{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.498591082380242,
  "eval_steps": 10,
  "global_step": 282,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0017680534836178794,
      "grad_norm": 107.08050537109375,
      "learning_rate": 2.6333333333333332e-06,
      "loss": 0.0511,
      "step": 1
    },
    {
      "epoch": 0.0035361069672357587,
      "grad_norm": 33.058345794677734,
      "learning_rate": 4.266666666666667e-06,
      "loss": 0.0073,
      "step": 2
    },
    {
      "epoch": 0.005304160450853639,
      "grad_norm": 11.785157203674316,
      "learning_rate": 5.9e-06,
      "loss": 0.0038,
      "step": 3
    },
    {
      "epoch": 0.0070722139344715175,
      "grad_norm": 3.531964063644409,
      "learning_rate": 7.533333333333334e-06,
      "loss": 0.002,
      "step": 4
    },
    {
      "epoch": 0.008840267418089397,
      "grad_norm": 1.3157356977462769,
      "learning_rate": 9.166666666666668e-06,
      "loss": 0.0012,
      "step": 5
    },
    {
      "epoch": 0.010608320901707277,
      "grad_norm": 0.22130531072616577,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.0009,
      "step": 6
    },
    {
      "epoch": 0.012376374385325156,
      "grad_norm": 0.16746075451374054,
      "learning_rate": 1.2433333333333335e-05,
      "loss": 0.0009,
      "step": 7
    },
    {
      "epoch": 0.014144427868943035,
      "grad_norm": 0.2485613077878952,
      "learning_rate": 1.4066666666666669e-05,
      "loss": 0.0008,
      "step": 8
    },
    {
      "epoch": 0.015912481352560914,
      "grad_norm": 0.19209939241409302,
      "learning_rate": 1.5700000000000002e-05,
      "loss": 0.0008,
      "step": 9
    },
    {
      "epoch": 0.017680534836178794,
      "grad_norm": 2.578878402709961,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.0008,
      "step": 10
    },
    {
      "epoch": 0.019448588319796675,
      "grad_norm": 19.695049285888672,
      "learning_rate": 1.896666666666667e-05,
      "loss": 0.0014,
      "step": 11
    },
    {
      "epoch": 0.021216641803414555,
      "grad_norm": 0.31556376814842224,
      "learning_rate": 2.0600000000000003e-05,
      "loss": 0.0007,
      "step": 12
    },
    {
      "epoch": 0.022984695287032432,
      "grad_norm": 34.17892074584961,
      "learning_rate": 2.2233333333333336e-05,
      "loss": 0.0008,
      "step": 13
    },
    {
      "epoch": 0.024752748770650312,
      "grad_norm": 0.9590745568275452,
      "learning_rate": 2.386666666666667e-05,
      "loss": 0.0002,
      "step": 14
    },
    {
      "epoch": 0.026520802254268193,
      "grad_norm": 7.85075044631958,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 0.0008,
      "step": 15
    },
    {
      "epoch": 0.02828885573788607,
      "grad_norm": 1.8065848350524902,
      "learning_rate": 2.7133333333333337e-05,
      "loss": 0.0003,
      "step": 16
    },
    {
      "epoch": 0.03005690922150395,
      "grad_norm": 0.3425370454788208,
      "learning_rate": 2.876666666666667e-05,
      "loss": 0.0,
      "step": 17
    },
    {
      "epoch": 0.03182496270512183,
      "grad_norm": 1.7299435138702393,
      "learning_rate": 3.0400000000000004e-05,
      "loss": 0.0001,
      "step": 18
    },
    {
      "epoch": 0.03359301618873971,
      "grad_norm": 0.37139496207237244,
      "learning_rate": 3.203333333333333e-05,
      "loss": 0.0,
      "step": 19
    },
    {
      "epoch": 0.03536106967235759,
      "grad_norm": 1.1884018182754517,
      "learning_rate": 3.366666666666667e-05,
      "loss": 0.0,
      "step": 20
    },
    {
      "epoch": 0.03712912315597547,
      "grad_norm": 0.018331818282604218,
      "learning_rate": 3.53e-05,
      "loss": 0.0,
      "step": 21
    },
    {
      "epoch": 0.03889717663959335,
      "grad_norm": 0.046326447278261185,
      "learning_rate": 3.6933333333333334e-05,
      "loss": 0.0,
      "step": 22
    },
    {
      "epoch": 0.04066523012321123,
      "grad_norm": 0.041994426399469376,
      "learning_rate": 3.856666666666667e-05,
      "loss": 0.0,
      "step": 23
    },
    {
      "epoch": 0.04243328360682911,
      "grad_norm": 5.286490440368652,
      "learning_rate": 4.02e-05,
      "loss": 0.0001,
      "step": 24
    },
    {
      "epoch": 0.044201337090446984,
      "grad_norm": 0.23539425432682037,
      "learning_rate": 4.183333333333334e-05,
      "loss": 0.0,
      "step": 25
    },
    {
      "epoch": 0.045969390574064864,
      "grad_norm": 22.727415084838867,
      "learning_rate": 4.346666666666667e-05,
      "loss": 0.0012,
      "step": 26
    },
    {
      "epoch": 0.047737444057682744,
      "grad_norm": 0.29610776901245117,
      "learning_rate": 4.5100000000000005e-05,
      "loss": 0.0001,
      "step": 27
    },
    {
      "epoch": 0.049505497541300625,
      "grad_norm": 319.07232666015625,
      "learning_rate": 4.6733333333333335e-05,
      "loss": 0.0006,
      "step": 28
    },
    {
      "epoch": 0.051273551024918505,
      "grad_norm": 430.6841125488281,
      "learning_rate": 4.836666666666667e-05,
      "loss": 0.0221,
      "step": 29
    },
    {
      "epoch": 0.053041604508536386,
      "grad_norm": 3.6946537494659424,
      "learning_rate": 5e-05,
      "loss": 0.0002,
      "step": 30
    },
    {
      "epoch": 0.05480965799215426,
      "grad_norm": 45.59299850463867,
      "learning_rate": 4.999834154609218e-05,
      "loss": 0.0035,
      "step": 31
    },
    {
      "epoch": 0.05657771147577214,
      "grad_norm": 32.24581527709961,
      "learning_rate": 4.999336640889681e-05,
      "loss": 0.0032,
      "step": 32
    },
    {
      "epoch": 0.05834576495939002,
      "grad_norm": 58.2963981628418,
      "learning_rate": 4.998507526196785e-05,
      "loss": 0.003,
      "step": 33
    },
    {
      "epoch": 0.0601138184430079,
      "grad_norm": 3.416010856628418,
      "learning_rate": 4.997346922779386e-05,
      "loss": 0.0005,
      "step": 34
    },
    {
      "epoch": 0.06188187192662578,
      "grad_norm": 0.8263393640518188,
      "learning_rate": 4.9958549877646073e-05,
      "loss": 0.0002,
      "step": 35
    },
    {
      "epoch": 0.06364992541024365,
      "grad_norm": 0.5320990681648254,
      "learning_rate": 4.994031923136569e-05,
      "loss": 0.0,
      "step": 36
    },
    {
      "epoch": 0.06541797889386154,
      "grad_norm": 0.006872811354696751,
      "learning_rate": 4.99187797570904e-05,
      "loss": 0.0,
      "step": 37
    },
    {
      "epoch": 0.06718603237747942,
      "grad_norm": 1.7324073314666748,
      "learning_rate": 4.9893934370920207e-05,
      "loss": 0.0001,
      "step": 38
    },
    {
      "epoch": 0.0689540858610973,
      "grad_norm": 17.383434295654297,
      "learning_rate": 4.98657864365227e-05,
      "loss": 0.0019,
      "step": 39
    },
    {
      "epoch": 0.07072213934471518,
      "grad_norm": 397.38128662109375,
      "learning_rate": 4.9834339764677606e-05,
      "loss": 0.0127,
      "step": 40
    },
    {
      "epoch": 0.07249019282833306,
      "grad_norm": 9.999519348144531,
      "learning_rate": 4.979959861276091e-05,
      "loss": 0.0006,
      "step": 41
    },
    {
      "epoch": 0.07425824631195094,
      "grad_norm": 1.2901943922042847,
      "learning_rate": 4.976156768416848e-05,
      "loss": 0.0001,
      "step": 42
    },
    {
      "epoch": 0.07602629979556881,
      "grad_norm": 0.0491662472486496,
      "learning_rate": 4.9720252127679233e-05,
      "loss": 0.0,
      "step": 43
    },
    {
      "epoch": 0.0777943532791867,
      "grad_norm": 0.012911037541925907,
      "learning_rate": 4.96756575367582e-05,
      "loss": 0.0,
      "step": 44
    },
    {
      "epoch": 0.07956240676280457,
      "grad_norm": 0.11798875033855438,
      "learning_rate": 4.96277899487991e-05,
      "loss": 0.0,
      "step": 45
    },
    {
      "epoch": 0.08133046024642246,
      "grad_norm": 4.875470161437988,
      "learning_rate": 4.957665584430713e-05,
      "loss": 0.0006,
      "step": 46
    },
    {
      "epoch": 0.08309851373004033,
      "grad_norm": 3.2761361598968506,
      "learning_rate": 4.9522262146021495e-05,
      "loss": 0.0002,
      "step": 47
    },
    {
      "epoch": 0.08486656721365822,
      "grad_norm": 1.1224943399429321,
      "learning_rate": 4.946461621797824e-05,
      "loss": 0.0,
      "step": 48
    },
    {
      "epoch": 0.0866346206972761,
      "grad_norm": 0.38790443539619446,
      "learning_rate": 4.940372586451325e-05,
      "loss": 0.0,
      "step": 49
    },
    {
      "epoch": 0.08840267418089397,
      "grad_norm": 0.6254971027374268,
      "learning_rate": 4.9339599329205686e-05,
      "loss": 0.0,
      "step": 50
    },
    {
      "epoch": 0.09017072766451185,
      "grad_norm": 0.11508096754550934,
      "learning_rate": 4.927224529376191e-05,
      "loss": 0.0,
      "step": 51
    },
    {
      "epoch": 0.09193878114812973,
      "grad_norm": 4.301520824432373,
      "learning_rate": 4.920167287684016e-05,
      "loss": 0.0001,
      "step": 52
    },
    {
      "epoch": 0.09370683463174762,
      "grad_norm": 0.008904535323381424,
      "learning_rate": 4.912789163281601e-05,
      "loss": 0.0,
      "step": 53
    },
    {
      "epoch": 0.09547488811536549,
      "grad_norm": 0.6658762097358704,
      "learning_rate": 4.905091155048882e-05,
      "loss": 0.0001,
      "step": 54
    },
    {
      "epoch": 0.09724294159898336,
      "grad_norm": 0.008907541632652283,
      "learning_rate": 4.897074305172948e-05,
      "loss": 0.0,
      "step": 55
    },
    {
      "epoch": 0.09901099508260125,
      "grad_norm": 1.161902904510498,
      "learning_rate": 4.8887396990069434e-05,
      "loss": 0.0,
      "step": 56
    },
    {
      "epoch": 0.10077904856621912,
      "grad_norm": 0.017062336206436157,
      "learning_rate": 4.8800884649231264e-05,
      "loss": 0.0,
      "step": 57
    },
    {
      "epoch": 0.10254710204983701,
      "grad_norm": 0.07215119153261185,
      "learning_rate": 4.871121774160107e-05,
      "loss": 0.0,
      "step": 58
    },
    {
      "epoch": 0.10431515553345488,
      "grad_norm": 0.29680925607681274,
      "learning_rate": 4.8618408406642795e-05,
      "loss": 0.0,
      "step": 59
    },
    {
      "epoch": 0.10608320901707277,
      "grad_norm": 1.7219104766845703,
      "learning_rate": 4.852246920925476e-05,
      "loss": 0.0,
      "step": 60
    },
    {
      "epoch": 0.10785126250069065,
      "grad_norm": 0.017782438546419144,
      "learning_rate": 4.842341313806852e-05,
      "loss": 0.0,
      "step": 61
    },
    {
      "epoch": 0.10961931598430852,
      "grad_norm": 26.54680061340332,
      "learning_rate": 4.832125360369049e-05,
      "loss": 0.0015,
      "step": 62
    },
    {
      "epoch": 0.1113873694679264,
      "grad_norm": 0.6519933342933655,
      "learning_rate": 4.82160044368863e-05,
      "loss": 0.0,
      "step": 63
    },
    {
      "epoch": 0.11315542295154428,
      "grad_norm": 7.924159049987793,
      "learning_rate": 4.810767988670834e-05,
      "loss": 0.0035,
      "step": 64
    },
    {
      "epoch": 0.11492347643516217,
      "grad_norm": 0.13337986171245575,
      "learning_rate": 4.799629461856672e-05,
      "loss": 0.0,
      "step": 65
    },
    {
      "epoch": 0.11669152991878004,
      "grad_norm": 19.419836044311523,
      "learning_rate": 4.788186371224372e-05,
      "loss": 0.0012,
      "step": 66
    },
    {
      "epoch": 0.11845958340239793,
      "grad_norm": 0.5985578894615173,
      "learning_rate": 4.776440265985233e-05,
      "loss": 0.0001,
      "step": 67
    },
    {
      "epoch": 0.1202276368860158,
      "grad_norm": 0.4132179319858551,
      "learning_rate": 4.764392736373876e-05,
      "loss": 0.0,
      "step": 68
    },
    {
      "epoch": 0.12199569036963367,
      "grad_norm": 0.14033225178718567,
      "learning_rate": 4.7520454134329594e-05,
      "loss": 0.0,
      "step": 69
    },
    {
      "epoch": 0.12376374385325156,
      "grad_norm": 0.5785059332847595,
      "learning_rate": 4.73939996879236e-05,
      "loss": 0.0001,
      "step": 70
    },
    {
      "epoch": 0.12553179733686945,
      "grad_norm": 0.14392127096652985,
      "learning_rate": 4.72645811444286e-05,
      "loss": 0.0,
      "step": 71
    },
    {
      "epoch": 0.1272998508204873,
      "grad_norm": 1.70956552028656,
      "learning_rate": 4.7132216025043714e-05,
      "loss": 0.0001,
      "step": 72
    },
    {
      "epoch": 0.1290679043041052,
      "grad_norm": 0.11939554661512375,
      "learning_rate": 4.699692224988726e-05,
      "loss": 0.0,
      "step": 73
    },
    {
      "epoch": 0.13083595778772308,
      "grad_norm": 0.33141276240348816,
      "learning_rate": 4.685871813557068e-05,
      "loss": 0.0001,
      "step": 74
    },
    {
      "epoch": 0.13260401127134097,
      "grad_norm": 0.1561572402715683,
      "learning_rate": 4.671762239271875e-05,
      "loss": 0.0,
      "step": 75
    },
    {
      "epoch": 0.13437206475495883,
      "grad_norm": 0.509835422039032,
      "learning_rate": 4.6573654123436456e-05,
      "loss": 0.0001,
      "step": 76
    },
    {
      "epoch": 0.13614011823857672,
      "grad_norm": 0.18685153126716614,
      "learning_rate": 4.642683281872288e-05,
      "loss": 0.0,
      "step": 77
    },
    {
      "epoch": 0.1379081717221946,
      "grad_norm": 0.10680101066827774,
      "learning_rate": 4.6277178355832434e-05,
      "loss": 0.0,
      "step": 78
    },
    {
      "epoch": 0.13967622520581247,
      "grad_norm": 0.0250838790088892,
      "learning_rate": 4.6124710995583807e-05,
      "loss": 0.0,
      "step": 79
    },
    {
      "epoch": 0.14144427868943035,
      "grad_norm": 0.04628512263298035,
      "learning_rate": 4.5969451379616945e-05,
      "loss": 0.0,
      "step": 80
    },
    {
      "epoch": 0.14321233217304824,
      "grad_norm": 1.1579428911209106,
      "learning_rate": 4.581142052759852e-05,
      "loss": 0.0007,
      "step": 81
    },
    {
      "epoch": 0.14498038565666613,
      "grad_norm": 0.3260582685470581,
      "learning_rate": 4.565063983437623e-05,
      "loss": 0.0,
      "step": 82
    },
    {
      "epoch": 0.146748439140284,
      "grad_norm": 0.14016196131706238,
      "learning_rate": 4.548713106708222e-05,
      "loss": 0.0,
      "step": 83
    },
    {
      "epoch": 0.14851649262390187,
      "grad_norm": 0.013404465280473232,
      "learning_rate": 4.532091636218621e-05,
      "loss": 0.0,
      "step": 84
    },
    {
      "epoch": 0.15028454610751976,
      "grad_norm": 0.058658480644226074,
      "learning_rate": 4.5152018222498574e-05,
      "loss": 0.0,
      "step": 85
    },
    {
      "epoch": 0.15205259959113762,
      "grad_norm": 0.058295607566833496,
      "learning_rate": 4.498045951412377e-05,
      "loss": 0.0,
      "step": 86
    },
    {
      "epoch": 0.1538206530747555,
      "grad_norm": 0.040826935321092606,
      "learning_rate": 4.480626346336469e-05,
      "loss": 0.0,
      "step": 87
    },
    {
      "epoch": 0.1555887065583734,
      "grad_norm": 0.5249735116958618,
      "learning_rate": 4.462945365357815e-05,
      "loss": 0.0001,
      "step": 88
    },
    {
      "epoch": 0.15735676004199128,
      "grad_norm": 0.027441998943686485,
      "learning_rate": 4.4450054021982115e-05,
      "loss": 0.0,
      "step": 89
    },
    {
      "epoch": 0.15912481352560914,
      "grad_norm": 0.16359610855579376,
      "learning_rate": 4.426808885641496e-05,
      "loss": 0.0,
      "step": 90
    },
    {
      "epoch": 0.16089286700922703,
      "grad_norm": 0.01867711916565895,
      "learning_rate": 4.408358279204729e-05,
      "loss": 0.0,
      "step": 91
    },
    {
      "epoch": 0.16266092049284492,
      "grad_norm": 0.026866458356380463,
      "learning_rate": 4.389656080804674e-05,
      "loss": 0.0,
      "step": 92
    },
    {
      "epoch": 0.16442897397646278,
      "grad_norm": 0.007205847185105085,
      "learning_rate": 4.370704822419616e-05,
      "loss": 0.0,
      "step": 93
    },
    {
      "epoch": 0.16619702746008067,
      "grad_norm": 0.027548780664801598,
      "learning_rate": 4.3515070697465805e-05,
      "loss": 0.0,
      "step": 94
    },
    {
      "epoch": 0.16796508094369855,
      "grad_norm": 0.03899214044213295,
      "learning_rate": 4.33206542185397e-05,
      "loss": 0.0,
      "step": 95
    },
    {
      "epoch": 0.16973313442731644,
      "grad_norm": 0.006059037055820227,
      "learning_rate": 4.3123825108296954e-05,
      "loss": 0.0,
      "step": 96
    },
    {
      "epoch": 0.1715011879109343,
      "grad_norm": 0.0009920188458636403,
      "learning_rate": 4.292461001424836e-05,
      "loss": 0.0,
      "step": 97
    },
    {
      "epoch": 0.1732692413945522,
      "grad_norm": 0.15524977445602417,
      "learning_rate": 4.272303590692872e-05,
      "loss": 0.0,
      "step": 98
    },
    {
      "epoch": 0.17503729487817007,
      "grad_norm": 0.0014381350483745337,
      "learning_rate": 4.251913007624543e-05,
      "loss": 0.0,
      "step": 99
    },
    {
      "epoch": 0.17680534836178793,
      "grad_norm": 0.2695734202861786,
      "learning_rate": 4.231292012778398e-05,
      "loss": 0.0001,
      "step": 100
    },
    {
      "epoch": 0.17857340184540582,
      "grad_norm": 42.32940673828125,
      "learning_rate": 4.210443397907045e-05,
      "loss": 0.0004,
      "step": 101
    },
    {
      "epoch": 0.1803414553290237,
      "grad_norm": 0.18144920468330383,
      "learning_rate": 4.1893699855792026e-05,
      "loss": 0.0,
      "step": 102
    },
    {
      "epoch": 0.1821095088126416,
      "grad_norm": 0.00621533440425992,
      "learning_rate": 4.1680746287975686e-05,
      "loss": 0.0,
      "step": 103
    },
    {
      "epoch": 0.18387756229625946,
      "grad_norm": 0.011037588119506836,
      "learning_rate": 4.1465602106125656e-05,
      "loss": 0.0,
      "step": 104
    },
    {
      "epoch": 0.18564561577987734,
      "grad_norm": 0.11516346037387848,
      "learning_rate": 4.1248296437320216e-05,
      "loss": 0.0,
      "step": 105
    },
    {
      "epoch": 0.18741366926349523,
      "grad_norm": 0.013171572238206863,
      "learning_rate": 4.10288587012684e-05,
      "loss": 0.0,
      "step": 106
    },
    {
      "epoch": 0.1891817227471131,
      "grad_norm": 0.1711127609014511,
      "learning_rate": 4.080731860632702e-05,
      "loss": 0.0,
      "step": 107
    },
    {
      "epoch": 0.19094977623073098,
      "grad_norm": 0.14865125715732574,
      "learning_rate": 4.0583706145478626e-05,
      "loss": 0.0,
      "step": 108
    },
    {
      "epoch": 0.19271782971434887,
      "grad_norm": 0.03242873027920723,
      "learning_rate": 4.035805159227094e-05,
      "loss": 0.0,
      "step": 109
    },
    {
      "epoch": 0.19448588319796672,
      "grad_norm": 0.14900478720664978,
      "learning_rate": 4.0130385496718264e-05,
      "loss": 0.0,
      "step": 110
    },
    {
      "epoch": 0.1962539366815846,
      "grad_norm": 0.004072085022926331,
      "learning_rate": 3.990073868116559e-05,
      "loss": 0.0,
      "step": 111
    },
    {
      "epoch": 0.1980219901652025,
      "grad_norm": 0.11495672166347504,
      "learning_rate": 3.966914223611567e-05,
      "loss": 0.0001,
      "step": 112
    },
    {
      "epoch": 0.1997900436488204,
      "grad_norm": 0.1237805113196373,
      "learning_rate": 3.943562751601987e-05,
      "loss": 0.0,
      "step": 113
    },
    {
      "epoch": 0.20155809713243825,
      "grad_norm": 0.044289521872997284,
      "learning_rate": 3.92002261350333e-05,
      "loss": 0.0,
      "step": 114
    },
    {
      "epoch": 0.20332615061605613,
      "grad_norm": 0.3086288869380951,
      "learning_rate": 3.896296996273475e-05,
      "loss": 0.0,
      "step": 115
    },
    {
      "epoch": 0.20509420409967402,
      "grad_norm": 0.7167790532112122,
      "learning_rate": 3.872389111981207e-05,
      "loss": 0.0002,
      "step": 116
    },
    {
      "epoch": 0.20686225758329188,
      "grad_norm": 0.01117889303714037,
      "learning_rate": 3.848302197371352e-05,
      "loss": 0.0,
      "step": 117
    },
    {
      "epoch": 0.20863031106690977,
      "grad_norm": 0.9482950568199158,
      "learning_rate": 3.8240395134265775e-05,
      "loss": 0.0002,
      "step": 118
    },
    {
      "epoch": 0.21039836455052766,
      "grad_norm": 0.01158567052334547,
      "learning_rate": 3.799604344925904e-05,
      "loss": 0.0,
      "step": 119
    },
    {
      "epoch": 0.21216641803414554,
      "grad_norm": 0.001482126535847783,
      "learning_rate": 3.775e-05,
      "loss": 0.0,
      "step": 120
    },
    {
      "epoch": 0.2139344715177634,
      "grad_norm": 0.2988652288913727,
      "learning_rate": 3.7502298096833135e-05,
      "loss": 0.0,
      "step": 121
    },
    {
      "epoch": 0.2157025250013813,
      "grad_norm": 0.011216375045478344,
      "learning_rate": 3.725297127463104e-05,
      "loss": 0.0,
      "step": 122
    },
    {
      "epoch": 0.21747057848499918,
      "grad_norm": 1.6700996160507202,
      "learning_rate": 3.7002053288254325e-05,
      "loss": 0.0002,
      "step": 123
    },
    {
      "epoch": 0.21923863196861704,
      "grad_norm": 0.17890863120555878,
      "learning_rate": 3.6749578107981784e-05,
      "loss": 0.0,
      "step": 124
    },
    {
      "epoch": 0.22100668545223492,
      "grad_norm": 0.004812731873244047,
      "learning_rate": 3.649557991491133e-05,
      "loss": 0.0,
      "step": 125
    },
    {
      "epoch": 0.2227747389358528,
      "grad_norm": 0.0008214990375563502,
      "learning_rate": 3.62400930963324e-05,
      "loss": 0.0,
      "step": 126
    },
    {
      "epoch": 0.2245427924194707,
      "grad_norm": 0.005143096670508385,
      "learning_rate": 3.598315224107053e-05,
      "loss": 0.0,
      "step": 127
    },
    {
      "epoch": 0.22631084590308856,
      "grad_norm": 0.14021626114845276,
      "learning_rate": 3.572479213480454e-05,
      "loss": 0.0,
      "step": 128
    },
    {
      "epoch": 0.22807889938670645,
      "grad_norm": 0.0075095584616065025,
      "learning_rate": 3.546504775535711e-05,
      "loss": 0.0,
      "step": 129
    },
    {
      "epoch": 0.22984695287032433,
      "grad_norm": 0.0022524860687553883,
      "learning_rate": 3.5203954267959345e-05,
      "loss": 0.0,
      "step": 130
    },
    {
      "epoch": 0.2316150063539422,
      "grad_norm": 0.0018451135838404298,
      "learning_rate": 3.4941547020489984e-05,
      "loss": 0.0,
      "step": 131
    },
    {
      "epoch": 0.23338305983756008,
      "grad_norm": 0.0029321014881134033,
      "learning_rate": 3.467786153868985e-05,
      "loss": 0.0,
      "step": 132
    },
    {
      "epoch": 0.23515111332117797,
      "grad_norm": 0.0029901517555117607,
      "learning_rate": 3.4412933521352215e-05,
      "loss": 0.0,
      "step": 133
    },
    {
      "epoch": 0.23691916680479586,
      "grad_norm": 0.1540692150592804,
      "learning_rate": 3.414679883548978e-05,
      "loss": 0.0,
      "step": 134
    },
    {
      "epoch": 0.23868722028841372,
      "grad_norm": 0.001123814843595028,
      "learning_rate": 3.387949351147889e-05,
      "loss": 0.0,
      "step": 135
    },
    {
      "epoch": 0.2404552737720316,
      "grad_norm": 0.00034733908250927925,
      "learning_rate": 3.3611053738181474e-05,
      "loss": 0.0,
      "step": 136
    },
    {
      "epoch": 0.2422233272556495,
      "grad_norm": 0.018943598493933678,
      "learning_rate": 3.33415158580458e-05,
      "loss": 0.0,
      "step": 137
    },
    {
      "epoch": 0.24399138073926735,
      "grad_norm": 0.0012388058239594102,
      "learning_rate": 3.307091636218621e-05,
      "loss": 0.0,
      "step": 138
    },
    {
      "epoch": 0.24575943422288524,
      "grad_norm": 0.579187273979187,
      "learning_rate": 3.2799291885442806e-05,
      "loss": 0.0,
      "step": 139
    },
    {
      "epoch": 0.24752748770650312,
      "grad_norm": 0.06531547755002975,
      "learning_rate": 3.2526679201421715e-05,
      "loss": 0.0,
      "step": 140
    },
    {
      "epoch": 0.249295541190121,
      "grad_norm": 0.0009087176294997334,
      "learning_rate": 3.225311521751648e-05,
      "loss": 0.0,
      "step": 141
    },
    {
      "epoch": 0.2510635946737389,
      "grad_norm": 0.0007331356173381209,
      "learning_rate": 3.197863696991146e-05,
      "loss": 0.0,
      "step": 142
    },
    {
      "epoch": 0.25283164815735676,
      "grad_norm": 0.0016905225347727537,
      "learning_rate": 3.170328161856765e-05,
      "loss": 0.0,
      "step": 143
    },
    {
      "epoch": 0.2545997016409746,
      "grad_norm": 0.8500795364379883,
      "learning_rate": 3.142708644219186e-05,
      "loss": 0.0001,
      "step": 144
    },
    {
      "epoch": 0.25636775512459253,
      "grad_norm": 0.0008820121875032783,
      "learning_rate": 3.1150088833189785e-05,
      "loss": 0.0,
      "step": 145
    },
    {
      "epoch": 0.2581358086082104,
      "grad_norm": 0.0022942032665014267,
      "learning_rate": 3.087232629260363e-05,
      "loss": 0.0,
      "step": 146
    },
    {
      "epoch": 0.25990386209182825,
      "grad_norm": 0.00030348575091920793,
      "learning_rate": 3.05938364250351e-05,
      "loss": 0.0,
      "step": 147
    },
    {
      "epoch": 0.26167191557544617,
      "grad_norm": 0.00010618369560688734,
      "learning_rate": 3.0314656933554343e-05,
      "loss": 0.0,
      "step": 148
    },
    {
      "epoch": 0.263439969059064,
      "grad_norm": 0.0016436876030638814,
      "learning_rate": 3.003482561459551e-05,
      "loss": 0.0,
      "step": 149
    },
    {
      "epoch": 0.26520802254268194,
      "grad_norm": 0.025613779202103615,
      "learning_rate": 2.97543803528398e-05,
      "loss": 0.0,
      "step": 150
    },
    {
      "epoch": 0.2669760760262998,
      "grad_norm": 0.0002844675909727812,
      "learning_rate": 2.947335911608641e-05,
      "loss": 0.0,
      "step": 151
    },
    {
      "epoch": 0.26874412950991766,
      "grad_norm": 0.0015938935102894902,
      "learning_rate": 2.9191799950112342e-05,
      "loss": 0.0,
      "step": 152
    },
    {
      "epoch": 0.2705121829935356,
      "grad_norm": 0.0003039448638446629,
      "learning_rate": 2.8909740973521605e-05,
      "loss": 0.0,
      "step": 153
    },
    {
      "epoch": 0.27228023647715344,
      "grad_norm": 0.0007424139184877276,
      "learning_rate": 2.8627220372584576e-05,
      "loss": 0.0,
      "step": 154
    },
    {
      "epoch": 0.2740482899607713,
      "grad_norm": 0.002882719039916992,
      "learning_rate": 2.8344276396068144e-05,
      "loss": 0.0,
      "step": 155
    },
    {
      "epoch": 0.2758163434443892,
      "grad_norm": 0.005102601367980242,
      "learning_rate": 2.8060947350057514e-05,
      "loss": 0.0,
      "step": 156
    },
    {
      "epoch": 0.27758439692800707,
      "grad_norm": 0.02509269304573536,
      "learning_rate": 2.7777271592770116e-05,
      "loss": 0.0,
      "step": 157
    },
    {
      "epoch": 0.27935245041162493,
      "grad_norm": 0.009309419430792332,
      "learning_rate": 2.749328752936253e-05,
      "loss": 0.0,
      "step": 158
    },
    {
      "epoch": 0.28112050389524285,
      "grad_norm": 0.0033299459610134363,
      "learning_rate": 2.720903360673107e-05,
      "loss": 0.0,
      "step": 159
    },
    {
      "epoch": 0.2828885573788607,
      "grad_norm": 0.009038392454385757,
      "learning_rate": 2.692454830830666e-05,
      "loss": 0.0,
      "step": 160
    },
    {
      "epoch": 0.28465661086247857,
      "grad_norm": 0.001115231541916728,
      "learning_rate": 2.6639870148844804e-05,
      "loss": 0.0,
      "step": 161
    },
    {
      "epoch": 0.2864246643460965,
      "grad_norm": 0.0021960127633064985,
      "learning_rate": 2.635503766921128e-05,
      "loss": 0.0,
      "step": 162
    },
    {
      "epoch": 0.28819271782971434,
      "grad_norm": 0.0012615895830094814,
      "learning_rate": 2.607008943116438e-05,
      "loss": 0.0,
      "step": 163
    },
    {
      "epoch": 0.28996077131333226,
      "grad_norm": 0.00011055181676056236,
      "learning_rate": 2.5785064012134235e-05,
      "loss": 0.0,
      "step": 164
    },
    {
      "epoch": 0.2917288247969501,
      "grad_norm": 1.1846184730529785,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 0.0001,
      "step": 165
    },
    {
      "epoch": 0.293496878280568,
      "grad_norm": 0.14394652843475342,
      "learning_rate": 2.5214935987865782e-05,
      "loss": 0.0,
      "step": 166
    },
    {
      "epoch": 0.2952649317641859,
      "grad_norm": 0.004770410712808371,
      "learning_rate": 2.4929910568835625e-05,
      "loss": 0.0,
      "step": 167
    },
    {
      "epoch": 0.29703298524780375,
      "grad_norm": 0.001502661150880158,
      "learning_rate": 2.4644962330788736e-05,
      "loss": 0.0,
      "step": 168
    },
    {
      "epoch": 0.2988010387314216,
      "grad_norm": 0.6833661794662476,
      "learning_rate": 2.436012985115521e-05,
      "loss": 0.0,
      "step": 169
    },
    {
      "epoch": 0.3005690922150395,
      "grad_norm": 0.000995868700556457,
      "learning_rate": 2.4075451691693345e-05,
      "loss": 0.0,
      "step": 170
    },
    {
      "epoch": 0.3023371456986574,
      "grad_norm": 0.28130462765693665,
      "learning_rate": 2.3790966393268934e-05,
      "loss": 0.0,
      "step": 171
    },
    {
      "epoch": 0.30410519918227524,
      "grad_norm": 0.0009763907874003053,
      "learning_rate": 2.3506712470637478e-05,
      "loss": 0.0,
      "step": 172
    },
    {
      "epoch": 0.30587325266589316,
      "grad_norm": 0.0002226334618171677,
      "learning_rate": 2.3222728407229894e-05,
      "loss": 0.0,
      "step": 173
    },
    {
      "epoch": 0.307641306149511,
      "grad_norm": 0.0005141818546690047,
      "learning_rate": 2.2939052649942496e-05,
      "loss": 0.0,
      "step": 174
    },
    {
      "epoch": 0.3094093596331289,
      "grad_norm": 0.000805897347163409,
      "learning_rate": 2.265572360393186e-05,
      "loss": 0.0,
      "step": 175
    },
    {
      "epoch": 0.3111774131167468,
      "grad_norm": 0.01516794040799141,
      "learning_rate": 2.2372779627415433e-05,
      "loss": 0.0,
      "step": 176
    },
    {
      "epoch": 0.31294546660036465,
      "grad_norm": 0.00015710192383266985,
      "learning_rate": 2.2090259026478405e-05,
      "loss": 0.0,
      "step": 177
    },
    {
      "epoch": 0.31471352008398257,
      "grad_norm": 0.0007961568771861494,
      "learning_rate": 2.1808200049887665e-05,
      "loss": 0.0,
      "step": 178
    },
    {
      "epoch": 0.3164815735676004,
      "grad_norm": 0.34228724241256714,
      "learning_rate": 2.1526640883913603e-05,
      "loss": 0.0001,
      "step": 179
    },
    {
      "epoch": 0.3182496270512183,
      "grad_norm": 0.00023374942247755826,
      "learning_rate": 2.124561964716021e-05,
      "loss": 0.0,
      "step": 180
    },
    {
      "epoch": 0.3200176805348362,
      "grad_norm": 0.0010924128582701087,
      "learning_rate": 2.09651743854045e-05,
      "loss": 0.0,
      "step": 181
    },
    {
      "epoch": 0.32178573401845406,
      "grad_norm": 0.0013544277753680944,
      "learning_rate": 2.0685343066445663e-05,
      "loss": 0.0,
      "step": 182
    },
    {
      "epoch": 0.3235537875020719,
      "grad_norm": 3.933921834686771e-05,
      "learning_rate": 2.0406163574964904e-05,
      "loss": 0.0,
      "step": 183
    },
    {
      "epoch": 0.32532184098568984,
      "grad_norm": 0.0008416796335950494,
      "learning_rate": 2.0127673707396367e-05,
      "loss": 0.0,
      "step": 184
    },
    {
      "epoch": 0.3270898944693077,
      "grad_norm": 8.908173185773194e-05,
      "learning_rate": 1.9849911166810218e-05,
      "loss": 0.0,
      "step": 185
    },
    {
      "epoch": 0.32885794795292556,
      "grad_norm": 7.215650839498267e-05,
      "learning_rate": 1.9572913557808146e-05,
      "loss": 0.0,
      "step": 186
    },
    {
      "epoch": 0.33062600143654347,
      "grad_norm": 0.009005086496472359,
      "learning_rate": 1.929671838143236e-05,
      "loss": 0.0,
      "step": 187
    },
    {
      "epoch": 0.33239405492016133,
      "grad_norm": 8.976953540695831e-05,
      "learning_rate": 1.9021363030088554e-05,
      "loss": 0.0,
      "step": 188
    },
    {
      "epoch": 0.3341621084037792,
      "grad_norm": 0.001722825225442648,
      "learning_rate": 1.8746884782483525e-05,
      "loss": 0.0,
      "step": 189
    },
    {
      "epoch": 0.3359301618873971,
      "grad_norm": 0.004825813230127096,
      "learning_rate": 1.8473320798578298e-05,
      "loss": 0.0,
      "step": 190
    },
    {
      "epoch": 0.33769821537101496,
      "grad_norm": 0.0007849527173675597,
      "learning_rate": 1.8200708114557193e-05,
      "loss": 0.0,
      "step": 191
    },
    {
      "epoch": 0.3394662688546329,
      "grad_norm": 0.000184956646990031,
      "learning_rate": 1.792908363781379e-05,
      "loss": 0.0,
      "step": 192
    },
    {
      "epoch": 0.34123432233825074,
      "grad_norm": 0.0004415436997078359,
      "learning_rate": 1.7658484141954197e-05,
      "loss": 0.0,
      "step": 193
    },
    {
      "epoch": 0.3430023758218686,
      "grad_norm": 0.0002726184611674398,
      "learning_rate": 1.7388946261818532e-05,
      "loss": 0.0,
      "step": 194
    },
    {
      "epoch": 0.3447704293054865,
      "grad_norm": 0.18765950202941895,
      "learning_rate": 1.712050648852112e-05,
      "loss": 0.0,
      "step": 195
    },
    {
      "epoch": 0.3465384827891044,
      "grad_norm": 5.5921867897268385e-05,
      "learning_rate": 1.6853201164510216e-05,
      "loss": 0.0,
      "step": 196
    },
    {
      "epoch": 0.34830653627272223,
      "grad_norm": 0.0006541347247548401,
      "learning_rate": 1.6587066478647798e-05,
      "loss": 0.0,
      "step": 197
    },
    {
      "epoch": 0.35007458975634015,
      "grad_norm": 0.06359551846981049,
      "learning_rate": 1.6322138461310157e-05,
      "loss": 0.0,
      "step": 198
    },
    {
      "epoch": 0.351842643239958,
      "grad_norm": 7.466107490472496e-05,
      "learning_rate": 1.6058452979510023e-05,
      "loss": 0.0,
      "step": 199
    },
    {
      "epoch": 0.35361069672357587,
      "grad_norm": 0.0012128509115427732,
      "learning_rate": 1.5796045732040658e-05,
      "loss": 0.0,
      "step": 200
    },
    {
      "epoch": 0.3553787502071938,
      "grad_norm": 0.0028565828688442707,
      "learning_rate": 1.55349522446429e-05,
      "loss": 0.0,
      "step": 201
    },
    {
      "epoch": 0.35714680369081164,
      "grad_norm": 0.0035199481062591076,
      "learning_rate": 1.5275207865195463e-05,
      "loss": 0.0,
      "step": 202
    },
    {
      "epoch": 0.3589148571744295,
      "grad_norm": 3.6958659620722756e-05,
      "learning_rate": 1.5016847758929478e-05,
      "loss": 0.0,
      "step": 203
    },
    {
      "epoch": 0.3606829106580474,
      "grad_norm": 0.013499101623892784,
      "learning_rate": 1.4759906903667603e-05,
      "loss": 0.0,
      "step": 204
    },
    {
      "epoch": 0.3624509641416653,
      "grad_norm": 0.00010528836719458923,
      "learning_rate": 1.4504420085088679e-05,
      "loss": 0.0,
      "step": 205
    },
    {
      "epoch": 0.3642190176252832,
      "grad_norm": 0.0005988091579638422,
      "learning_rate": 1.425042189201821e-05,
      "loss": 0.0,
      "step": 206
    },
    {
      "epoch": 0.36598707110890105,
      "grad_norm": 8.299437467940152e-05,
      "learning_rate": 1.3997946711745685e-05,
      "loss": 0.0,
      "step": 207
    },
    {
      "epoch": 0.3677551245925189,
      "grad_norm": 0.08108139038085938,
      "learning_rate": 1.3747028725368977e-05,
      "loss": 0.0,
      "step": 208
    },
    {
      "epoch": 0.3695231780761368,
      "grad_norm": 1.022919241222553e-05,
      "learning_rate": 1.3497701903166873e-05,
      "loss": 0.0,
      "step": 209
    },
    {
      "epoch": 0.3712912315597547,
      "grad_norm": 0.21460385620594025,
      "learning_rate": 1.3250000000000007e-05,
      "loss": 0.0001,
      "step": 210
    },
    {
      "epoch": 0.37305928504337255,
      "grad_norm": 0.00021967390784993768,
      "learning_rate": 1.3003956550740957e-05,
      "loss": 0.0,
      "step": 211
    },
    {
      "epoch": 0.37482733852699046,
      "grad_norm": 0.001171005773358047,
      "learning_rate": 1.2759604865734225e-05,
      "loss": 0.0,
      "step": 212
    },
    {
      "epoch": 0.3765953920106083,
      "grad_norm": 0.09991420805454254,
      "learning_rate": 1.2516978026286484e-05,
      "loss": 0.0,
      "step": 213
    },
    {
      "epoch": 0.3783634454942262,
      "grad_norm": 6.677448254777119e-05,
      "learning_rate": 1.2276108880187944e-05,
      "loss": 0.0,
      "step": 214
    },
    {
      "epoch": 0.3801314989778441,
      "grad_norm": 0.0013546834234148264,
      "learning_rate": 1.2037030037265254e-05,
      "loss": 0.0,
      "step": 215
    },
    {
      "epoch": 0.38189955246146196,
      "grad_norm": 0.0005404578987509012,
      "learning_rate": 1.1799773864966708e-05,
      "loss": 0.0,
      "step": 216
    },
    {
      "epoch": 0.3836676059450798,
      "grad_norm": 0.00040231485036201775,
      "learning_rate": 1.1564372483980137e-05,
      "loss": 0.0,
      "step": 217
    },
    {
      "epoch": 0.38543565942869773,
      "grad_norm": 1.2076519851689227e-05,
      "learning_rate": 1.1330857763884332e-05,
      "loss": 0.0,
      "step": 218
    },
    {
      "epoch": 0.3872037129123156,
      "grad_norm": 5.123095615999773e-05,
      "learning_rate": 1.1099261318834413e-05,
      "loss": 0.0,
      "step": 219
    },
    {
      "epoch": 0.38897176639593345,
      "grad_norm": 0.06918606162071228,
      "learning_rate": 1.0869614503281746e-05,
      "loss": 0.0,
      "step": 220
    },
    {
      "epoch": 0.39073981987955136,
      "grad_norm": 6.619005580432713e-05,
      "learning_rate": 1.064194840772908e-05,
      "loss": 0.0,
      "step": 221
    },
    {
      "epoch": 0.3925078733631692,
      "grad_norm": 0.00015157890447881073,
      "learning_rate": 1.0416293854521374e-05,
      "loss": 0.0,
      "step": 222
    },
    {
      "epoch": 0.39427592684678714,
      "grad_norm": 0.0037292453926056623,
      "learning_rate": 1.0192681393672982e-05,
      "loss": 0.0,
      "step": 223
    },
    {
      "epoch": 0.396043980330405,
      "grad_norm": 0.16160666942596436,
      "learning_rate": 9.971141298731597e-06,
      "loss": 0.0,
      "step": 224
    },
    {
      "epoch": 0.39781203381402286,
      "grad_norm": 0.06928429752588272,
      "learning_rate": 9.751703562679787e-06,
      "loss": 0.0,
      "step": 225
    },
    {
      "epoch": 0.3995800872976408,
      "grad_norm": 9.774271893547848e-05,
      "learning_rate": 9.53439789387435e-06,
      "loss": 0.0,
      "step": 226
    },
    {
      "epoch": 0.40134814078125863,
      "grad_norm": 0.0001532236929051578,
      "learning_rate": 9.319253712024317e-06,
      "loss": 0.0,
      "step": 227
    },
    {
      "epoch": 0.4031161942648765,
      "grad_norm": 0.00043412772356532514,
      "learning_rate": 9.106300144207982e-06,
      "loss": 0.0,
      "step": 228
    },
    {
      "epoch": 0.4048842477484944,
      "grad_norm": 0.0001928459678310901,
      "learning_rate": 8.895566020929561e-06,
      "loss": 0.0,
      "step": 229
    },
    {
      "epoch": 0.40665230123211227,
      "grad_norm": 0.00013222330017015338,
      "learning_rate": 8.68707987221603e-06,
      "loss": 0.0,
      "step": 230
    },
    {
      "epoch": 0.4084203547157301,
      "grad_norm": 0.00016468556714244187,
      "learning_rate": 8.480869923754565e-06,
      "loss": 0.0,
      "step": 231
    },
    {
      "epoch": 0.41018840819934804,
      "grad_norm": 1.7172653315356e-05,
      "learning_rate": 8.276964093071288e-06,
      "loss": 0.0,
      "step": 232
    },
    {
      "epoch": 0.4119564616829659,
      "grad_norm": 4.863995854975656e-05,
      "learning_rate": 8.075389985751642e-06,
      "loss": 0.0,
      "step": 233
    },
    {
      "epoch": 0.41372451516658376,
      "grad_norm": 1.6034869986469857e-05,
      "learning_rate": 7.876174891703052e-06,
      "loss": 0.0,
      "step": 234
    },
    {
      "epoch": 0.4154925686502017,
      "grad_norm": 0.020293768495321274,
      "learning_rate": 7.679345781460308e-06,
      "loss": 0.0,
      "step": 235
    },
    {
      "epoch": 0.41726062213381954,
      "grad_norm": 0.00036274161539040506,
      "learning_rate": 7.484929302534201e-06,
      "loss": 0.0,
      "step": 236
    },
    {
      "epoch": 0.41902867561743745,
      "grad_norm": 0.00016249797772616148,
      "learning_rate": 7.292951775803839e-06,
      "loss": 0.0,
      "step": 237
    },
    {
      "epoch": 0.4207967291010553,
      "grad_norm": 0.004641630221158266,
      "learning_rate": 7.10343919195327e-06,
      "loss": 0.0,
      "step": 238
    },
    {
      "epoch": 0.42256478258467317,
      "grad_norm": 0.0009545126813463867,
      "learning_rate": 6.916417207952713e-06,
      "loss": 0.0,
      "step": 239
    },
    {
      "epoch": 0.4243328360682911,
      "grad_norm": 0.00012078192958142608,
      "learning_rate": 6.731911143585042e-06,
      "loss": 0.0,
      "step": 240
    },
    {
      "epoch": 0.42610088955190895,
      "grad_norm": 0.0007262192084454,
      "learning_rate": 6.549945978017887e-06,
      "loss": 0.0,
      "step": 241
    },
    {
      "epoch": 0.4278689430355268,
      "grad_norm": 0.0009577589225955307,
      "learning_rate": 6.37054634642185e-06,
      "loss": 0.0,
      "step": 242
    },
    {
      "epoch": 0.4296369965191447,
      "grad_norm": 0.0585528202354908,
      "learning_rate": 6.193736536635314e-06,
      "loss": 0.0,
      "step": 243
    },
    {
      "epoch": 0.4314050500027626,
      "grad_norm": 7.095329783624038e-05,
      "learning_rate": 6.019540485876227e-06,
      "loss": 0.0,
      "step": 244
    },
    {
      "epoch": 0.43317310348638044,
      "grad_norm": 7.245037704706192e-05,
      "learning_rate": 5.847981777501426e-06,
      "loss": 0.0,
      "step": 245
    },
    {
      "epoch": 0.43494115696999835,
      "grad_norm": 5.6373453844571486e-05,
      "learning_rate": 5.6790836378137905e-06,
      "loss": 0.0,
      "step": 246
    },
    {
      "epoch": 0.4367092104536162,
      "grad_norm": 3.724493581103161e-05,
      "learning_rate": 5.5128689329177896e-06,
      "loss": 0.0,
      "step": 247
    },
    {
      "epoch": 0.4384772639372341,
      "grad_norm": 0.0006005808245390654,
      "learning_rate": 5.349360165623773e-06,
      "loss": 0.0,
      "step": 248
    },
    {
      "epoch": 0.440245317420852,
      "grad_norm": 3.0376237191376276e-05,
      "learning_rate": 5.188579472401481e-06,
      "loss": 0.0,
      "step": 249
    },
    {
      "epoch": 0.44201337090446985,
      "grad_norm": 0.00017754016153048724,
      "learning_rate": 5.0305486203830615e-06,
      "loss": 0.0,
      "step": 250
    },
    {
      "epoch": 0.44378142438808776,
      "grad_norm": 4.187566082691774e-05,
      "learning_rate": 4.8752890044161906e-06,
      "loss": 0.0,
      "step": 251
    },
    {
      "epoch": 0.4455494778717056,
      "grad_norm": 5.963912553852424e-05,
      "learning_rate": 4.722821644167564e-06,
      "loss": 0.0,
      "step": 252
    },
    {
      "epoch": 0.4473175313553235,
      "grad_norm": 0.00011405260738683864,
      "learning_rate": 4.573167181277125e-06,
      "loss": 0.0,
      "step": 253
    },
    {
      "epoch": 0.4490855848389414,
      "grad_norm": 7.98743076302344e-06,
      "learning_rate": 4.4263458765635525e-06,
      "loss": 0.0,
      "step": 254
    },
    {
      "epoch": 0.45085363832255926,
      "grad_norm": 0.00010863741772482172,
      "learning_rate": 4.282377607281252e-06,
      "loss": 0.0,
      "step": 255
    },
    {
      "epoch": 0.4526216918061771,
      "grad_norm": 9.043182217283174e-05,
      "learning_rate": 4.141281864429324e-06,
      "loss": 0.0,
      "step": 256
    },
    {
      "epoch": 0.45438974528979503,
      "grad_norm": 0.0006291704485192895,
      "learning_rate": 4.003077750112739e-06,
      "loss": 0.0,
      "step": 257
    },
    {
      "epoch": 0.4561577987734129,
      "grad_norm": 0.0001768024085322395,
      "learning_rate": 3.867783974956289e-06,
      "loss": 0.0,
      "step": 258
    },
    {
      "epoch": 0.45792585225703075,
      "grad_norm": 3.043546348635573e-05,
      "learning_rate": 3.735418855571405e-06,
      "loss": 0.0,
      "step": 259
    },
    {
      "epoch": 0.45969390574064867,
      "grad_norm": 1.1459591405582614e-05,
      "learning_rate": 3.6060003120764025e-06,
      "loss": 0.0,
      "step": 260
    },
    {
      "epoch": 0.4614619592242665,
      "grad_norm": 3.286511491751298e-05,
      "learning_rate": 3.479545865670413e-06,
      "loss": 0.0,
      "step": 261
    },
    {
      "epoch": 0.4632300127078844,
      "grad_norm": 1.5214385712170042e-05,
      "learning_rate": 3.356072636261248e-06,
      "loss": 0.0,
      "step": 262
    },
    {
      "epoch": 0.4649980661915023,
      "grad_norm": 2.3637854610569775e-05,
      "learning_rate": 3.235597340147676e-06,
      "loss": 0.0,
      "step": 263
    },
    {
      "epoch": 0.46676611967512016,
      "grad_norm": 0.00024508286151103675,
      "learning_rate": 3.1181362877562766e-06,
      "loss": 0.0,
      "step": 264
    },
    {
      "epoch": 0.4685341731587381,
      "grad_norm": 0.002545794239267707,
      "learning_rate": 3.00370538143329e-06,
      "loss": 0.0,
      "step": 265
    },
    {
      "epoch": 0.47030222664235594,
      "grad_norm": 2.2318992705550045e-05,
      "learning_rate": 2.8923201132916638e-06,
      "loss": 0.0,
      "step": 266
    },
    {
      "epoch": 0.4720702801259738,
      "grad_norm": 1.516962583991699e-05,
      "learning_rate": 2.783995563113711e-06,
      "loss": 0.0,
      "step": 267
    },
    {
      "epoch": 0.4738383336095917,
      "grad_norm": 0.000140973279485479,
      "learning_rate": 2.6787463963095105e-06,
      "loss": 0.0,
      "step": 268
    },
    {
      "epoch": 0.47560638709320957,
      "grad_norm": 0.01587841846048832,
      "learning_rate": 2.5765868619314814e-06,
      "loss": 0.0,
      "step": 269
    },
    {
      "epoch": 0.47737444057682743,
      "grad_norm": 1.07875666799373e-05,
      "learning_rate": 2.4775307907452464e-06,
      "loss": 0.0,
      "step": 270
    },
    {
      "epoch": 0.47914249406044535,
      "grad_norm": 0.0006835050880908966,
      "learning_rate": 2.381591593357204e-06,
      "loss": 0.0,
      "step": 271
    },
    {
      "epoch": 0.4809105475440632,
      "grad_norm": 1.7690043023321778e-05,
      "learning_rate": 2.2887822583989345e-06,
      "loss": 0.0,
      "step": 272
    },
    {
      "epoch": 0.48267860102768106,
      "grad_norm": 0.00046111404662951827,
      "learning_rate": 2.1991153507687385e-06,
      "loss": 0.0,
      "step": 273
    },
    {
      "epoch": 0.484446654511299,
      "grad_norm": 0.022714175283908844,
      "learning_rate": 2.104361506153305e-06,
      "loss": 0.0,
      "step": 274
    },
    {
      "epoch": 0.48621470799491684,
      "grad_norm": 4.023480505566113e-05,
      "learning_rate": 2.014008697185033e-06,
      "loss": 0.0,
      "step": 275
    },
    {
      "epoch": 0.4879827614785347,
      "grad_norm": 0.0869794487953186,
      "learning_rate": 1.9279975950776054e-06,
      "loss": 0.0,
      "step": 276
    },
    {
      "epoch": 0.4897508149621526,
      "grad_norm": 8.842632269079331e-06,
      "learning_rate": 1.8462681192674342e-06,
      "loss": 0.0,
      "step": 277
    },
    {
      "epoch": 0.4915188684457705,
      "grad_norm": 0.00010428605310153216,
      "learning_rate": 1.768759451931695e-06,
      "loss": 0.0,
      "step": 278
    },
    {
      "epoch": 0.4932869219293884,
      "grad_norm": 0.00014882163668517023,
      "learning_rate": 1.6954100526275048e-06,
      "loss": 0.0,
      "step": 279
    },
    {
      "epoch": 0.49505497541300625,
      "grad_norm": 2.4553623006795533e-05,
      "learning_rate": 1.6261576730494264e-06,
      "loss": 0.0,
      "step": 280
    },
    {
      "epoch": 0.4968230288966241,
      "grad_norm": 1.0228201063000597e-05,
      "learning_rate": 1.5609393719023522e-06,
      "loss": 0.0,
      "step": 281
    },
    {
      "epoch": 0.498591082380242,
      "grad_norm": 5.0795490096788853e-05,
      "learning_rate": 1.4996915298869766e-06,
      "loss": 0.0,
      "step": 282
    }
  ],
  "logging_steps": 1,
  "max_steps": 282,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}