{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 361,
  "global_step": 1444,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006925207756232687,
      "grad_norm": 0.46092769503593445,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.8461,
      "step": 1
    },
    {
      "epoch": 0.0006925207756232687,
      "eval_loss": 1.2960177659988403,
      "eval_runtime": 331.9286,
      "eval_samples_per_second": 4.266,
      "eval_steps_per_second": 0.533,
      "step": 1
    },
    {
      "epoch": 0.0013850415512465374,
      "grad_norm": 0.5322127342224121,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.9426,
      "step": 2
    },
    {
      "epoch": 0.002077562326869806,
      "grad_norm": 0.43637630343437195,
      "learning_rate": 6e-06,
      "loss": 1.7676,
      "step": 3
    },
    {
      "epoch": 0.002770083102493075,
      "grad_norm": 0.47813108563423157,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.8202,
      "step": 4
    },
    {
      "epoch": 0.0034626038781163434,
      "grad_norm": 0.4561654329299927,
      "learning_rate": 1e-05,
      "loss": 1.7695,
      "step": 5
    },
    {
      "epoch": 0.004155124653739612,
      "grad_norm": 0.5688905119895935,
      "learning_rate": 1.2e-05,
      "loss": 1.797,
      "step": 6
    },
    {
      "epoch": 0.004847645429362881,
      "grad_norm": 0.4705469012260437,
      "learning_rate": 1.4e-05,
      "loss": 1.7815,
      "step": 7
    },
    {
      "epoch": 0.00554016620498615,
      "grad_norm": 0.5055038332939148,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.8141,
      "step": 8
    },
    {
      "epoch": 0.006232686980609419,
      "grad_norm": 0.5161764621734619,
      "learning_rate": 1.8e-05,
      "loss": 1.6148,
      "step": 9
    },
    {
      "epoch": 0.006925207756232687,
      "grad_norm": 0.5202474594116211,
      "learning_rate": 2e-05,
      "loss": 1.6932,
      "step": 10
    },
    {
      "epoch": 0.007617728531855956,
      "grad_norm": 0.4921915829181671,
      "learning_rate": 1.9999976002208634e-05,
      "loss": 1.8003,
      "step": 11
    },
    {
      "epoch": 0.008310249307479225,
      "grad_norm": 0.44480928778648376,
      "learning_rate": 1.9999904008949705e-05,
      "loss": 1.7099,
      "step": 12
    },
    {
      "epoch": 0.009002770083102494,
      "grad_norm": 0.4888850748538971,
      "learning_rate": 1.9999784020568756e-05,
      "loss": 1.7404,
      "step": 13
    },
    {
      "epoch": 0.009695290858725761,
      "grad_norm": 0.4866941571235657,
      "learning_rate": 1.999961603764167e-05,
      "loss": 1.6584,
      "step": 14
    },
    {
      "epoch": 0.01038781163434903,
      "grad_norm": 0.5461321473121643,
      "learning_rate": 1.9999400060974704e-05,
      "loss": 1.797,
      "step": 15
    },
    {
      "epoch": 0.0110803324099723,
      "grad_norm": 0.536161482334137,
      "learning_rate": 1.9999136091604433e-05,
      "loss": 1.5394,
      "step": 16
    },
    {
      "epoch": 0.011772853185595568,
      "grad_norm": 0.6785447001457214,
      "learning_rate": 1.9998824130797813e-05,
      "loss": 1.5485,
      "step": 17
    },
    {
      "epoch": 0.012465373961218837,
      "grad_norm": 0.5797305107116699,
      "learning_rate": 1.99984641800521e-05,
      "loss": 1.4998,
      "step": 18
    },
    {
      "epoch": 0.013157894736842105,
      "grad_norm": 0.6252360343933105,
      "learning_rate": 1.9998056241094912e-05,
      "loss": 1.8451,
      "step": 19
    },
    {
      "epoch": 0.013850415512465374,
      "grad_norm": 0.5463359355926514,
      "learning_rate": 1.9997600315884166e-05,
      "loss": 1.6362,
      "step": 20
    },
    {
      "epoch": 0.014542936288088643,
      "grad_norm": 0.5481324791908264,
      "learning_rate": 1.9997096406608108e-05,
      "loss": 1.5417,
      "step": 21
    },
    {
      "epoch": 0.015235457063711912,
      "grad_norm": 0.4929562211036682,
      "learning_rate": 1.999654451568528e-05,
      "loss": 1.7517,
      "step": 22
    },
    {
      "epoch": 0.01592797783933518,
      "grad_norm": 0.547443151473999,
      "learning_rate": 1.9995944645764512e-05,
      "loss": 1.7079,
      "step": 23
    },
    {
      "epoch": 0.01662049861495845,
      "grad_norm": 0.497779905796051,
      "learning_rate": 1.9995296799724914e-05,
      "loss": 1.6336,
      "step": 24
    },
    {
      "epoch": 0.01731301939058172,
      "grad_norm": 0.5056659579277039,
      "learning_rate": 1.999460098067586e-05,
      "loss": 1.5759,
      "step": 25
    },
    {
      "epoch": 0.018005540166204988,
      "grad_norm": 0.5825318098068237,
      "learning_rate": 1.999385719195698e-05,
      "loss": 1.7271,
      "step": 26
    },
    {
      "epoch": 0.018698060941828253,
      "grad_norm": 0.5759397745132446,
      "learning_rate": 1.9993065437138126e-05,
      "loss": 1.6907,
      "step": 27
    },
    {
      "epoch": 0.019390581717451522,
      "grad_norm": 0.5402478575706482,
      "learning_rate": 1.9992225720019377e-05,
      "loss": 1.7526,
      "step": 28
    },
    {
      "epoch": 0.02008310249307479,
      "grad_norm": 0.5368712544441223,
      "learning_rate": 1.9991338044630996e-05,
      "loss": 1.6623,
      "step": 29
    },
    {
      "epoch": 0.02077562326869806,
      "grad_norm": 0.5437949895858765,
      "learning_rate": 1.9990402415233436e-05,
      "loss": 1.7876,
      "step": 30
    },
    {
      "epoch": 0.02146814404432133,
      "grad_norm": 0.6881880760192871,
      "learning_rate": 1.9989418836317304e-05,
      "loss": 1.5072,
      "step": 31
    },
    {
      "epoch": 0.0221606648199446,
      "grad_norm": 0.5721428990364075,
      "learning_rate": 1.998838731260335e-05,
      "loss": 1.5213,
      "step": 32
    },
    {
      "epoch": 0.022853185595567867,
      "grad_norm": 0.5982231497764587,
      "learning_rate": 1.9987307849042428e-05,
      "loss": 1.7325,
      "step": 33
    },
    {
      "epoch": 0.023545706371191136,
      "grad_norm": 0.6260273456573486,
      "learning_rate": 1.9986180450815485e-05,
      "loss": 1.6727,
      "step": 34
    },
    {
      "epoch": 0.024238227146814405,
      "grad_norm": 0.594240665435791,
      "learning_rate": 1.9985005123333536e-05,
      "loss": 1.7139,
      "step": 35
    },
    {
      "epoch": 0.024930747922437674,
      "grad_norm": 0.530646026134491,
      "learning_rate": 1.9983781872237634e-05,
      "loss": 1.6415,
      "step": 36
    },
    {
      "epoch": 0.025623268698060944,
      "grad_norm": 0.6217408776283264,
      "learning_rate": 1.9982510703398844e-05,
      "loss": 1.7185,
      "step": 37
    },
    {
      "epoch": 0.02631578947368421,
      "grad_norm": 0.537267804145813,
      "learning_rate": 1.9981191622918217e-05,
      "loss": 1.53,
      "step": 38
    },
    {
      "epoch": 0.027008310249307478,
      "grad_norm": 0.5151798129081726,
      "learning_rate": 1.9979824637126752e-05,
      "loss": 1.692,
      "step": 39
    },
    {
      "epoch": 0.027700831024930747,
      "grad_norm": 0.5791589021682739,
      "learning_rate": 1.997840975258538e-05,
      "loss": 1.6817,
      "step": 40
    },
    {
      "epoch": 0.028393351800554016,
      "grad_norm": 0.5138722062110901,
      "learning_rate": 1.9976946976084916e-05,
      "loss": 1.6058,
      "step": 41
    },
    {
      "epoch": 0.029085872576177285,
      "grad_norm": 0.5150821805000305,
      "learning_rate": 1.9975436314646052e-05,
      "loss": 1.6615,
      "step": 42
    },
    {
      "epoch": 0.029778393351800554,
      "grad_norm": 0.5633302927017212,
      "learning_rate": 1.9973877775519286e-05,
      "loss": 1.6728,
      "step": 43
    },
    {
      "epoch": 0.030470914127423823,
      "grad_norm": 0.7087507843971252,
      "learning_rate": 1.9972271366184922e-05,
      "loss": 1.3833,
      "step": 44
    },
    {
      "epoch": 0.031163434903047092,
      "grad_norm": 0.5197272896766663,
      "learning_rate": 1.9970617094353014e-05,
      "loss": 1.5722,
      "step": 45
    },
    {
      "epoch": 0.03185595567867036,
      "grad_norm": 0.5743720531463623,
      "learning_rate": 1.996891496796334e-05,
      "loss": 1.7185,
      "step": 46
    },
    {
      "epoch": 0.03254847645429363,
      "grad_norm": 0.6309240460395813,
      "learning_rate": 1.9967164995185345e-05,
      "loss": 1.6368,
      "step": 47
    },
    {
      "epoch": 0.0332409972299169,
      "grad_norm": 0.5736321806907654,
      "learning_rate": 1.9965367184418138e-05,
      "loss": 1.5741,
      "step": 48
    },
    {
      "epoch": 0.03393351800554017,
      "grad_norm": 0.6668252348899841,
      "learning_rate": 1.9963521544290402e-05,
      "loss": 1.6657,
      "step": 49
    },
    {
      "epoch": 0.03462603878116344,
      "grad_norm": 0.5815922617912292,
      "learning_rate": 1.9961628083660406e-05,
      "loss": 1.6007,
      "step": 50
    },
    {
      "epoch": 0.035318559556786706,
      "grad_norm": 0.5808799266815186,
      "learning_rate": 1.995968681161592e-05,
      "loss": 1.7115,
      "step": 51
    },
    {
      "epoch": 0.036011080332409975,
      "grad_norm": 0.7402616143226624,
      "learning_rate": 1.9957697737474198e-05,
      "loss": 1.4816,
      "step": 52
    },
    {
      "epoch": 0.036703601108033244,
      "grad_norm": 0.5336787104606628,
      "learning_rate": 1.9955660870781908e-05,
      "loss": 1.5755,
      "step": 53
    },
    {
      "epoch": 0.037396121883656507,
      "grad_norm": 0.6077079176902771,
      "learning_rate": 1.9953576221315116e-05,
      "loss": 1.6875,
      "step": 54
    },
    {
      "epoch": 0.038088642659279776,
      "grad_norm": 0.6559765934944153,
      "learning_rate": 1.9951443799079215e-05,
      "loss": 1.6986,
      "step": 55
    },
    {
      "epoch": 0.038781163434903045,
      "grad_norm": 0.7890306711196899,
      "learning_rate": 1.9949263614308894e-05,
      "loss": 1.2903,
      "step": 56
    },
    {
      "epoch": 0.039473684210526314,
      "grad_norm": 0.7925476431846619,
      "learning_rate": 1.9947035677468078e-05,
      "loss": 1.7279,
      "step": 57
    },
    {
      "epoch": 0.04016620498614958,
      "grad_norm": 0.7267352342605591,
      "learning_rate": 1.994475999924987e-05,
      "loss": 1.6833,
      "step": 58
    },
    {
      "epoch": 0.04085872576177285,
      "grad_norm": 0.657148003578186,
      "learning_rate": 1.9942436590576534e-05,
      "loss": 1.706,
      "step": 59
    },
    {
      "epoch": 0.04155124653739612,
      "grad_norm": 0.6420430541038513,
      "learning_rate": 1.9940065462599394e-05,
      "loss": 1.5163,
      "step": 60
    },
    {
      "epoch": 0.04224376731301939,
      "grad_norm": 0.6800001263618469,
      "learning_rate": 1.9937646626698823e-05,
      "loss": 1.6628,
      "step": 61
    },
    {
      "epoch": 0.04293628808864266,
      "grad_norm": 0.7497722506523132,
      "learning_rate": 1.9935180094484164e-05,
      "loss": 1.5616,
      "step": 62
    },
    {
      "epoch": 0.04362880886426593,
      "grad_norm": 0.7148786187171936,
      "learning_rate": 1.9932665877793678e-05,
      "loss": 1.585,
      "step": 63
    },
    {
      "epoch": 0.0443213296398892,
      "grad_norm": 0.6944183707237244,
      "learning_rate": 1.99301039886945e-05,
      "loss": 1.5801,
      "step": 64
    },
    {
      "epoch": 0.045013850415512466,
      "grad_norm": 0.7924667000770569,
      "learning_rate": 1.9927494439482566e-05,
      "loss": 1.3699,
      "step": 65
    },
    {
      "epoch": 0.045706371191135735,
      "grad_norm": 0.7302570343017578,
      "learning_rate": 1.992483724268255e-05,
      "loss": 1.644,
      "step": 66
    },
    {
      "epoch": 0.046398891966759004,
      "grad_norm": 0.7892882227897644,
      "learning_rate": 1.9922132411047835e-05,
      "loss": 1.5289,
      "step": 67
    },
    {
      "epoch": 0.04709141274238227,
      "grad_norm": 0.6914665699005127,
      "learning_rate": 1.9919379957560413e-05,
      "loss": 1.4982,
      "step": 68
    },
    {
      "epoch": 0.04778393351800554,
      "grad_norm": 0.7855063080787659,
      "learning_rate": 1.9916579895430845e-05,
      "loss": 1.5667,
      "step": 69
    },
    {
      "epoch": 0.04847645429362881,
      "grad_norm": 0.8000104427337646,
      "learning_rate": 1.991373223809819e-05,
      "loss": 1.7444,
      "step": 70
    },
    {
      "epoch": 0.04916897506925208,
      "grad_norm": 0.8270304799079895,
      "learning_rate": 1.991083699922995e-05,
      "loss": 1.5416,
      "step": 71
    },
    {
      "epoch": 0.04986149584487535,
      "grad_norm": 0.6860859394073486,
      "learning_rate": 1.990789419272199e-05,
      "loss": 1.5364,
      "step": 72
    },
    {
      "epoch": 0.05055401662049862,
      "grad_norm": 0.7422068119049072,
      "learning_rate": 1.9904903832698485e-05,
      "loss": 1.6074,
      "step": 73
    },
    {
      "epoch": 0.05124653739612189,
      "grad_norm": 0.9346948862075806,
      "learning_rate": 1.9901865933511834e-05,
      "loss": 1.3475,
      "step": 74
    },
    {
      "epoch": 0.05193905817174515,
      "grad_norm": 0.6729177832603455,
      "learning_rate": 1.989878050974262e-05,
      "loss": 1.6578,
      "step": 75
    },
    {
      "epoch": 0.05263157894736842,
      "grad_norm": 0.7014334797859192,
      "learning_rate": 1.9895647576199507e-05,
      "loss": 1.4947,
      "step": 76
    },
    {
      "epoch": 0.05332409972299169,
      "grad_norm": 0.7257154583930969,
      "learning_rate": 1.9892467147919196e-05,
      "loss": 1.615,
      "step": 77
    },
    {
      "epoch": 0.054016620498614956,
      "grad_norm": 0.7274760603904724,
      "learning_rate": 1.988923924016634e-05,
      "loss": 1.5596,
      "step": 78
    },
    {
      "epoch": 0.054709141274238225,
      "grad_norm": 0.8070396184921265,
      "learning_rate": 1.9885963868433463e-05,
      "loss": 1.4991,
      "step": 79
    },
    {
      "epoch": 0.055401662049861494,
      "grad_norm": 0.774266242980957,
      "learning_rate": 1.988264104844091e-05,
      "loss": 1.6116,
      "step": 80
    },
    {
      "epoch": 0.05609418282548476,
      "grad_norm": 0.5994683504104614,
      "learning_rate": 1.9879270796136746e-05,
      "loss": 0.9307,
      "step": 81
    },
    {
      "epoch": 0.05678670360110803,
      "grad_norm": 0.8213633298873901,
      "learning_rate": 1.987585312769669e-05,
      "loss": 1.541,
      "step": 82
    },
    {
      "epoch": 0.0574792243767313,
      "grad_norm": 0.7912551164627075,
      "learning_rate": 1.987238805952405e-05,
      "loss": 1.6115,
      "step": 83
    },
    {
      "epoch": 0.05817174515235457,
      "grad_norm": 0.8942985534667969,
      "learning_rate": 1.9868875608249613e-05,
      "loss": 1.4283,
      "step": 84
    },
    {
      "epoch": 0.05886426592797784,
      "grad_norm": 0.818084716796875,
      "learning_rate": 1.98653157907316e-05,
      "loss": 1.4575,
      "step": 85
    },
    {
      "epoch": 0.05955678670360111,
      "grad_norm": 0.756619393825531,
      "learning_rate": 1.986170862405556e-05,
      "loss": 1.5106,
      "step": 86
    },
    {
      "epoch": 0.06024930747922438,
      "grad_norm": 0.7861054539680481,
      "learning_rate": 1.9858054125534297e-05,
      "loss": 1.6412,
      "step": 87
    },
    {
      "epoch": 0.060941828254847646,
      "grad_norm": 0.767902672290802,
      "learning_rate": 1.98543523127078e-05,
      "loss": 1.6698,
      "step": 88
    },
    {
      "epoch": 0.061634349030470915,
      "grad_norm": 0.8469496369361877,
      "learning_rate": 1.985060320334312e-05,
      "loss": 1.6582,
      "step": 89
    },
    {
      "epoch": 0.062326869806094184,
      "grad_norm": 0.7616478800773621,
      "learning_rate": 1.984680681543434e-05,
      "loss": 1.5724,
      "step": 90
    },
    {
      "epoch": 0.06301939058171745,
      "grad_norm": 0.5673407912254333,
      "learning_rate": 1.9842963167202435e-05,
      "loss": 0.9085,
      "step": 91
    },
    {
      "epoch": 0.06371191135734072,
      "grad_norm": 0.9865372776985168,
      "learning_rate": 1.9839072277095222e-05,
      "loss": 1.551,
      "step": 92
    },
    {
      "epoch": 0.06440443213296398,
      "grad_norm": 0.8187092542648315,
      "learning_rate": 1.9835134163787257e-05,
      "loss": 1.5875,
      "step": 93
    },
    {
      "epoch": 0.06509695290858726,
      "grad_norm": 0.8418170809745789,
      "learning_rate": 1.9831148846179743e-05,
      "loss": 1.4789,
      "step": 94
    },
    {
      "epoch": 0.06578947368421052,
      "grad_norm": 0.8878125548362732,
      "learning_rate": 1.982711634340044e-05,
      "loss": 1.4319,
      "step": 95
    },
    {
      "epoch": 0.0664819944598338,
      "grad_norm": 0.9223076701164246,
      "learning_rate": 1.9823036674803585e-05,
      "loss": 1.3074,
      "step": 96
    },
    {
      "epoch": 0.06717451523545706,
      "grad_norm": 0.7803325057029724,
      "learning_rate": 1.981890985996979e-05,
      "loss": 1.6477,
      "step": 97
    },
    {
      "epoch": 0.06786703601108034,
      "grad_norm": 1.0602279901504517,
      "learning_rate": 1.981473591870593e-05,
      "loss": 1.3802,
      "step": 98
    },
    {
      "epoch": 0.0685595567867036,
      "grad_norm": 0.859002947807312,
      "learning_rate": 1.981051487104509e-05,
      "loss": 1.5765,
      "step": 99
    },
    {
      "epoch": 0.06925207756232687,
      "grad_norm": 0.9333771467208862,
      "learning_rate": 1.980624673724643e-05,
      "loss": 1.6703,
      "step": 100
    },
    {
      "epoch": 0.06994459833795014,
      "grad_norm": 1.0008572340011597,
      "learning_rate": 1.980193153779511e-05,
      "loss": 1.5619,
      "step": 101
    },
    {
      "epoch": 0.07063711911357341,
      "grad_norm": 0.9978685975074768,
      "learning_rate": 1.9797569293402174e-05,
      "loss": 1.6118,
      "step": 102
    },
    {
      "epoch": 0.07132963988919667,
      "grad_norm": 0.8422709107398987,
      "learning_rate": 1.9793160025004476e-05,
      "loss": 1.6786,
      "step": 103
    },
    {
      "epoch": 0.07202216066481995,
      "grad_norm": 0.8284188508987427,
      "learning_rate": 1.9788703753764554e-05,
      "loss": 1.4792,
      "step": 104
    },
    {
      "epoch": 0.07271468144044321,
      "grad_norm": 0.8767748475074768,
      "learning_rate": 1.9784200501070542e-05,
      "loss": 1.5439,
      "step": 105
    },
    {
      "epoch": 0.07340720221606649,
      "grad_norm": 0.9303686618804932,
      "learning_rate": 1.9779650288536057e-05,
      "loss": 1.5874,
      "step": 106
    },
    {
      "epoch": 0.07409972299168975,
      "grad_norm": 0.9373297095298767,
      "learning_rate": 1.9775053138000123e-05,
      "loss": 1.6427,
      "step": 107
    },
    {
      "epoch": 0.07479224376731301,
      "grad_norm": 0.9026633501052856,
      "learning_rate": 1.977040907152702e-05,
      "loss": 1.4498,
      "step": 108
    },
    {
      "epoch": 0.07548476454293629,
      "grad_norm": 0.9151142239570618,
      "learning_rate": 1.9765718111406217e-05,
      "loss": 1.3523,
      "step": 109
    },
    {
      "epoch": 0.07617728531855955,
      "grad_norm": 1.065004587173462,
      "learning_rate": 1.976098028015226e-05,
      "loss": 1.419,
      "step": 110
    },
    {
      "epoch": 0.07686980609418283,
      "grad_norm": 0.870905339717865,
      "learning_rate": 1.9756195600504637e-05,
      "loss": 1.589,
      "step": 111
    },
    {
      "epoch": 0.07756232686980609,
      "grad_norm": 0.9227341413497925,
      "learning_rate": 1.9751364095427694e-05,
      "loss": 1.4389,
      "step": 112
    },
    {
      "epoch": 0.07825484764542937,
      "grad_norm": 0.9902766942977905,
      "learning_rate": 1.974648578811053e-05,
      "loss": 1.54,
      "step": 113
    },
    {
      "epoch": 0.07894736842105263,
      "grad_norm": 1.0076311826705933,
      "learning_rate": 1.974156070196686e-05,
      "loss": 1.3828,
      "step": 114
    },
    {
      "epoch": 0.0796398891966759,
      "grad_norm": 0.8937874436378479,
      "learning_rate": 1.9736588860634928e-05,
      "loss": 1.5319,
      "step": 115
    },
    {
      "epoch": 0.08033240997229917,
      "grad_norm": 1.016831398010254,
      "learning_rate": 1.973157028797737e-05,
      "loss": 1.6604,
      "step": 116
    },
    {
      "epoch": 0.08102493074792244,
      "grad_norm": 0.8994125127792358,
      "learning_rate": 1.9726505008081113e-05,
      "loss": 1.532,
      "step": 117
    },
    {
      "epoch": 0.0817174515235457,
      "grad_norm": 0.8747804760932922,
      "learning_rate": 1.9721393045257277e-05,
      "loss": 1.5652,
      "step": 118
    },
    {
      "epoch": 0.08240997229916898,
      "grad_norm": 0.8889365196228027,
      "learning_rate": 1.9716234424041016e-05,
      "loss": 1.6021,
      "step": 119
    },
    {
      "epoch": 0.08310249307479224,
      "grad_norm": 0.9008545875549316,
      "learning_rate": 1.9711029169191437e-05,
      "loss": 1.4765,
      "step": 120
    },
    {
      "epoch": 0.08379501385041552,
      "grad_norm": 0.9812483191490173,
      "learning_rate": 1.9705777305691457e-05,
      "loss": 1.4359,
      "step": 121
    },
    {
      "epoch": 0.08448753462603878,
      "grad_norm": 0.9572108387947083,
      "learning_rate": 1.970047885874771e-05,
      "loss": 1.4399,
      "step": 122
    },
    {
      "epoch": 0.08518005540166206,
      "grad_norm": 1.038678526878357,
      "learning_rate": 1.9695133853790397e-05,
      "loss": 1.5133,
      "step": 123
    },
    {
      "epoch": 0.08587257617728532,
      "grad_norm": 0.898454487323761,
      "learning_rate": 1.968974231647318e-05,
      "loss": 1.4694,
      "step": 124
    },
    {
      "epoch": 0.0865650969529086,
      "grad_norm": 0.8905704021453857,
      "learning_rate": 1.9684304272673057e-05,
      "loss": 1.5121,
      "step": 125
    },
    {
      "epoch": 0.08725761772853186,
      "grad_norm": 0.9543448090553284,
      "learning_rate": 1.9678819748490236e-05,
      "loss": 1.4132,
      "step": 126
    },
    {
      "epoch": 0.08795013850415513,
      "grad_norm": 1.0110042095184326,
      "learning_rate": 1.9673288770248012e-05,
      "loss": 1.4191,
      "step": 127
    },
    {
      "epoch": 0.0886426592797784,
      "grad_norm": 0.9093568921089172,
      "learning_rate": 1.9667711364492638e-05,
      "loss": 1.4947,
      "step": 128
    },
    {
      "epoch": 0.08933518005540166,
      "grad_norm": 1.012677788734436,
      "learning_rate": 1.9662087557993193e-05,
      "loss": 1.4674,
      "step": 129
    },
    {
      "epoch": 0.09002770083102493,
      "grad_norm": 0.9102503657341003,
      "learning_rate": 1.965641737774147e-05,
      "loss": 1.4392,
      "step": 130
    },
    {
      "epoch": 0.0907202216066482,
      "grad_norm": 0.9984774589538574,
      "learning_rate": 1.9650700850951825e-05,
      "loss": 1.6278,
      "step": 131
    },
    {
      "epoch": 0.09141274238227147,
      "grad_norm": 0.9726704955101013,
      "learning_rate": 1.9644938005061062e-05,
      "loss": 1.548,
      "step": 132
    },
    {
      "epoch": 0.09210526315789473,
      "grad_norm": 1.1588407754898071,
      "learning_rate": 1.96391288677283e-05,
      "loss": 1.2959,
      "step": 133
    },
    {
      "epoch": 0.09279778393351801,
      "grad_norm": 0.9556019306182861,
      "learning_rate": 1.9633273466834826e-05,
      "loss": 1.4758,
      "step": 134
    },
    {
      "epoch": 0.09349030470914127,
      "grad_norm": 1.0594704151153564,
      "learning_rate": 1.9627371830483982e-05,
      "loss": 1.0755,
      "step": 135
    },
    {
      "epoch": 0.09418282548476455,
      "grad_norm": 1.4173860549926758,
      "learning_rate": 1.9621423987001013e-05,
      "loss": 1.4406,
      "step": 136
    },
    {
      "epoch": 0.09487534626038781,
      "grad_norm": 0.9821715950965881,
      "learning_rate": 1.9615429964932945e-05,
      "loss": 1.4767,
      "step": 137
    },
    {
      "epoch": 0.09556786703601108,
      "grad_norm": 1.2425581216812134,
      "learning_rate": 1.960938979304843e-05,
      "loss": 1.3854,
      "step": 138
    },
    {
      "epoch": 0.09626038781163435,
      "grad_norm": 1.0587480068206787,
      "learning_rate": 1.960330350033763e-05,
      "loss": 1.6481,
      "step": 139
    },
    {
      "epoch": 0.09695290858725762,
      "grad_norm": 0.9959594011306763,
      "learning_rate": 1.959717111601206e-05,
      "loss": 1.6265,
      "step": 140
    },
    {
      "epoch": 0.09764542936288088,
      "grad_norm": 0.9131922125816345,
      "learning_rate": 1.959099266950445e-05,
      "loss": 1.5011,
      "step": 141
    },
    {
      "epoch": 0.09833795013850416,
      "grad_norm": 0.9651609659194946,
      "learning_rate": 1.9584768190468624e-05,
      "loss": 1.7318,
      "step": 142
    },
    {
      "epoch": 0.09903047091412742,
      "grad_norm": 1.0025120973587036,
      "learning_rate": 1.9578497708779326e-05,
      "loss": 1.4781,
      "step": 143
    },
    {
      "epoch": 0.0997229916897507,
      "grad_norm": 0.9535677433013916,
      "learning_rate": 1.95721812545321e-05,
      "loss": 1.4097,
      "step": 144
    },
    {
      "epoch": 0.10041551246537396,
      "grad_norm": 0.9194809198379517,
      "learning_rate": 1.9565818858043136e-05,
      "loss": 1.6294,
      "step": 145
    },
    {
      "epoch": 0.10110803324099724,
      "grad_norm": 0.9537028074264526,
      "learning_rate": 1.9559410549849125e-05,
      "loss": 1.6033,
      "step": 146
    },
    {
      "epoch": 0.1018005540166205,
      "grad_norm": 0.8676417469978333,
      "learning_rate": 1.955295636070712e-05,
      "loss": 1.4449,
      "step": 147
    },
    {
      "epoch": 0.10249307479224377,
      "grad_norm": 0.9458538889884949,
      "learning_rate": 1.9546456321594374e-05,
      "loss": 1.5544,
      "step": 148
    },
    {
      "epoch": 0.10318559556786704,
      "grad_norm": 1.1934260129928589,
      "learning_rate": 1.9539910463708204e-05,
      "loss": 1.403,
      "step": 149
    },
    {
      "epoch": 0.1038781163434903,
      "grad_norm": 1.1710929870605469,
      "learning_rate": 1.9533318818465837e-05,
      "loss": 1.2662,
      "step": 150
    },
    {
      "epoch": 0.10457063711911357,
      "grad_norm": 1.031394362449646,
      "learning_rate": 1.952668141750426e-05,
      "loss": 1.4116,
      "step": 151
    },
    {
      "epoch": 0.10526315789473684,
      "grad_norm": 0.9981980919837952,
      "learning_rate": 1.9519998292680062e-05,
      "loss": 1.6333,
      "step": 152
    },
    {
      "epoch": 0.10595567867036011,
      "grad_norm": 1.0679597854614258,
      "learning_rate": 1.951326947606929e-05,
      "loss": 1.4244,
      "step": 153
    },
    {
      "epoch": 0.10664819944598337,
      "grad_norm": 1.1829801797866821,
      "learning_rate": 1.9506494999967298e-05,
      "loss": 1.4967,
      "step": 154
    },
    {
      "epoch": 0.10734072022160665,
      "grad_norm": 1.143169641494751,
      "learning_rate": 1.9499674896888573e-05,
      "loss": 1.3145,
      "step": 155
    },
    {
      "epoch": 0.10803324099722991,
      "grad_norm": 1.1118048429489136,
      "learning_rate": 1.94928091995666e-05,
      "loss": 1.5536,
      "step": 156
    },
    {
      "epoch": 0.10872576177285319,
      "grad_norm": 1.1905399560928345,
      "learning_rate": 1.948589794095369e-05,
      "loss": 1.7543,
      "step": 157
    },
    {
      "epoch": 0.10941828254847645,
      "grad_norm": 1.1353670358657837,
      "learning_rate": 1.9478941154220833e-05,
      "loss": 1.6226,
      "step": 158
    },
    {
      "epoch": 0.11011080332409973,
      "grad_norm": 1.1765538454055786,
      "learning_rate": 1.947193887275753e-05,
      "loss": 1.5584,
      "step": 159
    },
    {
      "epoch": 0.11080332409972299,
      "grad_norm": 1.1196644306182861,
      "learning_rate": 1.9464891130171647e-05,
      "loss": 1.6066,
      "step": 160
    },
    {
      "epoch": 0.11149584487534626,
      "grad_norm": 1.0848753452301025,
      "learning_rate": 1.945779796028923e-05,
      "loss": 1.6394,
      "step": 161
    },
    {
      "epoch": 0.11218836565096953,
      "grad_norm": 1.0297619104385376,
      "learning_rate": 1.9450659397154353e-05,
      "loss": 1.487,
      "step": 162
    },
    {
      "epoch": 0.1128808864265928,
      "grad_norm": 1.0936334133148193,
      "learning_rate": 1.9443475475028985e-05,
      "loss": 1.4577,
      "step": 163
    },
    {
      "epoch": 0.11357340720221606,
      "grad_norm": 1.1814393997192383,
      "learning_rate": 1.9436246228392762e-05,
      "loss": 1.3241,
      "step": 164
    },
    {
      "epoch": 0.11426592797783934,
      "grad_norm": 0.9540955424308777,
      "learning_rate": 1.942897169194288e-05,
      "loss": 1.555,
      "step": 165
    },
    {
      "epoch": 0.1149584487534626,
      "grad_norm": 1.0909806489944458,
      "learning_rate": 1.94216519005939e-05,
      "loss": 1.5422,
      "step": 166
    },
    {
      "epoch": 0.11565096952908588,
      "grad_norm": 1.1541738510131836,
      "learning_rate": 1.941428688947759e-05,
      "loss": 1.541,
      "step": 167
    },
    {
      "epoch": 0.11634349030470914,
      "grad_norm": 1.2583942413330078,
      "learning_rate": 1.9406876693942747e-05,
      "loss": 1.3919,
      "step": 168
    },
    {
      "epoch": 0.11703601108033242,
      "grad_norm": 1.1814335584640503,
      "learning_rate": 1.9399421349555037e-05,
      "loss": 1.4799,
      "step": 169
    },
    {
      "epoch": 0.11772853185595568,
      "grad_norm": 1.0238617658615112,
      "learning_rate": 1.939192089209682e-05,
      "loss": 1.462,
      "step": 170
    },
    {
      "epoch": 0.11842105263157894,
      "grad_norm": 0.9870756268501282,
      "learning_rate": 1.938437535756698e-05,
      "loss": 1.5175,
      "step": 171
    },
    {
      "epoch": 0.11911357340720222,
      "grad_norm": 0.9913883209228516,
      "learning_rate": 1.9376784782180747e-05,
      "loss": 1.547,
      "step": 172
    },
    {
      "epoch": 0.11980609418282548,
      "grad_norm": 1.0706313848495483,
      "learning_rate": 1.9369149202369532e-05,
      "loss": 1.5063,
      "step": 173
    },
    {
      "epoch": 0.12049861495844875,
      "grad_norm": 1.0937633514404297,
      "learning_rate": 1.9361468654780748e-05,
      "loss": 1.4345,
      "step": 174
    },
    {
      "epoch": 0.12119113573407202,
      "grad_norm": 1.341931700706482,
      "learning_rate": 1.9353743176277624e-05,
      "loss": 1.5701,
      "step": 175
    },
    {
      "epoch": 0.12188365650969529,
      "grad_norm": 1.1175850629806519,
      "learning_rate": 1.9345972803939046e-05,
      "loss": 1.4767,
      "step": 176
    },
    {
      "epoch": 0.12257617728531855,
      "grad_norm": 1.1408743858337402,
      "learning_rate": 1.933815757505937e-05,
      "loss": 1.4476,
      "step": 177
    },
    {
      "epoch": 0.12326869806094183,
      "grad_norm": 1.1984326839447021,
      "learning_rate": 1.9330297527148246e-05,
      "loss": 1.4915,
      "step": 178
    },
    {
      "epoch": 0.12396121883656509,
      "grad_norm": 1.1456125974655151,
      "learning_rate": 1.932239269793043e-05,
      "loss": 1.5642,
      "step": 179
    },
    {
      "epoch": 0.12465373961218837,
      "grad_norm": 1.0504119396209717,
      "learning_rate": 1.9314443125345606e-05,
      "loss": 1.494,
      "step": 180
    },
    {
      "epoch": 0.12534626038781163,
      "grad_norm": 1.188533067703247,
      "learning_rate": 1.9306448847548214e-05,
      "loss": 1.5529,
      "step": 181
    },
    {
      "epoch": 0.1260387811634349,
      "grad_norm": 1.1496820449829102,
      "learning_rate": 1.929840990290726e-05,
      "loss": 1.4418,
      "step": 182
    },
    {
      "epoch": 0.12673130193905818,
      "grad_norm": 1.0949493646621704,
      "learning_rate": 1.9290326330006125e-05,
      "loss": 1.58,
      "step": 183
    },
    {
      "epoch": 0.12742382271468145,
      "grad_norm": 1.0846694707870483,
      "learning_rate": 1.928219816764238e-05,
      "loss": 1.6893,
      "step": 184
    },
    {
      "epoch": 0.1281163434903047,
      "grad_norm": 1.086604118347168,
      "learning_rate": 1.9274025454827626e-05,
      "loss": 1.4686,
      "step": 185
    },
    {
      "epoch": 0.12880886426592797,
      "grad_norm": 1.2065285444259644,
      "learning_rate": 1.9265808230787265e-05,
      "loss": 1.4176,
      "step": 186
    },
    {
      "epoch": 0.12950138504155126,
      "grad_norm": 1.1812058687210083,
      "learning_rate": 1.9257546534960354e-05,
      "loss": 1.4403,
      "step": 187
    },
    {
      "epoch": 0.13019390581717452,
      "grad_norm": 1.1747177839279175,
      "learning_rate": 1.9249240406999366e-05,
      "loss": 1.551,
      "step": 188
    },
    {
      "epoch": 0.13088642659279778,
      "grad_norm": 1.2344166040420532,
      "learning_rate": 1.9240889886770065e-05,
      "loss": 1.496,
      "step": 189
    },
    {
      "epoch": 0.13157894736842105,
      "grad_norm": 1.1985383033752441,
      "learning_rate": 1.9232495014351248e-05,
      "loss": 1.534,
      "step": 190
    },
    {
      "epoch": 0.13227146814404434,
      "grad_norm": 1.2938690185546875,
      "learning_rate": 1.9224055830034597e-05,
      "loss": 1.3664,
      "step": 191
    },
    {
      "epoch": 0.1329639889196676,
      "grad_norm": 1.1214767694473267,
      "learning_rate": 1.921557237432447e-05,
      "loss": 1.4684,
      "step": 192
    },
    {
      "epoch": 0.13365650969529086,
      "grad_norm": 1.1654257774353027,
      "learning_rate": 1.9207044687937705e-05,
      "loss": 1.5447,
      "step": 193
    },
    {
      "epoch": 0.13434903047091412,
      "grad_norm": 1.1630734205245972,
      "learning_rate": 1.919847281180343e-05,
      "loss": 1.5108,
      "step": 194
    },
    {
      "epoch": 0.13504155124653738,
      "grad_norm": 1.125630497932434,
      "learning_rate": 1.918985678706287e-05,
      "loss": 1.5938,
      "step": 195
    },
    {
      "epoch": 0.13573407202216067,
      "grad_norm": 1.1549885272979736,
      "learning_rate": 1.9181196655069126e-05,
      "loss": 1.4511,
      "step": 196
    },
    {
      "epoch": 0.13642659279778394,
      "grad_norm": 1.4739876985549927,
      "learning_rate": 1.917249245738702e-05,
      "loss": 1.3833,
      "step": 197
    },
    {
      "epoch": 0.1371191135734072,
      "grad_norm": 1.3530839681625366,
      "learning_rate": 1.9163744235792845e-05,
      "loss": 1.4574,
      "step": 198
    },
    {
      "epoch": 0.13781163434903046,
      "grad_norm": 1.2195689678192139,
      "learning_rate": 1.915495203227421e-05,
      "loss": 1.4105,
      "step": 199
    },
    {
      "epoch": 0.13850415512465375,
      "grad_norm": 1.347271203994751,
      "learning_rate": 1.9146115889029793e-05,
      "loss": 1.6371,
      "step": 200
    },
    {
      "epoch": 0.139196675900277,
      "grad_norm": 1.3149245977401733,
      "learning_rate": 1.9137235848469196e-05,
      "loss": 1.3975,
      "step": 201
    },
    {
      "epoch": 0.13988919667590027,
      "grad_norm": 1.113451361656189,
      "learning_rate": 1.912831195321268e-05,
      "loss": 1.5483,
      "step": 202
    },
    {
      "epoch": 0.14058171745152354,
      "grad_norm": 1.2491871118545532,
      "learning_rate": 1.9119344246091e-05,
      "loss": 1.59,
      "step": 203
    },
    {
      "epoch": 0.14127423822714683,
      "grad_norm": 1.165940284729004,
      "learning_rate": 1.9110332770145198e-05,
      "loss": 1.4319,
      "step": 204
    },
    {
      "epoch": 0.1419667590027701,
      "grad_norm": 1.2933601140975952,
      "learning_rate": 1.9101277568626374e-05,
      "loss": 1.4285,
      "step": 205
    },
    {
      "epoch": 0.14265927977839335,
      "grad_norm": 1.3592076301574707,
      "learning_rate": 1.9092178684995487e-05,
      "loss": 1.435,
      "step": 206
    },
    {
      "epoch": 0.1433518005540166,
      "grad_norm": 1.3157401084899902,
      "learning_rate": 1.908303616292317e-05,
      "loss": 1.4506,
      "step": 207
    },
    {
      "epoch": 0.1440443213296399,
      "grad_norm": 1.179969072341919,
      "learning_rate": 1.9073850046289484e-05,
      "loss": 1.5505,
      "step": 208
    },
    {
      "epoch": 0.14473684210526316,
      "grad_norm": 1.3946419954299927,
      "learning_rate": 1.9064620379183733e-05,
      "loss": 1.4977,
      "step": 209
    },
    {
      "epoch": 0.14542936288088643,
      "grad_norm": 1.2105120420455933,
      "learning_rate": 1.9055347205904245e-05,
      "loss": 1.422,
      "step": 210
    },
    {
      "epoch": 0.1461218836565097,
      "grad_norm": 1.3983213901519775,
      "learning_rate": 1.9046030570958153e-05,
      "loss": 1.4853,
      "step": 211
    },
    {
      "epoch": 0.14681440443213298,
      "grad_norm": 1.2327667474746704,
      "learning_rate": 1.903667051906119e-05,
      "loss": 1.5621,
      "step": 212
    },
    {
      "epoch": 0.14750692520775624,
      "grad_norm": 1.0103579759597778,
      "learning_rate": 1.9027267095137466e-05,
      "loss": 0.7994,
      "step": 213
    },
    {
      "epoch": 0.1481994459833795,
      "grad_norm": 1.1519508361816406,
      "learning_rate": 1.901782034431927e-05,
      "loss": 1.4775,
      "step": 214
    },
    {
      "epoch": 0.14889196675900276,
      "grad_norm": 1.2052768468856812,
      "learning_rate": 1.9008330311946826e-05,
      "loss": 1.3932,
      "step": 215
    },
    {
      "epoch": 0.14958448753462603,
      "grad_norm": 1.217500925064087,
      "learning_rate": 1.8998797043568102e-05,
      "loss": 1.5393,
      "step": 216
    },
    {
      "epoch": 0.15027700831024932,
      "grad_norm": 1.2066594362258911,
      "learning_rate": 1.8989220584938574e-05,
      "loss": 1.5545,
      "step": 217
    },
    {
      "epoch": 0.15096952908587258,
      "grad_norm": 1.2560105323791504,
      "learning_rate": 1.8979600982021014e-05,
      "loss": 1.4585,
      "step": 218
    },
    {
      "epoch": 0.15166204986149584,
      "grad_norm": 1.2688912153244019,
      "learning_rate": 1.8969938280985264e-05,
      "loss": 1.3712,
      "step": 219
    },
    {
      "epoch": 0.1523545706371191,
      "grad_norm": 1.2653838396072388,
      "learning_rate": 1.896023252820802e-05,
      "loss": 1.5732,
      "step": 220
    },
    {
      "epoch": 0.1530470914127424,
      "grad_norm": 1.3516199588775635,
      "learning_rate": 1.8950483770272617e-05,
      "loss": 1.3206,
      "step": 221
    },
    {
      "epoch": 0.15373961218836565,
      "grad_norm": 1.3281147480010986,
      "learning_rate": 1.8940692053968773e-05,
      "loss": 1.3235,
      "step": 222
    },
    {
      "epoch": 0.15443213296398892,
      "grad_norm": 1.3650656938552856,
      "learning_rate": 1.893085742629241e-05,
      "loss": 1.3977,
      "step": 223
    },
    {
      "epoch": 0.15512465373961218,
      "grad_norm": 1.270662784576416,
      "learning_rate": 1.89209799344454e-05,
      "loss": 1.5181,
      "step": 224
    },
    {
      "epoch": 0.15581717451523547,
      "grad_norm": 1.3656113147735596,
      "learning_rate": 1.891105962583533e-05,
      "loss": 1.3082,
      "step": 225
    },
    {
      "epoch": 0.15650969529085873,
      "grad_norm": 1.2502739429473877,
      "learning_rate": 1.8901096548075305e-05,
      "loss": 1.5326,
      "step": 226
    },
    {
      "epoch": 0.157202216066482,
      "grad_norm": 1.3605083227157593,
      "learning_rate": 1.88910907489837e-05,
      "loss": 1.3527,
      "step": 227
    },
    {
      "epoch": 0.15789473684210525,
      "grad_norm": 1.2121204137802124,
      "learning_rate": 1.8881042276583924e-05,
      "loss": 1.3969,
      "step": 228
    },
    {
      "epoch": 0.15858725761772854,
      "grad_norm": 1.2529863119125366,
      "learning_rate": 1.8870951179104214e-05,
      "loss": 1.4062,
      "step": 229
    },
    {
      "epoch": 0.1592797783933518,
      "grad_norm": 1.1819360256195068,
      "learning_rate": 1.8860817504977374e-05,
      "loss": 1.3974,
      "step": 230
    },
    {
      "epoch": 0.15997229916897507,
      "grad_norm": 1.234582543373108,
      "learning_rate": 1.8850641302840565e-05,
      "loss": 1.4027,
      "step": 231
    },
    {
      "epoch": 0.16066481994459833,
      "grad_norm": 1.3239039182662964,
      "learning_rate": 1.8840422621535067e-05,
      "loss": 1.3196,
      "step": 232
    },
    {
      "epoch": 0.16135734072022162,
      "grad_norm": 1.1581454277038574,
      "learning_rate": 1.8830161510106027e-05,
      "loss": 1.4674,
      "step": 233
    },
    {
      "epoch": 0.16204986149584488,
      "grad_norm": 1.3907638788223267,
      "learning_rate": 1.881985801780225e-05,
      "loss": 1.391,
      "step": 234
    },
    {
      "epoch": 0.16274238227146814,
      "grad_norm": 1.2587774991989136,
      "learning_rate": 1.880951219407596e-05,
      "loss": 1.4388,
      "step": 235
    },
    {
      "epoch": 0.1634349030470914,
      "grad_norm": 1.3093003034591675,
      "learning_rate": 1.8799124088582523e-05,
      "loss": 1.6846,
      "step": 236
    },
    {
      "epoch": 0.16412742382271467,
      "grad_norm": 1.4729034900665283,
      "learning_rate": 1.878869375118027e-05,
      "loss": 1.2827,
      "step": 237
    },
    {
      "epoch": 0.16481994459833796,
      "grad_norm": 1.171746015548706,
      "learning_rate": 1.8778221231930204e-05,
      "loss": 1.517,
      "step": 238
    },
    {
      "epoch": 0.16551246537396122,
      "grad_norm": 1.384203314781189,
      "learning_rate": 1.8767706581095796e-05,
      "loss": 1.5171,
      "step": 239
    },
    {
      "epoch": 0.16620498614958448,
      "grad_norm": 1.3806113004684448,
      "learning_rate": 1.8757149849142724e-05,
      "loss": 1.4459,
      "step": 240
    },
    {
      "epoch": 0.16689750692520774,
      "grad_norm": 1.2819597721099854,
      "learning_rate": 1.874655108673864e-05,
      "loss": 1.5718,
      "step": 241
    },
    {
      "epoch": 0.16759002770083103,
      "grad_norm": 1.3466342687606812,
      "learning_rate": 1.8735910344752925e-05,
      "loss": 1.4098,
      "step": 242
    },
    {
      "epoch": 0.1682825484764543,
      "grad_norm": 1.301113247871399,
      "learning_rate": 1.872522767425643e-05,
      "loss": 1.1911,
      "step": 243
    },
    {
      "epoch": 0.16897506925207756,
      "grad_norm": 1.115594506263733,
      "learning_rate": 1.871450312652126e-05,
      "loss": 1.4129,
      "step": 244
    },
    {
      "epoch": 0.16966759002770082,
      "grad_norm": 1.236174464225769,
      "learning_rate": 1.870373675302051e-05,
      "loss": 1.3763,
      "step": 245
    },
    {
      "epoch": 0.1703601108033241,
      "grad_norm": 1.6457040309906006,
      "learning_rate": 1.8692928605428016e-05,
      "loss": 1.3063,
      "step": 246
    },
    {
      "epoch": 0.17105263157894737,
      "grad_norm": 1.3574951887130737,
      "learning_rate": 1.868207873561811e-05,
      "loss": 1.3223,
      "step": 247
    },
    {
      "epoch": 0.17174515235457063,
      "grad_norm": 1.3911534547805786,
      "learning_rate": 1.8671187195665373e-05,
      "loss": 1.3415,
      "step": 248
    },
    {
      "epoch": 0.1724376731301939,
      "grad_norm": 1.2706600427627563,
      "learning_rate": 1.866025403784439e-05,
      "loss": 1.4396,
      "step": 249
    },
    {
      "epoch": 0.1731301939058172,
      "grad_norm": 1.2637407779693604,
      "learning_rate": 1.8649279314629484e-05,
      "loss": 1.4115,
      "step": 250
    },
    {
      "epoch": 0.17382271468144045,
      "grad_norm": 1.3504443168640137,
      "learning_rate": 1.8638263078694483e-05,
      "loss": 1.333,
      "step": 251
    },
    {
      "epoch": 0.1745152354570637,
      "grad_norm": 1.1280931234359741,
      "learning_rate": 1.862720538291245e-05,
      "loss": 1.454,
      "step": 252
    },
    {
      "epoch": 0.17520775623268697,
      "grad_norm": 1.309478521347046,
      "learning_rate": 1.8616106280355443e-05,
      "loss": 1.455,
      "step": 253
    },
    {
      "epoch": 0.17590027700831026,
      "grad_norm": 1.4461839199066162,
      "learning_rate": 1.8604965824294253e-05,
      "loss": 1.1718,
      "step": 254
    },
    {
      "epoch": 0.17659279778393353,
      "grad_norm": 1.3986719846725464,
      "learning_rate": 1.859378406819814e-05,
      "loss": 1.4925,
      "step": 255
    },
    {
      "epoch": 0.1772853185595568,
      "grad_norm": 1.329654574394226,
      "learning_rate": 1.8582561065734602e-05,
      "loss": 1.4181,
      "step": 256
    },
    {
      "epoch": 0.17797783933518005,
      "grad_norm": 1.472766637802124,
      "learning_rate": 1.8571296870769093e-05,
      "loss": 1.2677,
      "step": 257
    },
    {
      "epoch": 0.1786703601108033,
      "grad_norm": 1.3047882318496704,
      "learning_rate": 1.8559991537364767e-05,
      "loss": 1.4402,
      "step": 258
    },
    {
      "epoch": 0.1793628808864266,
      "grad_norm": 1.4063674211502075,
      "learning_rate": 1.854864511978224e-05,
      "loss": 1.2174,
      "step": 259
    },
    {
      "epoch": 0.18005540166204986,
      "grad_norm": 1.3601303100585938,
      "learning_rate": 1.8537257672479293e-05,
      "loss": 1.5546,
      "step": 260
    },
    {
      "epoch": 0.18074792243767313,
      "grad_norm": 1.6353052854537964,
      "learning_rate": 1.8525829250110653e-05,
      "loss": 1.2166,
      "step": 261
    },
    {
      "epoch": 0.1814404432132964,
      "grad_norm": 1.4279634952545166,
      "learning_rate": 1.8514359907527693e-05,
      "loss": 1.4062,
      "step": 262
    },
    {
      "epoch": 0.18213296398891968,
      "grad_norm": 1.654395580291748,
      "learning_rate": 1.8502849699778193e-05,
      "loss": 1.2227,
      "step": 263
    },
    {
      "epoch": 0.18282548476454294,
      "grad_norm": 1.4442977905273438,
      "learning_rate": 1.8491298682106066e-05,
      "loss": 1.533,
      "step": 264
    },
    {
      "epoch": 0.1835180055401662,
      "grad_norm": 1.3365898132324219,
      "learning_rate": 1.8479706909951095e-05,
      "loss": 1.4334,
      "step": 265
    },
    {
      "epoch": 0.18421052631578946,
      "grad_norm": 1.3646080493927002,
      "learning_rate": 1.8468074438948664e-05,
      "loss": 1.2597,
      "step": 266
    },
    {
      "epoch": 0.18490304709141275,
      "grad_norm": 1.2578566074371338,
      "learning_rate": 1.84564013249295e-05,
      "loss": 1.5369,
      "step": 267
    },
    {
      "epoch": 0.18559556786703602,
      "grad_norm": 1.4007184505462646,
      "learning_rate": 1.8444687623919388e-05,
      "loss": 1.3263,
      "step": 268
    },
    {
      "epoch": 0.18628808864265928,
      "grad_norm": 1.362870216369629,
      "learning_rate": 1.8432933392138922e-05,
      "loss": 1.5152,
      "step": 269
    },
    {
      "epoch": 0.18698060941828254,
      "grad_norm": 1.277584433555603,
      "learning_rate": 1.842113868600322e-05,
      "loss": 1.5021,
      "step": 270
    },
    {
      "epoch": 0.18767313019390583,
      "grad_norm": 1.2652945518493652,
      "learning_rate": 1.8409303562121664e-05,
      "loss": 1.4268,
      "step": 271
    },
    {
      "epoch": 0.1883656509695291,
      "grad_norm": 1.5869050025939941,
      "learning_rate": 1.8397428077297622e-05,
      "loss": 1.4879,
      "step": 272
    },
    {
      "epoch": 0.18905817174515235,
      "grad_norm": 1.4051944017410278,
      "learning_rate": 1.838551228852817e-05,
      "loss": 1.4672,
      "step": 273
    },
    {
      "epoch": 0.18975069252077562,
      "grad_norm": 1.375993251800537,
      "learning_rate": 1.837355625300383e-05,
      "loss": 1.4808,
      "step": 274
    },
    {
      "epoch": 0.1904432132963989,
      "grad_norm": 1.6987355947494507,
      "learning_rate": 1.83615600281083e-05,
      "loss": 1.5659,
      "step": 275
    },
    {
      "epoch": 0.19113573407202217,
      "grad_norm": 1.5170179605484009,
      "learning_rate": 1.834952367141816e-05,
      "loss": 1.5005,
      "step": 276
    },
    {
      "epoch": 0.19182825484764543,
      "grad_norm": 1.5223069190979004,
      "learning_rate": 1.8337447240702596e-05,
      "loss": 1.4851,
      "step": 277
    },
    {
      "epoch": 0.1925207756232687,
      "grad_norm": 1.4131394624710083,
      "learning_rate": 1.8325330793923146e-05,
      "loss": 1.4095,
      "step": 278
    },
    {
      "epoch": 0.19321329639889195,
      "grad_norm": 1.3926132917404175,
      "learning_rate": 1.8313174389233403e-05,
      "loss": 1.3965,
      "step": 279
    },
    {
      "epoch": 0.19390581717451524,
      "grad_norm": 1.392206072807312,
      "learning_rate": 1.8300978084978736e-05,
      "loss": 1.4758,
      "step": 280
    },
    {
      "epoch": 0.1945983379501385,
      "grad_norm": 1.290574550628662,
      "learning_rate": 1.8288741939696023e-05,
      "loss": 1.4714,
      "step": 281
    },
    {
      "epoch": 0.19529085872576177,
      "grad_norm": 1.2657057046890259,
      "learning_rate": 1.8276466012113358e-05,
      "loss": 1.3736,
      "step": 282
    },
    {
      "epoch": 0.19598337950138503,
      "grad_norm": 1.5142226219177246,
      "learning_rate": 1.826415036114976e-05,
      "loss": 1.3636,
      "step": 283
    },
    {
      "epoch": 0.19667590027700832,
      "grad_norm": 1.4049171209335327,
      "learning_rate": 1.8251795045914922e-05,
      "loss": 1.4736,
      "step": 284
    },
    {
      "epoch": 0.19736842105263158,
      "grad_norm": 1.409709095954895,
      "learning_rate": 1.8239400125708902e-05,
      "loss": 1.5229,
      "step": 285
    },
    {
      "epoch": 0.19806094182825484,
      "grad_norm": 1.5270507335662842,
      "learning_rate": 1.8226965660021836e-05,
      "loss": 1.4371,
      "step": 286
    },
    {
      "epoch": 0.1987534626038781,
      "grad_norm": 1.3285179138183594,
      "learning_rate": 1.8214491708533667e-05,
      "loss": 1.4363,
      "step": 287
    },
    {
      "epoch": 0.1994459833795014,
      "grad_norm": 1.5918437242507935,
      "learning_rate": 1.8201978331113855e-05,
      "loss": 1.4965,
      "step": 288
    },
    {
      "epoch": 0.20013850415512466,
      "grad_norm": 1.3763656616210938,
      "learning_rate": 1.8189425587821083e-05,
      "loss": 1.6337,
      "step": 289
    },
    {
      "epoch": 0.20083102493074792,
      "grad_norm": 1.6876428127288818,
      "learning_rate": 1.817683353890297e-05,
      "loss": 1.1922,
      "step": 290
    },
    {
      "epoch": 0.20152354570637118,
      "grad_norm": 1.462172031402588,
      "learning_rate": 1.8164202244795795e-05,
      "loss": 1.4912,
      "step": 291
    },
    {
      "epoch": 0.20221606648199447,
      "grad_norm": 1.4951701164245605,
      "learning_rate": 1.8151531766124186e-05,
      "loss": 1.5338,
      "step": 292
    },
    {
      "epoch": 0.20290858725761773,
      "grad_norm": 1.4076398611068726,
      "learning_rate": 1.8138822163700847e-05,
      "loss": 1.382,
      "step": 293
    },
    {
      "epoch": 0.203601108033241,
      "grad_norm": 1.4428248405456543,
      "learning_rate": 1.8126073498526254e-05,
      "loss": 1.3539,
      "step": 294
    },
    {
      "epoch": 0.20429362880886426,
      "grad_norm": 1.4771924018859863,
      "learning_rate": 1.8113285831788364e-05,
      "loss": 1.4596,
      "step": 295
    },
    {
      "epoch": 0.20498614958448755,
      "grad_norm": 1.6791538000106812,
      "learning_rate": 1.8100459224862336e-05,
      "loss": 1.1887,
      "step": 296
    },
    {
      "epoch": 0.2056786703601108,
      "grad_norm": 1.3377101421356201,
      "learning_rate": 1.8087593739310214e-05,
      "loss": 1.5108,
      "step": 297
    },
    {
      "epoch": 0.20637119113573407,
      "grad_norm": 1.5619332790374756,
      "learning_rate": 1.8074689436880643e-05,
      "loss": 1.3056,
      "step": 298
    },
    {
      "epoch": 0.20706371191135733,
      "grad_norm": 1.4617923498153687,
      "learning_rate": 1.806174637950858e-05,
      "loss": 1.5482,
      "step": 299
    },
    {
      "epoch": 0.2077562326869806,
      "grad_norm": 1.582040786743164,
      "learning_rate": 1.804876462931498e-05,
      "loss": 1.4631,
      "step": 300
    },
    {
      "epoch": 0.2084487534626039,
      "grad_norm": 1.5981621742248535,
      "learning_rate": 1.803574424860651e-05,
      "loss": 1.4452,
      "step": 301
    },
    {
      "epoch": 0.20914127423822715,
      "grad_norm": 1.3639321327209473,
      "learning_rate": 1.8022685299875245e-05,
      "loss": 1.5022,
      "step": 302
    },
    {
      "epoch": 0.2098337950138504,
      "grad_norm": 1.5008395910263062,
      "learning_rate": 1.800958784579837e-05,
      "loss": 1.4582,
      "step": 303
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 1.4245600700378418,
      "learning_rate": 1.799645194923788e-05,
      "loss": 1.4573,
      "step": 304
    },
    {
      "epoch": 0.21121883656509696,
      "grad_norm": 1.5339776277542114,
      "learning_rate": 1.7983277673240277e-05,
      "loss": 1.5179,
      "step": 305
    },
    {
      "epoch": 0.21191135734072022,
      "grad_norm": 1.4085662364959717,
      "learning_rate": 1.7970065081036266e-05,
      "loss": 1.4609,
      "step": 306
    },
    {
      "epoch": 0.2126038781163435,
      "grad_norm": 1.540628433227539,
      "learning_rate": 1.795681423604045e-05,
      "loss": 1.4168,
      "step": 307
    },
    {
      "epoch": 0.21329639889196675,
      "grad_norm": 1.3457483053207397,
      "learning_rate": 1.7943525201851038e-05,
      "loss": 1.3792,
      "step": 308
    },
    {
      "epoch": 0.21398891966759004,
      "grad_norm": 1.6046233177185059,
      "learning_rate": 1.7930198042249523e-05,
      "loss": 1.4604,
      "step": 309
    },
    {
      "epoch": 0.2146814404432133,
      "grad_norm": 1.6902879476547241,
      "learning_rate": 1.7916832821200375e-05,
      "loss": 1.3465,
      "step": 310
    },
    {
      "epoch": 0.21537396121883656,
      "grad_norm": 1.4801162481307983,
      "learning_rate": 1.7903429602850765e-05,
      "loss": 1.4711,
      "step": 311
    },
    {
      "epoch": 0.21606648199445982,
      "grad_norm": 1.7086902856826782,
      "learning_rate": 1.7889988451530208e-05,
      "loss": 1.3191,
      "step": 312
    },
    {
      "epoch": 0.21675900277008311,
      "grad_norm": 1.6342074871063232,
      "learning_rate": 1.7876509431750303e-05,
      "loss": 1.3109,
      "step": 313
    },
    {
      "epoch": 0.21745152354570638,
      "grad_norm": 1.7456109523773193,
      "learning_rate": 1.7862992608204384e-05,
      "loss": 1.5732,
      "step": 314
    },
    {
      "epoch": 0.21814404432132964,
      "grad_norm": 1.527922511100769,
      "learning_rate": 1.7849438045767233e-05,
      "loss": 1.2966,
      "step": 315
    },
    {
      "epoch": 0.2188365650969529,
      "grad_norm": 1.6418479681015015,
      "learning_rate": 1.783584580949477e-05,
      "loss": 1.4165,
      "step": 316
    },
    {
      "epoch": 0.2195290858725762,
      "grad_norm": 1.5722501277923584,
      "learning_rate": 1.782221596462372e-05,
      "loss": 1.4292,
      "step": 317
    },
    {
      "epoch": 0.22022160664819945,
      "grad_norm": 1.8803280591964722,
      "learning_rate": 1.7808548576571314e-05,
      "loss": 1.1589,
      "step": 318
    },
    {
      "epoch": 0.22091412742382271,
      "grad_norm": 1.510886549949646,
      "learning_rate": 1.7794843710934982e-05,
      "loss": 1.3436,
      "step": 319
    },
    {
      "epoch": 0.22160664819944598,
      "grad_norm": 1.5350871086120605,
      "learning_rate": 1.7781101433492026e-05,
      "loss": 1.5063,
      "step": 320
    },
    {
      "epoch": 0.22229916897506924,
      "grad_norm": 1.538533329963684,
      "learning_rate": 1.7767321810199305e-05,
      "loss": 1.5124,
      "step": 321
    },
    {
      "epoch": 0.22299168975069253,
      "grad_norm": 1.6667261123657227,
      "learning_rate": 1.7753504907192923e-05,
      "loss": 1.318,
      "step": 322
    },
    {
      "epoch": 0.2236842105263158,
      "grad_norm": 1.5465984344482422,
      "learning_rate": 1.7739650790787915e-05,
      "loss": 1.4244,
      "step": 323
    },
    {
      "epoch": 0.22437673130193905,
      "grad_norm": 1.4366633892059326,
      "learning_rate": 1.7725759527477923e-05,
      "loss": 1.5743,
      "step": 324
    },
    {
      "epoch": 0.22506925207756232,
      "grad_norm": 1.5777736902236938,
      "learning_rate": 1.771183118393486e-05,
      "loss": 1.4462,
      "step": 325
    },
    {
      "epoch": 0.2257617728531856,
      "grad_norm": 1.6491996049880981,
      "learning_rate": 1.769786582700864e-05,
      "loss": 1.3709,
      "step": 326
    },
    {
      "epoch": 0.22645429362880887,
      "grad_norm": 1.3817274570465088,
      "learning_rate": 1.7683863523726798e-05,
      "loss": 1.4792,
      "step": 327
    },
    {
      "epoch": 0.22714681440443213,
      "grad_norm": 1.506965160369873,
      "learning_rate": 1.7669824341294203e-05,
      "loss": 1.3656,
      "step": 328
    },
    {
      "epoch": 0.2278393351800554,
      "grad_norm": 1.699218988418579,
      "learning_rate": 1.7655748347092735e-05,
      "loss": 1.4267,
      "step": 329
    },
    {
      "epoch": 0.22853185595567868,
      "grad_norm": 1.546741247177124,
      "learning_rate": 1.7641635608680942e-05,
      "loss": 1.5038,
      "step": 330
    },
    {
      "epoch": 0.22922437673130194,
      "grad_norm": 1.7047183513641357,
      "learning_rate": 1.7627486193793744e-05,
      "loss": 1.3509,
      "step": 331
    },
    {
      "epoch": 0.2299168975069252,
      "grad_norm": 1.6500473022460938,
      "learning_rate": 1.7613300170342073e-05,
      "loss": 1.3246,
      "step": 332
    },
    {
      "epoch": 0.23060941828254847,
      "grad_norm": 1.4320755004882812,
      "learning_rate": 1.7599077606412577e-05,
      "loss": 1.416,
      "step": 333
    },
    {
      "epoch": 0.23130193905817176,
      "grad_norm": 1.5516586303710938,
      "learning_rate": 1.7584818570267287e-05,
      "loss": 1.4455,
      "step": 334
    },
    {
      "epoch": 0.23199445983379502,
      "grad_norm": 1.444295883178711,
      "learning_rate": 1.757052313034327e-05,
      "loss": 1.4894,
      "step": 335
    },
    {
      "epoch": 0.23268698060941828,
      "grad_norm": 1.6120071411132812,
      "learning_rate": 1.755619135525233e-05,
      "loss": 1.3861,
      "step": 336
    },
    {
      "epoch": 0.23337950138504154,
      "grad_norm": 1.4913694858551025,
      "learning_rate": 1.754182331378065e-05,
      "loss": 1.4899,
      "step": 337
    },
    {
      "epoch": 0.23407202216066483,
      "grad_norm": 1.5728522539138794,
      "learning_rate": 1.7527419074888483e-05,
      "loss": 1.458,
      "step": 338
    },
    {
      "epoch": 0.2347645429362881,
      "grad_norm": 1.6220054626464844,
      "learning_rate": 1.7512978707709817e-05,
      "loss": 1.4327,
      "step": 339
    },
    {
      "epoch": 0.23545706371191136,
      "grad_norm": 1.4640010595321655,
      "learning_rate": 1.749850228155203e-05,
      "loss": 1.3738,
      "step": 340
    },
    {
      "epoch": 0.23614958448753462,
      "grad_norm": 1.6346479654312134,
      "learning_rate": 1.7483989865895582e-05,
      "loss": 1.4064,
      "step": 341
    },
    {
      "epoch": 0.23684210526315788,
      "grad_norm": 1.6318504810333252,
      "learning_rate": 1.7469441530393652e-05,
      "loss": 1.3148,
      "step": 342
    },
    {
      "epoch": 0.23753462603878117,
      "grad_norm": 1.7678024768829346,
      "learning_rate": 1.7454857344871823e-05,
      "loss": 1.2937,
      "step": 343
    },
    {
      "epoch": 0.23822714681440443,
      "grad_norm": 1.9176974296569824,
      "learning_rate": 1.7440237379327745e-05,
      "loss": 1.3801,
      "step": 344
    },
    {
      "epoch": 0.2389196675900277,
      "grad_norm": 1.8107141256332397,
      "learning_rate": 1.7425581703930793e-05,
      "loss": 1.2044,
      "step": 345
    },
    {
      "epoch": 0.23961218836565096,
      "grad_norm": 1.5848854780197144,
      "learning_rate": 1.7410890389021737e-05,
      "loss": 1.4562,
      "step": 346
    },
    {
      "epoch": 0.24030470914127425,
      "grad_norm": 1.6646206378936768,
      "learning_rate": 1.7396163505112398e-05,
      "loss": 1.279,
      "step": 347
    },
    {
      "epoch": 0.2409972299168975,
      "grad_norm": 1.407603144645691,
      "learning_rate": 1.7381401122885316e-05,
      "loss": 1.2988,
      "step": 348
    },
    {
      "epoch": 0.24168975069252077,
      "grad_norm": 1.5010101795196533,
      "learning_rate": 1.73666033131934e-05,
      "loss": 1.3755,
      "step": 349
    },
    {
      "epoch": 0.24238227146814403,
      "grad_norm": 1.5880669355392456,
      "learning_rate": 1.7351770147059604e-05,
      "loss": 1.5101,
      "step": 350
    },
    {
      "epoch": 0.24307479224376732,
      "grad_norm": 1.6278942823410034,
      "learning_rate": 1.733690169567657e-05,
      "loss": 1.3891,
      "step": 351
    },
    {
      "epoch": 0.24376731301939059,
      "grad_norm": 1.6076022386550903,
      "learning_rate": 1.7321998030406303e-05,
      "loss": 1.4076,
      "step": 352
    },
    {
      "epoch": 0.24445983379501385,
      "grad_norm": 1.4889235496520996,
      "learning_rate": 1.7307059222779806e-05,
      "loss": 1.3706,
      "step": 353
    },
    {
      "epoch": 0.2451523545706371,
      "grad_norm": 1.619982361793518,
      "learning_rate": 1.729208534449676e-05,
      "loss": 1.3843,
      "step": 354
    },
    {
      "epoch": 0.2458448753462604,
      "grad_norm": 1.5546656847000122,
      "learning_rate": 1.7277076467425163e-05,
      "loss": 1.4328,
      "step": 355
    },
    {
      "epoch": 0.24653739612188366,
      "grad_norm": 1.69061279296875,
      "learning_rate": 1.7262032663601003e-05,
      "loss": 1.3942,
      "step": 356
    },
    {
      "epoch": 0.24722991689750692,
      "grad_norm": 1.8314852714538574,
      "learning_rate": 1.7246954005227884e-05,
      "loss": 1.3588,
      "step": 357
    },
    {
      "epoch": 0.24792243767313019,
      "grad_norm": 1.7705515623092651,
      "learning_rate": 1.723184056467671e-05,
      "loss": 1.4111,
      "step": 358
    },
    {
      "epoch": 0.24861495844875348,
      "grad_norm": 1.7024461030960083,
      "learning_rate": 1.721669241448532e-05,
      "loss": 1.3465,
      "step": 359
    },
    {
      "epoch": 0.24930747922437674,
      "grad_norm": 1.7152822017669678,
      "learning_rate": 1.7201509627358143e-05,
      "loss": 1.3865,
      "step": 360
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9307787418365479,
      "learning_rate": 1.718629227616585e-05,
      "loss": 1.3747,
      "step": 361
    },
    {
      "epoch": 0.25,
      "eval_loss": 1.0831600427627563,
      "eval_runtime": 335.7463,
      "eval_samples_per_second": 4.217,
      "eval_steps_per_second": 0.527,
      "step": 361
    },
    {
      "epoch": 0.25069252077562326,
      "grad_norm": 2.011570692062378,
      "learning_rate": 1.7171040433945006e-05,
      "loss": 1.219,
      "step": 362
    },
    {
      "epoch": 0.2513850415512465,
      "grad_norm": 1.763128399848938,
      "learning_rate": 1.7155754173897718e-05,
      "loss": 1.4044,
      "step": 363
    },
    {
      "epoch": 0.2520775623268698,
      "grad_norm": 1.6575963497161865,
      "learning_rate": 1.7140433569391275e-05,
      "loss": 1.5222,
      "step": 364
    },
    {
      "epoch": 0.25277008310249305,
      "grad_norm": 1.6661745309829712,
      "learning_rate": 1.7125078693957817e-05,
      "loss": 1.5345,
      "step": 365
    },
    {
      "epoch": 0.25346260387811637,
      "grad_norm": 1.8055832386016846,
      "learning_rate": 1.710968962129396e-05,
      "loss": 1.5074,
      "step": 366
    },
    {
      "epoch": 0.25415512465373963,
      "grad_norm": 1.827330231666565,
      "learning_rate": 1.709426642526046e-05,
      "loss": 1.4927,
      "step": 367
    },
    {
      "epoch": 0.2548476454293629,
      "grad_norm": 1.907296895980835,
      "learning_rate": 1.7078809179881847e-05,
      "loss": 1.2675,
      "step": 368
    },
    {
      "epoch": 0.25554016620498615,
      "grad_norm": 1.5925551652908325,
      "learning_rate": 1.706331795934606e-05,
      "loss": 1.369,
      "step": 369
    },
    {
      "epoch": 0.2562326869806094,
      "grad_norm": 1.6697511672973633,
      "learning_rate": 1.704779283800412e-05,
      "loss": 1.4416,
      "step": 370
    },
    {
      "epoch": 0.2569252077562327,
      "grad_norm": 1.766982913017273,
      "learning_rate": 1.7032233890369755e-05,
      "loss": 1.4247,
      "step": 371
    },
    {
      "epoch": 0.25761772853185594,
      "grad_norm": 1.8494454622268677,
      "learning_rate": 1.701664119111904e-05,
      "loss": 1.4531,
      "step": 372
    },
    {
      "epoch": 0.2583102493074792,
      "grad_norm": 1.8136225938796997,
      "learning_rate": 1.700101481509004e-05,
      "loss": 1.4062,
      "step": 373
    },
    {
      "epoch": 0.2590027700831025,
      "grad_norm": 1.662751317024231,
      "learning_rate": 1.6985354837282462e-05,
      "loss": 1.4713,
      "step": 374
    },
    {
      "epoch": 0.2596952908587258,
      "grad_norm": 1.670058012008667,
      "learning_rate": 1.6969661332857278e-05,
      "loss": 1.5039,
      "step": 375
    },
    {
      "epoch": 0.26038781163434904,
      "grad_norm": 1.7041593790054321,
      "learning_rate": 1.6953934377136375e-05,
      "loss": 1.5144,
      "step": 376
    },
    {
      "epoch": 0.2610803324099723,
      "grad_norm": 1.9452157020568848,
      "learning_rate": 1.6938174045602203e-05,
      "loss": 1.5408,
      "step": 377
    },
    {
      "epoch": 0.26177285318559557,
      "grad_norm": 1.7609469890594482,
      "learning_rate": 1.6922380413897382e-05,
      "loss": 1.4979,
      "step": 378
    },
    {
      "epoch": 0.26246537396121883,
      "grad_norm": 1.777033805847168,
      "learning_rate": 1.6906553557824372e-05,
      "loss": 1.3503,
      "step": 379
    },
    {
      "epoch": 0.2631578947368421,
      "grad_norm": 1.6396095752716064,
      "learning_rate": 1.689069355334509e-05,
      "loss": 1.3918,
      "step": 380
    },
    {
      "epoch": 0.26385041551246535,
      "grad_norm": 1.517849087715149,
      "learning_rate": 1.6874800476580553e-05,
      "loss": 1.4,
      "step": 381
    },
    {
      "epoch": 0.26454293628808867,
      "grad_norm": 1.8570725917816162,
      "learning_rate": 1.6858874403810507e-05,
      "loss": 1.2621,
      "step": 382
    },
    {
      "epoch": 0.26523545706371193,
      "grad_norm": 1.9514487981796265,
      "learning_rate": 1.684291541147307e-05,
      "loss": 1.4256,
      "step": 383
    },
    {
      "epoch": 0.2659279778393352,
      "grad_norm": 1.9198554754257202,
      "learning_rate": 1.682692357616435e-05,
      "loss": 1.5005,
      "step": 384
    },
    {
      "epoch": 0.26662049861495846,
      "grad_norm": 1.7078975439071655,
      "learning_rate": 1.6810898974638098e-05,
      "loss": 1.3862,
      "step": 385
    },
    {
      "epoch": 0.2673130193905817,
      "grad_norm": 1.663574457168579,
      "learning_rate": 1.679484168380532e-05,
      "loss": 1.3513,
      "step": 386
    },
    {
      "epoch": 0.268005540166205,
      "grad_norm": 1.7911577224731445,
      "learning_rate": 1.677875178073392e-05,
      "loss": 1.4409,
      "step": 387
    },
    {
      "epoch": 0.26869806094182824,
      "grad_norm": 1.7094714641571045,
      "learning_rate": 1.676262934264832e-05,
      "loss": 1.5266,
      "step": 388
    },
    {
      "epoch": 0.2693905817174515,
      "grad_norm": 1.6942095756530762,
      "learning_rate": 1.674647444692911e-05,
      "loss": 1.4287,
      "step": 389
    },
    {
      "epoch": 0.27008310249307477,
      "grad_norm": 1.7591415643692017,
      "learning_rate": 1.6730287171112652e-05,
      "loss": 1.3419,
      "step": 390
    },
    {
      "epoch": 0.2707756232686981,
      "grad_norm": 1.739908218383789,
      "learning_rate": 1.6714067592890713e-05,
      "loss": 1.3593,
      "step": 391
    },
    {
      "epoch": 0.27146814404432135,
      "grad_norm": 1.729467749595642,
      "learning_rate": 1.669781579011011e-05,
      "loss": 1.4044,
      "step": 392
    },
    {
      "epoch": 0.2721606648199446,
      "grad_norm": 1.7394282817840576,
      "learning_rate": 1.6681531840772314e-05,
      "loss": 1.5558,
      "step": 393
    },
    {
      "epoch": 0.27285318559556787,
      "grad_norm": 1.867372989654541,
      "learning_rate": 1.666521582303309e-05,
      "loss": 1.4419,
      "step": 394
    },
    {
      "epoch": 0.27354570637119113,
      "grad_norm": 1.7546651363372803,
      "learning_rate": 1.664886781520212e-05,
      "loss": 1.4957,
      "step": 395
    },
    {
      "epoch": 0.2742382271468144,
      "grad_norm": 2.4754140377044678,
      "learning_rate": 1.6632487895742612e-05,
      "loss": 1.3339,
      "step": 396
    },
    {
      "epoch": 0.27493074792243766,
      "grad_norm": 1.6385406255722046,
      "learning_rate": 1.661607614327095e-05,
      "loss": 1.3994,
      "step": 397
    },
    {
      "epoch": 0.2756232686980609,
      "grad_norm": 1.951462984085083,
      "learning_rate": 1.6599632636556292e-05,
      "loss": 1.2803,
      "step": 398
    },
    {
      "epoch": 0.27631578947368424,
      "grad_norm": 1.989600419998169,
      "learning_rate": 1.6583157454520214e-05,
      "loss": 1.359,
      "step": 399
    },
    {
      "epoch": 0.2770083102493075,
      "grad_norm": 1.6627402305603027,
      "learning_rate": 1.6566650676236307e-05,
      "loss": 1.4155,
      "step": 400
    },
    {
      "epoch": 0.27770083102493076,
      "grad_norm": 1.7846615314483643,
      "learning_rate": 1.6550112380929814e-05,
      "loss": 1.6159,
      "step": 401
    },
    {
      "epoch": 0.278393351800554,
      "grad_norm": 1.8167567253112793,
      "learning_rate": 1.653354264797725e-05,
      "loss": 1.411,
      "step": 402
    },
    {
      "epoch": 0.2790858725761773,
      "grad_norm": 2.0551259517669678,
      "learning_rate": 1.651694155690601e-05,
      "loss": 1.0842,
      "step": 403
    },
    {
      "epoch": 0.27977839335180055,
      "grad_norm": 1.8748019933700562,
      "learning_rate": 1.6500309187394005e-05,
      "loss": 1.3191,
      "step": 404
    },
    {
      "epoch": 0.2804709141274238,
      "grad_norm": 1.9822982549667358,
      "learning_rate": 1.6483645619269256e-05,
      "loss": 1.3953,
      "step": 405
    },
    {
      "epoch": 0.28116343490304707,
      "grad_norm": 1.8638688325881958,
      "learning_rate": 1.6466950932509532e-05,
      "loss": 1.3538,
      "step": 406
    },
    {
      "epoch": 0.28185595567867033,
      "grad_norm": 1.7385188341140747,
      "learning_rate": 1.645022520724195e-05,
      "loss": 1.4492,
      "step": 407
    },
    {
      "epoch": 0.28254847645429365,
      "grad_norm": 1.7316495180130005,
      "learning_rate": 1.643346852374261e-05,
      "loss": 1.3456,
      "step": 408
    },
    {
      "epoch": 0.2832409972299169,
      "grad_norm": 1.8079771995544434,
      "learning_rate": 1.641668096243619e-05,
      "loss": 1.3072,
      "step": 409
    },
    {
      "epoch": 0.2839335180055402,
      "grad_norm": 2.2126691341400146,
      "learning_rate": 1.6399862603895563e-05,
      "loss": 1.327,
      "step": 410
    },
    {
      "epoch": 0.28462603878116344,
      "grad_norm": 1.7359458208084106,
      "learning_rate": 1.6383013528841424e-05,
      "loss": 1.5197,
      "step": 411
    },
    {
      "epoch": 0.2853185595567867,
      "grad_norm": 1.96955406665802,
      "learning_rate": 1.6366133818141893e-05,
      "loss": 1.3577,
      "step": 412
    },
    {
      "epoch": 0.28601108033240996,
      "grad_norm": 1.8084375858306885,
      "learning_rate": 1.6349223552812125e-05,
      "loss": 1.3055,
      "step": 413
    },
    {
      "epoch": 0.2867036011080332,
      "grad_norm": 1.7716041803359985,
      "learning_rate": 1.633228281401392e-05,
      "loss": 1.3429,
      "step": 414
    },
    {
      "epoch": 0.2873961218836565,
      "grad_norm": 2.1030023097991943,
      "learning_rate": 1.631531168305534e-05,
      "loss": 1.142,
      "step": 415
    },
    {
      "epoch": 0.2880886426592798,
      "grad_norm": 1.6409399509429932,
      "learning_rate": 1.6298310241390326e-05,
      "loss": 1.5097,
      "step": 416
    },
    {
      "epoch": 0.28878116343490307,
      "grad_norm": 1.7756257057189941,
      "learning_rate": 1.6281278570618276e-05,
      "loss": 1.3902,
      "step": 417
    },
    {
      "epoch": 0.2894736842105263,
      "grad_norm": 1.753067970275879,
      "learning_rate": 1.6264216752483697e-05,
      "loss": 1.358,
      "step": 418
    },
    {
      "epoch": 0.2901662049861496,
      "grad_norm": 2.262544631958008,
      "learning_rate": 1.6247124868875777e-05,
      "loss": 1.4347,
      "step": 419
    },
    {
      "epoch": 0.29085872576177285,
      "grad_norm": 1.791831135749817,
      "learning_rate": 1.6230003001828e-05,
      "loss": 1.4917,
      "step": 420
    },
    {
      "epoch": 0.2915512465373961,
      "grad_norm": 1.7962827682495117,
      "learning_rate": 1.6212851233517772e-05,
      "loss": 1.3176,
      "step": 421
    },
    {
      "epoch": 0.2922437673130194,
      "grad_norm": 1.8246115446090698,
      "learning_rate": 1.6195669646266003e-05,
      "loss": 1.2801,
      "step": 422
    },
    {
      "epoch": 0.29293628808864264,
      "grad_norm": 1.9331282377243042,
      "learning_rate": 1.6178458322536722e-05,
      "loss": 1.3653,
      "step": 423
    },
    {
      "epoch": 0.29362880886426596,
      "grad_norm": 1.9322010278701782,
      "learning_rate": 1.616121734493668e-05,
      "loss": 1.4005,
      "step": 424
    },
    {
      "epoch": 0.2943213296398892,
      "grad_norm": 1.946174144744873,
      "learning_rate": 1.614394679621495e-05,
      "loss": 1.5496,
      "step": 425
    },
    {
      "epoch": 0.2950138504155125,
      "grad_norm": 1.8700590133666992,
      "learning_rate": 1.6126646759262548e-05,
      "loss": 1.3886,
      "step": 426
    },
    {
      "epoch": 0.29570637119113574,
      "grad_norm": 2.0393857955932617,
      "learning_rate": 1.6109317317111996e-05,
      "loss": 1.2218,
      "step": 427
    },
    {
      "epoch": 0.296398891966759,
      "grad_norm": 1.9611023664474487,
      "learning_rate": 1.609195855293697e-05,
      "loss": 1.3157,
      "step": 428
    },
    {
      "epoch": 0.29709141274238227,
      "grad_norm": 2.472890853881836,
      "learning_rate": 1.6074570550051864e-05,
      "loss": 1.3484,
      "step": 429
    },
    {
      "epoch": 0.29778393351800553,
      "grad_norm": 1.8426192998886108,
      "learning_rate": 1.6057153391911422e-05,
      "loss": 1.3931,
      "step": 430
    },
    {
      "epoch": 0.2984764542936288,
      "grad_norm": 1.7728618383407593,
      "learning_rate": 1.6039707162110296e-05,
      "loss": 1.3515,
      "step": 431
    },
    {
      "epoch": 0.29916897506925205,
      "grad_norm": 2.0701711177825928,
      "learning_rate": 1.6022231944382693e-05,
      "loss": 1.3197,
      "step": 432
    },
    {
      "epoch": 0.29986149584487537,
      "grad_norm": 1.7944912910461426,
      "learning_rate": 1.6004727822601932e-05,
      "loss": 1.4434,
      "step": 433
    },
    {
      "epoch": 0.30055401662049863,
      "grad_norm": 1.810572862625122,
      "learning_rate": 1.598719488078007e-05,
      "loss": 1.331,
      "step": 434
    },
    {
      "epoch": 0.3012465373961219,
      "grad_norm": 1.7466157674789429,
      "learning_rate": 1.596963320306748e-05,
      "loss": 1.3863,
      "step": 435
    },
    {
      "epoch": 0.30193905817174516,
      "grad_norm": 1.9109450578689575,
      "learning_rate": 1.5952042873752463e-05,
      "loss": 1.3476,
      "step": 436
    },
    {
      "epoch": 0.3026315789473684,
      "grad_norm": 1.788562297821045,
      "learning_rate": 1.593442397726082e-05,
      "loss": 1.3668,
      "step": 437
    },
    {
      "epoch": 0.3033240997229917,
      "grad_norm": 1.8199845552444458,
      "learning_rate": 1.5916776598155478e-05,
      "loss": 1.3688,
      "step": 438
    },
    {
      "epoch": 0.30401662049861494,
      "grad_norm": 1.8265742063522339,
      "learning_rate": 1.5899100821136065e-05,
      "loss": 1.4673,
      "step": 439
    },
    {
      "epoch": 0.3047091412742382,
      "grad_norm": 1.729217290878296,
      "learning_rate": 1.5881396731038493e-05,
      "loss": 1.4452,
      "step": 440
    },
    {
      "epoch": 0.3054016620498615,
      "grad_norm": 1.8088419437408447,
      "learning_rate": 1.5863664412834582e-05,
      "loss": 1.5256,
      "step": 441
    },
    {
      "epoch": 0.3060941828254848,
      "grad_norm": 1.9562954902648926,
      "learning_rate": 1.584590395163162e-05,
      "loss": 1.3788,
      "step": 442
    },
    {
      "epoch": 0.30678670360110805,
      "grad_norm": 2.257169008255005,
      "learning_rate": 1.5828115432671984e-05,
      "loss": 1.2451,
      "step": 443
    },
    {
      "epoch": 0.3074792243767313,
      "grad_norm": 1.9341715574264526,
      "learning_rate": 1.5810298941332696e-05,
      "loss": 1.4096,
      "step": 444
    },
    {
      "epoch": 0.30817174515235457,
      "grad_norm": 2.0554494857788086,
      "learning_rate": 1.579245456312506e-05,
      "loss": 1.552,
      "step": 445
    },
    {
      "epoch": 0.30886426592797783,
      "grad_norm": 2.0175342559814453,
      "learning_rate": 1.5774582383694196e-05,
      "loss": 1.4782,
      "step": 446
    },
    {
      "epoch": 0.3095567867036011,
      "grad_norm": 1.9586889743804932,
      "learning_rate": 1.5756682488818674e-05,
      "loss": 1.401,
      "step": 447
    },
    {
      "epoch": 0.31024930747922436,
      "grad_norm": 2.00122332572937,
      "learning_rate": 1.5738754964410084e-05,
      "loss": 1.3873,
      "step": 448
    },
    {
      "epoch": 0.3109418282548476,
      "grad_norm": 1.9168496131896973,
      "learning_rate": 1.5720799896512625e-05,
      "loss": 1.329,
      "step": 449
    },
    {
      "epoch": 0.31163434903047094,
      "grad_norm": 2.23854923248291,
      "learning_rate": 1.5702817371302684e-05,
      "loss": 1.3508,
      "step": 450
    },
    {
      "epoch": 0.3123268698060942,
      "grad_norm": 1.7631722688674927,
      "learning_rate": 1.5684807475088453e-05,
      "loss": 1.4677,
      "step": 451
    },
    {
      "epoch": 0.31301939058171746,
      "grad_norm": 2.0044777393341064,
      "learning_rate": 1.5666770294309467e-05,
      "loss": 1.3986,
      "step": 452
    },
    {
      "epoch": 0.3137119113573407,
      "grad_norm": 1.8325257301330566,
      "learning_rate": 1.5648705915536227e-05,
      "loss": 1.2592,
      "step": 453
    },
    {
      "epoch": 0.314404432132964,
      "grad_norm": 2.203279495239258,
      "learning_rate": 1.5630614425469776e-05,
      "loss": 1.1724,
      "step": 454
    },
    {
      "epoch": 0.31509695290858725,
      "grad_norm": 1.7246652841567993,
      "learning_rate": 1.561249591094127e-05,
      "loss": 1.4664,
      "step": 455
    },
    {
      "epoch": 0.3157894736842105,
      "grad_norm": 2.057136058807373,
      "learning_rate": 1.5594350458911586e-05,
      "loss": 1.3707,
      "step": 456
    },
    {
      "epoch": 0.31648199445983377,
      "grad_norm": 1.8668776750564575,
      "learning_rate": 1.5576178156470863e-05,
      "loss": 1.4212,
      "step": 457
    },
    {
      "epoch": 0.3171745152354571,
      "grad_norm": 1.90840744972229,
      "learning_rate": 1.5557979090838136e-05,
      "loss": 1.529,
      "step": 458
    },
    {
      "epoch": 0.31786703601108035,
      "grad_norm": 2.1340227127075195,
      "learning_rate": 1.553975334936088e-05,
      "loss": 1.4705,
      "step": 459
    },
    {
      "epoch": 0.3185595567867036,
      "grad_norm": 2.315753698348999,
      "learning_rate": 1.55215010195146e-05,
      "loss": 1.2551,
      "step": 460
    },
    {
      "epoch": 0.3192520775623269,
      "grad_norm": 2.071791172027588,
      "learning_rate": 1.5503222188902412e-05,
      "loss": 1.2973,
      "step": 461
    },
    {
      "epoch": 0.31994459833795014,
      "grad_norm": 1.9834339618682861,
      "learning_rate": 1.5484916945254642e-05,
      "loss": 1.351,
      "step": 462
    },
    {
      "epoch": 0.3206371191135734,
      "grad_norm": 1.92372465133667,
      "learning_rate": 1.5466585376428365e-05,
      "loss": 1.3197,
      "step": 463
    },
    {
      "epoch": 0.32132963988919666,
      "grad_norm": 2.1652302742004395,
      "learning_rate": 1.5448227570407012e-05,
      "loss": 1.3793,
      "step": 464
    },
    {
      "epoch": 0.3220221606648199,
      "grad_norm": 2.0220046043395996,
      "learning_rate": 1.5429843615299953e-05,
      "loss": 1.3064,
      "step": 465
    },
    {
      "epoch": 0.32271468144044324,
      "grad_norm": 2.301950454711914,
      "learning_rate": 1.5411433599342038e-05,
      "loss": 1.2608,
      "step": 466
    },
    {
      "epoch": 0.3234072022160665,
      "grad_norm": 2.135833501815796,
      "learning_rate": 1.539299761089322e-05,
      "loss": 1.1929,
      "step": 467
    },
    {
      "epoch": 0.32409972299168976,
      "grad_norm": 1.8044958114624023,
      "learning_rate": 1.5374535738438105e-05,
      "loss": 1.3328,
      "step": 468
    },
    {
      "epoch": 0.324792243767313,
      "grad_norm": 1.8159520626068115,
      "learning_rate": 1.5356048070585513e-05,
      "loss": 1.2574,
      "step": 469
    },
    {
      "epoch": 0.3254847645429363,
      "grad_norm": 1.9937934875488281,
      "learning_rate": 1.5337534696068088e-05,
      "loss": 1.2826,
      "step": 470
    },
    {
      "epoch": 0.32617728531855955,
      "grad_norm": 2.1979124546051025,
      "learning_rate": 1.5318995703741854e-05,
      "loss": 1.4514,
      "step": 471
    },
    {
      "epoch": 0.3268698060941828,
      "grad_norm": 2.0025925636291504,
      "learning_rate": 1.5300431182585777e-05,
      "loss": 1.499,
      "step": 472
    },
    {
      "epoch": 0.3275623268698061,
      "grad_norm": 2.007138252258301,
      "learning_rate": 1.528184122170137e-05,
      "loss": 1.2578,
      "step": 473
    },
    {
      "epoch": 0.32825484764542934,
      "grad_norm": 1.976704478263855,
      "learning_rate": 1.5263225910312222e-05,
      "loss": 1.5246,
      "step": 474
    },
    {
      "epoch": 0.32894736842105265,
      "grad_norm": 2.6022000312805176,
      "learning_rate": 1.524458533776361e-05,
      "loss": 1.1411,
      "step": 475
    },
    {
      "epoch": 0.3296398891966759,
      "grad_norm": 2.1338884830474854,
      "learning_rate": 1.5225919593522049e-05,
      "loss": 1.3387,
      "step": 476
    },
    {
      "epoch": 0.3303324099722992,
      "grad_norm": 2.0747604370117188,
      "learning_rate": 1.5207228767174865e-05,
      "loss": 1.3682,
      "step": 477
    },
    {
      "epoch": 0.33102493074792244,
      "grad_norm": 2.2153992652893066,
      "learning_rate": 1.5188512948429765e-05,
      "loss": 1.2949,
      "step": 478
    },
    {
      "epoch": 0.3317174515235457,
      "grad_norm": 2.1263861656188965,
      "learning_rate": 1.516977222711442e-05,
      "loss": 1.3209,
      "step": 479
    },
    {
      "epoch": 0.33240997229916897,
      "grad_norm": 2.2222840785980225,
      "learning_rate": 1.5151006693176005e-05,
      "loss": 1.1555,
      "step": 480
    },
    {
      "epoch": 0.3331024930747922,
      "grad_norm": 2.1737382411956787,
      "learning_rate": 1.5132216436680798e-05,
      "loss": 1.2706,
      "step": 481
    },
    {
      "epoch": 0.3337950138504155,
      "grad_norm": 1.967319369316101,
      "learning_rate": 1.5113401547813732e-05,
      "loss": 1.3886,
      "step": 482
    },
    {
      "epoch": 0.3344875346260388,
      "grad_norm": 2.301297664642334,
      "learning_rate": 1.5094562116877955e-05,
      "loss": 1.2934,
      "step": 483
    },
    {
      "epoch": 0.33518005540166207,
      "grad_norm": 1.921366810798645,
      "learning_rate": 1.5075698234294424e-05,
      "loss": 1.4122,
      "step": 484
    },
    {
      "epoch": 0.33587257617728533,
      "grad_norm": 2.278940439224243,
      "learning_rate": 1.5056809990601434e-05,
      "loss": 1.3407,
      "step": 485
    },
    {
      "epoch": 0.3365650969529086,
      "grad_norm": 2.016507148742676,
      "learning_rate": 1.5037897476454219e-05,
      "loss": 1.4208,
      "step": 486
    },
    {
      "epoch": 0.33725761772853186,
      "grad_norm": 1.9419777393341064,
      "learning_rate": 1.5018960782624486e-05,
      "loss": 1.3269,
      "step": 487
    },
    {
      "epoch": 0.3379501385041551,
      "grad_norm": 1.8907411098480225,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.493,
      "step": 488
    },
    {
      "epoch": 0.3386426592797784,
      "grad_norm": 2.395200252532959,
      "learning_rate": 1.4981015219584149e-05,
      "loss": 1.2829,
      "step": 489
    },
    {
      "epoch": 0.33933518005540164,
      "grad_norm": 2.1097700595855713,
      "learning_rate": 1.496200653249549e-05,
      "loss": 1.4754,
      "step": 490
    },
    {
      "epoch": 0.3400277008310249,
      "grad_norm": 2.089921474456787,
      "learning_rate": 1.4942974029967319e-05,
      "loss": 1.389,
      "step": 491
    },
    {
      "epoch": 0.3407202216066482,
      "grad_norm": 2.3994808197021484,
      "learning_rate": 1.492391780334725e-05,
      "loss": 1.2998,
      "step": 492
    },
    {
      "epoch": 0.3414127423822715,
      "grad_norm": 2.2223410606384277,
      "learning_rate": 1.4904837944096744e-05,
      "loss": 1.3659,
      "step": 493
    },
    {
      "epoch": 0.34210526315789475,
      "grad_norm": 2.277879238128662,
      "learning_rate": 1.4885734543790707e-05,
      "loss": 1.3224,
      "step": 494
    },
    {
      "epoch": 0.342797783933518,
      "grad_norm": 2.726764440536499,
      "learning_rate": 1.4866607694117012e-05,
      "loss": 1.5039,
      "step": 495
    },
    {
      "epoch": 0.34349030470914127,
      "grad_norm": 2.2179274559020996,
      "learning_rate": 1.4847457486876097e-05,
      "loss": 1.3766,
      "step": 496
    },
    {
      "epoch": 0.34418282548476453,
      "grad_norm": 2.170088052749634,
      "learning_rate": 1.4828284013980492e-05,
      "loss": 1.5275,
      "step": 497
    },
    {
      "epoch": 0.3448753462603878,
      "grad_norm": 2.057279586791992,
      "learning_rate": 1.4809087367454402e-05,
      "loss": 1.409,
      "step": 498
    },
    {
      "epoch": 0.34556786703601106,
      "grad_norm": 2.0357258319854736,
      "learning_rate": 1.478986763943325e-05,
      "loss": 1.4436,
      "step": 499
    },
    {
      "epoch": 0.3462603878116344,
      "grad_norm": 2.172924518585205,
      "learning_rate": 1.4770624922163233e-05,
      "loss": 1.2419,
      "step": 500
    },
    {
      "epoch": 0.34695290858725764,
      "grad_norm": 2.027270793914795,
      "learning_rate": 1.4751359308000904e-05,
      "loss": 1.3761,
      "step": 501
    },
    {
      "epoch": 0.3476454293628809,
      "grad_norm": 2.094010591506958,
      "learning_rate": 1.4732070889412693e-05,
      "loss": 1.5432,
      "step": 502
    },
    {
      "epoch": 0.34833795013850416,
      "grad_norm": 1.9574612379074097,
      "learning_rate": 1.4712759758974496e-05,
      "loss": 1.3299,
      "step": 503
    },
    {
      "epoch": 0.3490304709141274,
      "grad_norm": 2.2416937351226807,
      "learning_rate": 1.4693426009371203e-05,
      "loss": 1.2703,
      "step": 504
    },
    {
      "epoch": 0.3497229916897507,
      "grad_norm": 2.2920377254486084,
      "learning_rate": 1.4674069733396277e-05,
      "loss": 1.359,
      "step": 505
    },
    {
      "epoch": 0.35041551246537395,
      "grad_norm": 2.218531847000122,
      "learning_rate": 1.4654691023951289e-05,
      "loss": 1.3286,
      "step": 506
    },
    {
      "epoch": 0.3511080332409972,
      "grad_norm": 2.079054117202759,
      "learning_rate": 1.4635289974045485e-05,
      "loss": 1.4356,
      "step": 507
    },
    {
      "epoch": 0.3518005540166205,
      "grad_norm": 1.972861886024475,
      "learning_rate": 1.4615866676795334e-05,
      "loss": 1.4091,
      "step": 508
    },
    {
      "epoch": 0.3524930747922438,
      "grad_norm": 2.4713618755340576,
      "learning_rate": 1.4596421225424084e-05,
      "loss": 1.28,
      "step": 509
    },
    {
      "epoch": 0.35318559556786705,
      "grad_norm": 1.9506804943084717,
      "learning_rate": 1.4576953713261313e-05,
      "loss": 1.3192,
      "step": 510
    },
    {
      "epoch": 0.3538781163434903,
      "grad_norm": 2.0880720615386963,
      "learning_rate": 1.4557464233742478e-05,
      "loss": 1.4332,
      "step": 511
    },
    {
      "epoch": 0.3545706371191136,
      "grad_norm": 2.2580926418304443,
      "learning_rate": 1.4537952880408472e-05,
      "loss": 1.3437,
      "step": 512
    },
    {
      "epoch": 0.35526315789473684,
      "grad_norm": 2.160332441329956,
      "learning_rate": 1.4518419746905174e-05,
      "loss": 1.2442,
      "step": 513
    },
    {
      "epoch": 0.3559556786703601,
      "grad_norm": 2.0894155502319336,
      "learning_rate": 1.4498864926982996e-05,
      "loss": 1.4602,
      "step": 514
    },
    {
      "epoch": 0.35664819944598336,
      "grad_norm": 2.108365058898926,
      "learning_rate": 1.4479288514496434e-05,
      "loss": 1.4086,
      "step": 515
    },
    {
      "epoch": 0.3573407202216066,
      "grad_norm": 1.9162973165512085,
      "learning_rate": 1.4459690603403623e-05,
      "loss": 1.3188,
      "step": 516
    },
    {
      "epoch": 0.35803324099722994,
      "grad_norm": 1.92423677444458,
      "learning_rate": 1.4440071287765876e-05,
      "loss": 1.3285,
      "step": 517
    },
    {
      "epoch": 0.3587257617728532,
      "grad_norm": 2.039736270904541,
      "learning_rate": 1.4420430661747245e-05,
      "loss": 1.4598,
      "step": 518
    },
    {
      "epoch": 0.35941828254847646,
      "grad_norm": 2.099224090576172,
      "learning_rate": 1.4400768819614056e-05,
      "loss": 1.3224,
      "step": 519
    },
    {
      "epoch": 0.3601108033240997,
      "grad_norm": 2.0194785594940186,
      "learning_rate": 1.4381085855734468e-05,
      "loss": 1.308,
      "step": 520
    },
    {
      "epoch": 0.360803324099723,
      "grad_norm": 1.9957362413406372,
      "learning_rate": 1.4361381864578014e-05,
      "loss": 1.3694,
      "step": 521
    },
    {
      "epoch": 0.36149584487534625,
      "grad_norm": 2.1399877071380615,
      "learning_rate": 1.4341656940715147e-05,
      "loss": 1.4561,
      "step": 522
    },
    {
      "epoch": 0.3621883656509695,
      "grad_norm": 2.405832052230835,
      "learning_rate": 1.432191117881679e-05,
      "loss": 1.3531,
      "step": 523
    },
    {
      "epoch": 0.3628808864265928,
      "grad_norm": 2.3182177543640137,
      "learning_rate": 1.4302144673653875e-05,
      "loss": 1.1962,
      "step": 524
    },
    {
      "epoch": 0.3635734072022161,
      "grad_norm": 2.29636287689209,
      "learning_rate": 1.4282357520096896e-05,
      "loss": 1.2889,
      "step": 525
    },
    {
      "epoch": 0.36426592797783935,
      "grad_norm": 2.095970630645752,
      "learning_rate": 1.426254981311545e-05,
      "loss": 1.3959,
      "step": 526
    },
    {
      "epoch": 0.3649584487534626,
      "grad_norm": 2.20633864402771,
      "learning_rate": 1.424272164777778e-05,
      "loss": 1.3896,
      "step": 527
    },
    {
      "epoch": 0.3656509695290859,
      "grad_norm": 2.5923547744750977,
      "learning_rate": 1.4222873119250325e-05,
      "loss": 1.2329,
      "step": 528
    },
    {
      "epoch": 0.36634349030470914,
      "grad_norm": 2.6066620349884033,
      "learning_rate": 1.4203004322797252e-05,
      "loss": 1.167,
      "step": 529
    },
    {
      "epoch": 0.3670360110803324,
      "grad_norm": 2.101332187652588,
      "learning_rate": 1.4183115353780001e-05,
      "loss": 1.3173,
      "step": 530
    },
    {
      "epoch": 0.36772853185595566,
      "grad_norm": 2.3599350452423096,
      "learning_rate": 1.4163206307656849e-05,
      "loss": 1.3849,
      "step": 531
    },
    {
      "epoch": 0.3684210526315789,
      "grad_norm": 2.4479289054870605,
      "learning_rate": 1.4143277279982415e-05,
      "loss": 1.3258,
      "step": 532
    },
    {
      "epoch": 0.3691135734072022,
      "grad_norm": 2.5400850772857666,
      "learning_rate": 1.4123328366407235e-05,
      "loss": 1.2734,
      "step": 533
    },
    {
      "epoch": 0.3698060941828255,
      "grad_norm": 2.1413142681121826,
      "learning_rate": 1.4103359662677276e-05,
      "loss": 1.3371,
      "step": 534
    },
    {
      "epoch": 0.37049861495844877,
      "grad_norm": 2.246459484100342,
      "learning_rate": 1.4083371264633496e-05,
      "loss": 1.3819,
      "step": 535
    },
    {
      "epoch": 0.37119113573407203,
      "grad_norm": 2.5155673027038574,
      "learning_rate": 1.406336326821138e-05,
      "loss": 1.3142,
      "step": 536
    },
    {
      "epoch": 0.3718836565096953,
      "grad_norm": 2.42018461227417,
      "learning_rate": 1.4043335769440471e-05,
      "loss": 1.5278,
      "step": 537
    },
    {
      "epoch": 0.37257617728531855,
      "grad_norm": 2.8900959491729736,
      "learning_rate": 1.4023288864443915e-05,
      "loss": 1.142,
      "step": 538
    },
    {
      "epoch": 0.3732686980609418,
      "grad_norm": 2.2948379516601562,
      "learning_rate": 1.4003222649438004e-05,
      "loss": 1.296,
      "step": 539
    },
    {
      "epoch": 0.3739612188365651,
      "grad_norm": 2.6425232887268066,
      "learning_rate": 1.3983137220731702e-05,
      "loss": 1.3842,
      "step": 540
    },
    {
      "epoch": 0.37465373961218834,
      "grad_norm": 2.036116123199463,
      "learning_rate": 1.3963032674726197e-05,
      "loss": 1.3431,
      "step": 541
    },
    {
      "epoch": 0.37534626038781166,
      "grad_norm": 2.152360200881958,
      "learning_rate": 1.3942909107914431e-05,
      "loss": 1.3037,
      "step": 542
    },
    {
      "epoch": 0.3760387811634349,
      "grad_norm": 2.285806894302368,
      "learning_rate": 1.392276661688063e-05,
      "loss": 1.2652,
      "step": 543
    },
    {
      "epoch": 0.3767313019390582,
      "grad_norm": 2.133744239807129,
      "learning_rate": 1.390260529829986e-05,
      "loss": 1.2983,
      "step": 544
    },
    {
      "epoch": 0.37742382271468145,
      "grad_norm": 2.4105823040008545,
      "learning_rate": 1.388242524893754e-05,
      "loss": 1.3611,
      "step": 545
    },
    {
      "epoch": 0.3781163434903047,
      "grad_norm": 2.275736093521118,
      "learning_rate": 1.3862226565648996e-05,
      "loss": 1.2985,
      "step": 546
    },
    {
      "epoch": 0.37880886426592797,
      "grad_norm": 2.5612637996673584,
      "learning_rate": 1.3842009345378977e-05,
      "loss": 1.3993,
      "step": 547
    },
    {
      "epoch": 0.37950138504155123,
      "grad_norm": 2.3111212253570557,
      "learning_rate": 1.3821773685161224e-05,
      "loss": 1.2687,
      "step": 548
    },
    {
      "epoch": 0.3801939058171745,
      "grad_norm": 2.2761833667755127,
      "learning_rate": 1.3801519682117957e-05,
      "loss": 1.3877,
      "step": 549
    },
    {
      "epoch": 0.3808864265927978,
      "grad_norm": 2.110255002975464,
      "learning_rate": 1.3781247433459447e-05,
      "loss": 1.2858,
      "step": 550
    },
    {
      "epoch": 0.3815789473684211,
      "grad_norm": 2.6079838275909424,
      "learning_rate": 1.3760957036483532e-05,
      "loss": 1.3634,
      "step": 551
    },
    {
      "epoch": 0.38227146814404434,
      "grad_norm": 2.3239288330078125,
      "learning_rate": 1.3740648588575156e-05,
      "loss": 1.234,
      "step": 552
    },
    {
      "epoch": 0.3829639889196676,
      "grad_norm": 2.26419997215271,
      "learning_rate": 1.3720322187205897e-05,
      "loss": 1.1981,
      "step": 553
    },
    {
      "epoch": 0.38365650969529086,
      "grad_norm": 2.4017229080200195,
      "learning_rate": 1.3699977929933503e-05,
      "loss": 1.3152,
      "step": 554
    },
    {
      "epoch": 0.3843490304709141,
      "grad_norm": 2.3292219638824463,
      "learning_rate": 1.3679615914401423e-05,
      "loss": 1.4073,
      "step": 555
    },
    {
      "epoch": 0.3850415512465374,
      "grad_norm": 2.4745140075683594,
      "learning_rate": 1.3659236238338339e-05,
      "loss": 1.1486,
      "step": 556
    },
    {
      "epoch": 0.38573407202216065,
      "grad_norm": 2.755042314529419,
      "learning_rate": 1.3638838999557687e-05,
      "loss": 1.3127,
      "step": 557
    },
    {
      "epoch": 0.3864265927977839,
      "grad_norm": 2.610299587249756,
      "learning_rate": 1.361842429595721e-05,
      "loss": 1.3348,
      "step": 558
    },
    {
      "epoch": 0.3871191135734072,
      "grad_norm": 2.1434555053710938,
      "learning_rate": 1.3597992225518466e-05,
      "loss": 1.3572,
      "step": 559
    },
    {
      "epoch": 0.3878116343490305,
      "grad_norm": 2.3972840309143066,
      "learning_rate": 1.3577542886306367e-05,
      "loss": 1.3792,
      "step": 560
    },
    {
      "epoch": 0.38850415512465375,
      "grad_norm": 2.316579818725586,
      "learning_rate": 1.355707637646871e-05,
      "loss": 1.4216,
      "step": 561
    },
    {
      "epoch": 0.389196675900277,
      "grad_norm": 2.352252721786499,
      "learning_rate": 1.3536592794235696e-05,
      "loss": 1.3232,
      "step": 562
    },
    {
      "epoch": 0.3898891966759003,
      "grad_norm": 2.1979241371154785,
      "learning_rate": 1.3516092237919479e-05,
      "loss": 1.2895,
      "step": 563
    },
    {
      "epoch": 0.39058171745152354,
      "grad_norm": 2.2471816539764404,
      "learning_rate": 1.3495574805913669e-05,
      "loss": 1.3374,
      "step": 564
    },
    {
      "epoch": 0.3912742382271468,
      "grad_norm": 2.787330150604248,
      "learning_rate": 1.3475040596692876e-05,
      "loss": 1.1684,
      "step": 565
    },
    {
      "epoch": 0.39196675900277006,
      "grad_norm": 2.628101348876953,
      "learning_rate": 1.3454489708812237e-05,
      "loss": 1.3537,
      "step": 566
    },
    {
      "epoch": 0.3926592797783934,
      "grad_norm": 2.726956605911255,
      "learning_rate": 1.3433922240906936e-05,
      "loss": 1.3914,
      "step": 567
    },
    {
      "epoch": 0.39335180055401664,
      "grad_norm": 2.6245710849761963,
      "learning_rate": 1.3413338291691726e-05,
      "loss": 1.3597,
      "step": 568
    },
    {
      "epoch": 0.3940443213296399,
      "grad_norm": 2.3888232707977295,
      "learning_rate": 1.3392737959960482e-05,
      "loss": 1.3567,
      "step": 569
    },
    {
      "epoch": 0.39473684210526316,
      "grad_norm": 2.635704755783081,
      "learning_rate": 1.3372121344585694e-05,
      "loss": 1.2514,
      "step": 570
    },
    {
      "epoch": 0.3954293628808864,
      "grad_norm": 2.6729440689086914,
      "learning_rate": 1.3351488544518003e-05,
      "loss": 1.1759,
      "step": 571
    },
    {
      "epoch": 0.3961218836565097,
      "grad_norm": 2.542128801345825,
      "learning_rate": 1.3330839658785739e-05,
      "loss": 1.27,
      "step": 572
    },
    {
      "epoch": 0.39681440443213295,
      "grad_norm": 2.387706756591797,
      "learning_rate": 1.3310174786494432e-05,
      "loss": 1.2951,
      "step": 573
    },
    {
      "epoch": 0.3975069252077562,
      "grad_norm": 2.5825939178466797,
      "learning_rate": 1.3289494026826337e-05,
      "loss": 1.3551,
      "step": 574
    },
    {
      "epoch": 0.3981994459833795,
      "grad_norm": 2.307957649230957,
      "learning_rate": 1.3268797479039977e-05,
      "loss": 1.3592,
      "step": 575
    },
    {
      "epoch": 0.3988919667590028,
      "grad_norm": 2.071687936782837,
      "learning_rate": 1.3248085242469629e-05,
      "loss": 1.28,
      "step": 576
    },
    {
      "epoch": 0.39958448753462605,
      "grad_norm": 2.5913336277008057,
      "learning_rate": 1.3227357416524878e-05,
      "loss": 1.1898,
      "step": 577
    },
    {
      "epoch": 0.4002770083102493,
      "grad_norm": 2.27807354927063,
      "learning_rate": 1.3206614100690139e-05,
      "loss": 1.3734,
      "step": 578
    },
    {
      "epoch": 0.4009695290858726,
      "grad_norm": 2.303732395172119,
      "learning_rate": 1.318585539452416e-05,
      "loss": 1.3414,
      "step": 579
    },
    {
      "epoch": 0.40166204986149584,
      "grad_norm": 2.587827682495117,
      "learning_rate": 1.3165081397659563e-05,
      "loss": 1.4122,
      "step": 580
    },
    {
      "epoch": 0.4023545706371191,
      "grad_norm": 2.292515516281128,
      "learning_rate": 1.3144292209802358e-05,
      "loss": 1.3282,
      "step": 581
    },
    {
      "epoch": 0.40304709141274236,
      "grad_norm": 2.2421929836273193,
      "learning_rate": 1.3123487930731464e-05,
      "loss": 1.4146,
      "step": 582
    },
    {
      "epoch": 0.4037396121883656,
      "grad_norm": 2.186304807662964,
      "learning_rate": 1.3102668660298229e-05,
      "loss": 1.3864,
      "step": 583
    },
    {
      "epoch": 0.40443213296398894,
      "grad_norm": 2.5028858184814453,
      "learning_rate": 1.3081834498425952e-05,
      "loss": 1.1858,
      "step": 584
    },
    {
      "epoch": 0.4051246537396122,
      "grad_norm": 2.309316635131836,
      "learning_rate": 1.3060985545109408e-05,
      "loss": 1.359,
      "step": 585
    },
    {
      "epoch": 0.40581717451523547,
      "grad_norm": 2.236069917678833,
      "learning_rate": 1.3040121900414371e-05,
      "loss": 1.3318,
      "step": 586
    },
    {
      "epoch": 0.40650969529085873,
      "grad_norm": 2.1627039909362793,
      "learning_rate": 1.3019243664477108e-05,
      "loss": 1.3227,
      "step": 587
    },
    {
      "epoch": 0.407202216066482,
      "grad_norm": 2.374979257583618,
      "learning_rate": 1.2998350937503939e-05,
      "loss": 1.2892,
      "step": 588
    },
    {
      "epoch": 0.40789473684210525,
      "grad_norm": 2.3773934841156006,
      "learning_rate": 1.2977443819770717e-05,
      "loss": 1.3139,
      "step": 589
    },
    {
      "epoch": 0.4085872576177285,
      "grad_norm": 2.3296051025390625,
      "learning_rate": 1.2956522411622377e-05,
      "loss": 1.267,
      "step": 590
    },
    {
      "epoch": 0.4092797783933518,
      "grad_norm": 2.764315605163574,
      "learning_rate": 1.2935586813472433e-05,
      "loss": 1.3891,
      "step": 591
    },
    {
      "epoch": 0.4099722991689751,
      "grad_norm": 2.6064682006835938,
      "learning_rate": 1.2914637125802514e-05,
      "loss": 1.2025,
      "step": 592
    },
    {
      "epoch": 0.41066481994459836,
      "grad_norm": 2.60998272895813,
      "learning_rate": 1.2893673449161859e-05,
      "loss": 1.3943,
      "step": 593
    },
    {
      "epoch": 0.4113573407202216,
      "grad_norm": 2.4021079540252686,
      "learning_rate": 1.287269588416686e-05,
      "loss": 1.2099,
      "step": 594
    },
    {
      "epoch": 0.4120498614958449,
      "grad_norm": 2.560509443283081,
      "learning_rate": 1.2851704531500564e-05,
      "loss": 1.3387,
      "step": 595
    },
    {
      "epoch": 0.41274238227146814,
      "grad_norm": 2.1831934452056885,
      "learning_rate": 1.2830699491912186e-05,
      "loss": 1.4038,
      "step": 596
    },
    {
      "epoch": 0.4134349030470914,
      "grad_norm": 2.457613945007324,
      "learning_rate": 1.2809680866216647e-05,
      "loss": 1.4022,
      "step": 597
    },
    {
      "epoch": 0.41412742382271467,
      "grad_norm": 2.413032054901123,
      "learning_rate": 1.2788648755294056e-05,
      "loss": 1.3812,
      "step": 598
    },
    {
      "epoch": 0.41481994459833793,
      "grad_norm": 3.021824359893799,
      "learning_rate": 1.2767603260089257e-05,
      "loss": 1.2631,
      "step": 599
    },
    {
      "epoch": 0.4155124653739612,
      "grad_norm": 2.2408480644226074,
      "learning_rate": 1.2746544481611336e-05,
      "loss": 1.3919,
      "step": 600
    },
    {
      "epoch": 0.4162049861495845,
      "grad_norm": 2.4218242168426514,
      "learning_rate": 1.2725472520933122e-05,
      "loss": 1.3804,
      "step": 601
    },
    {
      "epoch": 0.4168975069252078,
      "grad_norm": 2.305410146713257,
      "learning_rate": 1.270438747919072e-05,
      "loss": 1.3913,
      "step": 602
    },
    {
      "epoch": 0.41759002770083103,
      "grad_norm": 2.362793207168579,
      "learning_rate": 1.2683289457583018e-05,
      "loss": 1.3815,
      "step": 603
    },
    {
      "epoch": 0.4182825484764543,
      "grad_norm": 2.615732431411743,
      "learning_rate": 1.2662178557371198e-05,
      "loss": 1.449,
      "step": 604
    },
    {
      "epoch": 0.41897506925207756,
      "grad_norm": 2.481135606765747,
      "learning_rate": 1.264105487987826e-05,
      "loss": 1.2778,
      "step": 605
    },
    {
      "epoch": 0.4196675900277008,
      "grad_norm": 2.73952054977417,
      "learning_rate": 1.261991852648852e-05,
      "loss": 1.334,
      "step": 606
    },
    {
      "epoch": 0.4203601108033241,
      "grad_norm": 2.4660911560058594,
      "learning_rate": 1.2598769598647136e-05,
      "loss": 1.2903,
      "step": 607
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 2.5404601097106934,
      "learning_rate": 1.2577608197859627e-05,
      "loss": 1.274,
      "step": 608
    },
    {
      "epoch": 0.42174515235457066,
      "grad_norm": 2.2028279304504395,
      "learning_rate": 1.2556434425691363e-05,
      "loss": 1.3181,
      "step": 609
    },
    {
      "epoch": 0.4224376731301939,
      "grad_norm": 2.5085532665252686,
      "learning_rate": 1.2535248383767102e-05,
      "loss": 1.3525,
      "step": 610
    },
    {
      "epoch": 0.4231301939058172,
      "grad_norm": 2.7390897274017334,
      "learning_rate": 1.2514050173770484e-05,
      "loss": 1.394,
      "step": 611
    },
    {
      "epoch": 0.42382271468144045,
      "grad_norm": 2.349844455718994,
      "learning_rate": 1.2492839897443554e-05,
      "loss": 1.3215,
      "step": 612
    },
    {
      "epoch": 0.4245152354570637,
      "grad_norm": 2.703737735748291,
      "learning_rate": 1.247161765658627e-05,
      "loss": 1.4081,
      "step": 613
    },
    {
      "epoch": 0.425207756232687,
      "grad_norm": 2.4210963249206543,
      "learning_rate": 1.2450383553056011e-05,
      "loss": 1.182,
      "step": 614
    },
    {
      "epoch": 0.42590027700831024,
      "grad_norm": 2.9356367588043213,
      "learning_rate": 1.2429137688767095e-05,
      "loss": 1.2014,
      "step": 615
    },
    {
      "epoch": 0.4265927977839335,
      "grad_norm": 1.582685947418213,
      "learning_rate": 1.2407880165690289e-05,
      "loss": 0.7954,
      "step": 616
    },
    {
      "epoch": 0.42728531855955676,
      "grad_norm": 2.6504549980163574,
      "learning_rate": 1.238661108585231e-05,
      "loss": 1.385,
      "step": 617
    },
    {
      "epoch": 0.4279778393351801,
      "grad_norm": 2.2224066257476807,
      "learning_rate": 1.2365330551335348e-05,
      "loss": 1.3786,
      "step": 618
    },
    {
      "epoch": 0.42867036011080334,
      "grad_norm": 2.686243772506714,
      "learning_rate": 1.234403866427657e-05,
      "loss": 1.2678,
      "step": 619
    },
    {
      "epoch": 0.4293628808864266,
      "grad_norm": 2.7656173706054688,
      "learning_rate": 1.2322735526867624e-05,
      "loss": 1.2544,
      "step": 620
    },
    {
      "epoch": 0.43005540166204986,
      "grad_norm": 2.2233879566192627,
      "learning_rate": 1.2301421241354166e-05,
      "loss": 1.354,
      "step": 621
    },
    {
      "epoch": 0.4307479224376731,
      "grad_norm": 2.6432528495788574,
      "learning_rate": 1.2280095910035343e-05,
      "loss": 1.3564,
      "step": 622
    },
    {
      "epoch": 0.4314404432132964,
      "grad_norm": 2.741586685180664,
      "learning_rate": 1.2258759635263333e-05,
      "loss": 1.2194,
      "step": 623
    },
    {
      "epoch": 0.43213296398891965,
      "grad_norm": 2.5118141174316406,
      "learning_rate": 1.2237412519442828e-05,
      "loss": 1.2546,
      "step": 624
    },
    {
      "epoch": 0.4328254847645429,
      "grad_norm": 2.843817710876465,
      "learning_rate": 1.2216054665030552e-05,
      "loss": 1.2101,
      "step": 625
    },
    {
      "epoch": 0.43351800554016623,
      "grad_norm": 2.5602662563323975,
      "learning_rate": 1.2194686174534771e-05,
      "loss": 1.3588,
      "step": 626
    },
    {
      "epoch": 0.4342105263157895,
      "grad_norm": 2.5971269607543945,
      "learning_rate": 1.2173307150514809e-05,
      "loss": 1.3227,
      "step": 627
    },
    {
      "epoch": 0.43490304709141275,
      "grad_norm": 2.9508907794952393,
      "learning_rate": 1.2151917695580523e-05,
      "loss": 1.2234,
      "step": 628
    },
    {
      "epoch": 0.435595567867036,
      "grad_norm": 2.6319260597229004,
      "learning_rate": 1.2130517912391861e-05,
      "loss": 1.36,
      "step": 629
    },
    {
      "epoch": 0.4362880886426593,
      "grad_norm": 2.494126081466675,
      "learning_rate": 1.2109107903658326e-05,
      "loss": 1.3959,
      "step": 630
    },
    {
      "epoch": 0.43698060941828254,
      "grad_norm": 3.109811782836914,
      "learning_rate": 1.2087687772138501e-05,
      "loss": 1.0508,
      "step": 631
    },
    {
      "epoch": 0.4376731301939058,
      "grad_norm": 2.5065367221832275,
      "learning_rate": 1.2066257620639557e-05,
      "loss": 1.3455,
      "step": 632
    },
    {
      "epoch": 0.43836565096952906,
      "grad_norm": 2.493720769882202,
      "learning_rate": 1.2044817552016752e-05,
      "loss": 1.3614,
      "step": 633
    },
    {
      "epoch": 0.4390581717451524,
      "grad_norm": 2.4560728073120117,
      "learning_rate": 1.2023367669172947e-05,
      "loss": 1.4117,
      "step": 634
    },
    {
      "epoch": 0.43975069252077564,
      "grad_norm": 2.2620697021484375,
      "learning_rate": 1.2001908075058108e-05,
      "loss": 1.3028,
      "step": 635
    },
    {
      "epoch": 0.4404432132963989,
      "grad_norm": 2.5376715660095215,
      "learning_rate": 1.1980438872668803e-05,
      "loss": 1.1942,
      "step": 636
    },
    {
      "epoch": 0.44113573407202217,
      "grad_norm": 1.7369788885116577,
      "learning_rate": 1.1958960165047716e-05,
      "loss": 0.7623,
      "step": 637
    },
    {
      "epoch": 0.44182825484764543,
      "grad_norm": 2.8980188369750977,
      "learning_rate": 1.1937472055283168e-05,
      "loss": 1.3815,
      "step": 638
    },
    {
      "epoch": 0.4425207756232687,
      "grad_norm": 2.5378687381744385,
      "learning_rate": 1.1915974646508581e-05,
      "loss": 1.457,
      "step": 639
    },
    {
      "epoch": 0.44321329639889195,
      "grad_norm": 2.880448818206787,
      "learning_rate": 1.189446804190203e-05,
      "loss": 1.0968,
      "step": 640
    },
    {
      "epoch": 0.4439058171745152,
      "grad_norm": 2.571089506149292,
      "learning_rate": 1.1872952344685713e-05,
      "loss": 1.2438,
      "step": 641
    },
    {
      "epoch": 0.4445983379501385,
      "grad_norm": 2.754330635070801,
      "learning_rate": 1.1851427658125474e-05,
      "loss": 1.2729,
      "step": 642
    },
    {
      "epoch": 0.4452908587257618,
      "grad_norm": 2.6155052185058594,
      "learning_rate": 1.1829894085530297e-05,
      "loss": 1.2282,
      "step": 643
    },
    {
      "epoch": 0.44598337950138506,
      "grad_norm": 2.518204689025879,
      "learning_rate": 1.180835173025183e-05,
      "loss": 1.3603,
      "step": 644
    },
    {
      "epoch": 0.4466759002770083,
      "grad_norm": 2.3861498832702637,
      "learning_rate": 1.1786800695683847e-05,
      "loss": 1.3179,
      "step": 645
    },
    {
      "epoch": 0.4473684210526316,
      "grad_norm": 2.5852103233337402,
      "learning_rate": 1.1765241085261802e-05,
      "loss": 1.3163,
      "step": 646
    },
    {
      "epoch": 0.44806094182825484,
      "grad_norm": 2.024843215942383,
      "learning_rate": 1.1743673002462303e-05,
      "loss": 0.7131,
      "step": 647
    },
    {
      "epoch": 0.4487534626038781,
      "grad_norm": 2.430532932281494,
      "learning_rate": 1.172209655080262e-05,
      "loss": 1.2961,
      "step": 648
    },
    {
      "epoch": 0.44944598337950137,
      "grad_norm": 2.568700075149536,
      "learning_rate": 1.1700511833840186e-05,
      "loss": 1.3669,
      "step": 649
    },
    {
      "epoch": 0.45013850415512463,
      "grad_norm": 2.587846517562866,
      "learning_rate": 1.1678918955172112e-05,
      "loss": 1.2835,
      "step": 650
    },
    {
      "epoch": 0.45083102493074795,
      "grad_norm": 3.3329403400421143,
      "learning_rate": 1.1657318018434673e-05,
      "loss": 1.1834,
      "step": 651
    },
    {
      "epoch": 0.4515235457063712,
      "grad_norm": 3.202256917953491,
      "learning_rate": 1.163570912730283e-05,
      "loss": 1.3079,
      "step": 652
    },
    {
      "epoch": 0.45221606648199447,
      "grad_norm": 2.8685662746429443,
      "learning_rate": 1.1614092385489708e-05,
      "loss": 1.4066,
      "step": 653
    },
    {
      "epoch": 0.45290858725761773,
      "grad_norm": 2.6699678897857666,
      "learning_rate": 1.1592467896746122e-05,
      "loss": 1.302,
      "step": 654
    },
    {
      "epoch": 0.453601108033241,
      "grad_norm": 2.71836519241333,
      "learning_rate": 1.157083576486007e-05,
      "loss": 1.3545,
      "step": 655
    },
    {
      "epoch": 0.45429362880886426,
      "grad_norm": 2.4176559448242188,
      "learning_rate": 1.1549196093656223e-05,
      "loss": 1.2774,
      "step": 656
    },
    {
      "epoch": 0.4549861495844875,
      "grad_norm": 2.4645159244537354,
      "learning_rate": 1.1527548986995449e-05,
      "loss": 1.3403,
      "step": 657
    },
    {
      "epoch": 0.4556786703601108,
      "grad_norm": 2.757352352142334,
      "learning_rate": 1.1505894548774294e-05,
      "loss": 1.3216,
      "step": 658
    },
    {
      "epoch": 0.45637119113573404,
      "grad_norm": 2.649862051010132,
      "learning_rate": 1.1484232882924501e-05,
      "loss": 1.2886,
      "step": 659
    },
    {
      "epoch": 0.45706371191135736,
      "grad_norm": 2.594386100769043,
      "learning_rate": 1.1462564093412493e-05,
      "loss": 1.2525,
      "step": 660
    },
    {
      "epoch": 0.4577562326869806,
      "grad_norm": 2.5127627849578857,
      "learning_rate": 1.144088828423889e-05,
      "loss": 1.25,
      "step": 661
    },
    {
      "epoch": 0.4584487534626039,
      "grad_norm": 2.3333370685577393,
      "learning_rate": 1.1419205559437998e-05,
      "loss": 1.3152,
      "step": 662
    },
    {
      "epoch": 0.45914127423822715,
      "grad_norm": 2.4757964611053467,
      "learning_rate": 1.1397516023077326e-05,
      "loss": 1.3548,
      "step": 663
    },
    {
      "epoch": 0.4598337950138504,
      "grad_norm": 2.6083545684814453,
      "learning_rate": 1.1375819779257058e-05,
      "loss": 1.5082,
      "step": 664
    },
    {
      "epoch": 0.4605263157894737,
      "grad_norm": 2.9225447177886963,
      "learning_rate": 1.1354116932109586e-05,
      "loss": 1.3173,
      "step": 665
    },
    {
      "epoch": 0.46121883656509693,
      "grad_norm": 2.9521665573120117,
      "learning_rate": 1.1332407585798992e-05,
      "loss": 1.3909,
      "step": 666
    },
    {
      "epoch": 0.4619113573407202,
      "grad_norm": 2.658435583114624,
      "learning_rate": 1.1310691844520543e-05,
      "loss": 1.3327,
      "step": 667
    },
    {
      "epoch": 0.4626038781163435,
      "grad_norm": 2.4518954753875732,
      "learning_rate": 1.1288969812500209e-05,
      "loss": 1.3068,
      "step": 668
    },
    {
      "epoch": 0.4632963988919668,
      "grad_norm": 3.0496392250061035,
      "learning_rate": 1.1267241593994145e-05,
      "loss": 1.2705,
      "step": 669
    },
    {
      "epoch": 0.46398891966759004,
      "grad_norm": 2.8972740173339844,
      "learning_rate": 1.1245507293288204e-05,
      "loss": 1.2617,
      "step": 670
    },
    {
      "epoch": 0.4646814404432133,
      "grad_norm": 2.567683219909668,
      "learning_rate": 1.1223767014697428e-05,
      "loss": 1.2594,
      "step": 671
    },
    {
      "epoch": 0.46537396121883656,
      "grad_norm": 2.789400577545166,
      "learning_rate": 1.1202020862565555e-05,
      "loss": 1.4519,
      "step": 672
    },
    {
      "epoch": 0.4660664819944598,
      "grad_norm": 2.6831300258636475,
      "learning_rate": 1.11802689412645e-05,
      "loss": 1.2585,
      "step": 673
    },
    {
      "epoch": 0.4667590027700831,
      "grad_norm": 2.578671455383301,
      "learning_rate": 1.1158511355193888e-05,
      "loss": 1.2332,
      "step": 674
    },
    {
      "epoch": 0.46745152354570635,
      "grad_norm": 2.688199996948242,
      "learning_rate": 1.1136748208780512e-05,
      "loss": 1.318,
      "step": 675
    },
    {
      "epoch": 0.46814404432132967,
      "grad_norm": 2.3270740509033203,
      "learning_rate": 1.1114979606477867e-05,
      "loss": 1.3231,
      "step": 676
    },
    {
      "epoch": 0.46883656509695293,
      "grad_norm": 2.6363329887390137,
      "learning_rate": 1.1093205652765627e-05,
      "loss": 1.2892,
      "step": 677
    },
    {
      "epoch": 0.4695290858725762,
      "grad_norm": 3.152540445327759,
      "learning_rate": 1.1071426452149152e-05,
      "loss": 1.2353,
      "step": 678
    },
    {
      "epoch": 0.47022160664819945,
      "grad_norm": 2.9541869163513184,
      "learning_rate": 1.1049642109158982e-05,
      "loss": 1.4063,
      "step": 679
    },
    {
      "epoch": 0.4709141274238227,
      "grad_norm": 2.647003412246704,
      "learning_rate": 1.1027852728350343e-05,
      "loss": 1.2904,
      "step": 680
    },
    {
      "epoch": 0.471606648199446,
      "grad_norm": 2.9237818717956543,
      "learning_rate": 1.1006058414302637e-05,
      "loss": 1.159,
      "step": 681
    },
    {
      "epoch": 0.47229916897506924,
      "grad_norm": 3.276366710662842,
      "learning_rate": 1.0984259271618947e-05,
      "loss": 1.2979,
      "step": 682
    },
    {
      "epoch": 0.4729916897506925,
      "grad_norm": 3.14654541015625,
      "learning_rate": 1.0962455404925524e-05,
      "loss": 1.2838,
      "step": 683
    },
    {
      "epoch": 0.47368421052631576,
      "grad_norm": 3.0861613750457764,
      "learning_rate": 1.09406469188713e-05,
      "loss": 1.2665,
      "step": 684
    },
    {
      "epoch": 0.4743767313019391,
      "grad_norm": 2.8374383449554443,
      "learning_rate": 1.0918833918127377e-05,
      "loss": 1.3168,
      "step": 685
    },
    {
      "epoch": 0.47506925207756234,
      "grad_norm": 3.03313946723938,
      "learning_rate": 1.0897016507386513e-05,
      "loss": 1.3287,
      "step": 686
    },
    {
      "epoch": 0.4757617728531856,
      "grad_norm": 2.811178207397461,
      "learning_rate": 1.0875194791362656e-05,
      "loss": 1.1427,
      "step": 687
    },
    {
      "epoch": 0.47645429362880887,
      "grad_norm": 2.7737855911254883,
      "learning_rate": 1.0853368874790392e-05,
      "loss": 1.3189,
      "step": 688
    },
    {
      "epoch": 0.47714681440443213,
      "grad_norm": 2.610959053039551,
      "learning_rate": 1.0831538862424488e-05,
      "loss": 1.3865,
      "step": 689
    },
    {
      "epoch": 0.4778393351800554,
      "grad_norm": 1.9542778730392456,
      "learning_rate": 1.0809704859039357e-05,
      "loss": 0.8104,
      "step": 690
    },
    {
      "epoch": 0.47853185595567865,
      "grad_norm": 2.8905603885650635,
      "learning_rate": 1.0787866969428569e-05,
      "loss": 1.3594,
      "step": 691
    },
    {
      "epoch": 0.4792243767313019,
      "grad_norm": 3.1295323371887207,
      "learning_rate": 1.0766025298404346e-05,
      "loss": 1.3643,
      "step": 692
    },
    {
      "epoch": 0.47991689750692523,
      "grad_norm": 2.8047592639923096,
      "learning_rate": 1.0744179950797069e-05,
      "loss": 1.2713,
      "step": 693
    },
    {
      "epoch": 0.4806094182825485,
      "grad_norm": 2.5008580684661865,
      "learning_rate": 1.0722331031454749e-05,
      "loss": 1.3686,
      "step": 694
    },
    {
      "epoch": 0.48130193905817176,
      "grad_norm": 3.0125064849853516,
      "learning_rate": 1.0700478645242552e-05,
      "loss": 1.2412,
      "step": 695
    },
    {
      "epoch": 0.481994459833795,
      "grad_norm": 2.661039352416992,
      "learning_rate": 1.0678622897042279e-05,
      "loss": 1.3159,
      "step": 696
    },
    {
      "epoch": 0.4826869806094183,
      "grad_norm": 2.9381444454193115,
      "learning_rate": 1.0656763891751865e-05,
      "loss": 1.341,
      "step": 697
    },
    {
      "epoch": 0.48337950138504154,
      "grad_norm": 2.9836440086364746,
      "learning_rate": 1.063490173428488e-05,
      "loss": 1.3917,
      "step": 698
    },
    {
      "epoch": 0.4840720221606648,
      "grad_norm": 3.4271857738494873,
      "learning_rate": 1.0613036529570022e-05,
      "loss": 1.1885,
      "step": 699
    },
    {
      "epoch": 0.48476454293628807,
      "grad_norm": 2.6977245807647705,
      "learning_rate": 1.0591168382550617e-05,
      "loss": 1.453,
      "step": 700
    },
    {
      "epoch": 0.48545706371191133,
      "grad_norm": 3.1043028831481934,
      "learning_rate": 1.0569297398184111e-05,
      "loss": 1.2817,
      "step": 701
    },
    {
      "epoch": 0.48614958448753465,
      "grad_norm": 2.9746053218841553,
      "learning_rate": 1.0547423681441567e-05,
      "loss": 1.2756,
      "step": 702
    },
    {
      "epoch": 0.4868421052631579,
      "grad_norm": 2.829667091369629,
      "learning_rate": 1.052554733730716e-05,
      "loss": 1.3638,
      "step": 703
    },
    {
      "epoch": 0.48753462603878117,
      "grad_norm": 3.197615385055542,
      "learning_rate": 1.050366847077769e-05,
      "loss": 1.2731,
      "step": 704
    },
    {
      "epoch": 0.48822714681440443,
      "grad_norm": 2.7651829719543457,
      "learning_rate": 1.0481787186862036e-05,
      "loss": 1.1986,
      "step": 705
    },
    {
      "epoch": 0.4889196675900277,
      "grad_norm": 2.8822498321533203,
      "learning_rate": 1.0459903590580706e-05,
      "loss": 1.1886,
      "step": 706
    },
    {
      "epoch": 0.48961218836565096,
      "grad_norm": 2.9656972885131836,
      "learning_rate": 1.0438017786965295e-05,
      "loss": 1.2598,
      "step": 707
    },
    {
      "epoch": 0.4903047091412742,
      "grad_norm": 3.422245979309082,
      "learning_rate": 1.0416129881057987e-05,
      "loss": 0.9257,
      "step": 708
    },
    {
      "epoch": 0.4909972299168975,
      "grad_norm": 1.814936637878418,
      "learning_rate": 1.0394239977911069e-05,
      "loss": 0.733,
      "step": 709
    },
    {
      "epoch": 0.4916897506925208,
      "grad_norm": 3.0122458934783936,
      "learning_rate": 1.03723481825864e-05,
      "loss": 1.2547,
      "step": 710
    },
    {
      "epoch": 0.49238227146814406,
      "grad_norm": 3.0159964561462402,
      "learning_rate": 1.0350454600154932e-05,
      "loss": 1.2414,
      "step": 711
    },
    {
      "epoch": 0.4930747922437673,
      "grad_norm": 2.8420827388763428,
      "learning_rate": 1.0328559335696188e-05,
      "loss": 1.1437,
      "step": 712
    },
    {
      "epoch": 0.4937673130193906,
      "grad_norm": 3.323718547821045,
      "learning_rate": 1.0306662494297766e-05,
      "loss": 1.1508,
      "step": 713
    },
    {
      "epoch": 0.49445983379501385,
      "grad_norm": 2.852436065673828,
      "learning_rate": 1.028476418105483e-05,
      "loss": 1.2692,
      "step": 714
    },
    {
      "epoch": 0.4951523545706371,
      "grad_norm": 3.037775993347168,
      "learning_rate": 1.0262864501069619e-05,
      "loss": 1.2536,
      "step": 715
    },
    {
      "epoch": 0.49584487534626037,
      "grad_norm": 3.1405160427093506,
      "learning_rate": 1.0240963559450909e-05,
      "loss": 1.2877,
      "step": 716
    },
    {
      "epoch": 0.49653739612188363,
      "grad_norm": 2.7466113567352295,
      "learning_rate": 1.0219061461313557e-05,
      "loss": 1.2839,
      "step": 717
    },
    {
      "epoch": 0.49722991689750695,
      "grad_norm": 2.971862554550171,
      "learning_rate": 1.0197158311777957e-05,
      "loss": 1.4034,
      "step": 718
    },
    {
      "epoch": 0.4979224376731302,
      "grad_norm": 2.713968515396118,
      "learning_rate": 1.017525421596955e-05,
      "loss": 1.3094,
      "step": 719
    },
    {
      "epoch": 0.4986149584487535,
      "grad_norm": 2.7105119228363037,
      "learning_rate": 1.015334927901832e-05,
      "loss": 1.2357,
      "step": 720
    },
    {
      "epoch": 0.49930747922437674,
      "grad_norm": 2.978726387023926,
      "learning_rate": 1.0131443606058291e-05,
      "loss": 1.3151,
      "step": 721
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.9404547214508057,
      "learning_rate": 1.0109537302227012e-05,
      "loss": 1.1534,
      "step": 722
    },
    {
      "epoch": 0.5,
      "eval_loss": 1.0960729122161865,
      "eval_runtime": 338.4976,
      "eval_samples_per_second": 4.183,
      "eval_steps_per_second": 0.523,
      "step": 722
    },
    {
      "epoch": 0.5006925207756233,
      "grad_norm": 3.121488571166992,
      "learning_rate": 1.0087630472665071e-05,
      "loss": 1.3357,
      "step": 723
    },
    {
      "epoch": 0.5013850415512465,
      "grad_norm": 2.7625746726989746,
      "learning_rate": 1.0065723222515566e-05,
      "loss": 1.3157,
      "step": 724
    },
    {
      "epoch": 0.5020775623268698,
      "grad_norm": 2.6933093070983887,
      "learning_rate": 1.0043815656923627e-05,
      "loss": 1.2687,
      "step": 725
    },
    {
      "epoch": 0.502770083102493,
      "grad_norm": 3.369460344314575,
      "learning_rate": 1.0021907881035891e-05,
      "loss": 1.1404,
      "step": 726
    },
    {
      "epoch": 0.5034626038781164,
      "grad_norm": 3.114061117172241,
      "learning_rate": 1e-05,
      "loss": 1.3558,
      "step": 727
    },
    {
      "epoch": 0.5041551246537396,
      "grad_norm": 3.136636972427368,
      "learning_rate": 9.97809211896411e-06,
      "loss": 1.2918,
      "step": 728
    },
    {
      "epoch": 0.5048476454293629,
      "grad_norm": 2.9409475326538086,
      "learning_rate": 9.956184343076378e-06,
      "loss": 1.2745,
      "step": 729
    },
    {
      "epoch": 0.5055401662049861,
      "grad_norm": 3.3385579586029053,
      "learning_rate": 9.934276777484436e-06,
      "loss": 1.3101,
      "step": 730
    },
    {
      "epoch": 0.5062326869806094,
      "grad_norm": 3.108218193054199,
      "learning_rate": 9.91236952733493e-06,
      "loss": 1.3295,
      "step": 731
    },
    {
      "epoch": 0.5069252077562327,
      "grad_norm": 2.7342264652252197,
      "learning_rate": 9.89046269777299e-06,
      "loss": 1.4148,
      "step": 732
    },
    {
      "epoch": 0.5076177285318559,
      "grad_norm": 2.875964879989624,
      "learning_rate": 9.868556393941714e-06,
      "loss": 1.1803,
      "step": 733
    },
    {
      "epoch": 0.5083102493074793,
      "grad_norm": 3.1713013648986816,
      "learning_rate": 9.846650720981682e-06,
      "loss": 1.368,
      "step": 734
    },
    {
      "epoch": 0.5090027700831025,
      "grad_norm": 3.592292070388794,
      "learning_rate": 9.824745784030451e-06,
      "loss": 1.1784,
      "step": 735
    },
    {
      "epoch": 0.5096952908587258,
      "grad_norm": 2.91928768157959,
      "learning_rate": 9.802841688222043e-06,
      "loss": 1.3635,
      "step": 736
    },
    {
      "epoch": 0.510387811634349,
      "grad_norm": 3.0569591522216797,
      "learning_rate": 9.780938538686444e-06,
      "loss": 1.185,
      "step": 737
    },
    {
      "epoch": 0.5110803324099723,
      "grad_norm": 3.4262917041778564,
      "learning_rate": 9.759036440549093e-06,
      "loss": 1.2004,
      "step": 738
    },
    {
      "epoch": 0.5117728531855956,
      "grad_norm": 2.7363193035125732,
      "learning_rate": 9.737135498930385e-06,
      "loss": 1.1571,
      "step": 739
    },
    {
      "epoch": 0.5124653739612188,
      "grad_norm": 3.3447537422180176,
      "learning_rate": 9.715235818945171e-06,
      "loss": 1.309,
      "step": 740
    },
    {
      "epoch": 0.5131578947368421,
      "grad_norm": 1.9384816884994507,
      "learning_rate": 9.693337505702238e-06,
      "loss": 0.8006,
      "step": 741
    },
    {
      "epoch": 0.5138504155124654,
      "grad_norm": 3.3189609050750732,
      "learning_rate": 9.671440664303813e-06,
      "loss": 1.1473,
      "step": 742
    },
    {
      "epoch": 0.5145429362880887,
      "grad_norm": 3.228372097015381,
      "learning_rate": 9.64954539984507e-06,
      "loss": 1.2468,
      "step": 743
    },
    {
      "epoch": 0.5152354570637119,
      "grad_norm": 2.7112698554992676,
      "learning_rate": 9.627651817413605e-06,
      "loss": 1.394,
      "step": 744
    },
    {
      "epoch": 0.5159279778393352,
      "grad_norm": 3.023843288421631,
      "learning_rate": 9.605760022088935e-06,
      "loss": 1.3856,
      "step": 745
    },
    {
      "epoch": 0.5166204986149584,
      "grad_norm": 2.7612435817718506,
      "learning_rate": 9.583870118942014e-06,
      "loss": 1.2712,
      "step": 746
    },
    {
      "epoch": 0.5173130193905817,
      "grad_norm": 2.827341318130493,
      "learning_rate": 9.561982213034707e-06,
      "loss": 1.2756,
      "step": 747
    },
    {
      "epoch": 0.518005540166205,
      "grad_norm": 2.7056517601013184,
      "learning_rate": 9.540096409419295e-06,
      "loss": 1.4065,
      "step": 748
    },
    {
      "epoch": 0.5186980609418282,
      "grad_norm": 2.6263115406036377,
      "learning_rate": 9.518212813137966e-06,
      "loss": 1.282,
      "step": 749
    },
    {
      "epoch": 0.5193905817174516,
      "grad_norm": 3.2539684772491455,
      "learning_rate": 9.496331529222313e-06,
      "loss": 1.4131,
      "step": 750
    },
    {
      "epoch": 0.5200831024930748,
      "grad_norm": 2.7921719551086426,
      "learning_rate": 9.47445266269284e-06,
      "loss": 1.25,
      "step": 751
    },
    {
      "epoch": 0.5207756232686981,
      "grad_norm": 2.722177267074585,
      "learning_rate": 9.452576318558437e-06,
      "loss": 1.2422,
      "step": 752
    },
    {
      "epoch": 0.5214681440443213,
      "grad_norm": 3.5722434520721436,
      "learning_rate": 9.430702601815892e-06,
      "loss": 1.356,
      "step": 753
    },
    {
      "epoch": 0.5221606648199446,
      "grad_norm": 2.7773613929748535,
      "learning_rate": 9.408831617449385e-06,
      "loss": 1.2914,
      "step": 754
    },
    {
      "epoch": 0.5228531855955678,
      "grad_norm": 4.336549758911133,
      "learning_rate": 9.38696347042998e-06,
      "loss": 1.1878,
      "step": 755
    },
    {
      "epoch": 0.5235457063711911,
      "grad_norm": 2.752716541290283,
      "learning_rate": 9.365098265715124e-06,
      "loss": 1.2605,
      "step": 756
    },
    {
      "epoch": 0.5242382271468145,
      "grad_norm": 2.604163885116577,
      "learning_rate": 9.34323610824814e-06,
      "loss": 1.3817,
      "step": 757
    },
    {
      "epoch": 0.5249307479224377,
      "grad_norm": 3.2916760444641113,
      "learning_rate": 9.321377102957723e-06,
      "loss": 1.1103,
      "step": 758
    },
    {
      "epoch": 0.525623268698061,
      "grad_norm": 3.1670544147491455,
      "learning_rate": 9.29952135475745e-06,
      "loss": 1.328,
      "step": 759
    },
    {
      "epoch": 0.5263157894736842,
      "grad_norm": 2.944361925125122,
      "learning_rate": 9.277668968545253e-06,
      "loss": 1.3703,
      "step": 760
    },
    {
      "epoch": 0.5270083102493075,
      "grad_norm": 2.8937935829162598,
      "learning_rate": 9.255820049202933e-06,
      "loss": 1.3303,
      "step": 761
    },
    {
      "epoch": 0.5277008310249307,
      "grad_norm": 3.042147636413574,
      "learning_rate": 9.233974701595654e-06,
      "loss": 1.2626,
      "step": 762
    },
    {
      "epoch": 0.528393351800554,
      "grad_norm": 3.470090627670288,
      "learning_rate": 9.212133030571438e-06,
      "loss": 1.4438,
      "step": 763
    },
    {
      "epoch": 0.5290858725761773,
      "grad_norm": 2.5336360931396484,
      "learning_rate": 9.190295140960649e-06,
      "loss": 1.2959,
      "step": 764
    },
    {
      "epoch": 0.5297783933518005,
      "grad_norm": 2.6885414123535156,
      "learning_rate": 9.168461137575515e-06,
      "loss": 1.2422,
      "step": 765
    },
    {
      "epoch": 0.5304709141274239,
      "grad_norm": 3.321873664855957,
      "learning_rate": 9.146631125209608e-06,
      "loss": 1.1752,
      "step": 766
    },
    {
      "epoch": 0.5311634349030471,
      "grad_norm": 2.9666600227355957,
      "learning_rate": 9.124805208637349e-06,
      "loss": 1.3353,
      "step": 767
    },
    {
      "epoch": 0.5318559556786704,
      "grad_norm": 2.8593125343322754,
      "learning_rate": 9.102983492613489e-06,
      "loss": 1.2074,
      "step": 768
    },
    {
      "epoch": 0.5325484764542936,
      "grad_norm": 2.73651385307312,
      "learning_rate": 9.081166081872627e-06,
      "loss": 1.2614,
      "step": 769
    },
    {
      "epoch": 0.5332409972299169,
      "grad_norm": 2.727332592010498,
      "learning_rate": 9.059353081128702e-06,
      "loss": 1.2188,
      "step": 770
    },
    {
      "epoch": 0.5339335180055401,
      "grad_norm": 2.6417131423950195,
      "learning_rate": 9.037544595074479e-06,
      "loss": 1.2823,
      "step": 771
    },
    {
      "epoch": 0.5346260387811634,
      "grad_norm": 2.800295352935791,
      "learning_rate": 9.015740728381055e-06,
      "loss": 1.2197,
      "step": 772
    },
    {
      "epoch": 0.5353185595567868,
      "grad_norm": 3.024468421936035,
      "learning_rate": 8.993941585697363e-06,
      "loss": 1.2803,
      "step": 773
    },
    {
      "epoch": 0.53601108033241,
      "grad_norm": 3.050037384033203,
      "learning_rate": 8.972147271649662e-06,
      "loss": 1.2804,
      "step": 774
    },
    {
      "epoch": 0.5367036011080333,
      "grad_norm": 3.3923215866088867,
      "learning_rate": 8.950357890841021e-06,
      "loss": 1.3596,
      "step": 775
    },
    {
      "epoch": 0.5373961218836565,
      "grad_norm": 2.8710789680480957,
      "learning_rate": 8.928573547850852e-06,
      "loss": 1.3116,
      "step": 776
    },
    {
      "epoch": 0.5380886426592798,
      "grad_norm": 2.874081611633301,
      "learning_rate": 8.906794347234373e-06,
      "loss": 1.3111,
      "step": 777
    },
    {
      "epoch": 0.538781163434903,
      "grad_norm": 2.6735353469848633,
      "learning_rate": 8.885020393522136e-06,
      "loss": 1.2969,
      "step": 778
    },
    {
      "epoch": 0.5394736842105263,
      "grad_norm": 3.593169689178467,
      "learning_rate": 8.863251791219491e-06,
      "loss": 1.2788,
      "step": 779
    },
    {
      "epoch": 0.5401662049861495,
      "grad_norm": 3.3737025260925293,
      "learning_rate": 8.841488644806115e-06,
      "loss": 1.2895,
      "step": 780
    },
    {
      "epoch": 0.5408587257617729,
      "grad_norm": 2.32466721534729,
      "learning_rate": 8.819731058735502e-06,
      "loss": 0.7021,
      "step": 781
    },
    {
      "epoch": 0.5415512465373962,
      "grad_norm": 3.052380323410034,
      "learning_rate": 8.797979137434452e-06,
      "loss": 1.1673,
      "step": 782
    },
    {
      "epoch": 0.5422437673130194,
      "grad_norm": 3.2256016731262207,
      "learning_rate": 8.776232985302573e-06,
      "loss": 1.282,
      "step": 783
    },
    {
      "epoch": 0.5429362880886427,
      "grad_norm": 3.534031629562378,
      "learning_rate": 8.754492706711798e-06,
      "loss": 1.2677,
      "step": 784
    },
    {
      "epoch": 0.5436288088642659,
      "grad_norm": 3.229583263397217,
      "learning_rate": 8.732758406005855e-06,
      "loss": 1.2163,
      "step": 785
    },
    {
      "epoch": 0.5443213296398892,
      "grad_norm": 2.711229085922241,
      "learning_rate": 8.711030187499795e-06,
      "loss": 1.2238,
      "step": 786
    },
    {
      "epoch": 0.5450138504155124,
      "grad_norm": 3.2641870975494385,
      "learning_rate": 8.68930815547946e-06,
      "loss": 1.2413,
      "step": 787
    },
    {
      "epoch": 0.5457063711911357,
      "grad_norm": 3.3665711879730225,
      "learning_rate": 8.66759241420101e-06,
      "loss": 1.3301,
      "step": 788
    },
    {
      "epoch": 0.546398891966759,
      "grad_norm": 3.5201163291931152,
      "learning_rate": 8.645883067890416e-06,
      "loss": 1.4187,
      "step": 789
    },
    {
      "epoch": 0.5470914127423823,
      "grad_norm": 3.2204978466033936,
      "learning_rate": 8.624180220742945e-06,
      "loss": 1.0933,
      "step": 790
    },
    {
      "epoch": 0.5477839335180056,
      "grad_norm": 3.3322551250457764,
      "learning_rate": 8.602483976922678e-06,
      "loss": 1.2983,
      "step": 791
    },
    {
      "epoch": 0.5484764542936288,
      "grad_norm": 2.787132740020752,
      "learning_rate": 8.580794440562003e-06,
      "loss": 1.3391,
      "step": 792
    },
    {
      "epoch": 0.5491689750692521,
      "grad_norm": 2.921457290649414,
      "learning_rate": 8.559111715761115e-06,
      "loss": 1.3271,
      "step": 793
    },
    {
      "epoch": 0.5498614958448753,
      "grad_norm": 2.8040547370910645,
      "learning_rate": 8.53743590658751e-06,
      "loss": 1.3346,
      "step": 794
    },
    {
      "epoch": 0.5505540166204986,
      "grad_norm": 3.099111557006836,
      "learning_rate": 8.515767117075502e-06,
      "loss": 1.3953,
      "step": 795
    },
    {
      "epoch": 0.5512465373961218,
      "grad_norm": 3.060025453567505,
      "learning_rate": 8.494105451225706e-06,
      "loss": 1.3549,
      "step": 796
    },
    {
      "epoch": 0.5519390581717452,
      "grad_norm": 2.8175106048583984,
      "learning_rate": 8.472451013004554e-06,
      "loss": 1.386,
      "step": 797
    },
    {
      "epoch": 0.5526315789473685,
      "grad_norm": 3.159795045852661,
      "learning_rate": 8.45080390634378e-06,
      "loss": 1.2336,
      "step": 798
    },
    {
      "epoch": 0.5533240997229917,
      "grad_norm": 3.1138765811920166,
      "learning_rate": 8.429164235139931e-06,
      "loss": 1.3276,
      "step": 799
    },
    {
      "epoch": 0.554016620498615,
      "grad_norm": 3.215301513671875,
      "learning_rate": 8.407532103253878e-06,
      "loss": 1.1742,
      "step": 800
    },
    {
      "epoch": 0.5547091412742382,
      "grad_norm": 3.050508737564087,
      "learning_rate": 8.385907614510297e-06,
      "loss": 1.33,
      "step": 801
    },
    {
      "epoch": 0.5554016620498615,
      "grad_norm": 3.1976349353790283,
      "learning_rate": 8.364290872697175e-06,
      "loss": 1.3126,
      "step": 802
    },
    {
      "epoch": 0.5560941828254847,
      "grad_norm": 2.6836695671081543,
      "learning_rate": 8.342681981565329e-06,
      "loss": 1.3309,
      "step": 803
    },
    {
      "epoch": 0.556786703601108,
      "grad_norm": 3.17729115486145,
      "learning_rate": 8.321081044827894e-06,
      "loss": 1.3075,
      "step": 804
    },
    {
      "epoch": 0.5574792243767313,
      "grad_norm": 3.135615587234497,
      "learning_rate": 8.299488166159817e-06,
      "loss": 1.2755,
      "step": 805
    },
    {
      "epoch": 0.5581717451523546,
      "grad_norm": 3.6281352043151855,
      "learning_rate": 8.277903449197383e-06,
      "loss": 1.1755,
      "step": 806
    },
    {
      "epoch": 0.5588642659279779,
      "grad_norm": 3.361931085586548,
      "learning_rate": 8.256326997537695e-06,
      "loss": 1.4074,
      "step": 807
    },
    {
      "epoch": 0.5595567867036011,
      "grad_norm": 2.785465717315674,
      "learning_rate": 8.2347589147382e-06,
      "loss": 1.2723,
      "step": 808
    },
    {
      "epoch": 0.5602493074792244,
      "grad_norm": 3.313589572906494,
      "learning_rate": 8.213199304316156e-06,
      "loss": 1.1785,
      "step": 809
    },
    {
      "epoch": 0.5609418282548476,
      "grad_norm": 3.2528610229492188,
      "learning_rate": 8.191648269748173e-06,
      "loss": 1.1545,
      "step": 810
    },
    {
      "epoch": 0.5616343490304709,
      "grad_norm": 3.762922763824463,
      "learning_rate": 8.170105914469701e-06,
      "loss": 1.3754,
      "step": 811
    },
    {
      "epoch": 0.5623268698060941,
      "grad_norm": 3.4045610427856445,
      "learning_rate": 8.14857234187453e-06,
      "loss": 1.173,
      "step": 812
    },
    {
      "epoch": 0.5630193905817175,
      "grad_norm": 3.2984979152679443,
      "learning_rate": 8.127047655314289e-06,
      "loss": 1.2817,
      "step": 813
    },
    {
      "epoch": 0.5637119113573407,
      "grad_norm": 3.1819562911987305,
      "learning_rate": 8.105531958097973e-06,
      "loss": 1.1583,
      "step": 814
    },
    {
      "epoch": 0.564404432132964,
      "grad_norm": 3.3363871574401855,
      "learning_rate": 8.08402535349142e-06,
      "loss": 1.2589,
      "step": 815
    },
    {
      "epoch": 0.5650969529085873,
      "grad_norm": 3.924525022506714,
      "learning_rate": 8.062527944716837e-06,
      "loss": 1.2983,
      "step": 816
    },
    {
      "epoch": 0.5657894736842105,
      "grad_norm": 3.191929340362549,
      "learning_rate": 8.041039834952287e-06,
      "loss": 1.1706,
      "step": 817
    },
    {
      "epoch": 0.5664819944598338,
      "grad_norm": 3.0657148361206055,
      "learning_rate": 8.019561127331202e-06,
      "loss": 1.325,
      "step": 818
    },
    {
      "epoch": 0.567174515235457,
      "grad_norm": 3.4930810928344727,
      "learning_rate": 7.998091924941897e-06,
      "loss": 1.3118,
      "step": 819
    },
    {
      "epoch": 0.5678670360110804,
      "grad_norm": 3.066667079925537,
      "learning_rate": 7.976632330827056e-06,
      "loss": 1.2784,
      "step": 820
    },
    {
      "epoch": 0.5685595567867036,
      "grad_norm": 3.7067830562591553,
      "learning_rate": 7.95518244798325e-06,
      "loss": 1.269,
      "step": 821
    },
    {
      "epoch": 0.5692520775623269,
      "grad_norm": 2.3034355640411377,
      "learning_rate": 7.933742379360446e-06,
      "loss": 0.7615,
      "step": 822
    },
    {
      "epoch": 0.5699445983379502,
      "grad_norm": 3.537672996520996,
      "learning_rate": 7.912312227861504e-06,
      "loss": 1.0391,
      "step": 823
    },
    {
      "epoch": 0.5706371191135734,
      "grad_norm": 3.186122179031372,
      "learning_rate": 7.890892096341677e-06,
      "loss": 1.2197,
      "step": 824
    },
    {
      "epoch": 0.5713296398891967,
      "grad_norm": 3.3523175716400146,
      "learning_rate": 7.86948208760814e-06,
      "loss": 1.3319,
      "step": 825
    },
    {
      "epoch": 0.5720221606648199,
      "grad_norm": 3.0430562496185303,
      "learning_rate": 7.848082304419478e-06,
      "loss": 1.2969,
      "step": 826
    },
    {
      "epoch": 0.5727146814404432,
      "grad_norm": 3.475856304168701,
      "learning_rate": 7.826692849485198e-06,
      "loss": 1.2284,
      "step": 827
    },
    {
      "epoch": 0.5734072022160664,
      "grad_norm": 3.8289895057678223,
      "learning_rate": 7.805313825465232e-06,
      "loss": 1.1896,
      "step": 828
    },
    {
      "epoch": 0.5740997229916898,
      "grad_norm": 3.8057913780212402,
      "learning_rate": 7.783945334969452e-06,
      "loss": 1.2574,
      "step": 829
    },
    {
      "epoch": 0.574792243767313,
      "grad_norm": 3.7936198711395264,
      "learning_rate": 7.762587480557175e-06,
      "loss": 1.2529,
      "step": 830
    },
    {
      "epoch": 0.5754847645429363,
      "grad_norm": 3.666149377822876,
      "learning_rate": 7.74124036473667e-06,
      "loss": 1.3672,
      "step": 831
    },
    {
      "epoch": 0.5761772853185596,
      "grad_norm": 3.881035089492798,
      "learning_rate": 7.719904089964658e-06,
      "loss": 1.119,
      "step": 832
    },
    {
      "epoch": 0.5768698060941828,
      "grad_norm": 3.367007255554199,
      "learning_rate": 7.698578758645838e-06,
      "loss": 1.4159,
      "step": 833
    },
    {
      "epoch": 0.5775623268698061,
      "grad_norm": 3.0489461421966553,
      "learning_rate": 7.67726447313238e-06,
      "loss": 1.3289,
      "step": 834
    },
    {
      "epoch": 0.5782548476454293,
      "grad_norm": 3.567683696746826,
      "learning_rate": 7.655961335723432e-06,
      "loss": 1.0198,
      "step": 835
    },
    {
      "epoch": 0.5789473684210527,
      "grad_norm": 3.55120849609375,
      "learning_rate": 7.634669448664655e-06,
      "loss": 1.3868,
      "step": 836
    },
    {
      "epoch": 0.5796398891966759,
      "grad_norm": 3.100919246673584,
      "learning_rate": 7.61338891414769e-06,
      "loss": 1.3426,
      "step": 837
    },
    {
      "epoch": 0.5803324099722992,
      "grad_norm": 2.899195432662964,
      "learning_rate": 7.5921198343097145e-06,
      "loss": 1.2532,
      "step": 838
    },
    {
      "epoch": 0.5810249307479224,
      "grad_norm": 3.1293692588806152,
      "learning_rate": 7.570862311232909e-06,
      "loss": 1.2963,
      "step": 839
    },
    {
      "epoch": 0.5817174515235457,
      "grad_norm": 3.7493510246276855,
      "learning_rate": 7.549616446943992e-06,
      "loss": 1.2741,
      "step": 840
    },
    {
      "epoch": 0.582409972299169,
      "grad_norm": 3.714538335800171,
      "learning_rate": 7.528382343413734e-06,
      "loss": 1.3811,
      "step": 841
    },
    {
      "epoch": 0.5831024930747922,
      "grad_norm": 3.606914520263672,
      "learning_rate": 7.507160102556451e-06,
      "loss": 1.2921,
      "step": 842
    },
    {
      "epoch": 0.5837950138504155,
      "grad_norm": 3.9541878700256348,
      "learning_rate": 7.485949826229518e-06,
      "loss": 1.1239,
      "step": 843
    },
    {
      "epoch": 0.5844875346260388,
      "grad_norm": 3.4801037311553955,
      "learning_rate": 7.464751616232902e-06,
      "loss": 1.32,
      "step": 844
    },
    {
      "epoch": 0.5851800554016621,
      "grad_norm": 3.1288328170776367,
      "learning_rate": 7.443565574308638e-06,
      "loss": 1.208,
      "step": 845
    },
    {
      "epoch": 0.5858725761772853,
      "grad_norm": 3.5429086685180664,
      "learning_rate": 7.422391802140376e-06,
      "loss": 1.1675,
      "step": 846
    },
    {
      "epoch": 0.5865650969529086,
      "grad_norm": 3.190911293029785,
      "learning_rate": 7.401230401352867e-06,
      "loss": 1.2133,
      "step": 847
    },
    {
      "epoch": 0.5872576177285319,
      "grad_norm": 3.729435920715332,
      "learning_rate": 7.380081473511484e-06,
      "loss": 1.1991,
      "step": 848
    },
    {
      "epoch": 0.5879501385041551,
      "grad_norm": 3.054455280303955,
      "learning_rate": 7.358945120121743e-06,
      "loss": 1.1601,
      "step": 849
    },
    {
      "epoch": 0.5886426592797784,
      "grad_norm": 3.179671287536621,
      "learning_rate": 7.337821442628805e-06,
      "loss": 1.3278,
      "step": 850
    },
    {
      "epoch": 0.5893351800554016,
      "grad_norm": 3.9718713760375977,
      "learning_rate": 7.316710542416983e-06,
      "loss": 1.2899,
      "step": 851
    },
    {
      "epoch": 0.590027700831025,
      "grad_norm": 3.6802737712860107,
      "learning_rate": 7.295612520809281e-06,
      "loss": 1.3197,
      "step": 852
    },
    {
      "epoch": 0.5907202216066482,
      "grad_norm": 4.423649311065674,
      "learning_rate": 7.274527479066883e-06,
      "loss": 1.4049,
      "step": 853
    },
    {
      "epoch": 0.5914127423822715,
      "grad_norm": 3.5196075439453125,
      "learning_rate": 7.253455518388668e-06,
      "loss": 1.2989,
      "step": 854
    },
    {
      "epoch": 0.5921052631578947,
      "grad_norm": 3.266993284225464,
      "learning_rate": 7.232396739910746e-06,
      "loss": 1.2234,
      "step": 855
    },
    {
      "epoch": 0.592797783933518,
      "grad_norm": 3.2850778102874756,
      "learning_rate": 7.211351244705947e-06,
      "loss": 1.2039,
      "step": 856
    },
    {
      "epoch": 0.5934903047091413,
      "grad_norm": 3.7952563762664795,
      "learning_rate": 7.190319133783359e-06,
      "loss": 1.2871,
      "step": 857
    },
    {
      "epoch": 0.5941828254847645,
      "grad_norm": 3.411527633666992,
      "learning_rate": 7.169300508087815e-06,
      "loss": 1.3137,
      "step": 858
    },
    {
      "epoch": 0.5948753462603878,
      "grad_norm": 3.390535593032837,
      "learning_rate": 7.148295468499438e-06,
      "loss": 1.3028,
      "step": 859
    },
    {
      "epoch": 0.5955678670360111,
      "grad_norm": 3.034454107284546,
      "learning_rate": 7.127304115833141e-06,
      "loss": 1.1951,
      "step": 860
    },
    {
      "epoch": 0.5962603878116344,
      "grad_norm": 3.163402557373047,
      "learning_rate": 7.106326550838145e-06,
      "loss": 1.4047,
      "step": 861
    },
    {
      "epoch": 0.5969529085872576,
      "grad_norm": 3.626631498336792,
      "learning_rate": 7.08536287419749e-06,
      "loss": 1.1003,
      "step": 862
    },
    {
      "epoch": 0.5976454293628809,
      "grad_norm": 3.2359683513641357,
      "learning_rate": 7.0644131865275675e-06,
      "loss": 1.3011,
      "step": 863
    },
    {
      "epoch": 0.5983379501385041,
      "grad_norm": 2.9554173946380615,
      "learning_rate": 7.043477588377623e-06,
      "loss": 1.31,
      "step": 864
    },
    {
      "epoch": 0.5990304709141274,
      "grad_norm": 3.3498265743255615,
      "learning_rate": 7.022556180229285e-06,
      "loss": 1.1903,
      "step": 865
    },
    {
      "epoch": 0.5997229916897507,
      "grad_norm": 2.4006574153900146,
      "learning_rate": 7.001649062496065e-06,
      "loss": 0.8033,
      "step": 866
    },
    {
      "epoch": 0.600415512465374,
      "grad_norm": 3.5354299545288086,
      "learning_rate": 6.980756335522893e-06,
      "loss": 1.2626,
      "step": 867
    },
    {
      "epoch": 0.6011080332409973,
      "grad_norm": 2.9956421852111816,
      "learning_rate": 6.959878099585634e-06,
      "loss": 1.2707,
      "step": 868
    },
    {
      "epoch": 0.6018005540166205,
      "grad_norm": 3.3528342247009277,
      "learning_rate": 6.939014454890594e-06,
      "loss": 0.978,
      "step": 869
    },
    {
      "epoch": 0.6024930747922438,
      "grad_norm": 3.61551833152771,
      "learning_rate": 6.918165501574051e-06,
      "loss": 1.327,
      "step": 870
    },
    {
      "epoch": 0.603185595567867,
      "grad_norm": 3.31195068359375,
      "learning_rate": 6.897331339701776e-06,
      "loss": 1.1225,
      "step": 871
    },
    {
      "epoch": 0.6038781163434903,
      "grad_norm": 2.8019473552703857,
      "learning_rate": 6.876512069268541e-06,
      "loss": 1.2743,
      "step": 872
    },
    {
      "epoch": 0.6045706371191135,
      "grad_norm": 3.4144673347473145,
      "learning_rate": 6.855707790197644e-06,
      "loss": 1.2898,
      "step": 873
    },
    {
      "epoch": 0.6052631578947368,
      "grad_norm": 3.687757730484009,
      "learning_rate": 6.834918602340439e-06,
      "loss": 1.1372,
      "step": 874
    },
    {
      "epoch": 0.6059556786703602,
      "grad_norm": 3.283721923828125,
      "learning_rate": 6.814144605475842e-06,
      "loss": 1.3031,
      "step": 875
    },
    {
      "epoch": 0.6066481994459834,
      "grad_norm": 3.4214913845062256,
      "learning_rate": 6.793385899309866e-06,
      "loss": 1.1862,
      "step": 876
    },
    {
      "epoch": 0.6073407202216067,
      "grad_norm": 3.3713910579681396,
      "learning_rate": 6.772642583475126e-06,
      "loss": 1.281,
      "step": 877
    },
    {
      "epoch": 0.6080332409972299,
      "grad_norm": 3.9484262466430664,
      "learning_rate": 6.751914757530375e-06,
      "loss": 1.2102,
      "step": 878
    },
    {
      "epoch": 0.6087257617728532,
      "grad_norm": 3.3405299186706543,
      "learning_rate": 6.731202520960025e-06,
      "loss": 1.3257,
      "step": 879
    },
    {
      "epoch": 0.6094182825484764,
      "grad_norm": 3.453195810317993,
      "learning_rate": 6.7105059731736645e-06,
      "loss": 1.2954,
      "step": 880
    },
    {
      "epoch": 0.6101108033240997,
      "grad_norm": 3.2071444988250732,
      "learning_rate": 6.689825213505572e-06,
      "loss": 1.2726,
      "step": 881
    },
    {
      "epoch": 0.610803324099723,
      "grad_norm": 3.140890121459961,
      "learning_rate": 6.669160341214265e-06,
      "loss": 1.3308,
      "step": 882
    },
    {
      "epoch": 0.6114958448753463,
      "grad_norm": 3.415692090988159,
      "learning_rate": 6.648511455482003e-06,
      "loss": 1.1971,
      "step": 883
    },
    {
      "epoch": 0.6121883656509696,
      "grad_norm": 3.596778392791748,
      "learning_rate": 6.627878655414311e-06,
      "loss": 1.3473,
      "step": 884
    },
    {
      "epoch": 0.6128808864265928,
      "grad_norm": 3.2050392627716064,
      "learning_rate": 6.607262040039519e-06,
      "loss": 1.2164,
      "step": 885
    },
    {
      "epoch": 0.6135734072022161,
      "grad_norm": 3.888380765914917,
      "learning_rate": 6.586661708308273e-06,
      "loss": 1.2232,
      "step": 886
    },
    {
      "epoch": 0.6142659279778393,
      "grad_norm": 2.849752902984619,
      "learning_rate": 6.56607775909307e-06,
      "loss": 1.3409,
      "step": 887
    },
    {
      "epoch": 0.6149584487534626,
      "grad_norm": 2.974795341491699,
      "learning_rate": 6.5455102911877665e-06,
      "loss": 1.3312,
      "step": 888
    },
    {
      "epoch": 0.6156509695290858,
      "grad_norm": 3.666857957839966,
      "learning_rate": 6.524959403307125e-06,
      "loss": 1.1317,
      "step": 889
    },
    {
      "epoch": 0.6163434903047091,
      "grad_norm": 3.64629864692688,
      "learning_rate": 6.504425194086334e-06,
      "loss": 1.4063,
      "step": 890
    },
    {
      "epoch": 0.6170360110803325,
      "grad_norm": 3.5016064643859863,
      "learning_rate": 6.483907762080526e-06,
      "loss": 1.2191,
      "step": 891
    },
    {
      "epoch": 0.6177285318559557,
      "grad_norm": 3.065629243850708,
      "learning_rate": 6.4634072057643045e-06,
      "loss": 1.181,
      "step": 892
    },
    {
      "epoch": 0.618421052631579,
      "grad_norm": 4.297072887420654,
      "learning_rate": 6.442923623531294e-06,
      "loss": 1.1937,
      "step": 893
    },
    {
      "epoch": 0.6191135734072022,
      "grad_norm": 3.8993167877197266,
      "learning_rate": 6.422457113693633e-06,
      "loss": 1.1729,
      "step": 894
    },
    {
      "epoch": 0.6198060941828255,
      "grad_norm": 3.1856400966644287,
      "learning_rate": 6.402007774481536e-06,
      "loss": 1.2613,
      "step": 895
    },
    {
      "epoch": 0.6204986149584487,
      "grad_norm": 3.68772554397583,
      "learning_rate": 6.381575704042792e-06,
      "loss": 1.1512,
      "step": 896
    },
    {
      "epoch": 0.621191135734072,
      "grad_norm": 3.291123151779175,
      "learning_rate": 6.361161000442313e-06,
      "loss": 1.2249,
      "step": 897
    },
    {
      "epoch": 0.6218836565096952,
      "grad_norm": 3.7242109775543213,
      "learning_rate": 6.340763761661665e-06,
      "loss": 1.1158,
      "step": 898
    },
    {
      "epoch": 0.6225761772853186,
      "grad_norm": 3.25076961517334,
      "learning_rate": 6.320384085598581e-06,
      "loss": 1.2239,
      "step": 899
    },
    {
      "epoch": 0.6232686980609419,
      "grad_norm": 3.5171546936035156,
      "learning_rate": 6.3000220700664985e-06,
      "loss": 1.2539,
      "step": 900
    },
    {
      "epoch": 0.6239612188365651,
      "grad_norm": 3.7872190475463867,
      "learning_rate": 6.2796778127941025e-06,
      "loss": 1.3232,
      "step": 901
    },
    {
      "epoch": 0.6246537396121884,
      "grad_norm": 3.50396466255188,
      "learning_rate": 6.259351411424849e-06,
      "loss": 1.2813,
      "step": 902
    },
    {
      "epoch": 0.6253462603878116,
      "grad_norm": 3.1483330726623535,
      "learning_rate": 6.239042963516471e-06,
      "loss": 1.2651,
      "step": 903
    },
    {
      "epoch": 0.6260387811634349,
      "grad_norm": 3.2591357231140137,
      "learning_rate": 6.218752566540555e-06,
      "loss": 1.1838,
      "step": 904
    },
    {
      "epoch": 0.6267313019390581,
      "grad_norm": 3.9565536975860596,
      "learning_rate": 6.1984803178820456e-06,
      "loss": 1.0281,
      "step": 905
    },
    {
      "epoch": 0.6274238227146814,
      "grad_norm": 3.1557421684265137,
      "learning_rate": 6.17822631483878e-06,
      "loss": 1.2428,
      "step": 906
    },
    {
      "epoch": 0.6281163434903048,
      "grad_norm": 3.933734655380249,
      "learning_rate": 6.157990654621025e-06,
      "loss": 1.31,
      "step": 907
    },
    {
      "epoch": 0.628808864265928,
      "grad_norm": 3.057314395904541,
      "learning_rate": 6.137773434351009e-06,
      "loss": 1.2082,
      "step": 908
    },
    {
      "epoch": 0.6295013850415513,
      "grad_norm": 2.9611635208129883,
      "learning_rate": 6.11757475106246e-06,
      "loss": 1.2675,
      "step": 909
    },
    {
      "epoch": 0.6301939058171745,
      "grad_norm": 3.71175217628479,
      "learning_rate": 6.097394701700146e-06,
      "loss": 1.3373,
      "step": 910
    },
    {
      "epoch": 0.6308864265927978,
      "grad_norm": 3.4437453746795654,
      "learning_rate": 6.077233383119372e-06,
      "loss": 1.3137,
      "step": 911
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 3.245744466781616,
      "learning_rate": 6.057090892085571e-06,
      "loss": 1.1768,
      "step": 912
    },
    {
      "epoch": 0.6322714681440443,
      "grad_norm": 3.2878289222717285,
      "learning_rate": 6.036967325273807e-06,
      "loss": 1.344,
      "step": 913
    },
    {
      "epoch": 0.6329639889196675,
      "grad_norm": 3.1663408279418945,
      "learning_rate": 6.016862779268301e-06,
      "loss": 1.2041,
      "step": 914
    },
    {
      "epoch": 0.6336565096952909,
      "grad_norm": 3.6326308250427246,
      "learning_rate": 5.996777350561997e-06,
      "loss": 1.3128,
      "step": 915
    },
    {
      "epoch": 0.6343490304709142,
      "grad_norm": 3.5937588214874268,
      "learning_rate": 5.976711135556086e-06,
      "loss": 1.2553,
      "step": 916
    },
    {
      "epoch": 0.6350415512465374,
      "grad_norm": 3.206855535507202,
      "learning_rate": 5.956664230559532e-06,
      "loss": 1.2699,
      "step": 917
    },
    {
      "epoch": 0.6357340720221607,
      "grad_norm": 3.8456692695617676,
      "learning_rate": 5.936636731788621e-06,
      "loss": 1.2254,
      "step": 918
    },
    {
      "epoch": 0.6364265927977839,
      "grad_norm": 4.080446720123291,
      "learning_rate": 5.916628735366505e-06,
      "loss": 1.314,
      "step": 919
    },
    {
      "epoch": 0.6371191135734072,
      "grad_norm": 3.2942323684692383,
      "learning_rate": 5.896640337322725e-06,
      "loss": 1.2532,
      "step": 920
    },
    {
      "epoch": 0.6378116343490304,
      "grad_norm": 3.4992527961730957,
      "learning_rate": 5.876671633592769e-06,
      "loss": 1.4546,
      "step": 921
    },
    {
      "epoch": 0.6385041551246537,
      "grad_norm": 3.231581687927246,
      "learning_rate": 5.8567227200175865e-06,
      "loss": 1.1727,
      "step": 922
    },
    {
      "epoch": 0.639196675900277,
      "grad_norm": 3.426119565963745,
      "learning_rate": 5.836793692343154e-06,
      "loss": 1.322,
      "step": 923
    },
    {
      "epoch": 0.6398891966759003,
      "grad_norm": 3.254599094390869,
      "learning_rate": 5.816884646219997e-06,
      "loss": 1.3904,
      "step": 924
    },
    {
      "epoch": 0.6405817174515236,
      "grad_norm": 3.066844940185547,
      "learning_rate": 5.7969956772027535e-06,
      "loss": 1.3591,
      "step": 925
    },
    {
      "epoch": 0.6412742382271468,
      "grad_norm": 3.5856401920318604,
      "learning_rate": 5.7771268807496794e-06,
      "loss": 1.0619,
      "step": 926
    },
    {
      "epoch": 0.6419667590027701,
      "grad_norm": 3.263921022415161,
      "learning_rate": 5.757278352222218e-06,
      "loss": 1.1964,
      "step": 927
    },
    {
      "epoch": 0.6426592797783933,
      "grad_norm": 3.6725008487701416,
      "learning_rate": 5.737450186884555e-06,
      "loss": 1.1908,
      "step": 928
    },
    {
      "epoch": 0.6433518005540166,
      "grad_norm": 3.5194790363311768,
      "learning_rate": 5.71764247990311e-06,
      "loss": 1.296,
      "step": 929
    },
    {
      "epoch": 0.6440443213296398,
      "grad_norm": 4.3180413246154785,
      "learning_rate": 5.6978553263461265e-06,
      "loss": 1.1636,
      "step": 930
    },
    {
      "epoch": 0.6447368421052632,
      "grad_norm": 3.6422464847564697,
      "learning_rate": 5.678088821183212e-06,
      "loss": 1.221,
      "step": 931
    },
    {
      "epoch": 0.6454293628808865,
      "grad_norm": 3.5116426944732666,
      "learning_rate": 5.6583430592848565e-06,
      "loss": 1.1897,
      "step": 932
    },
    {
      "epoch": 0.6461218836565097,
      "grad_norm": 3.304812431335449,
      "learning_rate": 5.638618135421986e-06,
      "loss": 1.2334,
      "step": 933
    },
    {
      "epoch": 0.646814404432133,
      "grad_norm": 3.4453015327453613,
      "learning_rate": 5.618914144265532e-06,
      "loss": 1.347,
      "step": 934
    },
    {
      "epoch": 0.6475069252077562,
      "grad_norm": 4.002638816833496,
      "learning_rate": 5.599231180385945e-06,
      "loss": 1.337,
      "step": 935
    },
    {
      "epoch": 0.6481994459833795,
      "grad_norm": 4.314698696136475,
      "learning_rate": 5.579569338252758e-06,
      "loss": 1.1725,
      "step": 936
    },
    {
      "epoch": 0.6488919667590027,
      "grad_norm": 3.4308438301086426,
      "learning_rate": 5.559928712234126e-06,
      "loss": 1.3876,
      "step": 937
    },
    {
      "epoch": 0.649584487534626,
      "grad_norm": 3.4807419776916504,
      "learning_rate": 5.5403093965963806e-06,
      "loss": 1.1882,
      "step": 938
    },
    {
      "epoch": 0.6502770083102493,
      "grad_norm": 3.439375638961792,
      "learning_rate": 5.520711485503569e-06,
      "loss": 1.2106,
      "step": 939
    },
    {
      "epoch": 0.6509695290858726,
      "grad_norm": 3.515831708908081,
      "learning_rate": 5.501135073017008e-06,
      "loss": 1.2495,
      "step": 940
    },
    {
      "epoch": 0.6516620498614959,
      "grad_norm": 4.047608852386475,
      "learning_rate": 5.481580253094828e-06,
      "loss": 1.2475,
      "step": 941
    },
    {
      "epoch": 0.6523545706371191,
      "grad_norm": 3.4397003650665283,
      "learning_rate": 5.4620471195915304e-06,
      "loss": 1.259,
      "step": 942
    },
    {
      "epoch": 0.6530470914127424,
      "grad_norm": 3.3549153804779053,
      "learning_rate": 5.4425357662575255e-06,
      "loss": 1.2415,
      "step": 943
    },
    {
      "epoch": 0.6537396121883656,
      "grad_norm": 3.8778305053710938,
      "learning_rate": 5.42304628673869e-06,
      "loss": 1.2269,
      "step": 944
    },
    {
      "epoch": 0.6544321329639889,
      "grad_norm": 3.3690638542175293,
      "learning_rate": 5.403578774575919e-06,
      "loss": 1.295,
      "step": 945
    },
    {
      "epoch": 0.6551246537396122,
      "grad_norm": 3.7962167263031006,
      "learning_rate": 5.384133323204666e-06,
      "loss": 1.2703,
      "step": 946
    },
    {
      "epoch": 0.6558171745152355,
      "grad_norm": 3.735410451889038,
      "learning_rate": 5.36471002595452e-06,
      "loss": 1.2357,
      "step": 947
    },
    {
      "epoch": 0.6565096952908587,
      "grad_norm": 3.9717748165130615,
      "learning_rate": 5.345308976048715e-06,
      "loss": 1.17,
      "step": 948
    },
    {
      "epoch": 0.657202216066482,
      "grad_norm": 3.596519708633423,
      "learning_rate": 5.325930266603724e-06,
      "loss": 1.3412,
      "step": 949
    },
    {
      "epoch": 0.6578947368421053,
      "grad_norm": 3.551438331604004,
      "learning_rate": 5.306573990628796e-06,
      "loss": 1.1691,
      "step": 950
    },
    {
      "epoch": 0.6585872576177285,
      "grad_norm": 4.270081043243408,
      "learning_rate": 5.287240241025509e-06,
      "loss": 1.358,
      "step": 951
    },
    {
      "epoch": 0.6592797783933518,
      "grad_norm": 3.3043808937072754,
      "learning_rate": 5.267929110587308e-06,
      "loss": 1.2799,
      "step": 952
    },
    {
      "epoch": 0.659972299168975,
      "grad_norm": 3.735628604888916,
      "learning_rate": 5.248640691999099e-06,
      "loss": 1.1946,
      "step": 953
    },
    {
      "epoch": 0.6606648199445984,
      "grad_norm": 3.4741759300231934,
      "learning_rate": 5.229375077836769e-06,
      "loss": 1.3411,
      "step": 954
    },
    {
      "epoch": 0.6613573407202216,
      "grad_norm": 3.6146764755249023,
      "learning_rate": 5.210132360566756e-06,
      "loss": 1.1197,
      "step": 955
    },
    {
      "epoch": 0.6620498614958449,
      "grad_norm": 3.68607497215271,
      "learning_rate": 5.190912632545599e-06,
      "loss": 1.3933,
      "step": 956
    },
    {
      "epoch": 0.6627423822714681,
      "grad_norm": 3.1479551792144775,
      "learning_rate": 5.171715986019509e-06,
      "loss": 1.3279,
      "step": 957
    },
    {
      "epoch": 0.6634349030470914,
      "grad_norm": 3.4101524353027344,
      "learning_rate": 5.152542513123906e-06,
      "loss": 1.354,
      "step": 958
    },
    {
      "epoch": 0.6641274238227147,
      "grad_norm": 3.9616758823394775,
      "learning_rate": 5.13339230588299e-06,
      "loss": 1.2024,
      "step": 959
    },
    {
      "epoch": 0.6648199445983379,
      "grad_norm": 3.6341795921325684,
      "learning_rate": 5.114265456209297e-06,
      "loss": 1.0029,
      "step": 960
    },
    {
      "epoch": 0.6655124653739612,
      "grad_norm": 3.9162726402282715,
      "learning_rate": 5.095162055903257e-06,
      "loss": 1.1473,
      "step": 961
    },
    {
      "epoch": 0.6662049861495845,
      "grad_norm": 3.362164258956909,
      "learning_rate": 5.076082196652754e-06,
      "loss": 1.3375,
      "step": 962
    },
    {
      "epoch": 0.6668975069252078,
      "grad_norm": 3.8291680812835693,
      "learning_rate": 5.057025970032683e-06,
      "loss": 1.3385,
      "step": 963
    },
    {
      "epoch": 0.667590027700831,
      "grad_norm": 3.777529239654541,
      "learning_rate": 5.037993467504515e-06,
      "loss": 1.3639,
      "step": 964
    },
    {
      "epoch": 0.6682825484764543,
      "grad_norm": 3.950564384460449,
      "learning_rate": 5.01898478041585e-06,
      "loss": 1.2299,
      "step": 965
    },
    {
      "epoch": 0.6689750692520776,
      "grad_norm": 4.049495697021484,
      "learning_rate": 5.000000000000003e-06,
      "loss": 1.2618,
      "step": 966
    },
    {
      "epoch": 0.6696675900277008,
      "grad_norm": 3.7501418590545654,
      "learning_rate": 4.98103921737552e-06,
      "loss": 1.2776,
      "step": 967
    },
    {
      "epoch": 0.6703601108033241,
      "grad_norm": 3.297684669494629,
      "learning_rate": 4.962102523545782e-06,
      "loss": 1.2447,
      "step": 968
    },
    {
      "epoch": 0.6710526315789473,
      "grad_norm": 3.3546786308288574,
      "learning_rate": 4.9431900093985655e-06,
      "loss": 1.201,
      "step": 969
    },
    {
      "epoch": 0.6717451523545707,
      "grad_norm": 4.063179969787598,
      "learning_rate": 4.92430176570558e-06,
      "loss": 1.257,
      "step": 970
    },
    {
      "epoch": 0.6724376731301939,
      "grad_norm": 3.4167563915252686,
      "learning_rate": 4.905437883122044e-06,
      "loss": 1.2553,
      "step": 971
    },
    {
      "epoch": 0.6731301939058172,
      "grad_norm": 4.087886810302734,
      "learning_rate": 4.88659845218627e-06,
      "loss": 1.4056,
      "step": 972
    },
    {
      "epoch": 0.6738227146814404,
      "grad_norm": 3.643519163131714,
      "learning_rate": 4.867783563319206e-06,
      "loss": 1.1488,
      "step": 973
    },
    {
      "epoch": 0.6745152354570637,
      "grad_norm": 3.689683675765991,
      "learning_rate": 4.848993306823997e-06,
      "loss": 1.1962,
      "step": 974
    },
    {
      "epoch": 0.675207756232687,
      "grad_norm": 3.7413206100463867,
      "learning_rate": 4.8302277728855826e-06,
      "loss": 1.1533,
      "step": 975
    },
    {
      "epoch": 0.6759002770083102,
      "grad_norm": 2.7265217304229736,
      "learning_rate": 4.811487051570235e-06,
      "loss": 0.7533,
      "step": 976
    },
    {
      "epoch": 0.6765927977839336,
      "grad_norm": 3.708652973175049,
      "learning_rate": 4.792771232825137e-06,
      "loss": 1.3488,
      "step": 977
    },
    {
      "epoch": 0.6772853185595568,
      "grad_norm": 3.641941547393799,
      "learning_rate": 4.774080406477954e-06,
      "loss": 1.2899,
      "step": 978
    },
    {
      "epoch": 0.6779778393351801,
      "grad_norm": 3.654796838760376,
      "learning_rate": 4.755414662236392e-06,
      "loss": 1.268,
      "step": 979
    },
    {
      "epoch": 0.6786703601108033,
      "grad_norm": 3.293848752975464,
      "learning_rate": 4.736774089687781e-06,
      "loss": 1.253,
      "step": 980
    },
    {
      "epoch": 0.6793628808864266,
      "grad_norm": 3.528782367706299,
      "learning_rate": 4.718158778298635e-06,
      "loss": 1.1901,
      "step": 981
    },
    {
      "epoch": 0.6800554016620498,
      "grad_norm": 4.274374485015869,
      "learning_rate": 4.699568817414224e-06,
      "loss": 1.2461,
      "step": 982
    },
    {
      "epoch": 0.6807479224376731,
      "grad_norm": 3.8092503547668457,
      "learning_rate": 4.681004296258151e-06,
      "loss": 1.2979,
      "step": 983
    },
    {
      "epoch": 0.6814404432132964,
      "grad_norm": 3.6626265048980713,
      "learning_rate": 4.662465303931912e-06,
      "loss": 1.2795,
      "step": 984
    },
    {
      "epoch": 0.6821329639889196,
      "grad_norm": 3.7979207038879395,
      "learning_rate": 4.643951929414493e-06,
      "loss": 1.2645,
      "step": 985
    },
    {
      "epoch": 0.682825484764543,
      "grad_norm": 3.516573429107666,
      "learning_rate": 4.625464261561902e-06,
      "loss": 1.1531,
      "step": 986
    },
    {
      "epoch": 0.6835180055401662,
      "grad_norm": 3.2881338596343994,
      "learning_rate": 4.607002389106777e-06,
      "loss": 1.263,
      "step": 987
    },
    {
      "epoch": 0.6842105263157895,
      "grad_norm": 3.467747688293457,
      "learning_rate": 4.588566400657965e-06,
      "loss": 1.2436,
      "step": 988
    },
    {
      "epoch": 0.6849030470914127,
      "grad_norm": 3.760425090789795,
      "learning_rate": 4.570156384700053e-06,
      "loss": 1.324,
      "step": 989
    },
    {
      "epoch": 0.685595567867036,
      "grad_norm": 3.4610824584960938,
      "learning_rate": 4.551772429592987e-06,
      "loss": 1.1936,
      "step": 990
    },
    {
      "epoch": 0.6862880886426593,
      "grad_norm": 3.2501208782196045,
      "learning_rate": 4.533414623571637e-06,
      "loss": 1.2801,
      "step": 991
    },
    {
      "epoch": 0.6869806094182825,
      "grad_norm": 3.383007526397705,
      "learning_rate": 4.515083054745363e-06,
      "loss": 1.2488,
      "step": 992
    },
    {
      "epoch": 0.6876731301939059,
      "grad_norm": 3.851450204849243,
      "learning_rate": 4.496777811097588e-06,
      "loss": 1.1017,
      "step": 993
    },
    {
      "epoch": 0.6883656509695291,
      "grad_norm": 3.6036598682403564,
      "learning_rate": 4.4784989804854055e-06,
      "loss": 1.2557,
      "step": 994
    },
    {
      "epoch": 0.6890581717451524,
      "grad_norm": 3.6002819538116455,
      "learning_rate": 4.460246650639125e-06,
      "loss": 1.2561,
      "step": 995
    },
    {
      "epoch": 0.6897506925207756,
      "grad_norm": 3.8944315910339355,
      "learning_rate": 4.4420209091618675e-06,
      "loss": 1.1098,
      "step": 996
    },
    {
      "epoch": 0.6904432132963989,
      "grad_norm": 3.2508392333984375,
      "learning_rate": 4.423821843529139e-06,
      "loss": 1.1746,
      "step": 997
    },
    {
      "epoch": 0.6911357340720221,
      "grad_norm": 3.485062837600708,
      "learning_rate": 4.405649541088419e-06,
      "loss": 1.2441,
      "step": 998
    },
    {
      "epoch": 0.6918282548476454,
      "grad_norm": 4.058342456817627,
      "learning_rate": 4.38750408905873e-06,
      "loss": 1.0997,
      "step": 999
    },
    {
      "epoch": 0.6925207756232687,
      "grad_norm": 3.4277503490448,
      "learning_rate": 4.369385574530227e-06,
      "loss": 1.3599,
      "step": 1000
    },
    {
      "epoch": 0.693213296398892,
      "grad_norm": 4.020823955535889,
      "learning_rate": 4.351294084463776e-06,
      "loss": 1.0891,
      "step": 1001
    },
    {
      "epoch": 0.6939058171745153,
      "grad_norm": 3.4967589378356934,
      "learning_rate": 4.3332297056905385e-06,
      "loss": 1.292,
      "step": 1002
    },
    {
      "epoch": 0.6945983379501385,
      "grad_norm": 4.008986949920654,
      "learning_rate": 4.315192524911551e-06,
      "loss": 1.2815,
      "step": 1003
    },
    {
      "epoch": 0.6952908587257618,
      "grad_norm": 3.5449891090393066,
      "learning_rate": 4.2971826286973175e-06,
      "loss": 1.2599,
      "step": 1004
    },
    {
      "epoch": 0.695983379501385,
      "grad_norm": 3.296858310699463,
      "learning_rate": 4.279200103487381e-06,
      "loss": 1.1863,
      "step": 1005
    },
    {
      "epoch": 0.6966759002770083,
      "grad_norm": 3.776249408721924,
      "learning_rate": 4.261245035589917e-06,
      "loss": 1.0698,
      "step": 1006
    },
    {
      "epoch": 0.6973684210526315,
      "grad_norm": 3.578810214996338,
      "learning_rate": 4.2433175111813305e-06,
      "loss": 1.2139,
      "step": 1007
    },
    {
      "epoch": 0.6980609418282548,
      "grad_norm": 3.9288926124572754,
      "learning_rate": 4.225417616305809e-06,
      "loss": 1.321,
      "step": 1008
    },
    {
      "epoch": 0.6987534626038782,
      "grad_norm": 3.7822978496551514,
      "learning_rate": 4.207545436874941e-06,
      "loss": 1.212,
      "step": 1009
    },
    {
      "epoch": 0.6994459833795014,
      "grad_norm": 3.7697253227233887,
      "learning_rate": 4.189701058667301e-06,
      "loss": 1.2954,
      "step": 1010
    },
    {
      "epoch": 0.7001385041551247,
      "grad_norm": 3.6546411514282227,
      "learning_rate": 4.171884567328021e-06,
      "loss": 1.1255,
      "step": 1011
    },
    {
      "epoch": 0.7008310249307479,
      "grad_norm": 3.8267335891723633,
      "learning_rate": 4.15409604836838e-06,
      "loss": 1.2291,
      "step": 1012
    },
    {
      "epoch": 0.7015235457063712,
      "grad_norm": 3.5930862426757812,
      "learning_rate": 4.136335587165421e-06,
      "loss": 1.3605,
      "step": 1013
    },
    {
      "epoch": 0.7022160664819944,
      "grad_norm": 3.3282723426818848,
      "learning_rate": 4.118603268961509e-06,
      "loss": 1.2689,
      "step": 1014
    },
    {
      "epoch": 0.7029085872576177,
      "grad_norm": 3.3957300186157227,
      "learning_rate": 4.100899178863938e-06,
      "loss": 1.3074,
      "step": 1015
    },
    {
      "epoch": 0.703601108033241,
      "grad_norm": 4.040927410125732,
      "learning_rate": 4.083223401844523e-06,
      "loss": 1.2394,
      "step": 1016
    },
    {
      "epoch": 0.7042936288088643,
      "grad_norm": 3.9132001399993896,
      "learning_rate": 4.065576022739181e-06,
      "loss": 1.2725,
      "step": 1017
    },
    {
      "epoch": 0.7049861495844876,
      "grad_norm": 4.065426826477051,
      "learning_rate": 4.047957126247542e-06,
      "loss": 1.078,
      "step": 1018
    },
    {
      "epoch": 0.7056786703601108,
      "grad_norm": 3.6188275814056396,
      "learning_rate": 4.030366796932521e-06,
      "loss": 1.2997,
      "step": 1019
    },
    {
      "epoch": 0.7063711911357341,
      "grad_norm": 3.8960142135620117,
      "learning_rate": 4.0128051192199325e-06,
      "loss": 1.2801,
      "step": 1020
    },
    {
      "epoch": 0.7070637119113573,
      "grad_norm": 3.781181812286377,
      "learning_rate": 3.99527217739807e-06,
      "loss": 1.1776,
      "step": 1021
    },
    {
      "epoch": 0.7077562326869806,
      "grad_norm": 3.5254998207092285,
      "learning_rate": 3.97776805561731e-06,
      "loss": 1.3147,
      "step": 1022
    },
    {
      "epoch": 0.7084487534626038,
      "grad_norm": 3.9870445728302,
      "learning_rate": 3.960292837889707e-06,
      "loss": 1.2304,
      "step": 1023
    },
    {
      "epoch": 0.7091412742382271,
      "grad_norm": 3.8298988342285156,
      "learning_rate": 3.942846608088583e-06,
      "loss": 1.3364,
      "step": 1024
    },
    {
      "epoch": 0.7098337950138505,
      "grad_norm": 4.035478591918945,
      "learning_rate": 3.925429449948134e-06,
      "loss": 1.2999,
      "step": 1025
    },
    {
      "epoch": 0.7105263157894737,
      "grad_norm": 3.4640512466430664,
      "learning_rate": 3.908041447063034e-06,
      "loss": 1.309,
      "step": 1026
    },
    {
      "epoch": 0.711218836565097,
      "grad_norm": 3.686994791030884,
      "learning_rate": 3.8906826828880085e-06,
      "loss": 1.2533,
      "step": 1027
    },
    {
      "epoch": 0.7119113573407202,
      "grad_norm": 3.6794652938842773,
      "learning_rate": 3.8733532407374555e-06,
      "loss": 1.3765,
      "step": 1028
    },
    {
      "epoch": 0.7126038781163435,
      "grad_norm": 3.3836612701416016,
      "learning_rate": 3.856053203785049e-06,
      "loss": 1.2114,
      "step": 1029
    },
    {
      "epoch": 0.7132963988919667,
      "grad_norm": 3.39042329788208,
      "learning_rate": 3.838782655063326e-06,
      "loss": 1.2425,
      "step": 1030
    },
    {
      "epoch": 0.71398891966759,
      "grad_norm": 3.2604761123657227,
      "learning_rate": 3.82154167746328e-06,
      "loss": 1.3216,
      "step": 1031
    },
    {
      "epoch": 0.7146814404432132,
      "grad_norm": 3.7054238319396973,
      "learning_rate": 3.804330353733998e-06,
      "loss": 1.1362,
      "step": 1032
    },
    {
      "epoch": 0.7153739612188366,
      "grad_norm": 3.187572479248047,
      "learning_rate": 3.7871487664822326e-06,
      "loss": 1.2687,
      "step": 1033
    },
    {
      "epoch": 0.7160664819944599,
      "grad_norm": 2.562303304672241,
      "learning_rate": 3.769996998172002e-06,
      "loss": 0.7703,
      "step": 1034
    },
    {
      "epoch": 0.7167590027700831,
      "grad_norm": 3.7910683155059814,
      "learning_rate": 3.7528751311242273e-06,
      "loss": 1.1611,
      "step": 1035
    },
    {
      "epoch": 0.7174515235457064,
      "grad_norm": 3.786738157272339,
      "learning_rate": 3.735783247516305e-06,
      "loss": 1.2285,
      "step": 1036
    },
    {
      "epoch": 0.7181440443213296,
      "grad_norm": 3.5773112773895264,
      "learning_rate": 3.718721429381724e-06,
      "loss": 1.2091,
      "step": 1037
    },
    {
      "epoch": 0.7188365650969529,
      "grad_norm": 4.090264320373535,
      "learning_rate": 3.7016897586096778e-06,
      "loss": 1.2265,
      "step": 1038
    },
    {
      "epoch": 0.7195290858725761,
      "grad_norm": 4.514917850494385,
      "learning_rate": 3.6846883169446624e-06,
      "loss": 1.3236,
      "step": 1039
    },
    {
      "epoch": 0.7202216066481995,
      "grad_norm": 4.169983386993408,
      "learning_rate": 3.667717185986085e-06,
      "loss": 1.2036,
      "step": 1040
    },
    {
      "epoch": 0.7209141274238227,
      "grad_norm": 2.4571571350097656,
      "learning_rate": 3.6507764471878804e-06,
      "loss": 0.6646,
      "step": 1041
    },
    {
      "epoch": 0.721606648199446,
      "grad_norm": 3.5560426712036133,
      "learning_rate": 3.63386618185811e-06,
      "loss": 1.196,
      "step": 1042
    },
    {
      "epoch": 0.7222991689750693,
      "grad_norm": 3.649376153945923,
      "learning_rate": 3.616986471158579e-06,
      "loss": 1.2761,
      "step": 1043
    },
    {
      "epoch": 0.7229916897506925,
      "grad_norm": 3.541705846786499,
      "learning_rate": 3.6001373961044385e-06,
      "loss": 1.2419,
      "step": 1044
    },
    {
      "epoch": 0.7236842105263158,
      "grad_norm": 3.4428951740264893,
      "learning_rate": 3.583319037563816e-06,
      "loss": 1.2972,
      "step": 1045
    },
    {
      "epoch": 0.724376731301939,
      "grad_norm": 3.983161687850952,
      "learning_rate": 3.5665314762573933e-06,
      "loss": 1.2115,
      "step": 1046
    },
    {
      "epoch": 0.7250692520775623,
      "grad_norm": 3.9475114345550537,
      "learning_rate": 3.5497747927580495e-06,
      "loss": 1.1102,
      "step": 1047
    },
    {
      "epoch": 0.7257617728531855,
      "grad_norm": 3.721270799636841,
      "learning_rate": 3.5330490674904737e-06,
      "loss": 1.2487,
      "step": 1048
    },
    {
      "epoch": 0.7264542936288089,
      "grad_norm": 3.9590468406677246,
      "learning_rate": 3.516354380730749e-06,
      "loss": 1.1853,
      "step": 1049
    },
    {
      "epoch": 0.7271468144044322,
      "grad_norm": 3.30857515335083,
      "learning_rate": 3.499690812605997e-06,
      "loss": 1.0928,
      "step": 1050
    },
    {
      "epoch": 0.7278393351800554,
      "grad_norm": 3.6948113441467285,
      "learning_rate": 3.4830584430939896e-06,
      "loss": 1.3328,
      "step": 1051
    },
    {
      "epoch": 0.7285318559556787,
      "grad_norm": 3.596212148666382,
      "learning_rate": 3.4664573520227564e-06,
      "loss": 1.2525,
      "step": 1052
    },
    {
      "epoch": 0.7292243767313019,
      "grad_norm": 3.829021692276001,
      "learning_rate": 3.449887619070188e-06,
      "loss": 1.2454,
      "step": 1053
    },
    {
      "epoch": 0.7299168975069252,
      "grad_norm": 3.797069787979126,
      "learning_rate": 3.433349323763696e-06,
      "loss": 1.2858,
      "step": 1054
    },
    {
      "epoch": 0.7306094182825484,
      "grad_norm": 3.8871891498565674,
      "learning_rate": 3.4168425454797884e-06,
      "loss": 1.1735,
      "step": 1055
    },
    {
      "epoch": 0.7313019390581718,
      "grad_norm": 3.7732465267181396,
      "learning_rate": 3.4003673634437084e-06,
      "loss": 1.2215,
      "step": 1056
    },
    {
      "epoch": 0.731994459833795,
      "grad_norm": 3.984323024749756,
      "learning_rate": 3.3839238567290523e-06,
      "loss": 1.1836,
      "step": 1057
    },
    {
      "epoch": 0.7326869806094183,
      "grad_norm": 3.9881927967071533,
      "learning_rate": 3.367512104257391e-06,
      "loss": 1.1729,
      "step": 1058
    },
    {
      "epoch": 0.7333795013850416,
      "grad_norm": 3.331230640411377,
      "learning_rate": 3.351132184797884e-06,
      "loss": 1.1262,
      "step": 1059
    },
    {
      "epoch": 0.7340720221606648,
      "grad_norm": 3.5761208534240723,
      "learning_rate": 3.334784176966912e-06,
      "loss": 1.2797,
      "step": 1060
    },
    {
      "epoch": 0.7347645429362881,
      "grad_norm": 3.5858423709869385,
      "learning_rate": 3.318468159227689e-06,
      "loss": 1.3824,
      "step": 1061
    },
    {
      "epoch": 0.7354570637119113,
      "grad_norm": 3.6287832260131836,
      "learning_rate": 3.3021842098898938e-06,
      "loss": 1.3382,
      "step": 1062
    },
    {
      "epoch": 0.7361495844875346,
      "grad_norm": 3.945394277572632,
      "learning_rate": 3.2859324071092902e-06,
      "loss": 1.2343,
      "step": 1063
    },
    {
      "epoch": 0.7368421052631579,
      "grad_norm": 3.6410295963287354,
      "learning_rate": 3.2697128288873536e-06,
      "loss": 1.1632,
      "step": 1064
    },
    {
      "epoch": 0.7375346260387812,
      "grad_norm": 4.534976482391357,
      "learning_rate": 3.253525553070893e-06,
      "loss": 1.2305,
      "step": 1065
    },
    {
      "epoch": 0.7382271468144044,
      "grad_norm": 4.022517204284668,
      "learning_rate": 3.2373706573516795e-06,
      "loss": 1.3596,
      "step": 1066
    },
    {
      "epoch": 0.7389196675900277,
      "grad_norm": 3.2530150413513184,
      "learning_rate": 3.2212482192660876e-06,
      "loss": 1.235,
      "step": 1067
    },
    {
      "epoch": 0.739612188365651,
      "grad_norm": 3.64341402053833,
      "learning_rate": 3.2051583161946865e-06,
      "loss": 1.1267,
      "step": 1068
    },
    {
      "epoch": 0.7403047091412742,
      "grad_norm": 4.010177135467529,
      "learning_rate": 3.189101025361905e-06,
      "loss": 1.0941,
      "step": 1069
    },
    {
      "epoch": 0.7409972299168975,
      "grad_norm": 3.3516669273376465,
      "learning_rate": 3.1730764238356517e-06,
      "loss": 1.2421,
      "step": 1070
    },
    {
      "epoch": 0.7416897506925207,
      "grad_norm": 3.6677613258361816,
      "learning_rate": 3.1570845885269365e-06,
      "loss": 1.3087,
      "step": 1071
    },
    {
      "epoch": 0.7423822714681441,
      "grad_norm": 3.4726169109344482,
      "learning_rate": 3.141125596189494e-06,
      "loss": 1.2735,
      "step": 1072
    },
    {
      "epoch": 0.7430747922437673,
      "grad_norm": 4.0901079177856445,
      "learning_rate": 3.125199523419449e-06,
      "loss": 1.2798,
      "step": 1073
    },
    {
      "epoch": 0.7437673130193906,
      "grad_norm": 3.457503318786621,
      "learning_rate": 3.109306446654912e-06,
      "loss": 1.2239,
      "step": 1074
    },
    {
      "epoch": 0.7444598337950139,
      "grad_norm": 4.589022159576416,
      "learning_rate": 3.0934464421756306e-06,
      "loss": 1.2632,
      "step": 1075
    },
    {
      "epoch": 0.7451523545706371,
      "grad_norm": 4.605857849121094,
      "learning_rate": 3.0776195861026202e-06,
      "loss": 1.1476,
      "step": 1076
    },
    {
      "epoch": 0.7458448753462604,
      "grad_norm": 3.9767348766326904,
      "learning_rate": 3.0618259543978e-06,
      "loss": 1.0512,
      "step": 1077
    },
    {
      "epoch": 0.7465373961218836,
      "grad_norm": 3.565870761871338,
      "learning_rate": 3.0460656228636254e-06,
      "loss": 1.2016,
      "step": 1078
    },
    {
      "epoch": 0.747229916897507,
      "grad_norm": 3.557732105255127,
      "learning_rate": 3.030338667142726e-06,
      "loss": 1.2852,
      "step": 1079
    },
    {
      "epoch": 0.7479224376731302,
      "grad_norm": 3.5791172981262207,
      "learning_rate": 3.014645162717542e-06,
      "loss": 1.2961,
      "step": 1080
    },
    {
      "epoch": 0.7486149584487535,
      "grad_norm": 4.24458646774292,
      "learning_rate": 2.9989851849099595e-06,
      "loss": 1.1436,
      "step": 1081
    },
    {
      "epoch": 0.7493074792243767,
      "grad_norm": 3.791700601577759,
      "learning_rate": 2.9833588088809627e-06,
      "loss": 1.1667,
      "step": 1082
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.004382610321045,
      "learning_rate": 2.9677661096302467e-06,
      "loss": 1.1902,
      "step": 1083
    },
    {
      "epoch": 0.75,
      "eval_loss": 1.1204665899276733,
      "eval_runtime": 339.3016,
      "eval_samples_per_second": 4.173,
      "eval_steps_per_second": 0.522,
      "step": 1083
    },
    {
      "epoch": 0.7506925207756233,
      "grad_norm": 4.0300984382629395,
      "learning_rate": 2.952207161995879e-06,
      "loss": 1.1181,
      "step": 1084
    },
    {
      "epoch": 0.7513850415512465,
      "grad_norm": 4.063140392303467,
      "learning_rate": 2.936682040653942e-06,
      "loss": 1.2998,
      "step": 1085
    },
    {
      "epoch": 0.7520775623268698,
      "grad_norm": 3.3608922958374023,
      "learning_rate": 2.9211908201181592e-06,
      "loss": 1.2074,
      "step": 1086
    },
    {
      "epoch": 0.752770083102493,
      "grad_norm": 3.313950777053833,
      "learning_rate": 2.905733574739542e-06,
      "loss": 1.2484,
      "step": 1087
    },
    {
      "epoch": 0.7534626038781164,
      "grad_norm": 3.8276939392089844,
      "learning_rate": 2.8903103787060395e-06,
      "loss": 1.1704,
      "step": 1088
    },
    {
      "epoch": 0.7541551246537396,
      "grad_norm": 4.115742206573486,
      "learning_rate": 2.8749213060421854e-06,
      "loss": 1.2243,
      "step": 1089
    },
    {
      "epoch": 0.7548476454293629,
      "grad_norm": 4.2053022384643555,
      "learning_rate": 2.8595664306087313e-06,
      "loss": 1.2551,
      "step": 1090
    },
    {
      "epoch": 0.7555401662049861,
      "grad_norm": 3.9844744205474854,
      "learning_rate": 2.8442458261022867e-06,
      "loss": 1.1682,
      "step": 1091
    },
    {
      "epoch": 0.7562326869806094,
      "grad_norm": 3.5984349250793457,
      "learning_rate": 2.828959566054996e-06,
      "loss": 1.1658,
      "step": 1092
    },
    {
      "epoch": 0.7569252077562327,
      "grad_norm": 4.055597305297852,
      "learning_rate": 2.8137077238341526e-06,
      "loss": 1.1369,
      "step": 1093
    },
    {
      "epoch": 0.7576177285318559,
      "grad_norm": 3.7269349098205566,
      "learning_rate": 2.79849037264186e-06,
      "loss": 1.2058,
      "step": 1094
    },
    {
      "epoch": 0.7583102493074793,
      "grad_norm": 3.704430341720581,
      "learning_rate": 2.783307585514683e-06,
      "loss": 1.2248,
      "step": 1095
    },
    {
      "epoch": 0.7590027700831025,
      "grad_norm": 3.74150013923645,
      "learning_rate": 2.7681594353232934e-06,
      "loss": 1.2354,
      "step": 1096
    },
    {
      "epoch": 0.7596952908587258,
      "grad_norm": 4.019088268280029,
      "learning_rate": 2.75304599477212e-06,
      "loss": 1.2765,
      "step": 1097
    },
    {
      "epoch": 0.760387811634349,
      "grad_norm": 3.8129332065582275,
      "learning_rate": 2.737967336399002e-06,
      "loss": 1.2936,
      "step": 1098
    },
    {
      "epoch": 0.7610803324099723,
      "grad_norm": 4.70414400100708,
      "learning_rate": 2.7229235325748394e-06,
      "loss": 1.1368,
      "step": 1099
    },
    {
      "epoch": 0.7617728531855956,
      "grad_norm": 4.109380722045898,
      "learning_rate": 2.707914655503242e-06,
      "loss": 1.2475,
      "step": 1100
    },
    {
      "epoch": 0.7624653739612188,
      "grad_norm": 4.041932106018066,
      "learning_rate": 2.6929407772201986e-06,
      "loss": 1.1576,
      "step": 1101
    },
    {
      "epoch": 0.7631578947368421,
      "grad_norm": 3.5189266204833984,
      "learning_rate": 2.678001969593701e-06,
      "loss": 1.1453,
      "step": 1102
    },
    {
      "epoch": 0.7638504155124654,
      "grad_norm": 3.557849407196045,
      "learning_rate": 2.663098304323429e-06,
      "loss": 1.195,
      "step": 1103
    },
    {
      "epoch": 0.7645429362880887,
      "grad_norm": 4.041141510009766,
      "learning_rate": 2.6482298529403973e-06,
      "loss": 1.0855,
      "step": 1104
    },
    {
      "epoch": 0.7652354570637119,
      "grad_norm": 3.6010632514953613,
      "learning_rate": 2.633396686806604e-06,
      "loss": 1.1841,
      "step": 1105
    },
    {
      "epoch": 0.7659279778393352,
      "grad_norm": 3.5304527282714844,
      "learning_rate": 2.6185988771146864e-06,
      "loss": 1.208,
      "step": 1106
    },
    {
      "epoch": 0.7666204986149584,
      "grad_norm": 3.7058682441711426,
      "learning_rate": 2.603836494887603e-06,
      "loss": 1.2629,
      "step": 1107
    },
    {
      "epoch": 0.7673130193905817,
      "grad_norm": 4.019807815551758,
      "learning_rate": 2.5891096109782644e-06,
      "loss": 1.2371,
      "step": 1108
    },
    {
      "epoch": 0.768005540166205,
      "grad_norm": 4.549735069274902,
      "learning_rate": 2.5744182960692087e-06,
      "loss": 1.0124,
      "step": 1109
    },
    {
      "epoch": 0.7686980609418282,
      "grad_norm": 4.584209442138672,
      "learning_rate": 2.5597626206722583e-06,
      "loss": 1.134,
      "step": 1110
    },
    {
      "epoch": 0.7693905817174516,
      "grad_norm": 3.772904872894287,
      "learning_rate": 2.54514265512818e-06,
      "loss": 1.1927,
      "step": 1111
    },
    {
      "epoch": 0.7700831024930748,
      "grad_norm": 4.115574359893799,
      "learning_rate": 2.530558469606351e-06,
      "loss": 1.1982,
      "step": 1112
    },
    {
      "epoch": 0.7707756232686981,
      "grad_norm": 4.603262901306152,
      "learning_rate": 2.5160101341044195e-06,
      "loss": 1.0217,
      "step": 1113
    },
    {
      "epoch": 0.7714681440443213,
      "grad_norm": 3.315068483352661,
      "learning_rate": 2.5014977184479696e-06,
      "loss": 1.2076,
      "step": 1114
    },
    {
      "epoch": 0.7721606648199446,
      "grad_norm": 4.200598239898682,
      "learning_rate": 2.487021292290186e-06,
      "loss": 1.2153,
      "step": 1115
    },
    {
      "epoch": 0.7728531855955678,
      "grad_norm": 3.8972954750061035,
      "learning_rate": 2.4725809251115208e-06,
      "loss": 1.2619,
      "step": 1116
    },
    {
      "epoch": 0.7735457063711911,
      "grad_norm": 4.2333831787109375,
      "learning_rate": 2.4581766862193556e-06,
      "loss": 1.1756,
      "step": 1117
    },
    {
      "epoch": 0.7742382271468145,
      "grad_norm": 4.078391075134277,
      "learning_rate": 2.443808644747675e-06,
      "loss": 1.2905,
      "step": 1118
    },
    {
      "epoch": 0.7749307479224377,
      "grad_norm": 3.5866286754608154,
      "learning_rate": 2.4294768696567293e-06,
      "loss": 1.2961,
      "step": 1119
    },
    {
      "epoch": 0.775623268698061,
      "grad_norm": 4.606243133544922,
      "learning_rate": 2.4151814297327157e-06,
      "loss": 1.3393,
      "step": 1120
    },
    {
      "epoch": 0.7763157894736842,
      "grad_norm": 4.044128894805908,
      "learning_rate": 2.4009223935874237e-06,
      "loss": 1.1318,
      "step": 1121
    },
    {
      "epoch": 0.7770083102493075,
      "grad_norm": 3.6675806045532227,
      "learning_rate": 2.386699829657928e-06,
      "loss": 1.2092,
      "step": 1122
    },
    {
      "epoch": 0.7777008310249307,
      "grad_norm": 3.6052377223968506,
      "learning_rate": 2.372513806206258e-06,
      "loss": 1.1631,
      "step": 1123
    },
    {
      "epoch": 0.778393351800554,
      "grad_norm": 4.187111854553223,
      "learning_rate": 2.35836439131906e-06,
      "loss": 1.048,
      "step": 1124
    },
    {
      "epoch": 0.7790858725761773,
      "grad_norm": 3.9402425289154053,
      "learning_rate": 2.3442516529072677e-06,
      "loss": 1.1767,
      "step": 1125
    },
    {
      "epoch": 0.7797783933518005,
      "grad_norm": 4.503572940826416,
      "learning_rate": 2.330175658705799e-06,
      "loss": 1.1202,
      "step": 1126
    },
    {
      "epoch": 0.7804709141274239,
      "grad_norm": 3.9013688564300537,
      "learning_rate": 2.3161364762732095e-06,
      "loss": 1.1952,
      "step": 1127
    },
    {
      "epoch": 0.7811634349030471,
      "grad_norm": 3.9454150199890137,
      "learning_rate": 2.3021341729913625e-06,
      "loss": 1.2963,
      "step": 1128
    },
    {
      "epoch": 0.7818559556786704,
      "grad_norm": 4.15333890914917,
      "learning_rate": 2.2881688160651406e-06,
      "loss": 1.2271,
      "step": 1129
    },
    {
      "epoch": 0.7825484764542936,
      "grad_norm": 3.8262038230895996,
      "learning_rate": 2.2742404725220823e-06,
      "loss": 1.2192,
      "step": 1130
    },
    {
      "epoch": 0.7832409972299169,
      "grad_norm": 4.1932244300842285,
      "learning_rate": 2.260349209212086e-06,
      "loss": 1.2081,
      "step": 1131
    },
    {
      "epoch": 0.7839335180055401,
      "grad_norm": 3.71751070022583,
      "learning_rate": 2.246495092807077e-06,
      "loss": 1.3275,
      "step": 1132
    },
    {
      "epoch": 0.7846260387811634,
      "grad_norm": 4.331583023071289,
      "learning_rate": 2.232678189800698e-06,
      "loss": 1.279,
      "step": 1133
    },
    {
      "epoch": 0.7853185595567868,
      "grad_norm": 4.326261043548584,
      "learning_rate": 2.2188985665079777e-06,
      "loss": 1.3143,
      "step": 1134
    },
    {
      "epoch": 0.78601108033241,
      "grad_norm": 3.9819676876068115,
      "learning_rate": 2.20515628906502e-06,
      "loss": 1.1659,
      "step": 1135
    },
    {
      "epoch": 0.7867036011080333,
      "grad_norm": 3.6455228328704834,
      "learning_rate": 2.19145142342869e-06,
      "loss": 1.1818,
      "step": 1136
    },
    {
      "epoch": 0.7873961218836565,
      "grad_norm": 3.962939500808716,
      "learning_rate": 2.177784035376286e-06,
      "loss": 1.1648,
      "step": 1137
    },
    {
      "epoch": 0.7880886426592798,
      "grad_norm": 4.336730003356934,
      "learning_rate": 2.164154190505231e-06,
      "loss": 1.2002,
      "step": 1138
    },
    {
      "epoch": 0.788781163434903,
      "grad_norm": 3.8926541805267334,
      "learning_rate": 2.150561954232768e-06,
      "loss": 1.1932,
      "step": 1139
    },
    {
      "epoch": 0.7894736842105263,
      "grad_norm": 4.326043605804443,
      "learning_rate": 2.13700739179562e-06,
      "loss": 1.1593,
      "step": 1140
    },
    {
      "epoch": 0.7901662049861495,
      "grad_norm": 3.7204015254974365,
      "learning_rate": 2.1234905682496987e-06,
      "loss": 1.1474,
      "step": 1141
    },
    {
      "epoch": 0.7908587257617729,
      "grad_norm": 4.206140041351318,
      "learning_rate": 2.1100115484697946e-06,
      "loss": 1.1646,
      "step": 1142
    },
    {
      "epoch": 0.7915512465373962,
      "grad_norm": 4.0659308433532715,
      "learning_rate": 2.096570397149239e-06,
      "loss": 1.2382,
      "step": 1143
    },
    {
      "epoch": 0.7922437673130194,
      "grad_norm": 4.224088191986084,
      "learning_rate": 2.083167178799623e-06,
      "loss": 1.2274,
      "step": 1144
    },
    {
      "epoch": 0.7929362880886427,
      "grad_norm": 3.992633819580078,
      "learning_rate": 2.069801957750479e-06,
      "loss": 1.1936,
      "step": 1145
    },
    {
      "epoch": 0.7936288088642659,
      "grad_norm": 4.030457019805908,
      "learning_rate": 2.0564747981489643e-06,
      "loss": 1.0835,
      "step": 1146
    },
    {
      "epoch": 0.7943213296398892,
      "grad_norm": 4.128167629241943,
      "learning_rate": 2.043185763959549e-06,
      "loss": 1.0168,
      "step": 1147
    },
    {
      "epoch": 0.7950138504155124,
      "grad_norm": 4.305471897125244,
      "learning_rate": 2.029934918963735e-06,
      "loss": 1.1999,
      "step": 1148
    },
    {
      "epoch": 0.7957063711911357,
      "grad_norm": 3.8589119911193848,
      "learning_rate": 2.0167223267597246e-06,
      "loss": 1.208,
      "step": 1149
    },
    {
      "epoch": 0.796398891966759,
      "grad_norm": 3.898130416870117,
      "learning_rate": 2.0035480507621217e-06,
      "loss": 1.2731,
      "step": 1150
    },
    {
      "epoch": 0.7970914127423823,
      "grad_norm": 4.046352386474609,
      "learning_rate": 1.990412154201633e-06,
      "loss": 1.3092,
      "step": 1151
    },
    {
      "epoch": 0.7977839335180056,
      "grad_norm": 3.7665176391601562,
      "learning_rate": 1.9773147001247583e-06,
      "loss": 1.1797,
      "step": 1152
    },
    {
      "epoch": 0.7984764542936288,
      "grad_norm": 5.006435394287109,
      "learning_rate": 1.9642557513934935e-06,
      "loss": 1.0088,
      "step": 1153
    },
    {
      "epoch": 0.7991689750692521,
      "grad_norm": 3.8915364742279053,
      "learning_rate": 1.951235370685023e-06,
      "loss": 1.1999,
      "step": 1154
    },
    {
      "epoch": 0.7998614958448753,
      "grad_norm": 3.539031744003296,
      "learning_rate": 1.9382536204914214e-06,
      "loss": 1.2408,
      "step": 1155
    },
    {
      "epoch": 0.8005540166204986,
      "grad_norm": 3.7190475463867188,
      "learning_rate": 1.925310563119358e-06,
      "loss": 1.1743,
      "step": 1156
    },
    {
      "epoch": 0.8012465373961218,
      "grad_norm": 3.969273090362549,
      "learning_rate": 1.9124062606897884e-06,
      "loss": 1.2168,
      "step": 1157
    },
    {
      "epoch": 0.8019390581717452,
      "grad_norm": 4.360279083251953,
      "learning_rate": 1.899540775137666e-06,
      "loss": 1.3146,
      "step": 1158
    },
    {
      "epoch": 0.8026315789473685,
      "grad_norm": 3.9002370834350586,
      "learning_rate": 1.8867141682116373e-06,
      "loss": 1.2197,
      "step": 1159
    },
    {
      "epoch": 0.8033240997229917,
      "grad_norm": 4.110561370849609,
      "learning_rate": 1.8739265014737473e-06,
      "loss": 1.2671,
      "step": 1160
    },
    {
      "epoch": 0.804016620498615,
      "grad_norm": 4.611395835876465,
      "learning_rate": 1.861177836299155e-06,
      "loss": 1.1325,
      "step": 1161
    },
    {
      "epoch": 0.8047091412742382,
      "grad_norm": 3.9032986164093018,
      "learning_rate": 1.8484682338758154e-06,
      "loss": 1.3979,
      "step": 1162
    },
    {
      "epoch": 0.8054016620498615,
      "grad_norm": 4.64093542098999,
      "learning_rate": 1.835797755204205e-06,
      "loss": 1.1017,
      "step": 1163
    },
    {
      "epoch": 0.8060941828254847,
      "grad_norm": 4.68850564956665,
      "learning_rate": 1.8231664610970301e-06,
      "loss": 1.2508,
      "step": 1164
    },
    {
      "epoch": 0.806786703601108,
      "grad_norm": 3.6796510219573975,
      "learning_rate": 1.8105744121789226e-06,
      "loss": 1.3439,
      "step": 1165
    },
    {
      "epoch": 0.8074792243767313,
      "grad_norm": 3.682265043258667,
      "learning_rate": 1.7980216688861475e-06,
      "loss": 1.1941,
      "step": 1166
    },
    {
      "epoch": 0.8081717451523546,
      "grad_norm": 4.359194278717041,
      "learning_rate": 1.7855082914663346e-06,
      "loss": 1.2085,
      "step": 1167
    },
    {
      "epoch": 0.8088642659279779,
      "grad_norm": 4.421899318695068,
      "learning_rate": 1.773034339978167e-06,
      "loss": 1.2169,
      "step": 1168
    },
    {
      "epoch": 0.8095567867036011,
      "grad_norm": 3.645191192626953,
      "learning_rate": 1.7605998742911001e-06,
      "loss": 1.2932,
      "step": 1169
    },
    {
      "epoch": 0.8102493074792244,
      "grad_norm": 4.023017406463623,
      "learning_rate": 1.7482049540850788e-06,
      "loss": 1.145,
      "step": 1170
    },
    {
      "epoch": 0.8109418282548476,
      "grad_norm": 3.481224536895752,
      "learning_rate": 1.7358496388502422e-06,
      "loss": 1.2044,
      "step": 1171
    },
    {
      "epoch": 0.8116343490304709,
      "grad_norm": 3.7499403953552246,
      "learning_rate": 1.7235339878866474e-06,
      "loss": 1.1941,
      "step": 1172
    },
    {
      "epoch": 0.8123268698060941,
      "grad_norm": 3.887758255004883,
      "learning_rate": 1.7112580603039785e-06,
      "loss": 1.115,
      "step": 1173
    },
    {
      "epoch": 0.8130193905817175,
      "grad_norm": 3.799678087234497,
      "learning_rate": 1.6990219150212662e-06,
      "loss": 1.3243,
      "step": 1174
    },
    {
      "epoch": 0.8137119113573407,
      "grad_norm": 4.975963592529297,
      "learning_rate": 1.6868256107666015e-06,
      "loss": 1.1648,
      "step": 1175
    },
    {
      "epoch": 0.814404432132964,
      "grad_norm": 3.937077045440674,
      "learning_rate": 1.6746692060768576e-06,
      "loss": 1.2042,
      "step": 1176
    },
    {
      "epoch": 0.8150969529085873,
      "grad_norm": 4.324456691741943,
      "learning_rate": 1.6625527592974077e-06,
      "loss": 1.1491,
      "step": 1177
    },
    {
      "epoch": 0.8157894736842105,
      "grad_norm": 4.019317150115967,
      "learning_rate": 1.6504763285818437e-06,
      "loss": 1.2842,
      "step": 1178
    },
    {
      "epoch": 0.8164819944598338,
      "grad_norm": 3.58640718460083,
      "learning_rate": 1.6384399718916977e-06,
      "loss": 1.3005,
      "step": 1179
    },
    {
      "epoch": 0.817174515235457,
      "grad_norm": 4.04345703125,
      "learning_rate": 1.6264437469961703e-06,
      "loss": 1.1709,
      "step": 1180
    },
    {
      "epoch": 0.8178670360110804,
      "grad_norm": 3.9605162143707275,
      "learning_rate": 1.614487711471835e-06,
      "loss": 1.2854,
      "step": 1181
    },
    {
      "epoch": 0.8185595567867036,
      "grad_norm": 4.292003631591797,
      "learning_rate": 1.6025719227023817e-06,
      "loss": 1.1602,
      "step": 1182
    },
    {
      "epoch": 0.8192520775623269,
      "grad_norm": 4.0546417236328125,
      "learning_rate": 1.5906964378783373e-06,
      "loss": 1.3041,
      "step": 1183
    },
    {
      "epoch": 0.8199445983379502,
      "grad_norm": 4.742153644561768,
      "learning_rate": 1.5788613139967846e-06,
      "loss": 1.1946,
      "step": 1184
    },
    {
      "epoch": 0.8206371191135734,
      "grad_norm": 3.6266913414001465,
      "learning_rate": 1.5670666078610809e-06,
      "loss": 1.255,
      "step": 1185
    },
    {
      "epoch": 0.8213296398891967,
      "grad_norm": 3.7363014221191406,
      "learning_rate": 1.5553123760806144e-06,
      "loss": 1.2767,
      "step": 1186
    },
    {
      "epoch": 0.8220221606648199,
      "grad_norm": 3.828493118286133,
      "learning_rate": 1.5435986750705046e-06,
      "loss": 1.3323,
      "step": 1187
    },
    {
      "epoch": 0.8227146814404432,
      "grad_norm": 4.154706001281738,
      "learning_rate": 1.531925561051336e-06,
      "loss": 1.1205,
      "step": 1188
    },
    {
      "epoch": 0.8234072022160664,
      "grad_norm": 3.972463607788086,
      "learning_rate": 1.5202930900489056e-06,
      "loss": 1.2494,
      "step": 1189
    },
    {
      "epoch": 0.8240997229916898,
      "grad_norm": 4.380939960479736,
      "learning_rate": 1.5087013178939347e-06,
      "loss": 1.2353,
      "step": 1190
    },
    {
      "epoch": 0.824792243767313,
      "grad_norm": 3.84091854095459,
      "learning_rate": 1.497150300221808e-06,
      "loss": 1.1958,
      "step": 1191
    },
    {
      "epoch": 0.8254847645429363,
      "grad_norm": 3.6779820919036865,
      "learning_rate": 1.485640092472308e-06,
      "loss": 1.2238,
      "step": 1192
    },
    {
      "epoch": 0.8261772853185596,
      "grad_norm": 3.7688772678375244,
      "learning_rate": 1.4741707498893487e-06,
      "loss": 1.1069,
      "step": 1193
    },
    {
      "epoch": 0.8268698060941828,
      "grad_norm": 4.308027744293213,
      "learning_rate": 1.4627423275207075e-06,
      "loss": 1.1624,
      "step": 1194
    },
    {
      "epoch": 0.8275623268698061,
      "grad_norm": 3.8426570892333984,
      "learning_rate": 1.4513548802177635e-06,
      "loss": 1.3091,
      "step": 1195
    },
    {
      "epoch": 0.8282548476454293,
      "grad_norm": 3.7444517612457275,
      "learning_rate": 1.440008462635234e-06,
      "loss": 1.1742,
      "step": 1196
    },
    {
      "epoch": 0.8289473684210527,
      "grad_norm": 3.861121892929077,
      "learning_rate": 1.4287031292309105e-06,
      "loss": 1.0395,
      "step": 1197
    },
    {
      "epoch": 0.8296398891966759,
      "grad_norm": 3.4704298973083496,
      "learning_rate": 1.4174389342653972e-06,
      "loss": 1.2084,
      "step": 1198
    },
    {
      "epoch": 0.8303324099722992,
      "grad_norm": 3.99474835395813,
      "learning_rate": 1.406215931801862e-06,
      "loss": 1.1811,
      "step": 1199
    },
    {
      "epoch": 0.8310249307479224,
      "grad_norm": 3.7253665924072266,
      "learning_rate": 1.395034175705753e-06,
      "loss": 1.0946,
      "step": 1200
    },
    {
      "epoch": 0.8317174515235457,
      "grad_norm": 3.6922335624694824,
      "learning_rate": 1.383893719644558e-06,
      "loss": 1.183,
      "step": 1201
    },
    {
      "epoch": 0.832409972299169,
      "grad_norm": 3.5496158599853516,
      "learning_rate": 1.3727946170875538e-06,
      "loss": 1.2656,
      "step": 1202
    },
    {
      "epoch": 0.8331024930747922,
      "grad_norm": 3.911721706390381,
      "learning_rate": 1.361736921305522e-06,
      "loss": 1.2568,
      "step": 1203
    },
    {
      "epoch": 0.8337950138504155,
      "grad_norm": 4.524016380310059,
      "learning_rate": 1.3507206853705178e-06,
      "loss": 1.1337,
      "step": 1204
    },
    {
      "epoch": 0.8344875346260388,
      "grad_norm": 4.609461784362793,
      "learning_rate": 1.339745962155613e-06,
      "loss": 1.228,
      "step": 1205
    },
    {
      "epoch": 0.8351800554016621,
      "grad_norm": 4.127496242523193,
      "learning_rate": 1.3288128043346315e-06,
      "loss": 1.155,
      "step": 1206
    },
    {
      "epoch": 0.8358725761772853,
      "grad_norm": 3.739071846008301,
      "learning_rate": 1.3179212643818928e-06,
      "loss": 1.1306,
      "step": 1207
    },
    {
      "epoch": 0.8365650969529086,
      "grad_norm": 4.528608798980713,
      "learning_rate": 1.307071394571986e-06,
      "loss": 1.2538,
      "step": 1208
    },
    {
      "epoch": 0.8372576177285319,
      "grad_norm": 4.651935577392578,
      "learning_rate": 1.2962632469794901e-06,
      "loss": 1.241,
      "step": 1209
    },
    {
      "epoch": 0.8379501385041551,
      "grad_norm": 3.754818916320801,
      "learning_rate": 1.2854968734787398e-06,
      "loss": 1.1738,
      "step": 1210
    },
    {
      "epoch": 0.8386426592797784,
      "grad_norm": 4.032466888427734,
      "learning_rate": 1.2747723257435729e-06,
      "loss": 1.3379,
      "step": 1211
    },
    {
      "epoch": 0.8393351800554016,
      "grad_norm": 3.8999366760253906,
      "learning_rate": 1.2640896552470795e-06,
      "loss": 1.2087,
      "step": 1212
    },
    {
      "epoch": 0.840027700831025,
      "grad_norm": 5.117029666900635,
      "learning_rate": 1.2534489132613603e-06,
      "loss": 1.0604,
      "step": 1213
    },
    {
      "epoch": 0.8407202216066482,
      "grad_norm": 3.513732433319092,
      "learning_rate": 1.2428501508572765e-06,
      "loss": 1.2153,
      "step": 1214
    },
    {
      "epoch": 0.8414127423822715,
      "grad_norm": 4.629495620727539,
      "learning_rate": 1.232293418904207e-06,
      "loss": 1.0813,
      "step": 1215
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 3.8420591354370117,
      "learning_rate": 1.221778768069799e-06,
      "loss": 1.1894,
      "step": 1216
    },
    {
      "epoch": 0.842797783933518,
      "grad_norm": 4.045253753662109,
      "learning_rate": 1.2113062488197347e-06,
      "loss": 1.2621,
      "step": 1217
    },
    {
      "epoch": 0.8434903047091413,
      "grad_norm": 4.0366950035095215,
      "learning_rate": 1.2008759114174794e-06,
      "loss": 1.105,
      "step": 1218
    },
    {
      "epoch": 0.8441828254847645,
      "grad_norm": 4.136510372161865,
      "learning_rate": 1.1904878059240443e-06,
      "loss": 1.155,
      "step": 1219
    },
    {
      "epoch": 0.8448753462603878,
      "grad_norm": 3.633707284927368,
      "learning_rate": 1.1801419821977479e-06,
      "loss": 1.0978,
      "step": 1220
    },
    {
      "epoch": 0.8455678670360111,
      "grad_norm": 4.147572040557861,
      "learning_rate": 1.1698384898939774e-06,
      "loss": 1.2548,
      "step": 1221
    },
    {
      "epoch": 0.8462603878116344,
      "grad_norm": 4.219207286834717,
      "learning_rate": 1.159577378464939e-06,
      "loss": 1.0852,
      "step": 1222
    },
    {
      "epoch": 0.8469529085872576,
      "grad_norm": 3.746647596359253,
      "learning_rate": 1.149358697159435e-06,
      "loss": 1.2269,
      "step": 1223
    },
    {
      "epoch": 0.8476454293628809,
      "grad_norm": 3.936307430267334,
      "learning_rate": 1.1391824950226272e-06,
      "loss": 0.9683,
      "step": 1224
    },
    {
      "epoch": 0.8483379501385041,
      "grad_norm": 4.122901439666748,
      "learning_rate": 1.1290488208957894e-06,
      "loss": 1.1598,
      "step": 1225
    },
    {
      "epoch": 0.8490304709141274,
      "grad_norm": 4.317487716674805,
      "learning_rate": 1.1189577234160764e-06,
      "loss": 1.2321,
      "step": 1226
    },
    {
      "epoch": 0.8497229916897507,
      "grad_norm": 3.937451124191284,
      "learning_rate": 1.1089092510163025e-06,
      "loss": 1.253,
      "step": 1227
    },
    {
      "epoch": 0.850415512465374,
      "grad_norm": 3.7667171955108643,
      "learning_rate": 1.0989034519246956e-06,
      "loss": 1.1728,
      "step": 1228
    },
    {
      "epoch": 0.8511080332409973,
      "grad_norm": 4.663061618804932,
      "learning_rate": 1.088940374164672e-06,
      "loss": 1.3034,
      "step": 1229
    },
    {
      "epoch": 0.8518005540166205,
      "grad_norm": 4.142401218414307,
      "learning_rate": 1.0790200655546035e-06,
      "loss": 1.1516,
      "step": 1230
    },
    {
      "epoch": 0.8524930747922438,
      "grad_norm": 4.334724426269531,
      "learning_rate": 1.06914257370759e-06,
      "loss": 1.3624,
      "step": 1231
    },
    {
      "epoch": 0.853185595567867,
      "grad_norm": 5.013029098510742,
      "learning_rate": 1.0593079460312284e-06,
      "loss": 1.0803,
      "step": 1232
    },
    {
      "epoch": 0.8538781163434903,
      "grad_norm": 4.472379684448242,
      "learning_rate": 1.0495162297273876e-06,
      "loss": 0.9917,
      "step": 1233
    },
    {
      "epoch": 0.8545706371191135,
      "grad_norm": 4.156463146209717,
      "learning_rate": 1.0397674717919803e-06,
      "loss": 1.2364,
      "step": 1234
    },
    {
      "epoch": 0.8552631578947368,
      "grad_norm": 4.236593246459961,
      "learning_rate": 1.0300617190147388e-06,
      "loss": 1.2187,
      "step": 1235
    },
    {
      "epoch": 0.8559556786703602,
      "grad_norm": 4.179319381713867,
      "learning_rate": 1.0203990179789892e-06,
      "loss": 1.2905,
      "step": 1236
    },
    {
      "epoch": 0.8566481994459834,
      "grad_norm": 3.6013402938842773,
      "learning_rate": 1.010779415061428e-06,
      "loss": 1.1256,
      "step": 1237
    },
    {
      "epoch": 0.8573407202216067,
      "grad_norm": 4.2881340980529785,
      "learning_rate": 1.0012029564318993e-06,
      "loss": 1.1806,
      "step": 1238
    },
    {
      "epoch": 0.8580332409972299,
      "grad_norm": 3.446340322494507,
      "learning_rate": 9.916696880531739e-07,
      "loss": 1.1771,
      "step": 1239
    },
    {
      "epoch": 0.8587257617728532,
      "grad_norm": 3.927043914794922,
      "learning_rate": 9.82179655680734e-07,
      "loss": 1.1842,
      "step": 1240
    },
    {
      "epoch": 0.8594182825484764,
      "grad_norm": 4.7350616455078125,
      "learning_rate": 9.727329048625355e-07,
      "loss": 1.1911,
      "step": 1241
    },
    {
      "epoch": 0.8601108033240997,
      "grad_norm": 4.3509039878845215,
      "learning_rate": 9.633294809388127e-07,
      "loss": 1.1404,
      "step": 1242
    },
    {
      "epoch": 0.860803324099723,
      "grad_norm": 3.8866496086120605,
      "learning_rate": 9.539694290418488e-07,
      "loss": 1.1203,
      "step": 1243
    },
    {
      "epoch": 0.8614958448753463,
      "grad_norm": 4.465489387512207,
      "learning_rate": 9.446527940957573e-07,
      "loss": 1.2293,
      "step": 1244
    },
    {
      "epoch": 0.8621883656509696,
      "grad_norm": 4.468167304992676,
      "learning_rate": 9.353796208162669e-07,
      "loss": 1.1552,
      "step": 1245
    },
    {
      "epoch": 0.8628808864265928,
      "grad_norm": 4.596248626708984,
      "learning_rate": 9.261499537105179e-07,
      "loss": 1.252,
      "step": 1246
    },
    {
      "epoch": 0.8635734072022161,
      "grad_norm": 3.8985869884490967,
      "learning_rate": 9.169638370768341e-07,
      "loss": 1.1292,
      "step": 1247
    },
    {
      "epoch": 0.8642659279778393,
      "grad_norm": 5.140615463256836,
      "learning_rate": 9.078213150045145e-07,
      "loss": 1.2206,
      "step": 1248
    },
    {
      "epoch": 0.8649584487534626,
      "grad_norm": 4.030377388000488,
      "learning_rate": 8.987224313736309e-07,
      "loss": 1.2621,
      "step": 1249
    },
    {
      "epoch": 0.8656509695290858,
      "grad_norm": 4.131312370300293,
      "learning_rate": 8.896672298548037e-07,
      "loss": 1.2128,
      "step": 1250
    },
    {
      "epoch": 0.8663434903047091,
      "grad_norm": 4.123977184295654,
      "learning_rate": 8.806557539089988e-07,
      "loss": 1.2752,
      "step": 1251
    },
    {
      "epoch": 0.8670360110803325,
      "grad_norm": 4.873196125030518,
      "learning_rate": 8.716880467873235e-07,
      "loss": 1.1041,
      "step": 1252
    },
    {
      "epoch": 0.8677285318559557,
      "grad_norm": 3.93961238861084,
      "learning_rate": 8.627641515308072e-07,
      "loss": 1.2894,
      "step": 1253
    },
    {
      "epoch": 0.868421052631579,
      "grad_norm": 3.7824409008026123,
      "learning_rate": 8.538841109702079e-07,
      "loss": 1.2282,
      "step": 1254
    },
    {
      "epoch": 0.8691135734072022,
      "grad_norm": 4.6755757331848145,
      "learning_rate": 8.450479677257962e-07,
      "loss": 1.0459,
      "step": 1255
    },
    {
      "epoch": 0.8698060941828255,
      "grad_norm": 4.243119716644287,
      "learning_rate": 8.362557642071567e-07,
      "loss": 1.317,
      "step": 1256
    },
    {
      "epoch": 0.8704986149584487,
      "grad_norm": 4.0560479164123535,
      "learning_rate": 8.275075426129831e-07,
      "loss": 1.2543,
      "step": 1257
    },
    {
      "epoch": 0.871191135734072,
      "grad_norm": 3.3924214839935303,
      "learning_rate": 8.188033449308719e-07,
      "loss": 1.2128,
      "step": 1258
    },
    {
      "epoch": 0.8718836565096952,
      "grad_norm": 3.7904396057128906,
      "learning_rate": 8.101432129371345e-07,
      "loss": 1.1834,
      "step": 1259
    },
    {
      "epoch": 0.8725761772853186,
      "grad_norm": 3.563037395477295,
      "learning_rate": 8.015271881965714e-07,
      "loss": 1.2524,
      "step": 1260
    },
    {
      "epoch": 0.8732686980609419,
      "grad_norm": 3.5340735912323,
      "learning_rate": 7.929553120622968e-07,
      "loss": 1.1476,
      "step": 1261
    },
    {
      "epoch": 0.8739612188365651,
      "grad_norm": 4.195478916168213,
      "learning_rate": 7.844276256755345e-07,
      "loss": 1.2139,
      "step": 1262
    },
    {
      "epoch": 0.8746537396121884,
      "grad_norm": 4.329362392425537,
      "learning_rate": 7.759441699654068e-07,
      "loss": 1.1694,
      "step": 1263
    },
    {
      "epoch": 0.8753462603878116,
      "grad_norm": 3.519171953201294,
      "learning_rate": 7.675049856487549e-07,
      "loss": 1.1948,
      "step": 1264
    },
    {
      "epoch": 0.8760387811634349,
      "grad_norm": 3.8010988235473633,
      "learning_rate": 7.591101132299383e-07,
      "loss": 1.3021,
      "step": 1265
    },
    {
      "epoch": 0.8767313019390581,
      "grad_norm": 3.752256155014038,
      "learning_rate": 7.507595930006351e-07,
      "loss": 1.1363,
      "step": 1266
    },
    {
      "epoch": 0.8774238227146814,
      "grad_norm": 3.8671414852142334,
      "learning_rate": 7.42453465039652e-07,
      "loss": 1.1246,
      "step": 1267
    },
    {
      "epoch": 0.8781163434903048,
      "grad_norm": 4.748464584350586,
      "learning_rate": 7.341917692127354e-07,
      "loss": 1.2564,
      "step": 1268
    },
    {
      "epoch": 0.878808864265928,
      "grad_norm": 3.782492160797119,
      "learning_rate": 7.259745451723765e-07,
      "loss": 1.2389,
      "step": 1269
    },
    {
      "epoch": 0.8795013850415513,
      "grad_norm": 3.492922306060791,
      "learning_rate": 7.178018323576208e-07,
      "loss": 1.1598,
      "step": 1270
    },
    {
      "epoch": 0.8801939058171745,
      "grad_norm": 4.3590407371521,
      "learning_rate": 7.0967366999388e-07,
      "loss": 1.1275,
      "step": 1271
    },
    {
      "epoch": 0.8808864265927978,
      "grad_norm": 3.7808170318603516,
      "learning_rate": 7.015900970927414e-07,
      "loss": 1.2337,
      "step": 1272
    },
    {
      "epoch": 0.881578947368421,
      "grad_norm": 3.8861465454101562,
      "learning_rate": 6.935511524517835e-07,
      "loss": 1.2613,
      "step": 1273
    },
    {
      "epoch": 0.8822714681440443,
      "grad_norm": 4.897149562835693,
      "learning_rate": 6.85556874654395e-07,
      "loss": 1.0766,
      "step": 1274
    },
    {
      "epoch": 0.8829639889196675,
      "grad_norm": 3.961945056915283,
      "learning_rate": 6.776073020695728e-07,
      "loss": 1.178,
      "step": 1275
    },
    {
      "epoch": 0.8836565096952909,
      "grad_norm": 3.9338932037353516,
      "learning_rate": 6.697024728517531e-07,
      "loss": 1.2335,
      "step": 1276
    },
    {
      "epoch": 0.8843490304709142,
      "grad_norm": 4.2958598136901855,
      "learning_rate": 6.618424249406297e-07,
      "loss": 1.1101,
      "step": 1277
    },
    {
      "epoch": 0.8850415512465374,
      "grad_norm": 4.226284980773926,
      "learning_rate": 6.540271960609568e-07,
      "loss": 1.1793,
      "step": 1278
    },
    {
      "epoch": 0.8857340720221607,
      "grad_norm": 3.9522910118103027,
      "learning_rate": 6.462568237223787e-07,
      "loss": 1.2444,
      "step": 1279
    },
    {
      "epoch": 0.8864265927977839,
      "grad_norm": 4.591524600982666,
      "learning_rate": 6.385313452192554e-07,
      "loss": 1.233,
      "step": 1280
    },
    {
      "epoch": 0.8871191135734072,
      "grad_norm": 3.602266550064087,
      "learning_rate": 6.308507976304701e-07,
      "loss": 1.1108,
      "step": 1281
    },
    {
      "epoch": 0.8878116343490304,
      "grad_norm": 4.180389881134033,
      "learning_rate": 6.23215217819253e-07,
      "loss": 1.2063,
      "step": 1282
    },
    {
      "epoch": 0.8885041551246537,
      "grad_norm": 3.8051416873931885,
      "learning_rate": 6.156246424330215e-07,
      "loss": 1.3125,
      "step": 1283
    },
    {
      "epoch": 0.889196675900277,
      "grad_norm": 4.475870609283447,
      "learning_rate": 6.080791079031811e-07,
      "loss": 1.2985,
      "step": 1284
    },
    {
      "epoch": 0.8898891966759003,
      "grad_norm": 4.486598968505859,
      "learning_rate": 6.005786504449651e-07,
      "loss": 1.1294,
      "step": 1285
    },
    {
      "epoch": 0.8905817174515236,
      "grad_norm": 4.49049711227417,
      "learning_rate": 5.93123306057255e-07,
      "loss": 1.1075,
      "step": 1286
    },
    {
      "epoch": 0.8912742382271468,
      "grad_norm": 3.9002959728240967,
      "learning_rate": 5.857131105224123e-07,
      "loss": 1.2577,
      "step": 1287
    },
    {
      "epoch": 0.8919667590027701,
      "grad_norm": 5.098270416259766,
      "learning_rate": 5.783480994061019e-07,
      "loss": 1.137,
      "step": 1288
    },
    {
      "epoch": 0.8926592797783933,
      "grad_norm": 4.351301193237305,
      "learning_rate": 5.710283080571233e-07,
      "loss": 1.3018,
      "step": 1289
    },
    {
      "epoch": 0.8933518005540166,
      "grad_norm": 3.7987148761749268,
      "learning_rate": 5.637537716072416e-07,
      "loss": 1.322,
      "step": 1290
    },
    {
      "epoch": 0.8940443213296398,
      "grad_norm": 3.730034589767456,
      "learning_rate": 5.565245249710194e-07,
      "loss": 1.289,
      "step": 1291
    },
    {
      "epoch": 0.8947368421052632,
      "grad_norm": 3.9326162338256836,
      "learning_rate": 5.49340602845646e-07,
      "loss": 1.0907,
      "step": 1292
    },
    {
      "epoch": 0.8954293628808865,
      "grad_norm": 4.038126468658447,
      "learning_rate": 5.422020397107753e-07,
      "loss": 1.2528,
      "step": 1293
    },
    {
      "epoch": 0.8961218836565097,
      "grad_norm": 3.6822099685668945,
      "learning_rate": 5.351088698283557e-07,
      "loss": 1.2576,
      "step": 1294
    },
    {
      "epoch": 0.896814404432133,
      "grad_norm": 4.149168014526367,
      "learning_rate": 5.280611272424696e-07,
      "loss": 1.2217,
      "step": 1295
    },
    {
      "epoch": 0.8975069252077562,
      "grad_norm": 3.3951871395111084,
      "learning_rate": 5.21058845779171e-07,
      "loss": 1.27,
      "step": 1296
    },
    {
      "epoch": 0.8981994459833795,
      "grad_norm": 3.9425158500671387,
      "learning_rate": 5.141020590463142e-07,
      "loss": 1.1927,
      "step": 1297
    },
    {
      "epoch": 0.8988919667590027,
      "grad_norm": 3.875690460205078,
      "learning_rate": 5.071908004334025e-07,
      "loss": 1.2137,
      "step": 1298
    },
    {
      "epoch": 0.899584487534626,
      "grad_norm": 4.035533905029297,
      "learning_rate": 5.003251031114287e-07,
      "loss": 1.2056,
      "step": 1299
    },
    {
      "epoch": 0.9002770083102493,
      "grad_norm": 4.189031600952148,
      "learning_rate": 4.935050000327046e-07,
      "loss": 1.2325,
      "step": 1300
    },
    {
      "epoch": 0.9009695290858726,
      "grad_norm": 3.710629940032959,
      "learning_rate": 4.867305239307096e-07,
      "loss": 1.1215,
      "step": 1301
    },
    {
      "epoch": 0.9016620498614959,
      "grad_norm": 3.9204201698303223,
      "learning_rate": 4.800017073199403e-07,
      "loss": 1.2987,
      "step": 1302
    },
    {
      "epoch": 0.9023545706371191,
      "grad_norm": 4.115154266357422,
      "learning_rate": 4.73318582495742e-07,
      "loss": 1.1934,
      "step": 1303
    },
    {
      "epoch": 0.9030470914127424,
      "grad_norm": 4.008752822875977,
      "learning_rate": 4.666811815341643e-07,
      "loss": 0.9978,
      "step": 1304
    },
    {
      "epoch": 0.9037396121883656,
      "grad_norm": 4.517586708068848,
      "learning_rate": 4.6008953629179676e-07,
      "loss": 1.2414,
      "step": 1305
    },
    {
      "epoch": 0.9044321329639889,
      "grad_norm": 3.903789520263672,
      "learning_rate": 4.5354367840562684e-07,
      "loss": 1.2271,
      "step": 1306
    },
    {
      "epoch": 0.9051246537396122,
      "grad_norm": 3.9774749279022217,
      "learning_rate": 4.470436392928812e-07,
      "loss": 1.2613,
      "step": 1307
    },
    {
      "epoch": 0.9058171745152355,
      "grad_norm": 4.200412273406982,
      "learning_rate": 4.4058945015087453e-07,
      "loss": 1.2716,
      "step": 1308
    },
    {
      "epoch": 0.9065096952908587,
      "grad_norm": 4.395098686218262,
      "learning_rate": 4.341811419568653e-07,
      "loss": 1.1752,
      "step": 1309
    },
    {
      "epoch": 0.907202216066482,
      "grad_norm": 4.11729097366333,
      "learning_rate": 4.278187454679006e-07,
      "loss": 1.2215,
      "step": 1310
    },
    {
      "epoch": 0.9078947368421053,
      "grad_norm": 4.524949550628662,
      "learning_rate": 4.2150229122067565e-07,
      "loss": 1.1198,
      "step": 1311
    },
    {
      "epoch": 0.9085872576177285,
      "grad_norm": 3.813676595687866,
      "learning_rate": 4.1523180953137785e-07,
      "loss": 1.2266,
      "step": 1312
    },
    {
      "epoch": 0.9092797783933518,
      "grad_norm": 3.7839903831481934,
      "learning_rate": 4.090073304955511e-07,
      "loss": 1.2149,
      "step": 1313
    },
    {
      "epoch": 0.909972299168975,
      "grad_norm": 3.726961374282837,
      "learning_rate": 4.0282888398794353e-07,
      "loss": 1.2636,
      "step": 1314
    },
    {
      "epoch": 0.9106648199445984,
      "grad_norm": 3.8480069637298584,
      "learning_rate": 3.9669649966237347e-07,
      "loss": 1.1296,
      "step": 1315
    },
    {
      "epoch": 0.9113573407202216,
      "grad_norm": 4.21983528137207,
      "learning_rate": 3.906102069515727e-07,
      "loss": 1.181,
      "step": 1316
    },
    {
      "epoch": 0.9120498614958449,
      "grad_norm": 4.2239298820495605,
      "learning_rate": 3.845700350670567e-07,
      "loss": 1.1858,
      "step": 1317
    },
    {
      "epoch": 0.9127423822714681,
      "grad_norm": 4.391743183135986,
      "learning_rate": 3.785760129989868e-07,
      "loss": 1.1603,
      "step": 1318
    },
    {
      "epoch": 0.9134349030470914,
      "grad_norm": 4.167227745056152,
      "learning_rate": 3.7262816951602057e-07,
      "loss": 1.1548,
      "step": 1319
    },
    {
      "epoch": 0.9141274238227147,
      "grad_norm": 3.850712537765503,
      "learning_rate": 3.6672653316517595e-07,
      "loss": 1.2399,
      "step": 1320
    },
    {
      "epoch": 0.9148199445983379,
      "grad_norm": 4.206256866455078,
      "learning_rate": 3.6087113227170287e-07,
      "loss": 1.202,
      "step": 1321
    },
    {
      "epoch": 0.9155124653739612,
      "grad_norm": 4.473740577697754,
      "learning_rate": 3.5506199493894e-07,
      "loss": 1.2319,
      "step": 1322
    },
    {
      "epoch": 0.9162049861495845,
      "grad_norm": 3.7945244312286377,
      "learning_rate": 3.4929914904817677e-07,
      "loss": 1.2128,
      "step": 1323
    },
    {
      "epoch": 0.9168975069252078,
      "grad_norm": 3.6557018756866455,
      "learning_rate": 3.4358262225853255e-07,
      "loss": 1.1691,
      "step": 1324
    },
    {
      "epoch": 0.917590027700831,
      "grad_norm": 4.495970726013184,
      "learning_rate": 3.379124420068081e-07,
      "loss": 1.2153,
      "step": 1325
    },
    {
      "epoch": 0.9182825484764543,
      "grad_norm": 3.9685280323028564,
      "learning_rate": 3.32288635507364e-07,
      "loss": 1.2664,
      "step": 1326
    },
    {
      "epoch": 0.9189750692520776,
      "grad_norm": 3.6302640438079834,
      "learning_rate": 3.267112297519881e-07,
      "loss": 1.0025,
      "step": 1327
    },
    {
      "epoch": 0.9196675900277008,
      "grad_norm": 3.974372386932373,
      "learning_rate": 3.2118025150976394e-07,
      "loss": 1.1851,
      "step": 1328
    },
    {
      "epoch": 0.9203601108033241,
      "grad_norm": 3.8948326110839844,
      "learning_rate": 3.156957273269434e-07,
      "loss": 1.1475,
      "step": 1329
    },
    {
      "epoch": 0.9210526315789473,
      "grad_norm": 3.5848686695098877,
      "learning_rate": 3.102576835268212e-07,
      "loss": 1.2402,
      "step": 1330
    },
    {
      "epoch": 0.9217451523545707,
      "grad_norm": 3.9654040336608887,
      "learning_rate": 3.0486614620960476e-07,
      "loss": 1.1549,
      "step": 1331
    },
    {
      "epoch": 0.9224376731301939,
      "grad_norm": 3.918294668197632,
      "learning_rate": 2.995211412522914e-07,
      "loss": 1.1275,
      "step": 1332
    },
    {
      "epoch": 0.9231301939058172,
      "grad_norm": 4.397765159606934,
      "learning_rate": 2.942226943085424e-07,
      "loss": 1.1402,
      "step": 1333
    },
    {
      "epoch": 0.9238227146814404,
      "grad_norm": 4.527344703674316,
      "learning_rate": 2.8897083080856703e-07,
      "loss": 1.0616,
      "step": 1334
    },
    {
      "epoch": 0.9245152354570637,
      "grad_norm": 3.974189519882202,
      "learning_rate": 2.8376557595898635e-07,
      "loss": 1.1935,
      "step": 1335
    },
    {
      "epoch": 0.925207756232687,
      "grad_norm": 4.0205817222595215,
      "learning_rate": 2.786069547427239e-07,
      "loss": 1.2264,
      "step": 1336
    },
    {
      "epoch": 0.9259002770083102,
      "grad_norm": 4.807229518890381,
      "learning_rate": 2.7349499191888675e-07,
      "loss": 1.095,
      "step": 1337
    },
    {
      "epoch": 0.9265927977839336,
      "grad_norm": 4.34225606918335,
      "learning_rate": 2.6842971202263536e-07,
      "loss": 1.0819,
      "step": 1338
    },
    {
      "epoch": 0.9272853185595568,
      "grad_norm": 4.332900047302246,
      "learning_rate": 2.634111393650751e-07,
      "loss": 1.0125,
      "step": 1339
    },
    {
      "epoch": 0.9279778393351801,
      "grad_norm": 3.523615598678589,
      "learning_rate": 2.584392980331396e-07,
      "loss": 1.1952,
      "step": 1340
    },
    {
      "epoch": 0.9286703601108033,
      "grad_norm": 4.648331642150879,
      "learning_rate": 2.5351421188947287e-07,
      "loss": 0.9513,
      "step": 1341
    },
    {
      "epoch": 0.9293628808864266,
      "grad_norm": 4.426600456237793,
      "learning_rate": 2.4863590457230745e-07,
      "loss": 1.124,
      "step": 1342
    },
    {
      "epoch": 0.9300554016620498,
      "grad_norm": 4.5962233543396,
      "learning_rate": 2.438043994953687e-07,
      "loss": 1.1684,
      "step": 1343
    },
    {
      "epoch": 0.9307479224376731,
      "grad_norm": 3.893246650695801,
      "learning_rate": 2.3901971984774394e-07,
      "loss": 1.2581,
      "step": 1344
    },
    {
      "epoch": 0.9314404432132964,
      "grad_norm": 5.20170259475708,
      "learning_rate": 2.3428188859378253e-07,
      "loss": 0.9631,
      "step": 1345
    },
    {
      "epoch": 0.9321329639889196,
      "grad_norm": 3.8217201232910156,
      "learning_rate": 2.2959092847298358e-07,
      "loss": 1.1847,
      "step": 1346
    },
    {
      "epoch": 0.932825484764543,
      "grad_norm": 4.310759544372559,
      "learning_rate": 2.2494686199988069e-07,
      "loss": 1.2011,
      "step": 1347
    },
    {
      "epoch": 0.9335180055401662,
      "grad_norm": 3.6743648052215576,
      "learning_rate": 2.2034971146394302e-07,
      "loss": 1.324,
      "step": 1348
    },
    {
      "epoch": 0.9342105263157895,
      "grad_norm": 3.7658603191375732,
      "learning_rate": 2.1579949892946206e-07,
      "loss": 1.1968,
      "step": 1349
    },
    {
      "epoch": 0.9349030470914127,
      "grad_norm": 3.821629285812378,
      "learning_rate": 2.1129624623544843e-07,
      "loss": 1.3141,
      "step": 1350
    },
    {
      "epoch": 0.935595567867036,
      "grad_norm": 4.54724645614624,
      "learning_rate": 2.0683997499552632e-07,
      "loss": 1.0023,
      "step": 1351
    },
    {
      "epoch": 0.9362880886426593,
      "grad_norm": 3.9132909774780273,
      "learning_rate": 2.0243070659782705e-07,
      "loss": 1.186,
      "step": 1352
    },
    {
      "epoch": 0.9369806094182825,
      "grad_norm": 3.962965488433838,
      "learning_rate": 1.980684622048945e-07,
      "loss": 1.3023,
      "step": 1353
    },
    {
      "epoch": 0.9376731301939059,
      "grad_norm": 4.44710111618042,
      "learning_rate": 1.937532627535721e-07,
      "loss": 1.2159,
      "step": 1354
    },
    {
      "epoch": 0.9383656509695291,
      "grad_norm": 4.224870681762695,
      "learning_rate": 1.8948512895491156e-07,
      "loss": 1.2596,
      "step": 1355
    },
    {
      "epoch": 0.9390581717451524,
      "grad_norm": 4.837390422821045,
      "learning_rate": 1.8526408129407093e-07,
      "loss": 1.1943,
      "step": 1356
    },
    {
      "epoch": 0.9397506925207756,
      "grad_norm": 3.966322422027588,
      "learning_rate": 1.8109014003021453e-07,
      "loss": 1.3276,
      "step": 1357
    },
    {
      "epoch": 0.9404432132963989,
      "grad_norm": 4.658303260803223,
      "learning_rate": 1.7696332519641313e-07,
      "loss": 1.0963,
      "step": 1358
    },
    {
      "epoch": 0.9411357340720221,
      "grad_norm": 3.8974692821502686,
      "learning_rate": 1.7288365659956062e-07,
      "loss": 1.3048,
      "step": 1359
    },
    {
      "epoch": 0.9418282548476454,
      "grad_norm": 3.9904167652130127,
      "learning_rate": 1.6885115382026084e-07,
      "loss": 1.2654,
      "step": 1360
    },
    {
      "epoch": 0.9425207756232687,
      "grad_norm": 3.089048385620117,
      "learning_rate": 1.6486583621274532e-07,
      "loss": 0.6291,
      "step": 1361
    },
    {
      "epoch": 0.943213296398892,
      "grad_norm": 3.594970703125,
      "learning_rate": 1.609277229047801e-07,
      "loss": 1.2119,
      "step": 1362
    },
    {
      "epoch": 0.9439058171745153,
      "grad_norm": 4.219138145446777,
      "learning_rate": 1.5703683279756797e-07,
      "loss": 1.2162,
      "step": 1363
    },
    {
      "epoch": 0.9445983379501385,
      "grad_norm": 3.839979648590088,
      "learning_rate": 1.5319318456566424e-07,
      "loss": 1.246,
      "step": 1364
    },
    {
      "epoch": 0.9452908587257618,
      "grad_norm": 3.8669729232788086,
      "learning_rate": 1.49396796656881e-07,
      "loss": 1.3069,
      "step": 1365
    },
    {
      "epoch": 0.945983379501385,
      "grad_norm": 4.179749965667725,
      "learning_rate": 1.4564768729220414e-07,
      "loss": 1.2152,
      "step": 1366
    },
    {
      "epoch": 0.9466759002770083,
      "grad_norm": 3.3702120780944824,
      "learning_rate": 1.4194587446570206e-07,
      "loss": 1.1243,
      "step": 1367
    },
    {
      "epoch": 0.9473684210526315,
      "grad_norm": 4.31612491607666,
      "learning_rate": 1.3829137594444154e-07,
      "loss": 1.3198,
      "step": 1368
    },
    {
      "epoch": 0.9480609418282548,
      "grad_norm": 4.114274501800537,
      "learning_rate": 1.3468420926840198e-07,
      "loss": 1.2286,
      "step": 1369
    },
    {
      "epoch": 0.9487534626038782,
      "grad_norm": 3.7707936763763428,
      "learning_rate": 1.3112439175038794e-07,
      "loss": 1.1814,
      "step": 1370
    },
    {
      "epoch": 0.9494459833795014,
      "grad_norm": 4.673649787902832,
      "learning_rate": 1.276119404759535e-07,
      "loss": 1.159,
      "step": 1371
    },
    {
      "epoch": 0.9501385041551247,
      "grad_norm": 4.296148777008057,
      "learning_rate": 1.2414687230331124e-07,
      "loss": 1.2115,
      "step": 1372
    },
    {
      "epoch": 0.9508310249307479,
      "grad_norm": 4.370774269104004,
      "learning_rate": 1.2072920386325793e-07,
      "loss": 1.2973,
      "step": 1373
    },
    {
      "epoch": 0.9515235457063712,
      "grad_norm": 4.4123029708862305,
      "learning_rate": 1.1735895155909338e-07,
      "loss": 1.2401,
      "step": 1374
    },
    {
      "epoch": 0.9522160664819944,
      "grad_norm": 4.202178955078125,
      "learning_rate": 1.1403613156654058e-07,
      "loss": 1.1944,
      "step": 1375
    },
    {
      "epoch": 0.9529085872576177,
      "grad_norm": 3.994661808013916,
      "learning_rate": 1.1076075983366574e-07,
      "loss": 1.2576,
      "step": 1376
    },
    {
      "epoch": 0.953601108033241,
      "grad_norm": 3.8380768299102783,
      "learning_rate": 1.075328520808061e-07,
      "loss": 1.2041,
      "step": 1377
    },
    {
      "epoch": 0.9542936288088643,
      "grad_norm": 4.077901840209961,
      "learning_rate": 1.043524238004956e-07,
      "loss": 1.1962,
      "step": 1378
    },
    {
      "epoch": 0.9549861495844876,
      "grad_norm": 3.8877198696136475,
      "learning_rate": 1.0121949025738486e-07,
      "loss": 1.189,
      "step": 1379
    },
    {
      "epoch": 0.9556786703601108,
      "grad_norm": 3.7698280811309814,
      "learning_rate": 9.813406648816804e-08,
      "loss": 1.231,
      "step": 1380
    },
    {
      "epoch": 0.9563711911357341,
      "grad_norm": 3.8442647457122803,
      "learning_rate": 9.509616730151827e-08,
      "loss": 1.2865,
      "step": 1381
    },
    {
      "epoch": 0.9570637119113573,
      "grad_norm": 3.9559361934661865,
      "learning_rate": 9.21058072780101e-08,
      "loss": 1.1376,
      "step": 1382
    },
    {
      "epoch": 0.9577562326869806,
      "grad_norm": 3.7578396797180176,
      "learning_rate": 8.916300077005057e-08,
      "loss": 1.1069,
      "step": 1383
    },
    {
      "epoch": 0.9584487534626038,
      "grad_norm": 4.029587745666504,
      "learning_rate": 8.62677619018104e-08,
      "loss": 1.254,
      "step": 1384
    },
    {
      "epoch": 0.9591412742382271,
      "grad_norm": 4.258833408355713,
      "learning_rate": 8.342010456915739e-08,
      "loss": 1.2295,
      "step": 1385
    },
    {
      "epoch": 0.9598337950138505,
      "grad_norm": 4.724644660949707,
      "learning_rate": 8.062004243958866e-08,
      "loss": 1.0733,
      "step": 1386
    },
    {
      "epoch": 0.9605263157894737,
      "grad_norm": 4.294256210327148,
      "learning_rate": 7.786758895216629e-08,
      "loss": 1.1054,
      "step": 1387
    },
    {
      "epoch": 0.961218836565097,
      "grad_norm": 5.140510559082031,
      "learning_rate": 7.51627573174507e-08,
      "loss": 1.1778,
      "step": 1388
    },
    {
      "epoch": 0.9619113573407202,
      "grad_norm": 3.909797191619873,
      "learning_rate": 7.250556051743962e-08,
      "loss": 1.2818,
      "step": 1389
    },
    {
      "epoch": 0.9626038781163435,
      "grad_norm": 4.233206272125244,
      "learning_rate": 6.98960113055025e-08,
      "loss": 1.2068,
      "step": 1390
    },
    {
      "epoch": 0.9632963988919667,
      "grad_norm": 4.595902919769287,
      "learning_rate": 6.733412220632396e-08,
      "loss": 1.0369,
      "step": 1391
    },
    {
      "epoch": 0.96398891966759,
      "grad_norm": 3.9723401069641113,
      "learning_rate": 6.481990551583939e-08,
      "loss": 1.1125,
      "step": 1392
    },
    {
      "epoch": 0.9646814404432132,
      "grad_norm": 4.163870811462402,
      "learning_rate": 6.235337330117829e-08,
      "loss": 1.2698,
      "step": 1393
    },
    {
      "epoch": 0.9653739612188366,
      "grad_norm": 3.5928964614868164,
      "learning_rate": 5.993453740060773e-08,
      "loss": 1.2333,
      "step": 1394
    },
    {
      "epoch": 0.9660664819944599,
      "grad_norm": 4.822768211364746,
      "learning_rate": 5.756340942346894e-08,
      "loss": 1.0382,
      "step": 1395
    },
    {
      "epoch": 0.9667590027700831,
      "grad_norm": 3.9965407848358154,
      "learning_rate": 5.5240000750129695e-08,
      "loss": 1.0812,
      "step": 1396
    },
    {
      "epoch": 0.9674515235457064,
      "grad_norm": 3.666273832321167,
      "learning_rate": 5.29643225319254e-08,
      "loss": 1.2505,
      "step": 1397
    },
    {
      "epoch": 0.9681440443213296,
      "grad_norm": 4.149451732635498,
      "learning_rate": 5.0736385691106946e-08,
      "loss": 1.2226,
      "step": 1398
    },
    {
      "epoch": 0.9688365650969529,
      "grad_norm": 3.7079765796661377,
      "learning_rate": 4.8556200920786276e-08,
      "loss": 1.1604,
      "step": 1399
    },
    {
      "epoch": 0.9695290858725761,
      "grad_norm": 3.9731240272521973,
      "learning_rate": 4.642377868488646e-08,
      "loss": 1.2417,
      "step": 1400
    },
    {
      "epoch": 0.9702216066481995,
      "grad_norm": 4.004926681518555,
      "learning_rate": 4.4339129218095025e-08,
      "loss": 1.1581,
      "step": 1401
    },
    {
      "epoch": 0.9709141274238227,
      "grad_norm": 3.550807237625122,
      "learning_rate": 4.230226252580516e-08,
      "loss": 1.2245,
      "step": 1402
    },
    {
      "epoch": 0.971606648199446,
      "grad_norm": 3.742629289627075,
      "learning_rate": 4.031318838407905e-08,
      "loss": 1.2273,
      "step": 1403
    },
    {
      "epoch": 0.9722991689750693,
      "grad_norm": 3.9137470722198486,
      "learning_rate": 3.837191633959458e-08,
      "loss": 1.1456,
      "step": 1404
    },
    {
      "epoch": 0.9729916897506925,
      "grad_norm": 3.960930585861206,
      "learning_rate": 3.6478455709598735e-08,
      "loss": 1.1765,
      "step": 1405
    },
    {
      "epoch": 0.9736842105263158,
      "grad_norm": 3.7228150367736816,
      "learning_rate": 3.4632815581866484e-08,
      "loss": 1.1441,
      "step": 1406
    },
    {
      "epoch": 0.974376731301939,
      "grad_norm": 3.5775821208953857,
      "learning_rate": 3.28350048146564e-08,
      "loss": 1.2506,
      "step": 1407
    },
    {
      "epoch": 0.9750692520775623,
      "grad_norm": 3.720226526260376,
      "learning_rate": 3.108503203666402e-08,
      "loss": 1.1547,
      "step": 1408
    },
    {
      "epoch": 0.9757617728531855,
      "grad_norm": 3.89373517036438,
      "learning_rate": 2.9382905646986316e-08,
      "loss": 1.3248,
      "step": 1409
    },
    {
      "epoch": 0.9764542936288089,
      "grad_norm": 4.407764434814453,
      "learning_rate": 2.7728633815079508e-08,
      "loss": 1.2345,
      "step": 1410
    },
    {
      "epoch": 0.9771468144044322,
      "grad_norm": 4.202113628387451,
      "learning_rate": 2.6122224480715774e-08,
      "loss": 1.093,
      "step": 1411
    },
    {
      "epoch": 0.9778393351800554,
      "grad_norm": 4.07959508895874,
      "learning_rate": 2.456368535394993e-08,
      "loss": 1.3626,
      "step": 1412
    },
    {
      "epoch": 0.9785318559556787,
      "grad_norm": 3.8587090969085693,
      "learning_rate": 2.3053023915083904e-08,
      "loss": 1.3964,
      "step": 1413
    },
    {
      "epoch": 0.9792243767313019,
      "grad_norm": 3.3332502841949463,
      "learning_rate": 2.1590247414624566e-08,
      "loss": 1.1588,
      "step": 1414
    },
    {
      "epoch": 0.9799168975069252,
      "grad_norm": 3.622792959213257,
      "learning_rate": 2.0175362873250394e-08,
      "loss": 1.2278,
      "step": 1415
    },
    {
      "epoch": 0.9806094182825484,
      "grad_norm": 3.7608649730682373,
      "learning_rate": 1.8808377081785954e-08,
      "loss": 1.2254,
      "step": 1416
    },
    {
      "epoch": 0.9813019390581718,
      "grad_norm": 2.7306807041168213,
      "learning_rate": 1.7489296601156392e-08,
      "loss": 0.6271,
      "step": 1417
    },
    {
      "epoch": 0.981994459833795,
      "grad_norm": 4.163548946380615,
      "learning_rate": 1.62181277623652e-08,
      "loss": 1.2184,
      "step": 1418
    },
    {
      "epoch": 0.9826869806094183,
      "grad_norm": 3.852370023727417,
      "learning_rate": 1.4994876666464263e-08,
      "loss": 1.3356,
      "step": 1419
    },
    {
      "epoch": 0.9833795013850416,
      "grad_norm": 5.447317123413086,
      "learning_rate": 1.3819549184516113e-08,
      "loss": 1.1542,
      "step": 1420
    },
    {
      "epoch": 0.9840720221606648,
      "grad_norm": 4.445632457733154,
      "learning_rate": 1.2692150957572813e-08,
      "loss": 1.3178,
      "step": 1421
    },
    {
      "epoch": 0.9847645429362881,
      "grad_norm": 3.9901273250579834,
      "learning_rate": 1.1612687396650445e-08,
      "loss": 1.2509,
      "step": 1422
    },
    {
      "epoch": 0.9854570637119113,
      "grad_norm": 3.6384527683258057,
      "learning_rate": 1.0581163682695795e-08,
      "loss": 1.1513,
      "step": 1423
    },
    {
      "epoch": 0.9861495844875346,
      "grad_norm": 4.033609867095947,
      "learning_rate": 9.59758476656636e-09,
      "loss": 1.1859,
      "step": 1424
    },
    {
      "epoch": 0.9868421052631579,
      "grad_norm": 4.010962963104248,
      "learning_rate": 8.661955369007047e-09,
      "loss": 1.1084,
      "step": 1425
    },
    {
      "epoch": 0.9875346260387812,
      "grad_norm": 3.835806131362915,
      "learning_rate": 7.774279980626853e-09,
      "loss": 1.2522,
      "step": 1426
    },
    {
      "epoch": 0.9882271468144044,
      "grad_norm": 3.7668678760528564,
      "learning_rate": 6.9345628618744384e-09,
      "loss": 1.2265,
      "step": 1427
    },
    {
      "epoch": 0.9889196675900277,
      "grad_norm": 3.6999645233154297,
      "learning_rate": 6.142808043020365e-09,
      "loss": 1.166,
      "step": 1428
    },
    {
      "epoch": 0.989612188365651,
      "grad_norm": 3.9178340435028076,
      "learning_rate": 5.399019324139332e-09,
      "loss": 1.2671,
      "step": 1429
    },
    {
      "epoch": 0.9903047091412742,
      "grad_norm": 4.232564449310303,
      "learning_rate": 4.703200275087971e-09,
      "loss": 1.3597,
      "step": 1430
    },
    {
      "epoch": 0.9909972299168975,
      "grad_norm": 3.4954841136932373,
      "learning_rate": 4.055354235490416e-09,
      "loss": 1.2857,
      "step": 1431
    },
    {
      "epoch": 0.9916897506925207,
      "grad_norm": 4.343502998352051,
      "learning_rate": 3.4554843147216467e-09,
      "loss": 1.1973,
      "step": 1432
    },
    {
      "epoch": 0.9923822714681441,
      "grad_norm": 4.3146772384643555,
      "learning_rate": 2.9035933918919456e-09,
      "loss": 1.1719,
      "step": 1433
    },
    {
      "epoch": 0.9930747922437673,
      "grad_norm": 4.9093852043151855,
      "learning_rate": 2.3996841158346885e-09,
      "loss": 1.0808,
      "step": 1434
    },
    {
      "epoch": 0.9937673130193906,
      "grad_norm": 3.9903743267059326,
      "learning_rate": 1.943758905090798e-09,
      "loss": 1.1361,
      "step": 1435
    },
    {
      "epoch": 0.9944598337950139,
      "grad_norm": 4.14931058883667,
      "learning_rate": 1.535819947900974e-09,
      "loss": 1.1252,
      "step": 1436
    },
    {
      "epoch": 0.9951523545706371,
      "grad_norm": 3.4692118167877197,
      "learning_rate": 1.17586920219126e-09,
      "loss": 1.1593,
      "step": 1437
    },
    {
      "epoch": 0.9958448753462604,
      "grad_norm": 3.6907618045806885,
      "learning_rate": 8.639083955663819e-10,
      "loss": 1.1082,
      "step": 1438
    },
    {
      "epoch": 0.9965373961218836,
      "grad_norm": 3.5076091289520264,
      "learning_rate": 5.999390253008664e-10,
      "loss": 1.1992,
      "step": 1439
    },
    {
      "epoch": 0.997229916897507,
      "grad_norm": 4.372177600860596,
      "learning_rate": 3.839623583301588e-10,
      "loss": 1.1326,
      "step": 1440
    },
    {
      "epoch": 0.9979224376731302,
      "grad_norm": 3.999774932861328,
      "learning_rate": 2.1597943124729293e-10,
      "loss": 1.1831,
      "step": 1441
    },
    {
      "epoch": 0.9986149584487535,
      "grad_norm": 4.036330699920654,
      "learning_rate": 9.599105029622913e-11,
      "loss": 1.2234,
      "step": 1442
    },
    {
      "epoch": 0.9993074792243767,
      "grad_norm": 3.8736438751220703,
      "learning_rate": 2.3997791368524094e-11,
      "loss": 1.329,
      "step": 1443
    },
    {
      "epoch": 1.0,
      "grad_norm": 4.3855085372924805,
      "learning_rate": 0.0,
      "loss": 1.2446,
      "step": 1444
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.126600742340088,
      "eval_runtime": 337.1897,
      "eval_samples_per_second": 4.199,
      "eval_steps_per_second": 0.525,
      "step": 1444
    }
  ],
  "logging_steps": 1,
  "max_steps": 1444,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.9696935369116221e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}