678e-06,
      "loss": 0.1196,
      "step": 2874
    },
    {
      "epoch": 1.803355809941979,
      "grad_norm": 0.965158224105835,
      "learning_rate": 1.0028733605478115e-06,
      "loss": 0.1353,
      "step": 2875
    },
    {
      "epoch": 1.8039830641367414,
      "grad_norm": 0.9235615134239197,
      "learning_rate": 9.965291303416547e-07,
      "loss": 0.1256,
      "step": 2876
    },
    {
      "epoch": 1.8046103183315039,
      "grad_norm": 0.8588432669639587,
      "learning_rate": 9.902045180874699e-07,
      "loss": 0.1395,
      "step": 2877
    },
    {
      "epoch": 1.8052375725262664,
      "grad_norm": 0.8039693236351013,
      "learning_rate": 9.838995303143784e-07,
      "loss": 0.1267,
      "step": 2878
    },
    {
      "epoch": 1.8058648267210287,
      "grad_norm": 0.8312050700187683,
      "learning_rate": 9.776141735312428e-07,
      "loss": 0.0992,
      "step": 2879
    },
    {
      "epoch": 1.8064920809157912,
      "grad_norm": 0.9416694045066833,
      "learning_rate": 9.71348454226655e-07,
      "loss": 0.1207,
      "step": 2880
    },
    {
      "epoch": 1.8071193351105537,
      "grad_norm": 0.7612276673316956,
      "learning_rate": 9.651023788689406e-07,
      "loss": 0.1089,
      "step": 2881
    },
    {
      "epoch": 1.807746589305316,
      "grad_norm": 1.1019048690795898,
      "learning_rate": 9.58875953906142e-07,
      "loss": 0.1338,
      "step": 2882
    },
    {
      "epoch": 1.8083738435000785,
      "grad_norm": 0.7640385031700134,
      "learning_rate": 9.526691857660131e-07,
      "loss": 0.1003,
      "step": 2883
    },
    {
      "epoch": 1.809001097694841,
      "grad_norm": 0.9395763874053955,
      "learning_rate": 9.46482080856026e-07,
      "loss": 0.1448,
      "step": 2884
    },
    {
      "epoch": 1.8096283518896032,
      "grad_norm": 0.8155563473701477,
      "learning_rate": 9.403146455633405e-07,
      "loss": 0.1337,
      "step": 2885
    },
    {
      "epoch": 1.8102556060843655,
      "grad_norm": 0.8867181539535522,
      "learning_rate": 9.341668862548214e-07,
      "loss": 0.141,
      "step": 2886
    },
    {
      "epoch": 1.8108828602791283,
      "grad_norm": 0.7554093599319458,
      "learning_rate": 9.280388092770165e-07,
      "loss": 0.1071,
      "step": 2887
    },
    {
      "epoch": 1.8115101144738905,
      "grad_norm": 0.9735627770423889,
      "learning_rate": 9.219304209561541e-07,
      "loss": 0.1491,
      "step": 2888
    },
    {
      "epoch": 1.8121373686686528,
      "grad_norm": 0.9940028786659241,
      "learning_rate": 9.15841727598139e-07,
      "loss": 0.1551,
      "step": 2889
    },
    {
      "epoch": 1.8127646228634156,
      "grad_norm": 0.8162161707878113,
      "learning_rate": 9.097727354885432e-07,
      "loss": 0.1358,
      "step": 2890
    },
    {
      "epoch": 1.8133918770581778,
      "grad_norm": 0.7892768979072571,
      "learning_rate": 9.037234508926018e-07,
      "loss": 0.116,
      "step": 2891
    },
    {
      "epoch": 1.8140191312529401,
      "grad_norm": 0.8267726302146912,
      "learning_rate": 8.976938800552059e-07,
      "loss": 0.1225,
      "step": 2892
    },
    {
      "epoch": 1.8146463854477028,
      "grad_norm": 0.8965389728546143,
      "learning_rate": 8.916840292008921e-07,
      "loss": 0.1145,
      "step": 2893
    },
    {
      "epoch": 1.8152736396424651,
      "grad_norm": 1.0514887571334839,
      "learning_rate": 8.856939045338375e-07,
      "loss": 0.1642,
      "step": 2894
    },
    {
      "epoch": 1.8159008938372274,
      "grad_norm": 0.8502829074859619,
      "learning_rate": 8.797235122378623e-07,
      "loss": 0.1234,
      "step": 2895
    },
    {
      "epoch": 1.81652814803199,
      "grad_norm": 0.9405344724655151,
      "learning_rate": 8.737728584764116e-07,
      "loss": 0.1415,
      "step": 2896
    },
    {
      "epoch": 1.8171554022267524,
      "grad_norm": 0.867764949798584,
      "learning_rate": 8.67841949392556e-07,
      "loss": 0.1112,
      "step": 2897
    },
    {
      "epoch": 1.8177826564215147,
      "grad_norm": 0.9235755205154419,
      "learning_rate": 8.619307911089803e-07,
      "loss": 0.1453,
      "step": 2898
    },
    {
      "epoch": 1.8184099106162772,
      "grad_norm": 0.9043789505958557,
      "learning_rate": 8.560393897279851e-07,
      "loss": 0.1333,
      "step": 2899
    },
    {
      "epoch": 1.8190371648110397,
      "grad_norm": 0.7508955001831055,
      "learning_rate": 8.501677513314654e-07,
      "loss": 0.1076,
      "step": 2900
    },
    {
      "epoch": 1.819664419005802,
      "grad_norm": 0.8686279058456421,
      "learning_rate": 8.443158819809239e-07,
      "loss": 0.1297,
      "step": 2901
    },
    {
      "epoch": 1.8202916732005645,
      "grad_norm": 0.895136296749115,
      "learning_rate": 8.384837877174501e-07,
      "loss": 0.1353,
      "step": 2902
    },
    {
      "epoch": 1.820918927395327,
      "grad_norm": 0.8528735041618347,
      "learning_rate": 8.326714745617193e-07,
      "loss": 0.1298,
      "step": 2903
    },
    {
      "epoch": 1.8215461815900893,
      "grad_norm": 0.7849574089050293,
      "learning_rate": 8.268789485139916e-07,
      "loss": 0.1142,
      "step": 2904
    },
    {
      "epoch": 1.8221734357848518,
      "grad_norm": 0.9688342809677124,
      "learning_rate": 8.211062155540928e-07,
      "loss": 0.1088,
      "step": 2905
    },
    {
      "epoch": 1.8228006899796143,
      "grad_norm": 0.901724100112915,
      "learning_rate": 8.153532816414156e-07,
      "loss": 0.1269,
      "step": 2906
    },
    {
      "epoch": 1.8234279441743766,
      "grad_norm": 0.8248329758644104,
      "learning_rate": 8.096201527149183e-07,
      "loss": 0.1192,
      "step": 2907
    },
    {
      "epoch": 1.824055198369139,
      "grad_norm": 0.8081274628639221,
      "learning_rate": 8.039068346931134e-07,
      "loss": 0.1206,
      "step": 2908
    },
    {
      "epoch": 1.8246824525639016,
      "grad_norm": 0.88519686460495,
      "learning_rate": 7.982133334740583e-07,
      "loss": 0.1178,
      "step": 2909
    },
    {
      "epoch": 1.8253097067586639,
      "grad_norm": 0.9536784291267395,
      "learning_rate": 7.925396549353559e-07,
      "loss": 0.1379,
      "step": 2910
    },
    {
      "epoch": 1.8259369609534264,
      "grad_norm": 0.8239837288856506,
      "learning_rate": 7.868858049341432e-07,
      "loss": 0.1202,
      "step": 2911
    },
    {
      "epoch": 1.826564215148189,
      "grad_norm": 0.8505097031593323,
      "learning_rate": 7.812517893070892e-07,
      "loss": 0.124,
      "step": 2912
    },
    {
      "epoch": 1.8271914693429512,
      "grad_norm": 0.8420546650886536,
      "learning_rate": 7.756376138703859e-07,
      "loss": 0.134,
      "step": 2913
    },
    {
      "epoch": 1.8278187235377137,
      "grad_norm": 1.1077483892440796,
      "learning_rate": 7.700432844197436e-07,
      "loss": 0.1689,
      "step": 2914
    },
    {
      "epoch": 1.8284459777324762,
      "grad_norm": 0.8993534445762634,
      "learning_rate": 7.644688067303874e-07,
      "loss": 0.1239,
      "step": 2915
    },
    {
      "epoch": 1.8290732319272385,
      "grad_norm": 0.8536310195922852,
      "learning_rate": 7.589141865570493e-07,
      "loss": 0.1358,
      "step": 2916
    },
    {
      "epoch": 1.829700486122001,
      "grad_norm": 0.8410910964012146,
      "learning_rate": 7.533794296339536e-07,
      "loss": 0.1039,
      "step": 2917
    },
    {
      "epoch": 1.8303277403167635,
      "grad_norm": 0.916085422039032,
      "learning_rate": 7.478645416748298e-07,
      "loss": 0.1362,
      "step": 2918
    },
    {
      "epoch": 1.8309549945115258,
      "grad_norm": 0.8140565752983093,
      "learning_rate": 7.423695283728838e-07,
      "loss": 0.1255,
      "step": 2919
    },
    {
      "epoch": 1.8315822487062883,
      "grad_norm": 0.9491639137268066,
      "learning_rate": 7.368943954008179e-07,
      "loss": 0.1356,
      "step": 2920
    },
    {
      "epoch": 1.8322095029010508,
      "grad_norm": 0.8765885829925537,
      "learning_rate": 7.314391484108019e-07,
      "loss": 0.1404,
      "step": 2921
    },
    {
      "epoch": 1.832836757095813,
      "grad_norm": 0.9048905372619629,
      "learning_rate": 7.260037930344821e-07,
      "loss": 0.1324,
      "step": 2922
    },
    {
      "epoch": 1.8334640112905753,
      "grad_norm": 0.9376583099365234,
      "learning_rate": 7.205883348829612e-07,
      "loss": 0.1256,
      "step": 2923
    },
    {
      "epoch": 1.834091265485338,
      "grad_norm": 0.9348058104515076,
      "learning_rate": 7.151927795468139e-07,
      "loss": 0.1406,
      "step": 2924
    },
    {
      "epoch": 1.8347185196801004,
      "grad_norm": 0.8614555597305298,
      "learning_rate": 7.098171325960579e-07,
      "loss": 0.1179,
      "step": 2925
    },
    {
      "epoch": 1.8353457738748626,
      "grad_norm": 0.7639898061752319,
      "learning_rate": 7.04461399580163e-07,
      "loss": 0.1206,
      "step": 2926
    },
    {
      "epoch": 1.8359730280696254,
      "grad_norm": 0.7343938946723938,
      "learning_rate": 6.991255860280444e-07,
      "loss": 0.0943,
      "step": 2927
    },
    {
      "epoch": 1.8366002822643877,
      "grad_norm": 0.9248484969139099,
      "learning_rate": 6.938096974480468e-07,
      "loss": 0.1055,
      "step": 2928
    },
    {
      "epoch": 1.83722753645915,
      "grad_norm": 0.7805013656616211,
      "learning_rate": 6.885137393279495e-07,
      "loss": 0.1135,
      "step": 2929
    },
    {
      "epoch": 1.8378547906539127,
      "grad_norm": 0.9166563153266907,
      "learning_rate": 6.832377171349569e-07,
      "loss": 0.1192,
      "step": 2930
    },
    {
      "epoch": 1.838482044848675,
      "grad_norm": 1.0043163299560547,
      "learning_rate": 6.779816363156944e-07,
      "loss": 0.1586,
      "step": 2931
    },
    {
      "epoch": 1.8391092990434372,
      "grad_norm": 0.894653856754303,
      "learning_rate": 6.727455022961948e-07,
      "loss": 0.1547,
      "step": 2932
    },
    {
      "epoch": 1.8397365532381997,
      "grad_norm": 0.8928349614143372,
      "learning_rate": 6.675293204819078e-07,
      "loss": 0.1315,
      "step": 2933
    },
    {
      "epoch": 1.8403638074329622,
      "grad_norm": 0.883644700050354,
      "learning_rate": 6.623330962576835e-07,
      "loss": 0.1495,
      "step": 2934
    },
    {
      "epoch": 1.8409910616277245,
      "grad_norm": 0.9369957447052002,
      "learning_rate": 6.571568349877578e-07,
      "loss": 0.1354,
      "step": 2935
    },
    {
      "epoch": 1.841618315822487,
      "grad_norm": 0.8621444702148438,
      "learning_rate": 6.52000542015776e-07,
      "loss": 0.1201,
      "step": 2936
    },
    {
      "epoch": 1.8422455700172495,
      "grad_norm": 0.8870798349380493,
      "learning_rate": 6.468642226647582e-07,
      "loss": 0.151,
      "step": 2937
    },
    {
      "epoch": 1.8428728242120118,
      "grad_norm": 0.8756100535392761,
      "learning_rate": 6.417478822371071e-07,
      "loss": 0.1324,
      "step": 2938
    },
    {
      "epoch": 1.8435000784067743,
      "grad_norm": 0.8785619139671326,
      "learning_rate": 6.366515260146022e-07,
      "loss": 0.1262,
      "step": 2939
    },
    {
      "epoch": 1.8441273326015368,
      "grad_norm": 0.8960946798324585,
      "learning_rate": 6.315751592583908e-07,
      "loss": 0.1395,
      "step": 2940
    },
    {
      "epoch": 1.8447545867962991,
      "grad_norm": 0.9922255277633667,
      "learning_rate": 6.265187872089851e-07,
      "loss": 0.1409,
      "step": 2941
    },
    {
      "epoch": 1.8453818409910616,
      "grad_norm": 0.767423152923584,
      "learning_rate": 6.214824150862542e-07,
      "loss": 0.1037,
      "step": 2942
    },
    {
      "epoch": 1.8460090951858241,
      "grad_norm": 0.9177699685096741,
      "learning_rate": 6.164660480894235e-07,
      "loss": 0.1525,
      "step": 2943
    },
    {
      "epoch": 1.8466363493805864,
      "grad_norm": 0.9612424373626709,
      "learning_rate": 6.114696913970708e-07,
      "loss": 0.1493,
      "step": 2944
    },
    {
      "epoch": 1.847263603575349,
      "grad_norm": 0.6578683257102966,
      "learning_rate": 6.064933501671011e-07,
      "loss": 0.1368,
      "step": 2945
    },
    {
      "epoch": 1.8478908577701114,
      "grad_norm": 0.9469631910324097,
      "learning_rate": 6.015370295367739e-07,
      "loss": 0.1342,
      "step": 2946
    },
    {
      "epoch": 1.8485181119648737,
      "grad_norm": 0.9297729134559631,
      "learning_rate": 5.966007346226699e-07,
      "loss": 0.1198,
      "step": 2947
    },
    {
      "epoch": 1.8491453661596362,
      "grad_norm": 0.9222840666770935,
      "learning_rate": 5.916844705207014e-07,
      "loss": 0.1262,
      "step": 2948
    },
    {
      "epoch": 1.8497726203543987,
      "grad_norm": 0.8505162596702576,
      "learning_rate": 5.867882423061e-07,
      "loss": 0.1204,
      "step": 2949
    },
    {
      "epoch": 1.850399874549161,
      "grad_norm": 0.8304736018180847,
      "learning_rate": 5.819120550334157e-07,
      "loss": 0.12,
      "step": 2950
    },
    {
      "epoch": 1.8510271287439235,
      "grad_norm": 0.816307008266449,
      "learning_rate": 5.770559137365084e-07,
      "loss": 0.1164,
      "step": 2951
    },
    {
      "epoch": 1.851654382938686,
      "grad_norm": 0.8012158274650574,
      "learning_rate": 5.722198234285436e-07,
      "loss": 0.123,
      "step": 2952
    },
    {
      "epoch": 1.8522816371334483,
      "grad_norm": 0.7665334343910217,
      "learning_rate": 5.674037891019857e-07,
      "loss": 0.12,
      "step": 2953
    },
    {
      "epoch": 1.8529088913282108,
      "grad_norm": 0.87762451171875,
      "learning_rate": 5.626078157285996e-07,
      "loss": 0.116,
      "step": 2954
    },
    {
      "epoch": 1.8535361455229733,
      "grad_norm": 0.9224951863288879,
      "learning_rate": 5.578319082594341e-07,
      "loss": 0.1446,
      "step": 2955
    },
    {
      "epoch": 1.8541633997177356,
      "grad_norm": 0.8734859228134155,
      "learning_rate": 5.530760716248274e-07,
      "loss": 0.1242,
      "step": 2956
    },
    {
      "epoch": 1.854790653912498,
      "grad_norm": 0.9852175712585449,
      "learning_rate": 5.48340310734401e-07,
      "loss": 0.1414,
      "step": 2957
    },
    {
      "epoch": 1.8554179081072606,
      "grad_norm": 0.8510518670082092,
      "learning_rate": 5.43624630477042e-07,
      "loss": 0.1275,
      "step": 2958
    },
    {
      "epoch": 1.8560451623020229,
      "grad_norm": 0.8023516535758972,
      "learning_rate": 5.389290357209165e-07,
      "loss": 0.0947,
      "step": 2959
    },
    {
      "epoch": 1.8566724164967852,
      "grad_norm": 1.0212147235870361,
      "learning_rate": 5.342535313134511e-07,
      "loss": 0.1418,
      "step": 2960
    },
    {
      "epoch": 1.8572996706915479,
      "grad_norm": 0.8918972015380859,
      "learning_rate": 5.29598122081334e-07,
      "loss": 0.1208,
      "step": 2961
    },
    {
      "epoch": 1.8579269248863102,
      "grad_norm": 0.8254719376564026,
      "learning_rate": 5.249628128305073e-07,
      "loss": 0.1187,
      "step": 2962
    },
    {
      "epoch": 1.8585541790810725,
      "grad_norm": 0.8412727117538452,
      "learning_rate": 5.203476083461678e-07,
      "loss": 0.1292,
      "step": 2963
    },
    {
      "epoch": 1.8591814332758352,
      "grad_norm": 0.7970728874206543,
      "learning_rate": 5.157525133927465e-07,
      "loss": 0.1186,
      "step": 2964
    },
    {
      "epoch": 1.8598086874705975,
      "grad_norm": 0.8740736246109009,
      "learning_rate": 5.111775327139268e-07,
      "loss": 0.1425,
      "step": 2965
    },
    {
      "epoch": 1.8604359416653597,
      "grad_norm": 0.7795693278312683,
      "learning_rate": 5.066226710326216e-07,
      "loss": 0.1175,
      "step": 2966
    },
    {
      "epoch": 1.8610631958601225,
      "grad_norm": 0.770154595375061,
      "learning_rate": 5.02087933050972e-07,
      "loss": 0.1143,
      "step": 2967
    },
    {
      "epoch": 1.8616904500548848,
      "grad_norm": 0.9342362880706787,
      "learning_rate": 4.975733234503532e-07,
      "loss": 0.1372,
      "step": 2968
    },
    {
      "epoch": 1.862317704249647,
      "grad_norm": 0.9226587414741516,
      "learning_rate": 4.930788468913528e-07,
      "loss": 0.159,
      "step": 2969
    },
    {
      "epoch": 1.8629449584444095,
      "grad_norm": 0.8717913627624512,
      "learning_rate": 4.886045080137769e-07,
      "loss": 0.1161,
      "step": 2970
    },
    {
      "epoch": 1.863572212639172,
      "grad_norm": 0.8667058348655701,
      "learning_rate": 4.841503114366397e-07,
      "loss": 0.1362,
      "step": 2971
    },
    {
      "epoch": 1.8641994668339343,
      "grad_norm": 0.864131510257721,
      "learning_rate": 4.797162617581696e-07,
      "loss": 0.1247,
      "step": 2972
    },
    {
      "epoch": 1.8648267210286968,
      "grad_norm": 0.6426189541816711,
      "learning_rate": 4.7530236355579186e-07,
      "loss": 0.0818,
      "step": 2973
    },
    {
      "epoch": 1.8654539752234593,
      "grad_norm": 0.78322434425354,
      "learning_rate": 4.70908621386128e-07,
      "loss": 0.1062,
      "step": 2974
    },
    {
      "epoch": 1.8660812294182216,
      "grad_norm": 0.8293128609657288,
      "learning_rate": 4.665350397849922e-07,
      "loss": 0.1242,
      "step": 2975
    },
    {
      "epoch": 1.8667084836129841,
      "grad_norm": 0.8424291610717773,
      "learning_rate": 4.621816232673881e-07,
      "loss": 0.1294,
      "step": 2976
    },
    {
      "epoch": 1.8673357378077466,
      "grad_norm": 0.8814021348953247,
      "learning_rate": 4.578483763274988e-07,
      "loss": 0.1292,
      "step": 2977
    },
    {
      "epoch": 1.867962992002509,
      "grad_norm": 0.8511276245117188,
      "learning_rate": 4.5353530343869025e-07,
      "loss": 0.1195,
      "step": 2978
    },
    {
      "epoch": 1.8685902461972714,
      "grad_norm": 0.8883662819862366,
      "learning_rate": 4.4924240905349635e-07,
      "loss": 0.1269,
      "step": 2979
    },
    {
      "epoch": 1.869217500392034,
      "grad_norm": 0.899042546749115,
      "learning_rate": 4.449696976036277e-07,
      "loss": 0.1163,
      "step": 2980
    },
    {
      "epoch": 1.8698447545867962,
      "grad_norm": 0.888177216053009,
      "learning_rate": 4.4071717349995157e-07,
      "loss": 0.1195,
      "step": 2981
    },
    {
      "epoch": 1.8704720087815587,
      "grad_norm": 0.8384679555892944,
      "learning_rate": 4.364848411324962e-07,
      "loss": 0.1324,
      "step": 2982
    },
    {
      "epoch": 1.8710992629763212,
      "grad_norm": 0.9146760702133179,
      "learning_rate": 4.3227270487045114e-07,
      "loss": 0.121,
      "step": 2983
    },
    {
      "epoch": 1.8717265171710835,
      "grad_norm": 0.9058627486228943,
      "learning_rate": 4.2808076906215135e-07,
      "loss": 0.1149,
      "step": 2984
    },
    {
      "epoch": 1.872353771365846,
      "grad_norm": 0.8663889169692993,
      "learning_rate": 4.2390903803507965e-07,
      "loss": 0.133,
      "step": 2985
    },
    {
      "epoch": 1.8729810255606085,
      "grad_norm": 0.7832933664321899,
      "learning_rate": 4.1975751609586225e-07,
      "loss": 0.0965,
      "step": 2986
    },
    {
      "epoch": 1.8736082797553708,
      "grad_norm": 0.8505681157112122,
      "learning_rate": 4.1562620753025975e-07,
      "loss": 0.1207,
      "step": 2987
    },
    {
      "epoch": 1.8742355339501333,
      "grad_norm": 0.8499687910079956,
      "learning_rate": 4.115151166031672e-07,
      "loss": 0.1255,
      "step": 2988
    },
    {
      "epoch": 1.8748627881448958,
      "grad_norm": 0.8543071746826172,
      "learning_rate": 4.074242475586099e-07,
      "loss": 0.1346,
      "step": 2989
    },
    {
      "epoch": 1.875490042339658,
      "grad_norm": 0.8508833050727844,
      "learning_rate": 4.0335360461973395e-07,
      "loss": 0.115,
      "step": 2990
    },
    {
      "epoch": 1.8761172965344206,
      "grad_norm": 0.8167310953140259,
      "learning_rate": 3.993031919888113e-07,
      "loss": 0.1202,
      "step": 2991
    },
    {
      "epoch": 1.8767445507291831,
      "grad_norm": 0.8891164660453796,
      "learning_rate": 3.952730138472216e-07,
      "loss": 0.1417,
      "step": 2992
    },
    {
      "epoch": 1.8773718049239454,
      "grad_norm": 0.7716173529624939,
      "learning_rate": 3.912630743554613e-07,
      "loss": 0.1173,
      "step": 2993
    },
    {
      "epoch": 1.877999059118708,
      "grad_norm": 1.014568567276001,
      "learning_rate": 3.8727337765313013e-07,
      "loss": 0.143,
      "step": 2994
    },
    {
      "epoch": 1.8786263133134704,
      "grad_norm": 0.8445636630058289,
      "learning_rate": 3.833039278589379e-07,
      "loss": 0.1265,
      "step": 2995
    },
    {
      "epoch": 1.8792535675082327,
      "grad_norm": 0.7436814904212952,
      "learning_rate": 3.7935472907068005e-07,
      "loss": 0.1048,
      "step": 2996
    },
    {
      "epoch": 1.879880821702995,
      "grad_norm": 0.8433228135108948,
      "learning_rate": 3.7542578536525763e-07,
      "loss": 0.1154,
      "step": 2997
    },
    {
      "epoch": 1.8805080758977577,
      "grad_norm": 0.8996210098266602,
      "learning_rate": 3.715171007986551e-07,
      "loss": 0.133,
      "step": 2998
    },
    {
      "epoch": 1.88113533009252,
      "grad_norm": 0.83963543176651,
      "learning_rate": 3.676286794059514e-07,
      "loss": 0.1017,
      "step": 2999
    },
    {
      "epoch": 1.8817625842872823,
      "grad_norm": 0.823365330696106,
      "learning_rate": 3.637605252012932e-07,
      "loss": 0.1167,
      "step": 3000
    },
    {
      "epoch": 1.882389838482045,
      "grad_norm": 0.7450454235076904,
      "learning_rate": 3.5991264217791535e-07,
      "loss": 0.117,
      "step": 3001
    },
    {
      "epoch": 1.8830170926768073,
      "grad_norm": 0.8203656077384949,
      "learning_rate": 3.560850343081246e-07,
      "loss": 0.1154,
      "step": 3002
    },
    {
      "epoch": 1.8836443468715696,
      "grad_norm": 0.8934017419815063,
      "learning_rate": 3.5227770554329134e-07,
      "loss": 0.1252,
      "step": 3003
    },
    {
      "epoch": 1.8842716010663323,
      "grad_norm": 0.8821412324905396,
      "learning_rate": 3.484906598138604e-07,
      "loss": 0.1241,
      "step": 3004
    },
    {
      "epoch": 1.8848988552610946,
      "grad_norm": 0.8018887042999268,
      "learning_rate": 3.4472390102932904e-07,
      "loss": 0.1273,
      "step": 3005
    },
    {
      "epoch": 1.8855261094558569,
      "grad_norm": 1.057244896888733,
      "learning_rate": 3.409774330782534e-07,
      "loss": 0.1173,
      "step": 3006
    },
    {
      "epoch": 1.8861533636506194,
      "grad_norm": 0.8295458555221558,
      "learning_rate": 3.372512598282462e-07,
      "loss": 0.1212,
      "step": 3007
    },
    {
      "epoch": 1.8867806178453819,
      "grad_norm": 1.0016921758651733,
      "learning_rate": 3.3354538512596847e-07,
      "loss": 0.1569,
      "step": 3008
    },
    {
      "epoch": 1.8874078720401442,
      "grad_norm": 0.8008216023445129,
      "learning_rate": 3.298598127971242e-07,
      "loss": 0.1238,
      "step": 3009
    },
    {
      "epoch": 1.8880351262349067,
      "grad_norm": 0.8262543082237244,
      "learning_rate": 3.261945466464611e-07,
      "loss": 0.1132,
      "step": 3010
    },
    {
      "epoch": 1.8886623804296692,
      "grad_norm": 0.9967755675315857,
      "learning_rate": 3.225495904577569e-07,
      "loss": 0.1236,
      "step": 3011
    },
    {
      "epoch": 1.8892896346244314,
      "grad_norm": 0.8778575658798218,
      "learning_rate": 3.1892494799383276e-07,
      "loss": 0.1194,
      "step": 3012
    },
    {
      "epoch": 1.889916888819194,
      "grad_norm": 0.7905102968215942,
      "learning_rate": 3.1532062299653334e-07,
      "loss": 0.1027,
      "step": 3013
    },
    {
      "epoch": 1.8905441430139565,
      "grad_norm": 0.8325642347335815,
      "learning_rate": 3.1173661918673326e-07,
      "loss": 0.102,
      "step": 3014
    },
    {
      "epoch": 1.8911713972087187,
      "grad_norm": 0.9177772998809814,
      "learning_rate": 3.081729402643219e-07,
      "loss": 0.1741,
      "step": 3015
    },
    {
      "epoch": 1.8917986514034812,
      "grad_norm": 0.8657806515693665,
      "learning_rate": 3.046295899082119e-07,
      "loss": 0.1316,
      "step": 3016
    },
    {
      "epoch": 1.8924259055982438,
      "grad_norm": 0.834739089012146,
      "learning_rate": 3.0110657177632616e-07,
      "loss": 0.1177,
      "step": 3017
    },
    {
      "epoch": 1.893053159793006,
      "grad_norm": 0.7131056189537048,
      "learning_rate": 2.976038895056044e-07,
      "loss": 0.1066,
      "step": 3018
    },
    {
      "epoch": 1.8936804139877685,
      "grad_norm": 1.0009664297103882,
      "learning_rate": 2.9412154671198735e-07,
      "loss": 0.1511,
      "step": 3019
    },
    {
      "epoch": 1.894307668182531,
      "grad_norm": 0.7970536351203918,
      "learning_rate": 2.906595469904194e-07,
      "loss": 0.1213,
      "step": 3020
    },
    {
      "epoch": 1.8949349223772933,
      "grad_norm": 0.8229966163635254,
      "learning_rate": 2.8721789391484623e-07,
      "loss": 0.1176,
      "step": 3021
    },
    {
      "epoch": 1.8955621765720558,
      "grad_norm": 0.9330523610115051,
      "learning_rate": 2.837965910382057e-07,
      "loss": 0.13,
      "step": 3022
    },
    {
      "epoch": 1.8961894307668183,
      "grad_norm": 0.9967787265777588,
      "learning_rate": 2.8039564189243027e-07,
      "loss": 0.1529,
      "step": 3023
    },
    {
      "epoch": 1.8968166849615806,
      "grad_norm": 0.7935978174209595,
      "learning_rate": 2.7701504998843834e-07,
      "loss": 0.1405,
      "step": 3024
    },
    {
      "epoch": 1.8974439391563431,
      "grad_norm": 0.8954715132713318,
      "learning_rate": 2.7365481881613363e-07,
      "loss": 0.1269,
      "step": 3025
    },
    {
      "epoch": 1.8980711933511056,
      "grad_norm": 0.8750838041305542,
      "learning_rate": 2.703149518444037e-07,
      "loss": 0.1174,
      "step": 3026
    },
    {
      "epoch": 1.898698447545868,
      "grad_norm": 0.9578278660774231,
      "learning_rate": 2.6699545252110824e-07,
      "loss": 0.1426,
      "step": 3027
    },
    {
      "epoch": 1.8993257017406304,
      "grad_norm": 0.8749047517776489,
      "learning_rate": 2.636963242730861e-07,
      "loss": 0.1159,
      "step": 3028
    },
    {
      "epoch": 1.899952955935393,
      "grad_norm": 0.7772375345230103,
      "learning_rate": 2.604175705061396e-07,
      "loss": 0.1023,
      "step": 3029
    },
    {
      "epoch": 1.9005802101301552,
      "grad_norm": 0.9160566926002502,
      "learning_rate": 2.5715919460504336e-07,
      "loss": 0.1263,
      "step": 3030
    },
    {
      "epoch": 1.9012074643249177,
      "grad_norm": 0.7499338388442993,
      "learning_rate": 2.539211999335356e-07,
      "loss": 0.1117,
      "step": 3031
    },
    {
      "epoch": 1.9018347185196802,
      "grad_norm": 0.8376059532165527,
      "learning_rate": 2.5070358983430907e-07,
      "loss": 0.1137,
      "step": 3032
    },
    {
      "epoch": 1.9024619727144425,
      "grad_norm": 0.9330485463142395,
      "learning_rate": 2.475063676290179e-07,
      "loss": 0.1388,
      "step": 3033
    },
    {
      "epoch": 1.9030892269092048,
      "grad_norm": 0.926713228225708,
      "learning_rate": 2.443295366182663e-07,
      "loss": 0.12,
      "step": 3034
    },
    {
      "epoch": 1.9037164811039675,
      "grad_norm": 0.8909222483634949,
      "learning_rate": 2.411731000816109e-07,
      "loss": 0.1291,
      "step": 3035
    },
    {
      "epoch": 1.9043437352987298,
      "grad_norm": 0.9053573608398438,
      "learning_rate": 2.3803706127754732e-07,
      "loss": 0.1333,
      "step": 3036
    },
    {
      "epoch": 1.904970989493492,
      "grad_norm": 0.8453134894371033,
      "learning_rate": 2.3492142344352375e-07,
      "loss": 0.1175,
      "step": 3037
    },
    {
      "epoch": 1.9055982436882548,
      "grad_norm": 0.9282842874526978,
      "learning_rate": 2.3182618979592065e-07,
      "loss": 0.1426,
      "step": 3038
    },
    {
      "epoch": 1.906225497883017,
      "grad_norm": 0.7932876944541931,
      "learning_rate": 2.2875136353005755e-07,
      "loss": 0.113,
      "step": 3039
    },
    {
      "epoch": 1.9068527520777794,
      "grad_norm": 0.7805168032646179,
      "learning_rate": 2.2569694782018426e-07,
      "loss": 0.139,
      "step": 3040
    },
    {
      "epoch": 1.907480006272542,
      "grad_norm": 1.2216638326644897,
      "learning_rate": 2.226629458194851e-07,
      "loss": 0.1432,
      "step": 3041
    },
    {
      "epoch": 1.9081072604673044,
      "grad_norm": 0.9536857604980469,
      "learning_rate": 2.1964936066006138e-07,
      "loss": 0.1372,
      "step": 3042
    },
    {
      "epoch": 1.9087345146620667,
      "grad_norm": 0.8344295024871826,
      "learning_rate": 2.166561954529489e-07,
      "loss": 0.108,
      "step": 3043
    },
    {
      "epoch": 1.9093617688568292,
      "grad_norm": 0.7920276522636414,
      "learning_rate": 2.1368345328809382e-07,
      "loss": 0.1274,
      "step": 3044
    },
    {
      "epoch": 1.9099890230515917,
      "grad_norm": 0.8191457986831665,
      "learning_rate": 2.1073113723436566e-07,
      "loss": 0.1308,
      "step": 3045
    },
    {
      "epoch": 1.910616277246354,
      "grad_norm": 0.7653352618217468,
      "learning_rate": 2.0779925033953985e-07,
      "loss": 0.1087,
      "step": 3046
    },
    {
      "epoch": 1.9112435314411165,
      "grad_norm": 0.8056324124336243,
      "learning_rate": 2.048877956303108e-07,
      "loss": 0.1159,
      "step": 3047
    },
    {
      "epoch": 1.911870785635879,
      "grad_norm": 1.0009253025054932,
      "learning_rate": 2.0199677611227209e-07,
      "loss": 0.1403,
      "step": 3048
    },
    {
      "epoch": 1.9124980398306413,
      "grad_norm": 1.1077009439468384,
      "learning_rate": 1.9912619476992745e-07,
      "loss": 0.1588,
      "step": 3049
    },
    {
      "epoch": 1.9131252940254038,
      "grad_norm": 0.7939375638961792,
      "learning_rate": 1.9627605456667975e-07,
      "loss": 0.1107,
      "step": 3050
    },
    {
      "epoch": 1.9137525482201663,
      "grad_norm": 0.9140084981918335,
      "learning_rate": 1.934463584448265e-07,
      "loss": 0.1457,
      "step": 3051
    },
    {
      "epoch": 1.9143798024149286,
      "grad_norm": 0.8728494644165039,
      "learning_rate": 1.9063710932556435e-07,
      "loss": 0.1368,
      "step": 3052
    },
    {
      "epoch": 1.915007056609691,
      "grad_norm": 0.9430707693099976,
      "learning_rate": 1.8784831010897787e-07,
      "loss": 0.1518,
      "step": 3053
    },
    {
      "epoch": 1.9156343108044536,
      "grad_norm": 0.8481541872024536,
      "learning_rate": 1.8507996367404413e-07,
      "loss": 0.1127,
      "step": 3054
    },
    {
      "epoch": 1.9162615649992158,
      "grad_norm": 0.9993071556091309,
      "learning_rate": 1.8233207287862597e-07,
      "loss": 0.1533,
      "step": 3055
    },
    {
      "epoch": 1.9168888191939784,
      "grad_norm": 0.9163568019866943,
      "learning_rate": 1.7960464055946536e-07,
      "loss": 0.104,
      "step": 3056
    },
    {
      "epoch": 1.9175160733887409,
      "grad_norm": 1.0502631664276123,
      "learning_rate": 1.7689766953219e-07,
      "loss": 0.135,
      "step": 3057
    },
    {
      "epoch": 1.9181433275835031,
      "grad_norm": 0.837598443031311,
      "learning_rate": 1.742111625912979e-07,
      "loss": 0.1222,
      "step": 3058
    },
    {
      "epoch": 1.9187705817782656,
      "grad_norm": 0.865238606929779,
      "learning_rate": 1.7154512251016608e-07,
      "loss": 0.1226,
      "step": 3059
    },
    {
      "epoch": 1.9193978359730282,
      "grad_norm": 0.9549113512039185,
      "learning_rate": 1.688995520410397e-07,
      "loss": 0.1448,
      "step": 3060
    },
    {
      "epoch": 1.9200250901677904,
      "grad_norm": 0.9464938044548035,
      "learning_rate": 1.662744539150385e-07,
      "loss": 0.1243,
      "step": 3061
    },
    {
      "epoch": 1.920652344362553,
      "grad_norm": 0.9034533500671387,
      "learning_rate": 1.6366983084214139e-07,
      "loss": 0.147,
      "step": 3062
    },
    {
      "epoch": 1.9212795985573154,
      "grad_norm": 0.9751142263412476,
      "learning_rate": 1.6108568551119308e-07,
      "loss": 0.1495,
      "step": 3063
    },
    {
      "epoch": 1.9219068527520777,
      "grad_norm": 0.880317747592926,
      "learning_rate": 1.585220205898952e-07,
      "loss": 0.1256,
      "step": 3064
    },
    {
      "epoch": 1.9225341069468402,
      "grad_norm": 0.8028441667556763,
      "learning_rate": 1.559788387248129e-07,
      "loss": 0.1176,
      "step": 3065
    },
    {
      "epoch": 1.9231613611416027,
      "grad_norm": 0.8881433010101318,
      "learning_rate": 1.5345614254135944e-07,
      "loss": 0.1104,
      "step": 3066
    },
    {
      "epoch": 1.923788615336365,
      "grad_norm": 0.8313564658164978,
      "learning_rate": 1.5095393464380493e-07,
      "loss": 0.138,
      "step": 3067
    },
    {
      "epoch": 1.9244158695311275,
      "grad_norm": 0.9249715209007263,
      "learning_rate": 1.484722176152653e-07,
      "loss": 0.1298,
      "step": 3068
    },
    {
      "epoch": 1.92504312372589,
      "grad_norm": 1.0869942903518677,
      "learning_rate": 1.4601099401770458e-07,
      "loss": 0.115,
      "step": 3069
    },
    {
      "epoch": 1.9256703779206523,
      "grad_norm": 0.8350560069084167,
      "learning_rate": 1.435702663919325e-07,
      "loss": 0.1181,
      "step": 3070
    },
    {
      "epoch": 1.9262976321154148,
      "grad_norm": 0.8322076797485352,
      "learning_rate": 1.411500372575958e-07,
      "loss": 0.1208,
      "step": 3071
    },
    {
      "epoch": 1.9269248863101773,
      "grad_norm": 0.732231855392456,
      "learning_rate": 1.3875030911318256e-07,
      "loss": 0.1001,
      "step": 3072
    },
    {
      "epoch": 1.9275521405049396,
      "grad_norm": 0.8648920059204102,
      "learning_rate": 1.3637108443601776e-07,
      "loss": 0.1344,
      "step": 3073
    },
    {
      "epoch": 1.928179394699702,
      "grad_norm": 0.9535402059555054,
      "learning_rate": 1.3401236568225674e-07,
      "loss": 0.1525,
      "step": 3074
    },
    {
      "epoch": 1.9288066488944646,
      "grad_norm": 0.8213782906532288,
      "learning_rate": 1.3167415528689165e-07,
      "loss": 0.1296,
      "step": 3075
    },
    {
      "epoch": 1.929433903089227,
      "grad_norm": 0.8394836187362671,
      "learning_rate": 1.293564556637339e-07,
      "loss": 0.1157,
      "step": 3076
    },
    {
      "epoch": 1.9300611572839892,
      "grad_norm": 0.9124313592910767,
      "learning_rate": 1.2705926920543177e-07,
      "loss": 0.1365,
      "step": 3077
    },
    {
      "epoch": 1.930688411478752,
      "grad_norm": 0.9417162537574768,
      "learning_rate": 1.247825982834483e-07,
      "loss": 0.1227,
      "step": 3078
    },
    {
      "epoch": 1.9313156656735142,
      "grad_norm": 1.0487611293792725,
      "learning_rate": 1.2252644524807233e-07,
      "loss": 0.136,
      "step": 3079
    },
    {
      "epoch": 1.9319429198682765,
      "grad_norm": 0.7279402017593384,
      "learning_rate": 1.2029081242840968e-07,
      "loss": 0.0945,
      "step": 3080
    },
    {
      "epoch": 1.932570174063039,
      "grad_norm": 0.7787689566612244,
      "learning_rate": 1.1807570213238307e-07,
      "loss": 0.1133,
      "step": 3081
    },
    {
      "epoch": 1.9331974282578015,
      "grad_norm": 0.8247366547584534,
      "learning_rate": 1.1588111664672996e-07,
      "loss": 0.0978,
      "step": 3082
    },
    {
      "epoch": 1.9338246824525638,
      "grad_norm": 0.9361110329627991,
      "learning_rate": 1.137070582369959e-07,
      "loss": 0.1098,
      "step": 3083
    },
    {
      "epoch": 1.9344519366473263,
      "grad_norm": 0.896119236946106,
      "learning_rate": 1.1155352914753892e-07,
      "loss": 0.1405,
      "step": 3084
    },
    {
      "epoch": 1.9350791908420888,
      "grad_norm": 0.973166823387146,
      "learning_rate": 1.0942053160152289e-07,
      "loss": 0.1345,
      "step": 3085
    },
    {
      "epoch": 1.935706445036851,
      "grad_norm": 0.8656417727470398,
      "learning_rate": 1.0730806780091751e-07,
      "loss": 0.1383,
      "step": 3086
    },
    {
      "epoch": 1.9363336992316136,
      "grad_norm": 0.8609120845794678,
      "learning_rate": 1.0521613992649171e-07,
      "loss": 0.1187,
      "step": 3087
    },
    {
      "epoch": 1.936960953426376,
      "grad_norm": 0.8523600697517395,
      "learning_rate": 1.0314475013781578e-07,
      "loss": 0.1143,
      "step": 3088
    },
    {
      "epoch": 1.9375882076211384,
      "grad_norm": 0.8735218048095703,
      "learning_rate": 1.010939005732614e-07,
      "loss": 0.1231,
      "step": 3089
    },
    {
      "epoch": 1.9382154618159009,
      "grad_norm": 0.8490827679634094,
      "learning_rate": 9.906359334999061e-08,
      "loss": 0.1266,
      "step": 3090
    },
    {
      "epoch": 1.9388427160106634,
      "grad_norm": 0.8786249160766602,
      "learning_rate": 9.705383056396012e-08,
      "loss": 0.124,
      "step": 3091
    },
    {
      "epoch": 1.9394699702054257,
      "grad_norm": 0.9569567441940308,
      "learning_rate": 9.506461428992142e-08,
      "loss": 0.1358,
      "step": 3092
    },
    {
      "epoch": 1.9400972244001882,
      "grad_norm": 0.7432516813278198,
      "learning_rate": 9.309594658141185e-08,
      "loss": 0.1143,
      "step": 3093
    },
    {
      "epoch": 1.9407244785949507,
      "grad_norm": 0.9143378138542175,
      "learning_rate": 9.114782947075462e-08,
      "loss": 0.1389,
      "step": 3094
    },
    {
      "epoch": 1.941351732789713,
      "grad_norm": 0.8691182732582092,
      "learning_rate": 8.922026496905878e-08,
      "loss": 0.1154,
      "step": 3095
    },
    {
      "epoch": 1.9419789869844755,
      "grad_norm": 0.9375434517860413,
      "learning_rate": 8.731325506622145e-08,
      "loss": 0.1266,
      "step": 3096
    },
    {
      "epoch": 1.942606241179238,
      "grad_norm": 0.7888829112052917,
      "learning_rate": 8.542680173091233e-08,
      "loss": 0.1137,
      "step": 3097
    },
    {
      "epoch": 1.9432334953740003,
      "grad_norm": 0.8019591569900513,
      "learning_rate": 8.356090691058693e-08,
      "loss": 0.1047,
      "step": 3098
    },
    {
      "epoch": 1.9438607495687628,
      "grad_norm": 1.0162184238433838,
      "learning_rate": 8.171557253146889e-08,
      "loss": 0.153,
      "step": 3099
    },
    {
      "epoch": 1.9444880037635253,
      "grad_norm": 0.8386593461036682,
      "learning_rate": 7.989080049856546e-08,
      "loss": 0.1238,
      "step": 3100
    },
    {
      "epoch": 1.9451152579582875,
      "grad_norm": 0.9435861110687256,
      "learning_rate": 7.808659269565422e-08,
      "loss": 0.1662,
      "step": 3101
    },
    {
      "epoch": 1.94574251215305,
      "grad_norm": 0.7505217790603638,
      "learning_rate": 7.630295098527862e-08,
      "loss": 0.0964,
      "step": 3102
    },
    {
      "epoch": 1.9463697663478126,
      "grad_norm": 0.9808968901634216,
      "learning_rate": 7.45398772087591e-08,
      "loss": 0.153,
      "step": 3103
    },
    {
      "epoch": 1.9469970205425748,
      "grad_norm": 0.8741431832313538,
      "learning_rate": 7.27973731861753e-08,
      "loss": 0.1229,
      "step": 3104
    },
    {
      "epoch": 1.9476242747373373,
      "grad_norm": 0.9125285148620605,
      "learning_rate": 7.107544071637718e-08,
      "loss": 0.1276,
      "step": 3105
    },
    {
      "epoch": 1.9482515289320999,
      "grad_norm": 0.9317863583564758,
      "learning_rate": 6.937408157697612e-08,
      "loss": 0.1529,
      "step": 3106
    },
    {
      "epoch": 1.9488787831268621,
      "grad_norm": 0.9088873267173767,
      "learning_rate": 6.769329752434272e-08,
      "loss": 0.151,
      "step": 3107
    },
    {
      "epoch": 1.9495060373216246,
      "grad_norm": 0.9075539708137512,
      "learning_rate": 6.603309029361349e-08,
      "loss": 0.1226,
      "step": 3108
    },
    {
      "epoch": 1.9501332915163871,
      "grad_norm": 0.9500503540039062,
      "learning_rate": 6.439346159867743e-08,
      "loss": 0.1379,
      "step": 3109
    },
    {
      "epoch": 1.9507605457111494,
      "grad_norm": 0.9206661581993103,
      "learning_rate": 6.277441313218058e-08,
      "loss": 0.1397,
      "step": 3110
    },
    {
      "epoch": 1.9513877999059117,
      "grad_norm": 0.8864201307296753,
      "learning_rate": 6.11759465655215e-08,
      "loss": 0.111,
      "step": 3111
    },
    {
      "epoch": 1.9520150541006744,
      "grad_norm": 0.7336342334747314,
      "learning_rate": 5.9598063548855776e-08,
      "loss": 0.1047,
      "step": 3112
    },
    {
      "epoch": 1.9526423082954367,
      "grad_norm": 0.8321946859359741,
      "learning_rate": 5.80407657110893e-08,
      "loss": 0.1122,
      "step": 3113
    },
    {
      "epoch": 1.953269562490199,
      "grad_norm": 0.8356947302818298,
      "learning_rate": 5.650405465987163e-08,
      "loss": 0.1242,
      "step": 3114
    },
    {
      "epoch": 1.9538968166849617,
      "grad_norm": 0.838462233543396,
      "learning_rate": 5.4987931981607124e-08,
      "loss": 0.1213,
      "step": 3115
    },
    {
      "epoch": 1.954524070879724,
      "grad_norm": 1.0194828510284424,
      "learning_rate": 5.349239924143934e-08,
      "loss": 0.167,
      "step": 3116
    },
    {
      "epoch": 1.9551513250744863,
      "grad_norm": 0.7172075510025024,
      "learning_rate": 5.2017457983259964e-08,
      "loss": 0.1136,
      "step": 3117
    },
    {
      "epoch": 1.955778579269249,
      "grad_norm": 0.8290290236473083,
      "learning_rate": 5.0563109729706575e-08,
      "loss": 0.1233,
      "step": 3118
    },
    {
      "epoch": 1.9564058334640113,
      "grad_norm": 0.8827482461929321,
      "learning_rate": 4.9129355982149294e-08,
      "loss": 0.127,
      "step": 3119
    },
    {
      "epoch": 1.9570330876587736,
      "grad_norm": 0.8358291983604431,
      "learning_rate": 4.771619822070639e-08,
      "loss": 0.128,
      "step": 3120
    },
    {
      "epoch": 1.957660341853536,
      "grad_norm": 0.8537582159042358,
      "learning_rate": 4.632363790422867e-08,
      "loss": 0.1241,
      "step": 3121
    },
    {
      "epoch": 1.9582875960482986,
      "grad_norm": 0.8248356580734253,
      "learning_rate": 4.495167647030396e-08,
      "loss": 0.1158,
      "step": 3122
    },
    {
      "epoch": 1.958914850243061,
      "grad_norm": 0.7904495000839233,
      "learning_rate": 4.36003153352571e-08,
      "loss": 0.1081,
      "step": 3123
    },
    {
      "epoch": 1.9595421044378234,
      "grad_norm": 0.8422828316688538,
      "learning_rate": 4.226955589414772e-08,
      "loss": 0.1284,
      "step": 3124
    },
    {
      "epoch": 1.960169358632586,
      "grad_norm": 0.8301834464073181,
      "learning_rate": 4.095939952076355e-08,
      "loss": 0.1059,
      "step": 3125
    },
    {
      "epoch": 1.9607966128273482,
      "grad_norm": 0.8255082368850708,
      "learning_rate": 3.966984756762493e-08,
      "loss": 0.1056,
      "step": 3126
    },
    {
      "epoch": 1.9614238670221107,
      "grad_norm": 0.8115298748016357,
      "learning_rate": 3.8400901365982514e-08,
      "loss": 0.1245,
      "step": 3127
    },
    {
      "epoch": 1.9620511212168732,
      "grad_norm": 0.7331147193908691,
      "learning_rate": 3.7152562225815094e-08,
      "loss": 0.0873,
      "step": 3128
    },
    {
      "epoch": 1.9626783754116355,
      "grad_norm": 0.8912408947944641,
      "learning_rate": 3.592483143582515e-08,
      "loss": 0.1511,
      "step": 3129
    },
    {
      "epoch": 1.963305629606398,
      "grad_norm": 0.8669161796569824,
      "learning_rate": 3.471771026344328e-08,
      "loss": 0.1191,
      "step": 3130
    },
    {
      "epoch": 1.9639328838011605,
      "grad_norm": 0.9340421557426453,
      "learning_rate": 3.353119995482379e-08,
      "loss": 0.1274,
      "step": 3131
    },
    {
      "epoch": 1.9645601379959228,
      "grad_norm": 0.9183062314987183,
      "learning_rate": 3.2365301734842427e-08,
      "loss": 0.1455,
      "step": 3132
    },
    {
      "epoch": 1.9651873921906853,
      "grad_norm": 0.9617796540260315,
      "learning_rate": 3.122001680709863e-08,
      "loss": 0.1298,
      "step": 3133
    },
    {
      "epoch": 1.9658146463854478,
      "grad_norm": 0.8618515133857727,
      "learning_rate": 3.0095346353906654e-08,
      "loss": 0.1308,
      "step": 3134
    },
    {
      "epoch": 1.96644190058021,
      "grad_norm": 0.944823682308197,
      "learning_rate": 2.899129153630664e-08,
      "loss": 0.1459,
      "step": 3135
    },
    {
      "epoch": 1.9670691547749726,
      "grad_norm": 0.9516502022743225,
      "learning_rate": 2.790785349405134e-08,
      "loss": 0.1519,
      "step": 3136
    },
    {
      "epoch": 1.967696408969735,
      "grad_norm": 0.8677295446395874,
      "learning_rate": 2.6845033345612726e-08,
      "loss": 0.1106,
      "step": 3137
    },
    {
      "epoch": 1.9683236631644974,
      "grad_norm": 0.7925218939781189,
      "learning_rate": 2.5802832188177583e-08,
      "loss": 0.1048,
      "step": 3138
    },
    {
      "epoch": 1.9689509173592599,
      "grad_norm": 0.9727739691734314,
      "learning_rate": 2.4781251097647506e-08,
      "loss": 0.1396,
      "step": 3139
    },
    {
      "epoch": 1.9695781715540224,
      "grad_norm": 0.9012995362281799,
      "learning_rate": 2.378029112863445e-08,
      "loss": 0.1292,
      "step": 3140
    },
    {
      "epoch": 1.9702054257487847,
      "grad_norm": 0.8515464663505554,
      "learning_rate": 2.279995331446738e-08,
      "loss": 0.1184,
      "step": 3141
    },
    {
      "epoch": 1.9708326799435472,
      "grad_norm": 0.8222053050994873,
      "learning_rate": 2.1840238667181213e-08,
      "loss": 0.1283,
      "step": 3142
    },
    {
      "epoch": 1.9714599341383097,
      "grad_norm": 0.948520302772522,
      "learning_rate": 2.0901148177525642e-08,
      "loss": 0.1469,
      "step": 3143
    },
    {
      "epoch": 1.972087188333072,
      "grad_norm": 0.8818579316139221,
      "learning_rate": 1.998268281495408e-08,
      "loss": 0.1415,
      "step": 3144
    },
    {
      "epoch": 1.9727144425278345,
      "grad_norm": 0.8317146897315979,
      "learning_rate": 1.9084843527632514e-08,
      "loss": 0.1138,
      "step": 3145
    },
    {
      "epoch": 1.973341696722597,
      "grad_norm": 0.9792901873588562,
      "learning_rate": 1.8207631242430635e-08,
      "loss": 0.12,
      "step": 3146
    },
    {
      "epoch": 1.9739689509173592,
      "grad_norm": 0.9085427522659302,
      "learning_rate": 1.735104686492628e-08,
      "loss": 0.1239,
      "step": 3147
    },
    {
      "epoch": 1.9745962051121215,
      "grad_norm": 0.8495365381240845,
      "learning_rate": 1.6515091279403206e-08,
      "loss": 0.1564,
      "step": 3148
    },
    {
      "epoch": 1.9752234593068843,
      "grad_norm": 0.931058943271637,
      "learning_rate": 1.569976534884443e-08,
      "loss": 0.1342,
      "step": 3149
    },
    {
      "epoch": 1.9758507135016465,
      "grad_norm": 0.7746802568435669,
      "learning_rate": 1.4905069914943336e-08,
      "loss": 0.1118,
      "step": 3150
    },
    {
      "epoch": 1.9764779676964088,
      "grad_norm": 0.7811020612716675,
      "learning_rate": 1.4131005798090347e-08,
      "loss": 0.1278,
      "step": 3151
    },
    {
      "epoch": 1.9771052218911715,
      "grad_norm": 0.9032891392707825,
      "learning_rate": 1.3377573797377363e-08,
      "loss": 0.1304,
      "step": 3152
    },
    {
      "epoch": 1.9777324760859338,
      "grad_norm": 0.8794789910316467,
      "learning_rate": 1.2644774690602214e-08,
      "loss": 0.152,
      "step": 3153
    },
    {
      "epoch": 1.9783597302806961,
      "grad_norm": 0.8922184705734253,
      "learning_rate": 1.193260923425532e-08,
      "loss": 0.129,
      "step": 3154
    },
    {
      "epoch": 1.9789869844754588,
      "grad_norm": 0.84577476978302,
      "learning_rate": 1.1241078163533038e-08,
      "loss": 0.1283,
      "step": 3155
    },
    {
      "epoch": 1.9796142386702211,
      "grad_norm": 0.8250657916069031,
      "learning_rate": 1.0570182192326528e-08,
      "loss": 0.1193,
      "step": 3156
    },
    {
      "epoch": 1.9802414928649834,
      "grad_norm": 0.9372608661651611,
      "learning_rate": 9.919922013226225e-09,
      "loss": 0.1617,
      "step": 3157
    },
    {
      "epoch": 1.980868747059746,
      "grad_norm": 0.8537949919700623,
      "learning_rate": 9.290298297515154e-09,
      "loss": 0.1299,
      "step": 3158
    },
    {
      "epoch": 1.9814960012545084,
      "grad_norm": 0.8301073908805847,
      "learning_rate": 8.68131169518005e-09,
      "loss": 0.1278,
      "step": 3159
    },
    {
      "epoch": 1.9821232554492707,
      "grad_norm": 0.940880537033081,
      "learning_rate": 8.092962834898022e-09,
      "loss": 0.1358,
      "step": 3160
    },
    {
      "epoch": 1.9827505096440332,
      "grad_norm": 0.8299143314361572,
      "learning_rate": 7.525252324041e-09,
      "loss": 0.112,
      "step": 3161
    },
    {
      "epoch": 1.9833777638387957,
      "grad_norm": 0.9687228798866272,
      "learning_rate": 6.978180748677954e-09,
      "loss": 0.1346,
      "step": 3162
    },
    {
      "epoch": 1.984005018033558,
      "grad_norm": 0.8207062482833862,
      "learning_rate": 6.4517486735682366e-09,
      "loss": 0.1169,
      "step": 3163
    },
    {
      "epoch": 1.9846322722283205,
      "grad_norm": 0.9281582236289978,
      "learning_rate": 5.945956642168238e-09,
      "loss": 0.1194,
      "step": 3164
    },
    {
      "epoch": 1.985259526423083,
      "grad_norm": 0.8668698668479919,
      "learning_rate": 5.460805176624729e-09,
      "loss": 0.1248,
      "step": 3165
    },
    {
      "epoch": 1.9858867806178453,
      "grad_norm": 0.9078335165977478,
      "learning_rate": 4.9962947777770775e-09,
      "loss": 0.1331,
      "step": 3166
    },
    {
      "epoch": 1.9865140348126078,
      "grad_norm": 0.911410391330719,
      "learning_rate": 4.552425925152815e-09,
      "loss": 0.1121,
      "step": 3167
    },
    {
      "epoch": 1.9871412890073703,
      "grad_norm": 1.0109318494796753,
      "learning_rate": 4.129199076976509e-09,
      "loss": 0.1333,
      "step": 3168
    },
    {
      "epoch": 1.9877685432021326,
      "grad_norm": 0.7333261966705322,
      "learning_rate": 3.726614670158668e-09,
      "loss": 0.1175,
      "step": 3169
    },
    {
      "epoch": 1.988395797396895,
      "grad_norm": 0.8047865033149719,
      "learning_rate": 3.3446731203023995e-09,
      "loss": 0.1159,
      "step": 3170
    },
    {
      "epoch": 1.9890230515916576,
      "grad_norm": 0.7697649598121643,
      "learning_rate": 2.9833748216967494e-09,
      "loss": 0.1023,
      "step": 3171
    },
    {
      "epoch": 1.9896503057864199,
      "grad_norm": 0.8524652719497681,
      "learning_rate": 2.642720147325584e-09,
      "loss": 0.125,
      "step": 3172
    },
    {
      "epoch": 1.9902775599811824,
      "grad_norm": 0.9473152756690979,
      "learning_rate": 2.3227094488587066e-09,
      "loss": 0.1342,
      "step": 3173
    },
    {
      "epoch": 1.990904814175945,
      "grad_norm": 0.9332677125930786,
      "learning_rate": 2.023343056651861e-09,
      "loss": 0.1433,
      "step": 3174
    },
    {
      "epoch": 1.9915320683707072,
      "grad_norm": 0.8399542570114136,
      "learning_rate": 1.744621279753389e-09,
      "loss": 0.1079,
      "step": 3175
    },
    {
      "epoch": 1.9921593225654697,
      "grad_norm": 0.9552322626113892,
      "learning_rate": 1.4865444058997924e-09,
      "loss": 0.1257,
      "step": 3176
    },
    {
      "epoch": 1.9927865767602322,
      "grad_norm": 1.0563658475875854,
      "learning_rate": 1.2491127015068494e-09,
      "loss": 0.1199,
      "step": 3177
    },
    {
      "epoch": 1.9934138309549945,
      "grad_norm": 0.8615870475769043,
      "learning_rate": 1.0323264116895991e-09,
      "loss": 0.1241,
      "step": 3178
    },
    {
      "epoch": 1.994041085149757,
      "grad_norm": 0.9080129265785217,
      "learning_rate": 8.36185760240138e-10,
      "loss": 0.1463,
      "step": 3179
    },
    {
      "epoch": 1.9946683393445195,
      "grad_norm": 0.9104377627372742,
      "learning_rate": 6.606909496453817e-10,
      "loss": 0.1264,
      "step": 3180
    },
    {
      "epoch": 1.9952955935392818,
      "grad_norm": 0.9355060458183289,
      "learning_rate": 5.058421610715237e-10,
      "loss": 0.1359,
      "step": 3181
    },
    {
      "epoch": 1.9959228477340443,
      "grad_norm": 0.8101538419723511,
      "learning_rate": 3.7163955437291567e-10,
      "loss": 0.1157,
      "step": 3182
    },
    {
      "epoch": 1.9965501019288068,
      "grad_norm": 0.9194393754005432,
      "learning_rate": 2.5808326809650954e-10,
      "loss": 0.1287,
      "step": 3183
    },
    {
      "epoch": 1.997177356123569,
      "grad_norm": 0.7908146977424622,
      "learning_rate": 1.6517341946853394e-10,
      "loss": 0.1039,
      "step": 3184
    },
    {
      "epoch": 1.9978046103183313,
      "grad_norm": 0.9933124780654907,
      "learning_rate": 9.2910104401156e-11,
      "loss": 0.1204,
      "step": 3185
    },
    {
      "epoch": 1.998431864513094,
      "grad_norm": 0.7302822470664978,
      "learning_rate": 4.1293397494701624e-11,
      "loss": 0.1045,
      "step": 3186
    },
    {
      "epoch": 1.9990591187078564,
      "grad_norm": 0.8666970133781433,
      "learning_rate": 1.0323352037655555e-11,
      "loss": 0.1419,
      "step": 3187
    },
    {
      "epoch": 1.9996863729026186,
      "grad_norm": 0.8657733201980591,
      "learning_rate": 0.0,
      "loss": 0.1267,
      "step": 3188
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 3188,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 200,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.579466809106206e+22,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}