{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1334069953743239,
  "eval_steps": 5888,
  "global_step": 19635,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.79434659405775e-06,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0,
      "loss": 3.0684,
      "step": 1
    },
    {
      "epoch": 1.35886931881155e-05,
      "grad_norm": 1.078125,
      "learning_rate": 8.488964346349746e-07,
      "loss": 3.2931,
      "step": 2
    },
    {
      "epoch": 2.0383039782173247e-05,
      "grad_norm": 1.953125,
      "learning_rate": 1.6977928692699491e-06,
      "loss": 3.3896,
      "step": 3
    },
    {
      "epoch": 2.7177386376231e-05,
      "grad_norm": 1.1015625,
      "learning_rate": 2.546689303904924e-06,
      "loss": 3.2248,
      "step": 4
    },
    {
      "epoch": 3.397173297028875e-05,
      "grad_norm": 1.140625,
      "learning_rate": 3.3955857385398982e-06,
      "loss": 3.0969,
      "step": 5
    },
    {
      "epoch": 4.0766079564346494e-05,
      "grad_norm": 1.1328125,
      "learning_rate": 4.244482173174873e-06,
      "loss": 3.1374,
      "step": 6
    },
    {
      "epoch": 4.7560426158404245e-05,
      "grad_norm": 1.125,
      "learning_rate": 5.093378607809848e-06,
      "loss": 3.1244,
      "step": 7
    },
    {
      "epoch": 5.4354772752462e-05,
      "grad_norm": 1.09375,
      "learning_rate": 5.942275042444822e-06,
      "loss": 3.2879,
      "step": 8
    },
    {
      "epoch": 6.114911934651975e-05,
      "grad_norm": 1.140625,
      "learning_rate": 6.7911714770797965e-06,
      "loss": 3.3371,
      "step": 9
    },
    {
      "epoch": 6.79434659405775e-05,
      "grad_norm": 1.03125,
      "learning_rate": 7.640067911714771e-06,
      "loss": 3.2245,
      "step": 10
    },
    {
      "epoch": 7.473781253463524e-05,
      "grad_norm": 1.171875,
      "learning_rate": 8.488964346349745e-06,
      "loss": 3.2075,
      "step": 11
    },
    {
      "epoch": 8.153215912869299e-05,
      "grad_norm": 1.109375,
      "learning_rate": 9.33786078098472e-06,
      "loss": 3.251,
      "step": 12
    },
    {
      "epoch": 8.832650572275074e-05,
      "grad_norm": 0.984375,
      "learning_rate": 1.0186757215619695e-05,
      "loss": 3.158,
      "step": 13
    },
    {
      "epoch": 9.512085231680849e-05,
      "grad_norm": 1.0390625,
      "learning_rate": 1.103565365025467e-05,
      "loss": 3.1302,
      "step": 14
    },
    {
      "epoch": 0.00010191519891086624,
      "grad_norm": 0.9453125,
      "learning_rate": 1.1884550084889643e-05,
      "loss": 3.067,
      "step": 15
    },
    {
      "epoch": 0.000108709545504924,
      "grad_norm": 1.0390625,
      "learning_rate": 1.2733446519524619e-05,
      "loss": 3.1482,
      "step": 16
    },
    {
      "epoch": 0.00011550389209898174,
      "grad_norm": 1.0,
      "learning_rate": 1.3582342954159593e-05,
      "loss": 3.238,
      "step": 17
    },
    {
      "epoch": 0.0001222982386930395,
      "grad_norm": 0.9296875,
      "learning_rate": 1.4431239388794567e-05,
      "loss": 3.2459,
      "step": 18
    },
    {
      "epoch": 0.00012909258528709725,
      "grad_norm": 0.9296875,
      "learning_rate": 1.5280135823429543e-05,
      "loss": 3.1294,
      "step": 19
    },
    {
      "epoch": 0.000135886931881155,
      "grad_norm": 0.80859375,
      "learning_rate": 1.6129032258064517e-05,
      "loss": 3.2102,
      "step": 20
    },
    {
      "epoch": 0.00014268127847521275,
      "grad_norm": 0.8125,
      "learning_rate": 1.697792869269949e-05,
      "loss": 3.1589,
      "step": 21
    },
    {
      "epoch": 0.00014947562506927047,
      "grad_norm": 0.828125,
      "learning_rate": 1.7826825127334468e-05,
      "loss": 3.1508,
      "step": 22
    },
    {
      "epoch": 0.00015626997166332823,
      "grad_norm": 0.69921875,
      "learning_rate": 1.867572156196944e-05,
      "loss": 3.2411,
      "step": 23
    },
    {
      "epoch": 0.00016306431825738598,
      "grad_norm": 0.7265625,
      "learning_rate": 1.9524617996604416e-05,
      "loss": 2.9424,
      "step": 24
    },
    {
      "epoch": 0.00016985866485144373,
      "grad_norm": 0.625,
      "learning_rate": 2.037351443123939e-05,
      "loss": 3.1513,
      "step": 25
    },
    {
      "epoch": 0.00017665301144550148,
      "grad_norm": 0.5390625,
      "learning_rate": 2.1222410865874364e-05,
      "loss": 3.1654,
      "step": 26
    },
    {
      "epoch": 0.00018344735803955923,
      "grad_norm": 0.484375,
      "learning_rate": 2.207130730050934e-05,
      "loss": 3.0458,
      "step": 27
    },
    {
      "epoch": 0.00019024170463361698,
      "grad_norm": 0.4921875,
      "learning_rate": 2.2920203735144312e-05,
      "loss": 3.1279,
      "step": 28
    },
    {
      "epoch": 0.00019703605122767473,
      "grad_norm": 0.48046875,
      "learning_rate": 2.3769100169779286e-05,
      "loss": 3.1902,
      "step": 29
    },
    {
      "epoch": 0.00020383039782173248,
      "grad_norm": 0.412109375,
      "learning_rate": 2.461799660441426e-05,
      "loss": 3.0434,
      "step": 30
    },
    {
      "epoch": 0.00021062474441579024,
      "grad_norm": 0.4140625,
      "learning_rate": 2.5466893039049238e-05,
      "loss": 3.1478,
      "step": 31
    },
    {
      "epoch": 0.000217419091009848,
      "grad_norm": 0.4140625,
      "learning_rate": 2.631578947368421e-05,
      "loss": 3.0531,
      "step": 32
    },
    {
      "epoch": 0.00022421343760390574,
      "grad_norm": 0.416015625,
      "learning_rate": 2.7164685908319186e-05,
      "loss": 3.123,
      "step": 33
    },
    {
      "epoch": 0.0002310077841979635,
      "grad_norm": 0.357421875,
      "learning_rate": 2.801358234295416e-05,
      "loss": 3.1384,
      "step": 34
    },
    {
      "epoch": 0.00023780213079202124,
      "grad_norm": 0.447265625,
      "learning_rate": 2.8862478777589134e-05,
      "loss": 3.1636,
      "step": 35
    },
    {
      "epoch": 0.000244596477386079,
      "grad_norm": 0.400390625,
      "learning_rate": 2.9711375212224108e-05,
      "loss": 3.0499,
      "step": 36
    },
    {
      "epoch": 0.0002513908239801367,
      "grad_norm": 0.337890625,
      "learning_rate": 3.0560271646859086e-05,
      "loss": 2.8693,
      "step": 37
    },
    {
      "epoch": 0.0002581851705741945,
      "grad_norm": 0.328125,
      "learning_rate": 3.140916808149406e-05,
      "loss": 3.0906,
      "step": 38
    },
    {
      "epoch": 0.0002649795171682522,
      "grad_norm": 0.3359375,
      "learning_rate": 3.2258064516129034e-05,
      "loss": 3.2466,
      "step": 39
    },
    {
      "epoch": 0.00027177386376231,
      "grad_norm": 0.458984375,
      "learning_rate": 3.310696095076401e-05,
      "loss": 3.1212,
      "step": 40
    },
    {
      "epoch": 0.0002785682103563677,
      "grad_norm": 0.294921875,
      "learning_rate": 3.395585738539898e-05,
      "loss": 2.9368,
      "step": 41
    },
    {
      "epoch": 0.0002853625569504255,
      "grad_norm": 0.302734375,
      "learning_rate": 3.4804753820033956e-05,
      "loss": 3.0046,
      "step": 42
    },
    {
      "epoch": 0.0002921569035444832,
      "grad_norm": 0.3203125,
      "learning_rate": 3.5653650254668936e-05,
      "loss": 2.9756,
      "step": 43
    },
    {
      "epoch": 0.00029895125013854095,
      "grad_norm": 0.29296875,
      "learning_rate": 3.6502546689303904e-05,
      "loss": 2.8354,
      "step": 44
    },
    {
      "epoch": 0.0003057455967325987,
      "grad_norm": 0.333984375,
      "learning_rate": 3.735144312393888e-05,
      "loss": 3.0328,
      "step": 45
    },
    {
      "epoch": 0.00031253994332665645,
      "grad_norm": 0.287109375,
      "learning_rate": 3.820033955857385e-05,
      "loss": 3.0337,
      "step": 46
    },
    {
      "epoch": 0.00031933428992071423,
      "grad_norm": 0.26171875,
      "learning_rate": 3.904923599320883e-05,
      "loss": 3.0472,
      "step": 47
    },
    {
      "epoch": 0.00032612863651477195,
      "grad_norm": 0.291015625,
      "learning_rate": 3.989813242784381e-05,
      "loss": 3.0383,
      "step": 48
    },
    {
      "epoch": 0.00033292298310882973,
      "grad_norm": 0.265625,
      "learning_rate": 4.074702886247878e-05,
      "loss": 2.9754,
      "step": 49
    },
    {
      "epoch": 0.00033971732970288746,
      "grad_norm": 0.26953125,
      "learning_rate": 4.159592529711375e-05,
      "loss": 2.9579,
      "step": 50
    },
    {
      "epoch": 0.00034651167629694523,
      "grad_norm": 0.298828125,
      "learning_rate": 4.244482173174873e-05,
      "loss": 2.7848,
      "step": 51
    },
    {
      "epoch": 0.00035330602289100296,
      "grad_norm": 0.466796875,
      "learning_rate": 4.32937181663837e-05,
      "loss": 2.953,
      "step": 52
    },
    {
      "epoch": 0.00036010036948506074,
      "grad_norm": 0.3125,
      "learning_rate": 4.414261460101868e-05,
      "loss": 3.0392,
      "step": 53
    },
    {
      "epoch": 0.00036689471607911846,
      "grad_norm": 0.267578125,
      "learning_rate": 4.499151103565365e-05,
      "loss": 2.9993,
      "step": 54
    },
    {
      "epoch": 0.00037368906267317624,
      "grad_norm": 0.28125,
      "learning_rate": 4.5840407470288625e-05,
      "loss": 3.0826,
      "step": 55
    },
    {
      "epoch": 0.00038048340926723396,
      "grad_norm": 0.291015625,
      "learning_rate": 4.6689303904923606e-05,
      "loss": 2.8721,
      "step": 56
    },
    {
      "epoch": 0.00038727775586129174,
      "grad_norm": 0.306640625,
      "learning_rate": 4.753820033955857e-05,
      "loss": 3.0875,
      "step": 57
    },
    {
      "epoch": 0.00039407210245534947,
      "grad_norm": 0.24609375,
      "learning_rate": 4.838709677419355e-05,
      "loss": 2.9739,
      "step": 58
    },
    {
      "epoch": 0.0004008664490494072,
      "grad_norm": 0.263671875,
      "learning_rate": 4.923599320882852e-05,
      "loss": 3.1005,
      "step": 59
    },
    {
      "epoch": 0.00040766079564346497,
      "grad_norm": 0.2490234375,
      "learning_rate": 5.00848896434635e-05,
      "loss": 2.9004,
      "step": 60
    },
    {
      "epoch": 0.0004144551422375227,
      "grad_norm": 0.279296875,
      "learning_rate": 5.0933786078098476e-05,
      "loss": 2.933,
      "step": 61
    },
    {
      "epoch": 0.00042124948883158047,
      "grad_norm": 0.275390625,
      "learning_rate": 5.178268251273345e-05,
      "loss": 2.9386,
      "step": 62
    },
    {
      "epoch": 0.0004280438354256382,
      "grad_norm": 0.296875,
      "learning_rate": 5.263157894736842e-05,
      "loss": 2.9116,
      "step": 63
    },
    {
      "epoch": 0.000434838182019696,
      "grad_norm": 0.298828125,
      "learning_rate": 5.34804753820034e-05,
      "loss": 3.0344,
      "step": 64
    },
    {
      "epoch": 0.0004416325286137537,
      "grad_norm": 0.2421875,
      "learning_rate": 5.432937181663837e-05,
      "loss": 2.9883,
      "step": 65
    },
    {
      "epoch": 0.0004484268752078115,
      "grad_norm": 0.2734375,
      "learning_rate": 5.5178268251273346e-05,
      "loss": 2.9767,
      "step": 66
    },
    {
      "epoch": 0.0004552212218018692,
      "grad_norm": 0.2421875,
      "learning_rate": 5.602716468590832e-05,
      "loss": 2.9793,
      "step": 67
    },
    {
      "epoch": 0.000462015568395927,
      "grad_norm": 0.30078125,
      "learning_rate": 5.6876061120543294e-05,
      "loss": 3.0236,
      "step": 68
    },
    {
      "epoch": 0.0004688099149899847,
      "grad_norm": 0.294921875,
      "learning_rate": 5.772495755517827e-05,
      "loss": 3.0205,
      "step": 69
    },
    {
      "epoch": 0.0004756042615840425,
      "grad_norm": 0.248046875,
      "learning_rate": 5.857385398981324e-05,
      "loss": 2.897,
      "step": 70
    },
    {
      "epoch": 0.0004823986081781002,
      "grad_norm": 0.275390625,
      "learning_rate": 5.9422750424448216e-05,
      "loss": 3.078,
      "step": 71
    },
    {
      "epoch": 0.000489192954772158,
      "grad_norm": 0.267578125,
      "learning_rate": 6.027164685908319e-05,
      "loss": 2.8938,
      "step": 72
    },
    {
      "epoch": 0.0004959873013662157,
      "grad_norm": 0.26953125,
      "learning_rate": 6.112054329371817e-05,
      "loss": 2.8192,
      "step": 73
    },
    {
      "epoch": 0.0005027816479602734,
      "grad_norm": 0.2421875,
      "learning_rate": 6.196943972835314e-05,
      "loss": 2.9967,
      "step": 74
    },
    {
      "epoch": 0.0005095759945543312,
      "grad_norm": 0.25,
      "learning_rate": 6.281833616298812e-05,
      "loss": 2.9123,
      "step": 75
    },
    {
      "epoch": 0.000516370341148389,
      "grad_norm": 0.2490234375,
      "learning_rate": 6.36672325976231e-05,
      "loss": 3.0417,
      "step": 76
    },
    {
      "epoch": 0.0005231646877424467,
      "grad_norm": 0.248046875,
      "learning_rate": 6.451612903225807e-05,
      "loss": 2.9966,
      "step": 77
    },
    {
      "epoch": 0.0005299590343365044,
      "grad_norm": 0.25390625,
      "learning_rate": 6.536502546689305e-05,
      "loss": 3.0096,
      "step": 78
    },
    {
      "epoch": 0.0005367533809305622,
      "grad_norm": 0.251953125,
      "learning_rate": 6.621392190152802e-05,
      "loss": 2.7807,
      "step": 79
    },
    {
      "epoch": 0.00054354772752462,
      "grad_norm": 0.232421875,
      "learning_rate": 6.706281833616298e-05,
      "loss": 2.9113,
      "step": 80
    },
    {
      "epoch": 0.0005503420741186777,
      "grad_norm": 0.255859375,
      "learning_rate": 6.791171477079796e-05,
      "loss": 2.8832,
      "step": 81
    },
    {
      "epoch": 0.0005571364207127354,
      "grad_norm": 0.28125,
      "learning_rate": 6.876061120543293e-05,
      "loss": 2.9627,
      "step": 82
    },
    {
      "epoch": 0.0005639307673067932,
      "grad_norm": 0.25,
      "learning_rate": 6.960950764006791e-05,
      "loss": 2.8434,
      "step": 83
    },
    {
      "epoch": 0.000570725113900851,
      "grad_norm": 0.2373046875,
      "learning_rate": 7.045840407470289e-05,
      "loss": 2.9362,
      "step": 84
    },
    {
      "epoch": 0.0005775194604949087,
      "grad_norm": 0.2490234375,
      "learning_rate": 7.130730050933787e-05,
      "loss": 2.866,
      "step": 85
    },
    {
      "epoch": 0.0005843138070889664,
      "grad_norm": 0.28515625,
      "learning_rate": 7.215619694397284e-05,
      "loss": 2.8126,
      "step": 86
    },
    {
      "epoch": 0.0005911081536830242,
      "grad_norm": 0.291015625,
      "learning_rate": 7.300509337860781e-05,
      "loss": 3.1331,
      "step": 87
    },
    {
      "epoch": 0.0005979025002770819,
      "grad_norm": 0.2421875,
      "learning_rate": 7.385398981324279e-05,
      "loss": 2.9356,
      "step": 88
    },
    {
      "epoch": 0.0006046968468711397,
      "grad_norm": 0.2470703125,
      "learning_rate": 7.470288624787776e-05,
      "loss": 3.0422,
      "step": 89
    },
    {
      "epoch": 0.0006114911934651975,
      "grad_norm": 0.248046875,
      "learning_rate": 7.555178268251274e-05,
      "loss": 3.019,
      "step": 90
    },
    {
      "epoch": 0.0006182855400592552,
      "grad_norm": 0.2578125,
      "learning_rate": 7.64006791171477e-05,
      "loss": 2.8655,
      "step": 91
    },
    {
      "epoch": 0.0006250798866533129,
      "grad_norm": 0.240234375,
      "learning_rate": 7.724957555178267e-05,
      "loss": 2.9504,
      "step": 92
    },
    {
      "epoch": 0.0006318742332473707,
      "grad_norm": 0.2431640625,
      "learning_rate": 7.809847198641767e-05,
      "loss": 2.9709,
      "step": 93
    },
    {
      "epoch": 0.0006386685798414285,
      "grad_norm": 0.2255859375,
      "learning_rate": 7.894736842105263e-05,
      "loss": 2.9155,
      "step": 94
    },
    {
      "epoch": 0.0006454629264354862,
      "grad_norm": 0.2451171875,
      "learning_rate": 7.979626485568761e-05,
      "loss": 3.0536,
      "step": 95
    },
    {
      "epoch": 0.0006522572730295439,
      "grad_norm": 0.2373046875,
      "learning_rate": 8.064516129032258e-05,
      "loss": 2.8648,
      "step": 96
    },
    {
      "epoch": 0.0006590516196236017,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.149405772495756e-05,
      "loss": 2.8653,
      "step": 97
    },
    {
      "epoch": 0.0006658459662176595,
      "grad_norm": 0.251953125,
      "learning_rate": 8.234295415959253e-05,
      "loss": 2.8352,
      "step": 98
    },
    {
      "epoch": 0.0006726403128117172,
      "grad_norm": 0.25,
      "learning_rate": 8.31918505942275e-05,
      "loss": 3.0179,
      "step": 99
    },
    {
      "epoch": 0.0006794346594057749,
      "grad_norm": 0.2373046875,
      "learning_rate": 8.404074702886248e-05,
      "loss": 2.8047,
      "step": 100
    },
    {
      "epoch": 0.0006862290059998327,
      "grad_norm": 0.279296875,
      "learning_rate": 8.488964346349746e-05,
      "loss": 2.9678,
      "step": 101
    },
    {
      "epoch": 0.0006930233525938905,
      "grad_norm": 0.2392578125,
      "learning_rate": 8.573853989813244e-05,
      "loss": 2.8967,
      "step": 102
    },
    {
      "epoch": 0.0006998176991879481,
      "grad_norm": 0.2294921875,
      "learning_rate": 8.65874363327674e-05,
      "loss": 2.8714,
      "step": 103
    },
    {
      "epoch": 0.0007066120457820059,
      "grad_norm": 0.234375,
      "learning_rate": 8.743633276740239e-05,
      "loss": 2.9967,
      "step": 104
    },
    {
      "epoch": 0.0007134063923760637,
      "grad_norm": 0.291015625,
      "learning_rate": 8.828522920203735e-05,
      "loss": 2.9365,
      "step": 105
    },
    {
      "epoch": 0.0007202007389701215,
      "grad_norm": 0.251953125,
      "learning_rate": 8.913412563667232e-05,
      "loss": 2.9259,
      "step": 106
    },
    {
      "epoch": 0.0007269950855641791,
      "grad_norm": 0.322265625,
      "learning_rate": 8.99830220713073e-05,
      "loss": 2.9475,
      "step": 107
    },
    {
      "epoch": 0.0007337894321582369,
      "grad_norm": 0.25,
      "learning_rate": 9.083191850594227e-05,
      "loss": 2.8322,
      "step": 108
    },
    {
      "epoch": 0.0007405837787522947,
      "grad_norm": 0.236328125,
      "learning_rate": 9.168081494057725e-05,
      "loss": 2.8687,
      "step": 109
    },
    {
      "epoch": 0.0007473781253463525,
      "grad_norm": 0.2333984375,
      "learning_rate": 9.252971137521223e-05,
      "loss": 2.9295,
      "step": 110
    },
    {
      "epoch": 0.0007541724719404101,
      "grad_norm": 0.2373046875,
      "learning_rate": 9.337860780984721e-05,
      "loss": 2.9451,
      "step": 111
    },
    {
      "epoch": 0.0007609668185344679,
      "grad_norm": 0.2314453125,
      "learning_rate": 9.422750424448218e-05,
      "loss": 2.8909,
      "step": 112
    },
    {
      "epoch": 0.0007677611651285257,
      "grad_norm": 0.2255859375,
      "learning_rate": 9.507640067911715e-05,
      "loss": 2.9054,
      "step": 113
    },
    {
      "epoch": 0.0007745555117225835,
      "grad_norm": 0.2119140625,
      "learning_rate": 9.592529711375213e-05,
      "loss": 2.8414,
      "step": 114
    },
    {
      "epoch": 0.0007813498583166412,
      "grad_norm": 0.25390625,
      "learning_rate": 9.67741935483871e-05,
      "loss": 2.9126,
      "step": 115
    },
    {
      "epoch": 0.0007881442049106989,
      "grad_norm": 0.26171875,
      "learning_rate": 9.762308998302207e-05,
      "loss": 2.9298,
      "step": 116
    },
    {
      "epoch": 0.0007949385515047567,
      "grad_norm": 0.23046875,
      "learning_rate": 9.847198641765704e-05,
      "loss": 2.8057,
      "step": 117
    },
    {
      "epoch": 0.0008017328980988144,
      "grad_norm": 0.22265625,
      "learning_rate": 9.932088285229201e-05,
      "loss": 2.8567,
      "step": 118
    },
    {
      "epoch": 0.0008085272446928722,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.000100169779286927,
      "loss": 2.967,
      "step": 119
    },
    {
      "epoch": 0.0008153215912869299,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.00010101867572156197,
      "loss": 2.8596,
      "step": 120
    },
    {
      "epoch": 0.0008221159378809877,
      "grad_norm": 0.23046875,
      "learning_rate": 0.00010186757215619695,
      "loss": 2.9425,
      "step": 121
    },
    {
      "epoch": 0.0008289102844750454,
      "grad_norm": 0.2578125,
      "learning_rate": 0.00010271646859083192,
      "loss": 3.0152,
      "step": 122
    },
    {
      "epoch": 0.0008357046310691032,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0001035653650254669,
      "loss": 2.8321,
      "step": 123
    },
    {
      "epoch": 0.0008424989776631609,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00010441426146010187,
      "loss": 2.997,
      "step": 124
    },
    {
      "epoch": 0.0008492933242572187,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00010526315789473683,
      "loss": 2.9626,
      "step": 125
    },
    {
      "epoch": 0.0008560876708512764,
      "grad_norm": 2.046875,
      "learning_rate": 0.00010611205432937182,
      "loss": 2.7554,
      "step": 126
    },
    {
      "epoch": 0.0008628820174453342,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0001069609507640068,
      "loss": 2.9007,
      "step": 127
    },
    {
      "epoch": 0.000869676364039392,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00010780984719864178,
      "loss": 2.763,
      "step": 128
    },
    {
      "epoch": 0.0008764707106334497,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00010865874363327674,
      "loss": 2.8598,
      "step": 129
    },
    {
      "epoch": 0.0008832650572275074,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.00010950764006791172,
      "loss": 2.914,
      "step": 130
    },
    {
      "epoch": 0.0008900594038215652,
      "grad_norm": 0.248046875,
      "learning_rate": 0.00011035653650254669,
      "loss": 3.0302,
      "step": 131
    },
    {
      "epoch": 0.000896853750415623,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00011120543293718166,
      "loss": 2.8569,
      "step": 132
    },
    {
      "epoch": 0.0009036480970096806,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00011205432937181664,
      "loss": 2.8314,
      "step": 133
    },
    {
      "epoch": 0.0009104424436037384,
      "grad_norm": 0.228515625,
      "learning_rate": 0.00011290322580645161,
      "loss": 2.9452,
      "step": 134
    },
    {
      "epoch": 0.0009172367901977962,
      "grad_norm": 0.240234375,
      "learning_rate": 0.00011375212224108659,
      "loss": 2.9752,
      "step": 135
    },
    {
      "epoch": 0.000924031136791854,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.00011460101867572157,
      "loss": 2.9466,
      "step": 136
    },
    {
      "epoch": 0.0009308254833859116,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00011544991511035654,
      "loss": 2.8777,
      "step": 137
    },
    {
      "epoch": 0.0009376198299799694,
      "grad_norm": 0.244140625,
      "learning_rate": 0.00011629881154499152,
      "loss": 2.7599,
      "step": 138
    },
    {
      "epoch": 0.0009444141765740272,
      "grad_norm": 0.259765625,
      "learning_rate": 0.00011714770797962648,
      "loss": 2.8072,
      "step": 139
    },
    {
      "epoch": 0.000951208523168085,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00011799660441426147,
      "loss": 2.8173,
      "step": 140
    },
    {
      "epoch": 0.0009580028697621426,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.00011884550084889643,
      "loss": 2.7847,
      "step": 141
    },
    {
      "epoch": 0.0009647972163562004,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00011969439728353141,
      "loss": 2.8518,
      "step": 142
    },
    {
      "epoch": 0.0009715915629502582,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00012054329371816638,
      "loss": 2.9365,
      "step": 143
    },
    {
      "epoch": 0.000978385909544316,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00012139219015280135,
      "loss": 2.8499,
      "step": 144
    },
    {
      "epoch": 0.0009851802561383737,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00012224108658743634,
      "loss": 2.9442,
      "step": 145
    },
    {
      "epoch": 0.0009919746027324313,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0001230899830220713,
      "loss": 2.8394,
      "step": 146
    },
    {
      "epoch": 0.000998768949326489,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00012393887945670628,
      "loss": 2.8312,
      "step": 147
    },
    {
      "epoch": 0.0010055632959205469,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00012478777589134126,
      "loss": 2.9447,
      "step": 148
    },
    {
      "epoch": 0.0010123576425146046,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00012563667232597624,
      "loss": 2.8649,
      "step": 149
    },
    {
      "epoch": 0.0010191519891086624,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00012648556876061122,
      "loss": 2.8586,
      "step": 150
    },
    {
      "epoch": 0.0010259463357027202,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001273344651952462,
      "loss": 3.004,
      "step": 151
    },
    {
      "epoch": 0.001032740682296778,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00012818336162988115,
      "loss": 2.8926,
      "step": 152
    },
    {
      "epoch": 0.0010395350288908358,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.00012903225806451613,
      "loss": 2.8719,
      "step": 153
    },
    {
      "epoch": 0.0010463293754848933,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00012988115449915112,
      "loss": 2.8017,
      "step": 154
    },
    {
      "epoch": 0.001053123722078951,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0001307300509337861,
      "loss": 2.7739,
      "step": 155
    },
    {
      "epoch": 0.0010599180686730089,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00013157894736842105,
      "loss": 2.8823,
      "step": 156
    },
    {
      "epoch": 0.0010667124152670667,
      "grad_norm": 0.291015625,
      "learning_rate": 0.00013242784380305603,
      "loss": 2.9602,
      "step": 157
    },
    {
      "epoch": 0.0010735067618611244,
      "grad_norm": 0.240234375,
      "learning_rate": 0.000133276740237691,
      "loss": 2.8337,
      "step": 158
    },
    {
      "epoch": 0.0010803011084551822,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.00013412563667232596,
      "loss": 2.8303,
      "step": 159
    },
    {
      "epoch": 0.00108709545504924,
      "grad_norm": 0.236328125,
      "learning_rate": 0.00013497453310696095,
      "loss": 2.887,
      "step": 160
    },
    {
      "epoch": 0.0010938898016432975,
      "grad_norm": 0.236328125,
      "learning_rate": 0.00013582342954159593,
      "loss": 2.9231,
      "step": 161
    },
    {
      "epoch": 0.0011006841482373553,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0001366723259762309,
      "loss": 2.8289,
      "step": 162
    },
    {
      "epoch": 0.001107478494831413,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00013752122241086586,
      "loss": 2.9314,
      "step": 163
    },
    {
      "epoch": 0.0011142728414254709,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00013837011884550084,
      "loss": 2.8158,
      "step": 164
    },
    {
      "epoch": 0.0011210671880195287,
      "grad_norm": 0.23828125,
      "learning_rate": 0.00013921901528013582,
      "loss": 2.7784,
      "step": 165
    },
    {
      "epoch": 0.0011278615346135864,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0001400679117147708,
      "loss": 2.9911,
      "step": 166
    },
    {
      "epoch": 0.0011346558812076442,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.00014091680814940578,
      "loss": 2.9111,
      "step": 167
    },
    {
      "epoch": 0.001141450227801702,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00014176570458404077,
      "loss": 2.7743,
      "step": 168
    },
    {
      "epoch": 0.0011482445743957596,
      "grad_norm": 0.23828125,
      "learning_rate": 0.00014261460101867575,
      "loss": 2.8628,
      "step": 169
    },
    {
      "epoch": 0.0011550389209898173,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0001434634974533107,
      "loss": 2.9613,
      "step": 170
    },
    {
      "epoch": 0.0011618332675838751,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.00014431239388794568,
      "loss": 2.9482,
      "step": 171
    },
    {
      "epoch": 0.001168627614177933,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.00014516129032258066,
      "loss": 2.8264,
      "step": 172
    },
    {
      "epoch": 0.0011754219607719907,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00014601018675721562,
      "loss": 2.7461,
      "step": 173
    },
    {
      "epoch": 0.0011822163073660485,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0001468590831918506,
      "loss": 2.9618,
      "step": 174
    },
    {
      "epoch": 0.0011890106539601062,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.00014770797962648558,
      "loss": 2.8125,
      "step": 175
    },
    {
      "epoch": 0.0011958050005541638,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00014855687606112053,
      "loss": 2.8985,
      "step": 176
    },
    {
      "epoch": 0.0012025993471482216,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0001494057724957555,
      "loss": 2.8462,
      "step": 177
    },
    {
      "epoch": 0.0012093936937422793,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0001502546689303905,
      "loss": 2.8099,
      "step": 178
    },
    {
      "epoch": 0.0012161880403363371,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00015110356536502547,
      "loss": 2.8606,
      "step": 179
    },
    {
      "epoch": 0.001222982386930395,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.00015195246179966043,
      "loss": 2.771,
      "step": 180
    },
    {
      "epoch": 0.0012297767335244527,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0001528013582342954,
      "loss": 2.8421,
      "step": 181
    },
    {
      "epoch": 0.0012365710801185105,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0001536502546689304,
      "loss": 2.818,
      "step": 182
    },
    {
      "epoch": 0.0012433654267125682,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00015449915110356534,
      "loss": 2.8839,
      "step": 183
    },
    {
      "epoch": 0.0012501597733066258,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.00015534804753820035,
      "loss": 2.8036,
      "step": 184
    },
    {
      "epoch": 0.0012569541199006836,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00015619694397283533,
      "loss": 2.8591,
      "step": 185
    },
    {
      "epoch": 0.0012637484664947414,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0001570458404074703,
      "loss": 2.7169,
      "step": 186
    },
    {
      "epoch": 0.0012705428130887991,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00015789473684210527,
      "loss": 2.9075,
      "step": 187
    },
    {
      "epoch": 0.001277337159682857,
      "grad_norm": 0.234375,
      "learning_rate": 0.00015874363327674025,
      "loss": 2.8527,
      "step": 188
    },
    {
      "epoch": 0.0012841315062769147,
      "grad_norm": 0.265625,
      "learning_rate": 0.00015959252971137523,
      "loss": 2.8741,
      "step": 189
    },
    {
      "epoch": 0.0012909258528709725,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00016044142614601018,
      "loss": 2.8626,
      "step": 190
    },
    {
      "epoch": 0.00129772019946503,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00016129032258064516,
      "loss": 2.7825,
      "step": 191
    },
    {
      "epoch": 0.0013045145460590878,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00016213921901528014,
      "loss": 2.7023,
      "step": 192
    },
    {
      "epoch": 0.0013113088926531456,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00016298811544991512,
      "loss": 2.7873,
      "step": 193
    },
    {
      "epoch": 0.0013181032392472034,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.00016383701188455008,
      "loss": 3.0498,
      "step": 194
    },
    {
      "epoch": 0.0013248975858412611,
      "grad_norm": 0.240234375,
      "learning_rate": 0.00016468590831918506,
      "loss": 2.7902,
      "step": 195
    },
    {
      "epoch": 0.001331691932435319,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00016553480475382004,
      "loss": 2.9395,
      "step": 196
    },
    {
      "epoch": 0.0013384862790293767,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.000166383701188455,
      "loss": 2.7656,
      "step": 197
    },
    {
      "epoch": 0.0013452806256234345,
      "grad_norm": 0.259765625,
      "learning_rate": 0.00016723259762308997,
      "loss": 2.7662,
      "step": 198
    },
    {
      "epoch": 0.001352074972217492,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00016808149405772495,
      "loss": 2.8428,
      "step": 199
    },
    {
      "epoch": 0.0013588693188115498,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00016893039049235993,
      "loss": 2.8119,
      "step": 200
    },
    {
      "epoch": 0.0013656636654056076,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00016977928692699492,
      "loss": 2.875,
      "step": 201
    },
    {
      "epoch": 0.0013724580119996654,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0001706281833616299,
      "loss": 2.949,
      "step": 202
    },
    {
      "epoch": 0.0013792523585937232,
      "grad_norm": 0.236328125,
      "learning_rate": 0.00017147707979626488,
      "loss": 2.6981,
      "step": 203
    },
    {
      "epoch": 0.001386046705187781,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00017232597623089983,
      "loss": 3.0539,
      "step": 204
    },
    {
      "epoch": 0.0013928410517818387,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0001731748726655348,
      "loss": 2.6783,
      "step": 205
    },
    {
      "epoch": 0.0013996353983758963,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0001740237691001698,
      "loss": 2.6671,
      "step": 206
    },
    {
      "epoch": 0.001406429744969954,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00017487266553480477,
      "loss": 2.7022,
      "step": 207
    },
    {
      "epoch": 0.0014132240915640118,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00017572156196943973,
      "loss": 2.9025,
      "step": 208
    },
    {
      "epoch": 0.0014200184381580696,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0001765704584040747,
      "loss": 2.8172,
      "step": 209
    },
    {
      "epoch": 0.0014268127847521274,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0001774193548387097,
      "loss": 2.826,
      "step": 210
    },
    {
      "epoch": 0.0014336071313461852,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.00017826825127334464,
      "loss": 2.7315,
      "step": 211
    },
    {
      "epoch": 0.001440401477940243,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00017911714770797962,
      "loss": 2.8726,
      "step": 212
    },
    {
      "epoch": 0.0014471958245343007,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0001799660441426146,
      "loss": 3.0269,
      "step": 213
    },
    {
      "epoch": 0.0014539901711283583,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00018081494057724958,
      "loss": 2.8767,
      "step": 214
    },
    {
      "epoch": 0.001460784517722416,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00018166383701188454,
      "loss": 2.7111,
      "step": 215
    },
    {
      "epoch": 0.0014675788643164738,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00018251273344651952,
      "loss": 2.8907,
      "step": 216
    },
    {
      "epoch": 0.0014743732109105316,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0001833616298811545,
      "loss": 2.7534,
      "step": 217
    },
    {
      "epoch": 0.0014811675575045894,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00018421052631578948,
      "loss": 2.6244,
      "step": 218
    },
    {
      "epoch": 0.0014879619040986472,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00018505942275042446,
      "loss": 2.79,
      "step": 219
    },
    {
      "epoch": 0.001494756250692705,
      "grad_norm": 0.23828125,
      "learning_rate": 0.00018590831918505944,
      "loss": 2.8551,
      "step": 220
    },
    {
      "epoch": 0.0015015505972867625,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00018675721561969442,
      "loss": 2.7289,
      "step": 221
    },
    {
      "epoch": 0.0015083449438808203,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00018760611205432938,
      "loss": 2.8731,
      "step": 222
    },
    {
      "epoch": 0.001515139290474878,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00018845500848896436,
      "loss": 2.9027,
      "step": 223
    },
    {
      "epoch": 0.0015219336370689359,
      "grad_norm": 0.23046875,
      "learning_rate": 0.00018930390492359934,
      "loss": 2.8081,
      "step": 224
    },
    {
      "epoch": 0.0015287279836629936,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0001901528013582343,
      "loss": 2.8094,
      "step": 225
    },
    {
      "epoch": 0.0015355223302570514,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00019100169779286927,
      "loss": 2.8386,
      "step": 226
    },
    {
      "epoch": 0.0015423166768511092,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00019185059422750425,
      "loss": 2.7869,
      "step": 227
    },
    {
      "epoch": 0.001549111023445167,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0001926994906621392,
      "loss": 2.8567,
      "step": 228
    },
    {
      "epoch": 0.0015559053700392245,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0001935483870967742,
      "loss": 2.8412,
      "step": 229
    },
    {
      "epoch": 0.0015626997166332823,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00019439728353140917,
      "loss": 2.7561,
      "step": 230
    },
    {
      "epoch": 0.00156949406322734,
      "grad_norm": 0.23046875,
      "learning_rate": 0.00019524617996604415,
      "loss": 2.9081,
      "step": 231
    },
    {
      "epoch": 0.0015762884098213979,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0001960950764006791,
      "loss": 2.9366,
      "step": 232
    },
    {
      "epoch": 0.0015830827564154556,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00019694397283531408,
      "loss": 2.8806,
      "step": 233
    },
    {
      "epoch": 0.0015898771030095134,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00019779286926994906,
      "loss": 2.9277,
      "step": 234
    },
    {
      "epoch": 0.0015966714496035712,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00019864176570458402,
      "loss": 2.8793,
      "step": 235
    },
    {
      "epoch": 0.0016034657961976288,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00019949066213921903,
      "loss": 2.9344,
      "step": 236
    },
    {
      "epoch": 0.0016102601427916865,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.000200339558573854,
      "loss": 2.6976,
      "step": 237
    },
    {
      "epoch": 0.0016170544893857443,
      "grad_norm": 0.2734375,
      "learning_rate": 0.000201188455008489,
      "loss": 2.8692,
      "step": 238
    },
    {
      "epoch": 0.001623848835979802,
      "grad_norm": 0.20703125,
      "learning_rate": 0.00020203735144312394,
      "loss": 2.7494,
      "step": 239
    },
    {
      "epoch": 0.0016306431825738599,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00020288624787775892,
      "loss": 2.8449,
      "step": 240
    },
    {
      "epoch": 0.0016374375291679177,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0002037351443123939,
      "loss": 2.8963,
      "step": 241
    },
    {
      "epoch": 0.0016442318757619754,
      "grad_norm": 0.212890625,
      "learning_rate": 0.00020458404074702886,
      "loss": 2.8083,
      "step": 242
    },
    {
      "epoch": 0.0016510262223560332,
      "grad_norm": 0.26953125,
      "learning_rate": 0.00020543293718166384,
      "loss": 2.726,
      "step": 243
    },
    {
      "epoch": 0.0016578205689500908,
      "grad_norm": 0.265625,
      "learning_rate": 0.00020628183361629882,
      "loss": 2.9426,
      "step": 244
    },
    {
      "epoch": 0.0016646149155441485,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0002071307300509338,
      "loss": 2.9489,
      "step": 245
    },
    {
      "epoch": 0.0016714092621382063,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00020797962648556875,
      "loss": 2.646,
      "step": 246
    },
    {
      "epoch": 0.001678203608732264,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00020882852292020373,
      "loss": 2.7717,
      "step": 247
    },
    {
      "epoch": 0.0016849979553263219,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00020967741935483871,
      "loss": 2.6454,
      "step": 248
    },
    {
      "epoch": 0.0016917923019203797,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00021052631578947367,
      "loss": 2.8217,
      "step": 249
    },
    {
      "epoch": 0.0016985866485144374,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00021137521222410865,
      "loss": 2.8441,
      "step": 250
    },
    {
      "epoch": 0.001705380995108495,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00021222410865874363,
      "loss": 2.8293,
      "step": 251
    },
    {
      "epoch": 0.0017121753417025528,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0002130730050933786,
      "loss": 2.735,
      "step": 252
    },
    {
      "epoch": 0.0017189696882966106,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0002139219015280136,
      "loss": 2.8529,
      "step": 253
    },
    {
      "epoch": 0.0017257640348906683,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00021477079796264857,
      "loss": 2.8153,
      "step": 254
    },
    {
      "epoch": 0.0017325583814847261,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00021561969439728355,
      "loss": 2.8546,
      "step": 255
    },
    {
      "epoch": 0.001739352728078784,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002164685908319185,
      "loss": 2.9242,
      "step": 256
    },
    {
      "epoch": 0.0017461470746728417,
      "grad_norm": 0.234375,
      "learning_rate": 0.0002173174872665535,
      "loss": 2.7238,
      "step": 257
    },
    {
      "epoch": 0.0017529414212668995,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00021816638370118847,
      "loss": 2.8765,
      "step": 258
    },
    {
      "epoch": 0.001759735767860957,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.00021901528013582345,
      "loss": 2.888,
      "step": 259
    },
    {
      "epoch": 0.0017665301144550148,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002198641765704584,
      "loss": 2.6205,
      "step": 260
    },
    {
      "epoch": 0.0017733244610490726,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00022071307300509338,
      "loss": 2.6971,
      "step": 261
    },
    {
      "epoch": 0.0017801188076431303,
      "grad_norm": 0.244140625,
      "learning_rate": 0.00022156196943972836,
      "loss": 2.8212,
      "step": 262
    },
    {
      "epoch": 0.0017869131542371881,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00022241086587436332,
      "loss": 2.8072,
      "step": 263
    },
    {
      "epoch": 0.001793707500831246,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002232597623089983,
      "loss": 2.8601,
      "step": 264
    },
    {
      "epoch": 0.0018005018474253037,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00022410865874363328,
      "loss": 2.735,
      "step": 265
    },
    {
      "epoch": 0.0018072961940193612,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00022495755517826826,
      "loss": 2.8426,
      "step": 266
    },
    {
      "epoch": 0.001814090540613419,
      "grad_norm": 0.234375,
      "learning_rate": 0.00022580645161290321,
      "loss": 2.8843,
      "step": 267
    },
    {
      "epoch": 0.0018208848872074768,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002266553480475382,
      "loss": 2.8107,
      "step": 268
    },
    {
      "epoch": 0.0018276792338015346,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00022750424448217318,
      "loss": 2.7304,
      "step": 269
    },
    {
      "epoch": 0.0018344735803955924,
      "grad_norm": 0.201171875,
      "learning_rate": 0.00022835314091680816,
      "loss": 2.8788,
      "step": 270
    },
    {
      "epoch": 0.0018412679269896501,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00022920203735144314,
      "loss": 2.8424,
      "step": 271
    },
    {
      "epoch": 0.001848062273583708,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00023005093378607812,
      "loss": 2.8285,
      "step": 272
    },
    {
      "epoch": 0.0018548566201777657,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00023089983022071307,
      "loss": 2.8132,
      "step": 273
    },
    {
      "epoch": 0.0018616509667718233,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00023174872665534805,
      "loss": 2.7512,
      "step": 274
    },
    {
      "epoch": 0.001868445313365881,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00023259762308998303,
      "loss": 2.8947,
      "step": 275
    },
    {
      "epoch": 0.0018752396599599388,
      "grad_norm": 0.2578125,
      "learning_rate": 0.00023344651952461801,
      "loss": 2.8927,
      "step": 276
    },
    {
      "epoch": 0.0018820340065539966,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00023429541595925297,
      "loss": 2.7743,
      "step": 277
    },
    {
      "epoch": 0.0018888283531480544,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.00023514431239388795,
      "loss": 2.9656,
      "step": 278
    },
    {
      "epoch": 0.0018956226997421121,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00023599320882852293,
      "loss": 2.8434,
      "step": 279
    },
    {
      "epoch": 0.00190241704633617,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.00023684210526315788,
      "loss": 2.7542,
      "step": 280
    },
    {
      "epoch": 0.0019092113929302275,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00023769100169779286,
      "loss": 3.057,
      "step": 281
    },
    {
      "epoch": 0.0019160057395242853,
      "grad_norm": 0.21875,
      "learning_rate": 0.00023853989813242785,
      "loss": 2.7557,
      "step": 282
    },
    {
      "epoch": 0.001922800086118343,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00023938879456706283,
      "loss": 2.8017,
      "step": 283
    },
    {
      "epoch": 0.0019295944327124008,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00024023769100169778,
      "loss": 2.6527,
      "step": 284
    },
    {
      "epoch": 0.0019363887793064586,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00024108658743633276,
      "loss": 2.8017,
      "step": 285
    },
    {
      "epoch": 0.0019431831259005164,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00024193548387096774,
      "loss": 2.73,
      "step": 286
    },
    {
      "epoch": 0.0019499774724945742,
      "grad_norm": 0.283203125,
      "learning_rate": 0.0002427843803056027,
      "loss": 2.7771,
      "step": 287
    },
    {
      "epoch": 0.001956771819088632,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002436332767402377,
      "loss": 2.7042,
      "step": 288
    },
    {
      "epoch": 0.0019635661656826897,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002444821731748727,
      "loss": 2.7226,
      "step": 289
    },
    {
      "epoch": 0.0019703605122767475,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00024533106960950766,
      "loss": 2.8574,
      "step": 290
    },
    {
      "epoch": 0.0019771548588708053,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002461799660441426,
      "loss": 2.7581,
      "step": 291
    },
    {
      "epoch": 0.0019839492054648626,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.00024702886247877757,
      "loss": 2.8614,
      "step": 292
    },
    {
      "epoch": 0.0019907435520589204,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00024787775891341255,
      "loss": 2.6991,
      "step": 293
    },
    {
      "epoch": 0.001997537898652978,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.00024872665534804753,
      "loss": 2.7034,
      "step": 294
    },
    {
      "epoch": 0.002004332245247036,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002495755517826825,
      "loss": 2.7641,
      "step": 295
    },
    {
      "epoch": 0.0020111265918410937,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002504244482173175,
      "loss": 2.7944,
      "step": 296
    },
    {
      "epoch": 0.0020179209384351515,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0002512733446519525,
      "loss": 2.8025,
      "step": 297
    },
    {
      "epoch": 0.0020247152850292093,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00025212224108658746,
      "loss": 2.8246,
      "step": 298
    },
    {
      "epoch": 0.002031509631623267,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00025297113752122244,
      "loss": 2.7541,
      "step": 299
    },
    {
      "epoch": 0.002038303978217325,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0002538200339558574,
      "loss": 2.7473,
      "step": 300
    },
    {
      "epoch": 0.0020450983248113826,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0002546689303904924,
      "loss": 2.6191,
      "step": 301
    },
    {
      "epoch": 0.0020518926714054404,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002555178268251274,
      "loss": 2.826,
      "step": 302
    },
    {
      "epoch": 0.002058687017999498,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0002563667232597623,
      "loss": 2.7581,
      "step": 303
    },
    {
      "epoch": 0.002065481364593556,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002572156196943973,
      "loss": 2.7693,
      "step": 304
    },
    {
      "epoch": 0.0020722757111876137,
      "grad_norm": 0.267578125,
      "learning_rate": 0.00025806451612903227,
      "loss": 2.7811,
      "step": 305
    },
    {
      "epoch": 0.0020790700577816715,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00025891341256366725,
      "loss": 2.6706,
      "step": 306
    },
    {
      "epoch": 0.002085864404375729,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00025976230899830223,
      "loss": 2.8993,
      "step": 307
    },
    {
      "epoch": 0.0020926587509697866,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0002606112054329372,
      "loss": 2.8017,
      "step": 308
    },
    {
      "epoch": 0.0020994530975638444,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0002614601018675722,
      "loss": 2.829,
      "step": 309
    },
    {
      "epoch": 0.002106247444157902,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002623089983022071,
      "loss": 2.71,
      "step": 310
    },
    {
      "epoch": 0.00211304179075196,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0002631578947368421,
      "loss": 2.7595,
      "step": 311
    },
    {
      "epoch": 0.0021198361373460177,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0002640067911714771,
      "loss": 2.7617,
      "step": 312
    },
    {
      "epoch": 0.0021266304839400755,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00026485568760611206,
      "loss": 2.6043,
      "step": 313
    },
    {
      "epoch": 0.0021334248305341333,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00026570458404074704,
      "loss": 2.8243,
      "step": 314
    },
    {
      "epoch": 0.002140219177128191,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.000266553480475382,
      "loss": 2.8651,
      "step": 315
    },
    {
      "epoch": 0.002147013523722249,
      "grad_norm": 0.22265625,
      "learning_rate": 0.000267402376910017,
      "loss": 2.661,
      "step": 316
    },
    {
      "epoch": 0.0021538078703163066,
      "grad_norm": 0.2734375,
      "learning_rate": 0.00026825127334465193,
      "loss": 2.7639,
      "step": 317
    },
    {
      "epoch": 0.0021606022169103644,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0002691001697792869,
      "loss": 2.6235,
      "step": 318
    },
    {
      "epoch": 0.002167396563504422,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0002699490662139219,
      "loss": 2.7566,
      "step": 319
    },
    {
      "epoch": 0.00217419091009848,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00027079796264855687,
      "loss": 2.7299,
      "step": 320
    },
    {
      "epoch": 0.0021809852566925378,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00027164685908319185,
      "loss": 2.751,
      "step": 321
    },
    {
      "epoch": 0.002187779603286595,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00027249575551782683,
      "loss": 2.8527,
      "step": 322
    },
    {
      "epoch": 0.002194573949880653,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002733446519524618,
      "loss": 2.8027,
      "step": 323
    },
    {
      "epoch": 0.0022013682964747107,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00027419354838709674,
      "loss": 2.7229,
      "step": 324
    },
    {
      "epoch": 0.0022081626430687684,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0002750424448217317,
      "loss": 2.7974,
      "step": 325
    },
    {
      "epoch": 0.002214956989662826,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0002758913412563667,
      "loss": 2.7243,
      "step": 326
    },
    {
      "epoch": 0.002221751336256884,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0002767402376910017,
      "loss": 2.7873,
      "step": 327
    },
    {
      "epoch": 0.0022285456828509418,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00027758913412563666,
      "loss": 2.8559,
      "step": 328
    },
    {
      "epoch": 0.0022353400294449995,
      "grad_norm": 0.23046875,
      "learning_rate": 0.00027843803056027165,
      "loss": 2.821,
      "step": 329
    },
    {
      "epoch": 0.0022421343760390573,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0002792869269949067,
      "loss": 2.7314,
      "step": 330
    },
    {
      "epoch": 0.002248928722633115,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0002801358234295416,
      "loss": 2.72,
      "step": 331
    },
    {
      "epoch": 0.002255723069227173,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0002809847198641766,
      "loss": 2.715,
      "step": 332
    },
    {
      "epoch": 0.0022625174158212307,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00028183361629881157,
      "loss": 2.9223,
      "step": 333
    },
    {
      "epoch": 0.0022693117624152884,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.00028268251273344655,
      "loss": 2.8483,
      "step": 334
    },
    {
      "epoch": 0.0022761061090093462,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00028353140916808153,
      "loss": 2.7509,
      "step": 335
    },
    {
      "epoch": 0.002282900455603404,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0002843803056027165,
      "loss": 3.0235,
      "step": 336
    },
    {
      "epoch": 0.0022896948021974613,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002852292020373515,
      "loss": 2.8863,
      "step": 337
    },
    {
      "epoch": 0.002296489148791519,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0002860780984719864,
      "loss": 2.7432,
      "step": 338
    },
    {
      "epoch": 0.002303283495385577,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002869269949066214,
      "loss": 2.8468,
      "step": 339
    },
    {
      "epoch": 0.0023100778419796347,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0002877758913412564,
      "loss": 2.6698,
      "step": 340
    },
    {
      "epoch": 0.0023168721885736925,
      "grad_norm": 0.296875,
      "learning_rate": 0.00028862478777589136,
      "loss": 2.743,
      "step": 341
    },
    {
      "epoch": 0.0023236665351677502,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00028947368421052634,
      "loss": 2.7341,
      "step": 342
    },
    {
      "epoch": 0.002330460881761808,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0002903225806451613,
      "loss": 2.697,
      "step": 343
    },
    {
      "epoch": 0.002337255228355866,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002911714770797963,
      "loss": 2.848,
      "step": 344
    },
    {
      "epoch": 0.0023440495749499236,
      "grad_norm": 0.236328125,
      "learning_rate": 0.00029202037351443123,
      "loss": 2.6398,
      "step": 345
    },
    {
      "epoch": 0.0023508439215439813,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0002928692699490662,
      "loss": 2.8525,
      "step": 346
    },
    {
      "epoch": 0.002357638268138039,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0002937181663837012,
      "loss": 2.8375,
      "step": 347
    },
    {
      "epoch": 0.002364432614732097,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00029456706281833617,
      "loss": 2.6662,
      "step": 348
    },
    {
      "epoch": 0.0023712269613261547,
      "grad_norm": 0.25390625,
      "learning_rate": 0.00029541595925297115,
      "loss": 2.9168,
      "step": 349
    },
    {
      "epoch": 0.0023780213079202125,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00029626485568760613,
      "loss": 2.863,
      "step": 350
    },
    {
      "epoch": 0.0023848156545142702,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00029711375212224106,
      "loss": 2.7344,
      "step": 351
    },
    {
      "epoch": 0.0023916100011083276,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00029796264855687604,
      "loss": 2.6762,
      "step": 352
    },
    {
      "epoch": 0.0023984043477023854,
      "grad_norm": 0.244140625,
      "learning_rate": 0.000298811544991511,
      "loss": 2.7263,
      "step": 353
    },
    {
      "epoch": 0.002405198694296443,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.000299660441426146,
      "loss": 2.7859,
      "step": 354
    },
    {
      "epoch": 0.002411993040890501,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.000300509337860781,
      "loss": 2.8497,
      "step": 355
    },
    {
      "epoch": 0.0024187873874845587,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.00030135823429541596,
      "loss": 2.8111,
      "step": 356
    },
    {
      "epoch": 0.0024255817340786165,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00030220713073005095,
      "loss": 2.6183,
      "step": 357
    },
    {
      "epoch": 0.0024323760806726743,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00030305602716468587,
      "loss": 2.8038,
      "step": 358
    },
    {
      "epoch": 0.002439170427266732,
      "grad_norm": 0.2421875,
      "learning_rate": 0.00030390492359932085,
      "loss": 2.748,
      "step": 359
    },
    {
      "epoch": 0.00244596477386079,
      "grad_norm": 0.234375,
      "learning_rate": 0.00030475382003395583,
      "loss": 2.8563,
      "step": 360
    },
    {
      "epoch": 0.0024527591204548476,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0003056027164685908,
      "loss": 2.6595,
      "step": 361
    },
    {
      "epoch": 0.0024595534670489054,
      "grad_norm": 0.21875,
      "learning_rate": 0.0003064516129032258,
      "loss": 2.786,
      "step": 362
    },
    {
      "epoch": 0.002466347813642963,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0003073005093378608,
      "loss": 2.847,
      "step": 363
    },
    {
      "epoch": 0.002473142160237021,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.00030814940577249576,
      "loss": 2.8375,
      "step": 364
    },
    {
      "epoch": 0.0024799365068310787,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0003089983022071307,
      "loss": 2.6996,
      "step": 365
    },
    {
      "epoch": 0.0024867308534251365,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0003098471986417657,
      "loss": 2.7109,
      "step": 366
    },
    {
      "epoch": 0.002493525200019194,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0003106960950764007,
      "loss": 2.8857,
      "step": 367
    },
    {
      "epoch": 0.0025003195466132516,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0003115449915110357,
      "loss": 2.7875,
      "step": 368
    },
    {
      "epoch": 0.0025071138932073094,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00031239388794567066,
      "loss": 2.7761,
      "step": 369
    },
    {
      "epoch": 0.002513908239801367,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00031324278438030564,
      "loss": 2.6483,
      "step": 370
    },
    {
      "epoch": 0.002520702586395425,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0003140916808149406,
      "loss": 2.7672,
      "step": 371
    },
    {
      "epoch": 0.0025274969329894827,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00031494057724957555,
      "loss": 2.8583,
      "step": 372
    },
    {
      "epoch": 0.0025342912795835405,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00031578947368421053,
      "loss": 2.6976,
      "step": 373
    },
    {
      "epoch": 0.0025410856261775983,
      "grad_norm": 0.265625,
      "learning_rate": 0.0003166383701188455,
      "loss": 2.9443,
      "step": 374
    },
    {
      "epoch": 0.002547879972771656,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0003174872665534805,
      "loss": 2.8704,
      "step": 375
    },
    {
      "epoch": 0.002554674319365714,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00031833616298811547,
      "loss": 2.749,
      "step": 376
    },
    {
      "epoch": 0.0025614686659597716,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00031918505942275045,
      "loss": 2.7277,
      "step": 377
    },
    {
      "epoch": 0.0025682630125538294,
      "grad_norm": 0.240234375,
      "learning_rate": 0.00032003395585738543,
      "loss": 2.6732,
      "step": 378
    },
    {
      "epoch": 0.002575057359147887,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00032088285229202036,
      "loss": 2.6804,
      "step": 379
    },
    {
      "epoch": 0.002581851705741945,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.00032173174872665534,
      "loss": 2.6694,
      "step": 380
    },
    {
      "epoch": 0.0025886460523360027,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0003225806451612903,
      "loss": 2.767,
      "step": 381
    },
    {
      "epoch": 0.00259544039893006,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0003234295415959253,
      "loss": 2.8604,
      "step": 382
    },
    {
      "epoch": 0.002602234745524118,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0003242784380305603,
      "loss": 2.7404,
      "step": 383
    },
    {
      "epoch": 0.0026090290921181756,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00032512733446519526,
      "loss": 2.6912,
      "step": 384
    },
    {
      "epoch": 0.0026158234387122334,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00032597623089983025,
      "loss": 2.7596,
      "step": 385
    },
    {
      "epoch": 0.002622617785306291,
      "grad_norm": 0.21484375,
      "learning_rate": 0.00032682512733446517,
      "loss": 2.6758,
      "step": 386
    },
    {
      "epoch": 0.002629412131900349,
      "grad_norm": 0.21875,
      "learning_rate": 0.00032767402376910015,
      "loss": 2.721,
      "step": 387
    },
    {
      "epoch": 0.0026362064784944067,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.00032852292020373513,
      "loss": 2.8233,
      "step": 388
    },
    {
      "epoch": 0.0026430008250884645,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0003293718166383701,
      "loss": 2.8032,
      "step": 389
    },
    {
      "epoch": 0.0026497951716825223,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0003302207130730051,
      "loss": 2.7195,
      "step": 390
    },
    {
      "epoch": 0.00265658951827658,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0003310696095076401,
      "loss": 2.7351,
      "step": 391
    },
    {
      "epoch": 0.002663383864870638,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00033191850594227506,
      "loss": 2.6993,
      "step": 392
    },
    {
      "epoch": 0.0026701782114646956,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00033276740237691,
      "loss": 2.9046,
      "step": 393
    },
    {
      "epoch": 0.0026769725580587534,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00033361629881154496,
      "loss": 2.8229,
      "step": 394
    },
    {
      "epoch": 0.002683766904652811,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00033446519524617995,
      "loss": 2.7646,
      "step": 395
    },
    {
      "epoch": 0.002690561251246869,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0003353140916808149,
      "loss": 2.8116,
      "step": 396
    },
    {
      "epoch": 0.0026973555978409263,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0003361629881154499,
      "loss": 2.6848,
      "step": 397
    },
    {
      "epoch": 0.002704149944434984,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0003370118845500849,
      "loss": 2.7007,
      "step": 398
    },
    {
      "epoch": 0.002710944291029042,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00033786078098471987,
      "loss": 2.8143,
      "step": 399
    },
    {
      "epoch": 0.0027177386376230996,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00033870967741935485,
      "loss": 2.685,
      "step": 400
    },
    {
      "epoch": 0.0027245329842171574,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00033955857385398983,
      "loss": 2.8318,
      "step": 401
    },
    {
      "epoch": 0.002731327330811215,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0003404074702886248,
      "loss": 2.7458,
      "step": 402
    },
    {
      "epoch": 0.002738121677405273,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0003412563667232598,
      "loss": 2.86,
      "step": 403
    },
    {
      "epoch": 0.0027449160239993308,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00034210526315789477,
      "loss": 2.7335,
      "step": 404
    },
    {
      "epoch": 0.0027517103705933885,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.00034295415959252975,
      "loss": 2.6935,
      "step": 405
    },
    {
      "epoch": 0.0027585047171874463,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00034380305602716473,
      "loss": 2.707,
      "step": 406
    },
    {
      "epoch": 0.002765299063781504,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00034465195246179966,
      "loss": 2.7663,
      "step": 407
    },
    {
      "epoch": 0.002772093410375562,
      "grad_norm": 0.22265625,
      "learning_rate": 0.00034550084889643464,
      "loss": 2.8401,
      "step": 408
    },
    {
      "epoch": 0.0027788877569696196,
      "grad_norm": 0.296875,
      "learning_rate": 0.0003463497453310696,
      "loss": 2.7457,
      "step": 409
    },
    {
      "epoch": 0.0027856821035636774,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0003471986417657046,
      "loss": 2.6785,
      "step": 410
    },
    {
      "epoch": 0.002792476450157735,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0003480475382003396,
      "loss": 2.8392,
      "step": 411
    },
    {
      "epoch": 0.0027992707967517926,
      "grad_norm": 0.3125,
      "learning_rate": 0.00034889643463497456,
      "loss": 2.7914,
      "step": 412
    },
    {
      "epoch": 0.0028060651433458503,
      "grad_norm": 0.234375,
      "learning_rate": 0.00034974533106960955,
      "loss": 2.789,
      "step": 413
    },
    {
      "epoch": 0.002812859489939908,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00035059422750424447,
      "loss": 2.7387,
      "step": 414
    },
    {
      "epoch": 0.002819653836533966,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00035144312393887945,
      "loss": 2.8155,
      "step": 415
    },
    {
      "epoch": 0.0028264481831280237,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00035229202037351443,
      "loss": 2.723,
      "step": 416
    },
    {
      "epoch": 0.0028332425297220814,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0003531409168081494,
      "loss": 2.7192,
      "step": 417
    },
    {
      "epoch": 0.0028400368763161392,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0003539898132427844,
      "loss": 2.8419,
      "step": 418
    },
    {
      "epoch": 0.002846831222910197,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0003548387096774194,
      "loss": 2.6921,
      "step": 419
    },
    {
      "epoch": 0.0028536255695042548,
      "grad_norm": 0.240234375,
      "learning_rate": 0.00035568760611205436,
      "loss": 2.6149,
      "step": 420
    },
    {
      "epoch": 0.0028604199160983126,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0003565365025466893,
      "loss": 2.6579,
      "step": 421
    },
    {
      "epoch": 0.0028672142626923703,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.00035738539898132426,
      "loss": 2.7197,
      "step": 422
    },
    {
      "epoch": 0.002874008609286428,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00035823429541595925,
      "loss": 2.7502,
      "step": 423
    },
    {
      "epoch": 0.002880802955880486,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0003590831918505942,
      "loss": 2.7456,
      "step": 424
    },
    {
      "epoch": 0.0028875973024745437,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0003599320882852292,
      "loss": 2.8111,
      "step": 425
    },
    {
      "epoch": 0.0028943916490686014,
      "grad_norm": 0.28125,
      "learning_rate": 0.0003607809847198642,
      "loss": 2.7795,
      "step": 426
    },
    {
      "epoch": 0.002901185995662659,
      "grad_norm": 0.21875,
      "learning_rate": 0.00036162988115449917,
      "loss": 2.7733,
      "step": 427
    },
    {
      "epoch": 0.0029079803422567166,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0003624787775891341,
      "loss": 2.6609,
      "step": 428
    },
    {
      "epoch": 0.0029147746888507743,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0003633276740237691,
      "loss": 2.8458,
      "step": 429
    },
    {
      "epoch": 0.002921569035444832,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00036417657045840406,
      "loss": 2.7201,
      "step": 430
    },
    {
      "epoch": 0.00292836338203889,
      "grad_norm": 0.244140625,
      "learning_rate": 0.00036502546689303904,
      "loss": 2.8471,
      "step": 431
    },
    {
      "epoch": 0.0029351577286329477,
      "grad_norm": 0.232421875,
      "learning_rate": 0.000365874363327674,
      "loss": 2.8068,
      "step": 432
    },
    {
      "epoch": 0.0029419520752270055,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.000366723259762309,
      "loss": 2.689,
      "step": 433
    },
    {
      "epoch": 0.0029487464218210632,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00036757215619694403,
      "loss": 2.6078,
      "step": 434
    },
    {
      "epoch": 0.002955540768415121,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00036842105263157896,
      "loss": 2.8075,
      "step": 435
    },
    {
      "epoch": 0.002962335115009179,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00036926994906621394,
      "loss": 2.7959,
      "step": 436
    },
    {
      "epoch": 0.0029691294616032366,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0003701188455008489,
      "loss": 2.846,
      "step": 437
    },
    {
      "epoch": 0.0029759238081972944,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0003709677419354839,
      "loss": 2.786,
      "step": 438
    },
    {
      "epoch": 0.002982718154791352,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0003718166383701189,
      "loss": 2.6625,
      "step": 439
    },
    {
      "epoch": 0.00298951250138541,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00037266553480475386,
      "loss": 2.8853,
      "step": 440
    },
    {
      "epoch": 0.0029963068479794677,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00037351443123938885,
      "loss": 2.8381,
      "step": 441
    },
    {
      "epoch": 0.003003101194573525,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00037436332767402377,
      "loss": 2.7845,
      "step": 442
    },
    {
      "epoch": 0.003009895541167583,
      "grad_norm": 0.2265625,
      "learning_rate": 0.00037521222410865875,
      "loss": 2.7517,
      "step": 443
    },
    {
      "epoch": 0.0030166898877616406,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.00037606112054329373,
      "loss": 2.7208,
      "step": 444
    },
    {
      "epoch": 0.0030234842343556984,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0003769100169779287,
      "loss": 2.7503,
      "step": 445
    },
    {
      "epoch": 0.003030278580949756,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0003777589134125637,
      "loss": 2.8548,
      "step": 446
    },
    {
      "epoch": 0.003037072927543814,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0003786078098471987,
      "loss": 2.6796,
      "step": 447
    },
    {
      "epoch": 0.0030438672741378717,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0003794567062818336,
      "loss": 2.8325,
      "step": 448
    },
    {
      "epoch": 0.0030506616207319295,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0003803056027164686,
      "loss": 2.7429,
      "step": 449
    },
    {
      "epoch": 0.0030574559673259873,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00038115449915110356,
      "loss": 2.7617,
      "step": 450
    },
    {
      "epoch": 0.003064250313920045,
      "grad_norm": 0.2421875,
      "learning_rate": 0.00038200339558573855,
      "loss": 2.7129,
      "step": 451
    },
    {
      "epoch": 0.003071044660514103,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0003828522920203735,
      "loss": 2.6435,
      "step": 452
    },
    {
      "epoch": 0.0030778390071081606,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0003837011884550085,
      "loss": 2.7998,
      "step": 453
    },
    {
      "epoch": 0.0030846333537022184,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0003845500848896435,
      "loss": 2.7881,
      "step": 454
    },
    {
      "epoch": 0.003091427700296276,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0003853989813242784,
      "loss": 2.7409,
      "step": 455
    },
    {
      "epoch": 0.003098222046890334,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0003862478777589134,
      "loss": 2.832,
      "step": 456
    },
    {
      "epoch": 0.0031050163934843913,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0003870967741935484,
      "loss": 2.7522,
      "step": 457
    },
    {
      "epoch": 0.003111810740078449,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00038794567062818336,
      "loss": 2.6642,
      "step": 458
    },
    {
      "epoch": 0.003118605086672507,
      "grad_norm": 0.244140625,
      "learning_rate": 0.00038879456706281834,
      "loss": 2.6079,
      "step": 459
    },
    {
      "epoch": 0.0031253994332665646,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0003896434634974533,
      "loss": 2.84,
      "step": 460
    },
    {
      "epoch": 0.0031321937798606224,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0003904923599320883,
      "loss": 2.6928,
      "step": 461
    },
    {
      "epoch": 0.00313898812645468,
      "grad_norm": 0.25,
      "learning_rate": 0.0003913412563667232,
      "loss": 2.834,
      "step": 462
    },
    {
      "epoch": 0.003145782473048738,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0003921901528013582,
      "loss": 2.7723,
      "step": 463
    },
    {
      "epoch": 0.0031525768196427957,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0003930390492359932,
      "loss": 2.746,
      "step": 464
    },
    {
      "epoch": 0.0031593711662368535,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.00039388794567062817,
      "loss": 2.7141,
      "step": 465
    },
    {
      "epoch": 0.0031661655128309113,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00039473684210526315,
      "loss": 2.6586,
      "step": 466
    },
    {
      "epoch": 0.003172959859424969,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00039558573853989813,
      "loss": 2.7943,
      "step": 467
    },
    {
      "epoch": 0.003179754206019027,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0003964346349745331,
      "loss": 2.7278,
      "step": 468
    },
    {
      "epoch": 0.0031865485526130846,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00039728353140916804,
      "loss": 2.7563,
      "step": 469
    },
    {
      "epoch": 0.0031933428992071424,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00039813242784380307,
      "loss": 2.6818,
      "step": 470
    },
    {
      "epoch": 0.0032001372458012,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.00039898132427843805,
      "loss": 2.7672,
      "step": 471
    },
    {
      "epoch": 0.0032069315923952575,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.00039983022071307303,
      "loss": 2.655,
      "step": 472
    },
    {
      "epoch": 0.0032137259389893153,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.000400679117147708,
      "loss": 2.8879,
      "step": 473
    },
    {
      "epoch": 0.003220520285583373,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.000401528013582343,
      "loss": 2.7247,
      "step": 474
    },
    {
      "epoch": 0.003227314632177431,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.000402376910016978,
      "loss": 2.6308,
      "step": 475
    },
    {
      "epoch": 0.0032341089787714886,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0004032258064516129,
      "loss": 2.8005,
      "step": 476
    },
    {
      "epoch": 0.0032409033253655464,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0004040747028862479,
      "loss": 2.8125,
      "step": 477
    },
    {
      "epoch": 0.003247697671959604,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00040492359932088286,
      "loss": 2.8673,
      "step": 478
    },
    {
      "epoch": 0.003254492018553662,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00040577249575551785,
      "loss": 2.645,
      "step": 479
    },
    {
      "epoch": 0.0032612863651477197,
      "grad_norm": 0.294921875,
      "learning_rate": 0.0004066213921901528,
      "loss": 2.5551,
      "step": 480
    },
    {
      "epoch": 0.0032680807117417775,
      "grad_norm": 0.234375,
      "learning_rate": 0.0004074702886247878,
      "loss": 2.7148,
      "step": 481
    },
    {
      "epoch": 0.0032748750583358353,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0004083191850594228,
      "loss": 2.5837,
      "step": 482
    },
    {
      "epoch": 0.003281669404929893,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0004091680814940577,
      "loss": 2.801,
      "step": 483
    },
    {
      "epoch": 0.003288463751523951,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0004100169779286927,
      "loss": 2.7153,
      "step": 484
    },
    {
      "epoch": 0.0032952580981180086,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0004108658743633277,
      "loss": 2.6054,
      "step": 485
    },
    {
      "epoch": 0.0033020524447120664,
      "grad_norm": 0.2421875,
      "learning_rate": 0.00041171477079796266,
      "loss": 2.8503,
      "step": 486
    },
    {
      "epoch": 0.0033088467913061238,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00041256366723259764,
      "loss": 2.6904,
      "step": 487
    },
    {
      "epoch": 0.0033156411379001815,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0004134125636672326,
      "loss": 2.8046,
      "step": 488
    },
    {
      "epoch": 0.0033224354844942393,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0004142614601018676,
      "loss": 2.8032,
      "step": 489
    },
    {
      "epoch": 0.003329229831088297,
      "grad_norm": 0.265625,
      "learning_rate": 0.0004151103565365025,
      "loss": 2.8368,
      "step": 490
    },
    {
      "epoch": 0.003336024177682355,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0004159592529711375,
      "loss": 2.7086,
      "step": 491
    },
    {
      "epoch": 0.0033428185242764127,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0004168081494057725,
      "loss": 2.5956,
      "step": 492
    },
    {
      "epoch": 0.0033496128708704704,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00041765704584040747,
      "loss": 2.6951,
      "step": 493
    },
    {
      "epoch": 0.003356407217464528,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.00041850594227504245,
      "loss": 2.7198,
      "step": 494
    },
    {
      "epoch": 0.003363201564058586,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00041935483870967743,
      "loss": 2.7343,
      "step": 495
    },
    {
      "epoch": 0.0033699959106526438,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0004202037351443124,
      "loss": 2.6633,
      "step": 496
    },
    {
      "epoch": 0.0033767902572467015,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00042105263157894734,
      "loss": 2.7655,
      "step": 497
    },
    {
      "epoch": 0.0033835846038407593,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0004219015280135823,
      "loss": 2.8281,
      "step": 498
    },
    {
      "epoch": 0.003390378950434817,
      "grad_norm": 0.234375,
      "learning_rate": 0.0004227504244482173,
      "loss": 2.8182,
      "step": 499
    },
    {
      "epoch": 0.003397173297028875,
      "grad_norm": 0.234375,
      "learning_rate": 0.0004235993208828523,
      "loss": 2.7966,
      "step": 500
    },
    {
      "epoch": 0.0034039676436229327,
      "grad_norm": 0.23828125,
      "learning_rate": 0.00042444821731748726,
      "loss": 2.8709,
      "step": 501
    },
    {
      "epoch": 0.00341076199021699,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00042529711375212224,
      "loss": 2.6545,
      "step": 502
    },
    {
      "epoch": 0.0034175563368110478,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0004261460101867572,
      "loss": 2.7414,
      "step": 503
    },
    {
      "epoch": 0.0034243506834051056,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0004269949066213922,
      "loss": 2.6076,
      "step": 504
    },
    {
      "epoch": 0.0034311450299991633,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0004278438030560272,
      "loss": 2.838,
      "step": 505
    },
    {
      "epoch": 0.003437939376593221,
      "grad_norm": 0.240234375,
      "learning_rate": 0.00042869269949066216,
      "loss": 2.8161,
      "step": 506
    },
    {
      "epoch": 0.003444733723187279,
      "grad_norm": 0.248046875,
      "learning_rate": 0.00042954159592529715,
      "loss": 2.6522,
      "step": 507
    },
    {
      "epoch": 0.0034515280697813367,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0004303904923599321,
      "loss": 2.7026,
      "step": 508
    },
    {
      "epoch": 0.0034583224163753945,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0004312393887945671,
      "loss": 2.8099,
      "step": 509
    },
    {
      "epoch": 0.0034651167629694522,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0004320882852292021,
      "loss": 2.8633,
      "step": 510
    },
    {
      "epoch": 0.00347191110956351,
      "grad_norm": 0.216796875,
      "learning_rate": 0.000432937181663837,
      "loss": 2.7772,
      "step": 511
    },
    {
      "epoch": 0.003478705456157568,
      "grad_norm": 0.248046875,
      "learning_rate": 0.000433786078098472,
      "loss": 2.6461,
      "step": 512
    },
    {
      "epoch": 0.0034854998027516256,
      "grad_norm": 0.220703125,
      "learning_rate": 0.000434634974533107,
      "loss": 2.7583,
      "step": 513
    },
    {
      "epoch": 0.0034922941493456833,
      "grad_norm": 0.259765625,
      "learning_rate": 0.00043548387096774196,
      "loss": 2.673,
      "step": 514
    },
    {
      "epoch": 0.003499088495939741,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00043633276740237694,
      "loss": 2.8223,
      "step": 515
    },
    {
      "epoch": 0.003505882842533799,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0004371816638370119,
      "loss": 2.7105,
      "step": 516
    },
    {
      "epoch": 0.0035126771891278562,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0004380305602716469,
      "loss": 2.6331,
      "step": 517
    },
    {
      "epoch": 0.003519471535721914,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0004388794567062818,
      "loss": 2.8422,
      "step": 518
    },
    {
      "epoch": 0.003526265882315972,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0004397283531409168,
      "loss": 2.6904,
      "step": 519
    },
    {
      "epoch": 0.0035330602289100296,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0004405772495755518,
      "loss": 2.6825,
      "step": 520
    },
    {
      "epoch": 0.0035398545755040874,
      "grad_norm": 0.24609375,
      "learning_rate": 0.00044142614601018677,
      "loss": 2.9447,
      "step": 521
    },
    {
      "epoch": 0.003546648922098145,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.00044227504244482175,
      "loss": 2.7153,
      "step": 522
    },
    {
      "epoch": 0.003553443268692203,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.00044312393887945673,
      "loss": 2.7943,
      "step": 523
    },
    {
      "epoch": 0.0035602376152862607,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0004439728353140917,
      "loss": 2.8223,
      "step": 524
    },
    {
      "epoch": 0.0035670319618803185,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00044482173174872664,
      "loss": 2.6467,
      "step": 525
    },
    {
      "epoch": 0.0035738263084743763,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0004456706281833616,
      "loss": 2.7443,
      "step": 526
    },
    {
      "epoch": 0.003580620655068434,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0004465195246179966,
      "loss": 2.5825,
      "step": 527
    },
    {
      "epoch": 0.003587415001662492,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0004473684210526316,
      "loss": 2.7163,
      "step": 528
    },
    {
      "epoch": 0.0035942093482565496,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00044821731748726656,
      "loss": 2.7924,
      "step": 529
    },
    {
      "epoch": 0.0036010036948506074,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00044906621392190154,
      "loss": 2.6458,
      "step": 530
    },
    {
      "epoch": 0.003607798041444665,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0004499151103565365,
      "loss": 2.8297,
      "step": 531
    },
    {
      "epoch": 0.0036145923880387225,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00045076400679117145,
      "loss": 2.7683,
      "step": 532
    },
    {
      "epoch": 0.0036213867346327803,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00045161290322580643,
      "loss": 2.6723,
      "step": 533
    },
    {
      "epoch": 0.003628181081226838,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0004524617996604414,
      "loss": 2.8365,
      "step": 534
    },
    {
      "epoch": 0.003634975427820896,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0004533106960950764,
      "loss": 2.6503,
      "step": 535
    },
    {
      "epoch": 0.0036417697744149536,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00045415959252971137,
      "loss": 2.7509,
      "step": 536
    },
    {
      "epoch": 0.0036485641210090114,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00045500848896434635,
      "loss": 2.7616,
      "step": 537
    },
    {
      "epoch": 0.003655358467603069,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0004558573853989814,
      "loss": 2.7464,
      "step": 538
    },
    {
      "epoch": 0.003662152814197127,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0004567062818336163,
      "loss": 2.6732,
      "step": 539
    },
    {
      "epoch": 0.0036689471607911847,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0004575551782682513,
      "loss": 2.7842,
      "step": 540
    },
    {
      "epoch": 0.0036757415073852425,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0004584040747028863,
      "loss": 2.7093,
      "step": 541
    },
    {
      "epoch": 0.0036825358539793003,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00045925297113752126,
      "loss": 2.6809,
      "step": 542
    },
    {
      "epoch": 0.003689330200573358,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.00046010186757215624,
      "loss": 2.8686,
      "step": 543
    },
    {
      "epoch": 0.003696124547167416,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0004609507640067912,
      "loss": 2.689,
      "step": 544
    },
    {
      "epoch": 0.0037029188937614736,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.00046179966044142615,
      "loss": 2.7051,
      "step": 545
    },
    {
      "epoch": 0.0037097132403555314,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0004626485568760611,
      "loss": 2.7897,
      "step": 546
    },
    {
      "epoch": 0.0037165075869495887,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0004634974533106961,
      "loss": 2.6042,
      "step": 547
    },
    {
      "epoch": 0.0037233019335436465,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0004643463497453311,
      "loss": 2.7947,
      "step": 548
    },
    {
      "epoch": 0.0037300962801377043,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00046519524617996607,
      "loss": 2.9252,
      "step": 549
    },
    {
      "epoch": 0.003736890626731762,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00046604414261460105,
      "loss": 2.6229,
      "step": 550
    },
    {
      "epoch": 0.00374368497332582,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00046689303904923603,
      "loss": 2.6598,
      "step": 551
    },
    {
      "epoch": 0.0037504793199198776,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00046774193548387096,
      "loss": 2.7296,
      "step": 552
    },
    {
      "epoch": 0.0037572736665139354,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00046859083191850594,
      "loss": 2.6887,
      "step": 553
    },
    {
      "epoch": 0.003764068013107993,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0004694397283531409,
      "loss": 2.6979,
      "step": 554
    },
    {
      "epoch": 0.003770862359702051,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0004702886247877759,
      "loss": 2.7702,
      "step": 555
    },
    {
      "epoch": 0.0037776567062961087,
      "grad_norm": 0.27734375,
      "learning_rate": 0.0004711375212224109,
      "loss": 2.5591,
      "step": 556
    },
    {
      "epoch": 0.0037844510528901665,
      "grad_norm": 0.23828125,
      "learning_rate": 0.00047198641765704586,
      "loss": 2.7675,
      "step": 557
    },
    {
      "epoch": 0.0037912453994842243,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.00047283531409168084,
      "loss": 2.6893,
      "step": 558
    },
    {
      "epoch": 0.003798039746078282,
      "grad_norm": 0.228515625,
      "learning_rate": 0.00047368421052631577,
      "loss": 2.7511,
      "step": 559
    },
    {
      "epoch": 0.00380483409267234,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.00047453310696095075,
      "loss": 2.6323,
      "step": 560
    },
    {
      "epoch": 0.0038116284392663976,
      "grad_norm": 0.25390625,
      "learning_rate": 0.00047538200339558573,
      "loss": 2.742,
      "step": 561
    },
    {
      "epoch": 0.003818422785860455,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0004762308998302207,
      "loss": 2.8532,
      "step": 562
    },
    {
      "epoch": 0.0038252171324545127,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0004770797962648557,
      "loss": 2.7018,
      "step": 563
    },
    {
      "epoch": 0.0038320114790485705,
      "grad_norm": 0.216796875,
      "learning_rate": 0.00047792869269949067,
      "loss": 2.708,
      "step": 564
    },
    {
      "epoch": 0.0038388058256426283,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.00047877758913412565,
      "loss": 2.5849,
      "step": 565
    },
    {
      "epoch": 0.003845600172236686,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0004796264855687606,
      "loss": 2.6362,
      "step": 566
    },
    {
      "epoch": 0.003852394518830744,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.00048047538200339556,
      "loss": 2.7675,
      "step": 567
    },
    {
      "epoch": 0.0038591888654248016,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.00048132427843803054,
      "loss": 2.703,
      "step": 568
    },
    {
      "epoch": 0.0038659832120188594,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0004821731748726655,
      "loss": 2.7609,
      "step": 569
    },
    {
      "epoch": 0.003872777558612917,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0004830220713073005,
      "loss": 2.7263,
      "step": 570
    },
    {
      "epoch": 0.003879571905206975,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0004838709677419355,
      "loss": 2.7888,
      "step": 571
    },
    {
      "epoch": 0.0038863662518010328,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00048471986417657046,
      "loss": 2.6969,
      "step": 572
    },
    {
      "epoch": 0.0038931605983950905,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0004855687606112054,
      "loss": 2.83,
      "step": 573
    },
    {
      "epoch": 0.0038999549449891483,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0004864176570458404,
      "loss": 2.634,
      "step": 574
    },
    {
      "epoch": 0.003906749291583206,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0004872665534804754,
      "loss": 2.8033,
      "step": 575
    },
    {
      "epoch": 0.003913543638177264,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0004881154499151104,
      "loss": 2.6705,
      "step": 576
    },
    {
      "epoch": 0.003920337984771321,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0004889643463497454,
      "loss": 2.6236,
      "step": 577
    },
    {
      "epoch": 0.003927132331365379,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0004898132427843803,
      "loss": 2.6851,
      "step": 578
    },
    {
      "epoch": 0.003933926677959437,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0004906621392190153,
      "loss": 2.5893,
      "step": 579
    },
    {
      "epoch": 0.003940721024553495,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0004915110356536502,
      "loss": 2.5657,
      "step": 580
    },
    {
      "epoch": 0.003947515371147552,
      "grad_norm": 0.234375,
      "learning_rate": 0.0004923599320882852,
      "loss": 2.7216,
      "step": 581
    },
    {
      "epoch": 0.0039543097177416105,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0004932088285229202,
      "loss": 2.8507,
      "step": 582
    },
    {
      "epoch": 0.003961104064335668,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0004940577249575551,
      "loss": 2.6517,
      "step": 583
    },
    {
      "epoch": 0.003967898410929725,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0004949066213921901,
      "loss": 2.6146,
      "step": 584
    },
    {
      "epoch": 0.0039746927575237834,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0004957555178268251,
      "loss": 2.5833,
      "step": 585
    },
    {
      "epoch": 0.003981487104117841,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0004966044142614601,
      "loss": 2.6827,
      "step": 586
    },
    {
      "epoch": 0.003988281450711899,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0004974533106960951,
      "loss": 2.7295,
      "step": 587
    },
    {
      "epoch": 0.003995075797305956,
      "grad_norm": 0.224609375,
      "learning_rate": 0.00049830220713073,
      "loss": 2.648,
      "step": 588
    },
    {
      "epoch": 0.0040018701439000146,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.000499151103565365,
      "loss": 2.5983,
      "step": 589
    },
    {
      "epoch": 0.004008664490494072,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0005,
      "loss": 2.7231,
      "step": 590
    },
    {
      "epoch": 0.00401545883708813,
      "grad_norm": 0.255859375,
      "learning_rate": 0.000500848896434635,
      "loss": 2.812,
      "step": 591
    },
    {
      "epoch": 0.0040222531836821875,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00050169779286927,
      "loss": 2.7403,
      "step": 592
    },
    {
      "epoch": 0.004029047530276246,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.000502546689303905,
      "loss": 2.7413,
      "step": 593
    },
    {
      "epoch": 0.004035841876870303,
      "grad_norm": 0.263671875,
      "learning_rate": 0.0005033955857385399,
      "loss": 2.7604,
      "step": 594
    },
    {
      "epoch": 0.004042636223464361,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0005042444821731749,
      "loss": 2.7835,
      "step": 595
    },
    {
      "epoch": 0.004049430570058419,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0005050933786078099,
      "loss": 2.8442,
      "step": 596
    },
    {
      "epoch": 0.004056224916652477,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005059422750424449,
      "loss": 2.7553,
      "step": 597
    },
    {
      "epoch": 0.004063019263246534,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005067911714770797,
      "loss": 2.7603,
      "step": 598
    },
    {
      "epoch": 0.0040698136098405915,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005076400679117148,
      "loss": 2.711,
      "step": 599
    },
    {
      "epoch": 0.00407660795643465,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0005084889643463497,
      "loss": 2.7101,
      "step": 600
    },
    {
      "epoch": 0.004083402303028707,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005093378607809848,
      "loss": 2.6875,
      "step": 601
    },
    {
      "epoch": 0.004090196649622765,
      "grad_norm": 0.259765625,
      "learning_rate": 0.0005101867572156197,
      "loss": 2.6999,
      "step": 602
    },
    {
      "epoch": 0.004096990996216823,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0005110356536502548,
      "loss": 2.8277,
      "step": 603
    },
    {
      "epoch": 0.004103785342810881,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005118845500848896,
      "loss": 2.6893,
      "step": 604
    },
    {
      "epoch": 0.004110579689404938,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005127334465195246,
      "loss": 2.7045,
      "step": 605
    },
    {
      "epoch": 0.004117374035998996,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005135823429541596,
      "loss": 2.7179,
      "step": 606
    },
    {
      "epoch": 0.004124168382593054,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005144312393887946,
      "loss": 2.807,
      "step": 607
    },
    {
      "epoch": 0.004130962729187112,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005152801358234296,
      "loss": 2.7774,
      "step": 608
    },
    {
      "epoch": 0.004137757075781169,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005161290322580645,
      "loss": 2.6906,
      "step": 609
    },
    {
      "epoch": 0.0041445514223752275,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005169779286926995,
      "loss": 2.7598,
      "step": 610
    },
    {
      "epoch": 0.004151345768969285,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005178268251273345,
      "loss": 2.6129,
      "step": 611
    },
    {
      "epoch": 0.004158140115563343,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005186757215619694,
      "loss": 2.7048,
      "step": 612
    },
    {
      "epoch": 0.0041649344621574,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005195246179966045,
      "loss": 2.765,
      "step": 613
    },
    {
      "epoch": 0.004171728808751458,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0005203735144312393,
      "loss": 2.7921,
      "step": 614
    },
    {
      "epoch": 0.004178523155345516,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005212224108658744,
      "loss": 2.7739,
      "step": 615
    },
    {
      "epoch": 0.004185317501939573,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0005220713073005093,
      "loss": 2.7838,
      "step": 616
    },
    {
      "epoch": 0.0041921118485336315,
      "grad_norm": 0.21875,
      "learning_rate": 0.0005229202037351444,
      "loss": 2.5916,
      "step": 617
    },
    {
      "epoch": 0.004198906195127689,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0005237691001697793,
      "loss": 2.762,
      "step": 618
    },
    {
      "epoch": 0.004205700541721747,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0005246179966044142,
      "loss": 2.702,
      "step": 619
    },
    {
      "epoch": 0.004212494888315804,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005254668930390492,
      "loss": 2.777,
      "step": 620
    },
    {
      "epoch": 0.004219289234909863,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005263157894736842,
      "loss": 2.728,
      "step": 621
    },
    {
      "epoch": 0.00422608358150392,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005271646859083192,
      "loss": 2.695,
      "step": 622
    },
    {
      "epoch": 0.004232877928097978,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005280135823429542,
      "loss": 2.7207,
      "step": 623
    },
    {
      "epoch": 0.0042396722746920355,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0005288624787775892,
      "loss": 2.8051,
      "step": 624
    },
    {
      "epoch": 0.004246466621286094,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0005297113752122241,
      "loss": 2.6741,
      "step": 625
    },
    {
      "epoch": 0.004253260967880151,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0005305602716468591,
      "loss": 2.5562,
      "step": 626
    },
    {
      "epoch": 0.004260055314474209,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005314091680814941,
      "loss": 2.5001,
      "step": 627
    },
    {
      "epoch": 0.004266849661068267,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0005322580645161291,
      "loss": 2.5031,
      "step": 628
    },
    {
      "epoch": 0.004273644007662324,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.000533106960950764,
      "loss": 2.6428,
      "step": 629
    },
    {
      "epoch": 0.004280438354256382,
      "grad_norm": 0.25390625,
      "learning_rate": 0.000533955857385399,
      "loss": 2.5933,
      "step": 630
    },
    {
      "epoch": 0.0042872327008504395,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.000534804753820034,
      "loss": 2.7548,
      "step": 631
    },
    {
      "epoch": 0.004294027047444498,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.000535653650254669,
      "loss": 2.6056,
      "step": 632
    },
    {
      "epoch": 0.004300821394038555,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0005365025466893039,
      "loss": 2.7628,
      "step": 633
    },
    {
      "epoch": 0.004307615740632613,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.000537351443123939,
      "loss": 2.7694,
      "step": 634
    },
    {
      "epoch": 0.004314410087226671,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005382003395585738,
      "loss": 2.8436,
      "step": 635
    },
    {
      "epoch": 0.004321204433820729,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005390492359932089,
      "loss": 2.6722,
      "step": 636
    },
    {
      "epoch": 0.004327998780414786,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005398981324278438,
      "loss": 2.6684,
      "step": 637
    },
    {
      "epoch": 0.004334793127008844,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0005407470288624789,
      "loss": 2.7524,
      "step": 638
    },
    {
      "epoch": 0.004341587473602902,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0005415959252971137,
      "loss": 2.6915,
      "step": 639
    },
    {
      "epoch": 0.00434838182019696,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0005424448217317487,
      "loss": 2.7064,
      "step": 640
    },
    {
      "epoch": 0.004355176166791017,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005432937181663837,
      "loss": 2.5447,
      "step": 641
    },
    {
      "epoch": 0.0043619705133850755,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005441426146010187,
      "loss": 2.6814,
      "step": 642
    },
    {
      "epoch": 0.004368764859979133,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005449915110356537,
      "loss": 2.6467,
      "step": 643
    },
    {
      "epoch": 0.00437555920657319,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0005458404074702886,
      "loss": 2.7787,
      "step": 644
    },
    {
      "epoch": 0.004382353553167248,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005466893039049236,
      "loss": 2.6295,
      "step": 645
    },
    {
      "epoch": 0.004389147899761306,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005475382003395586,
      "loss": 2.8704,
      "step": 646
    },
    {
      "epoch": 0.004395942246355364,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005483870967741935,
      "loss": 2.6983,
      "step": 647
    },
    {
      "epoch": 0.004402736592949421,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005492359932088286,
      "loss": 2.6485,
      "step": 648
    },
    {
      "epoch": 0.0044095309395434795,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005500848896434634,
      "loss": 2.6586,
      "step": 649
    },
    {
      "epoch": 0.004416325286137537,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005509337860780985,
      "loss": 2.7097,
      "step": 650
    },
    {
      "epoch": 0.004423119632731595,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005517826825127334,
      "loss": 2.7079,
      "step": 651
    },
    {
      "epoch": 0.004429913979325652,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005526315789473685,
      "loss": 2.7504,
      "step": 652
    },
    {
      "epoch": 0.004436708325919711,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005534804753820034,
      "loss": 2.7222,
      "step": 653
    },
    {
      "epoch": 0.004443502672513768,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0005543293718166383,
      "loss": 2.7533,
      "step": 654
    },
    {
      "epoch": 0.004450297019107826,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0005551782682512733,
      "loss": 2.5781,
      "step": 655
    },
    {
      "epoch": 0.0044570913657018835,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005560271646859083,
      "loss": 2.6304,
      "step": 656
    },
    {
      "epoch": 0.004463885712295942,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005568760611205433,
      "loss": 2.7458,
      "step": 657
    },
    {
      "epoch": 0.004470680058889999,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005577249575551783,
      "loss": 2.5912,
      "step": 658
    },
    {
      "epoch": 0.0044774744054840564,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005585738539898134,
      "loss": 2.7247,
      "step": 659
    },
    {
      "epoch": 0.004484268752078115,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0005594227504244482,
      "loss": 2.6686,
      "step": 660
    },
    {
      "epoch": 0.004491063098672172,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0005602716468590832,
      "loss": 2.7565,
      "step": 661
    },
    {
      "epoch": 0.00449785744526623,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005611205432937182,
      "loss": 2.7212,
      "step": 662
    },
    {
      "epoch": 0.0045046517918602876,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005619694397283532,
      "loss": 2.8652,
      "step": 663
    },
    {
      "epoch": 0.004511446138454346,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005628183361629882,
      "loss": 2.6927,
      "step": 664
    },
    {
      "epoch": 0.004518240485048403,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0005636672325976231,
      "loss": 2.6832,
      "step": 665
    },
    {
      "epoch": 0.004525034831642461,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0005645161290322581,
      "loss": 2.5526,
      "step": 666
    },
    {
      "epoch": 0.004531829178236519,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0005653650254668931,
      "loss": 2.7227,
      "step": 667
    },
    {
      "epoch": 0.004538623524830577,
      "grad_norm": 0.236328125,
      "learning_rate": 0.000566213921901528,
      "loss": 2.826,
      "step": 668
    },
    {
      "epoch": 0.004545417871424634,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0005670628183361631,
      "loss": 2.8351,
      "step": 669
    },
    {
      "epoch": 0.0045522122180186924,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005679117147707979,
      "loss": 2.6059,
      "step": 670
    },
    {
      "epoch": 0.00455900656461275,
      "grad_norm": 0.236328125,
      "learning_rate": 0.000568760611205433,
      "loss": 2.7179,
      "step": 671
    },
    {
      "epoch": 0.004565800911206808,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005696095076400679,
      "loss": 2.5671,
      "step": 672
    },
    {
      "epoch": 0.004572595257800865,
      "grad_norm": 0.244140625,
      "learning_rate": 0.000570458404074703,
      "loss": 2.6308,
      "step": 673
    },
    {
      "epoch": 0.004579389604394923,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005713073005093379,
      "loss": 2.7009,
      "step": 674
    },
    {
      "epoch": 0.004586183950988981,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0005721561969439728,
      "loss": 2.7293,
      "step": 675
    },
    {
      "epoch": 0.004592978297583038,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005730050933786078,
      "loss": 2.8392,
      "step": 676
    },
    {
      "epoch": 0.0045997726441770964,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005738539898132428,
      "loss": 2.766,
      "step": 677
    },
    {
      "epoch": 0.004606566990771154,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005747028862478778,
      "loss": 2.5818,
      "step": 678
    },
    {
      "epoch": 0.004613361337365212,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0005755517826825128,
      "loss": 2.695,
      "step": 679
    },
    {
      "epoch": 0.004620155683959269,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005764006791171477,
      "loss": 2.7303,
      "step": 680
    },
    {
      "epoch": 0.0046269500305533276,
      "grad_norm": 0.234375,
      "learning_rate": 0.0005772495755517827,
      "loss": 2.7454,
      "step": 681
    },
    {
      "epoch": 0.004633744377147385,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0005780984719864176,
      "loss": 2.6909,
      "step": 682
    },
    {
      "epoch": 0.004640538723741443,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0005789473684210527,
      "loss": 2.596,
      "step": 683
    },
    {
      "epoch": 0.0046473330703355005,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005797962648556876,
      "loss": 2.6357,
      "step": 684
    },
    {
      "epoch": 0.004654127416929559,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0005806451612903226,
      "loss": 2.681,
      "step": 685
    },
    {
      "epoch": 0.004660921763523616,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.0005814940577249575,
      "loss": 2.6606,
      "step": 686
    },
    {
      "epoch": 0.004667716110117674,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0005823429541595926,
      "loss": 2.8379,
      "step": 687
    },
    {
      "epoch": 0.004674510456711732,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0005831918505942275,
      "loss": 2.7499,
      "step": 688
    },
    {
      "epoch": 0.004681304803305789,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0005840407470288625,
      "loss": 2.6424,
      "step": 689
    },
    {
      "epoch": 0.004688099149899847,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0005848896434634974,
      "loss": 2.7506,
      "step": 690
    },
    {
      "epoch": 0.0046948934964939045,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0005857385398981324,
      "loss": 2.7554,
      "step": 691
    },
    {
      "epoch": 0.004701687843087963,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.0005865874363327674,
      "loss": 2.7674,
      "step": 692
    },
    {
      "epoch": 0.00470848218968202,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0005874363327674024,
      "loss": 2.5977,
      "step": 693
    },
    {
      "epoch": 0.004715276536276078,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0005882852292020373,
      "loss": 2.5311,
      "step": 694
    },
    {
      "epoch": 0.004722070882870136,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0005891341256366723,
      "loss": 2.7068,
      "step": 695
    },
    {
      "epoch": 0.004728865229464194,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0005899830220713073,
      "loss": 2.7174,
      "step": 696
    },
    {
      "epoch": 0.004735659576058251,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005908319185059423,
      "loss": 2.7153,
      "step": 697
    },
    {
      "epoch": 0.004742453922652309,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005916808149405773,
      "loss": 2.5655,
      "step": 698
    },
    {
      "epoch": 0.004749248269246367,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0005925297113752123,
      "loss": 2.7996,
      "step": 699
    },
    {
      "epoch": 0.004756042615840425,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0005933786078098472,
      "loss": 2.874,
      "step": 700
    },
    {
      "epoch": 0.004762836962434482,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0005942275042444821,
      "loss": 2.6724,
      "step": 701
    },
    {
      "epoch": 0.0047696313090285405,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0005950764006791172,
      "loss": 2.6396,
      "step": 702
    },
    {
      "epoch": 0.004776425655622598,
      "grad_norm": 0.234375,
      "learning_rate": 0.0005959252971137521,
      "loss": 2.9655,
      "step": 703
    },
    {
      "epoch": 0.004783220002216655,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0005967741935483872,
      "loss": 2.8812,
      "step": 704
    },
    {
      "epoch": 0.004790014348810713,
      "grad_norm": 0.2265625,
      "learning_rate": 0.000597623089983022,
      "loss": 2.7859,
      "step": 705
    },
    {
      "epoch": 0.004796808695404771,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0005984719864176571,
      "loss": 2.5998,
      "step": 706
    },
    {
      "epoch": 0.004803603041998829,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.000599320882852292,
      "loss": 2.6433,
      "step": 707
    },
    {
      "epoch": 0.004810397388592886,
      "grad_norm": 0.205078125,
      "learning_rate": 0.000600169779286927,
      "loss": 2.6792,
      "step": 708
    },
    {
      "epoch": 0.0048171917351869445,
      "grad_norm": 0.228515625,
      "learning_rate": 0.000601018675721562,
      "loss": 2.6883,
      "step": 709
    },
    {
      "epoch": 0.004823986081781002,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.000601867572156197,
      "loss": 2.6964,
      "step": 710
    },
    {
      "epoch": 0.00483078042837506,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0006027164685908319,
      "loss": 2.7208,
      "step": 711
    },
    {
      "epoch": 0.004837574774969117,
      "grad_norm": 0.21875,
      "learning_rate": 0.0006035653650254669,
      "loss": 2.7044,
      "step": 712
    },
    {
      "epoch": 0.004844369121563176,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0006044142614601019,
      "loss": 2.6756,
      "step": 713
    },
    {
      "epoch": 0.004851163468157233,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0006052631578947369,
      "loss": 2.7007,
      "step": 714
    },
    {
      "epoch": 0.004857957814751291,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0006061120543293717,
      "loss": 2.889,
      "step": 715
    },
    {
      "epoch": 0.0048647521613453485,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0006069609507640068,
      "loss": 2.8127,
      "step": 716
    },
    {
      "epoch": 0.004871546507939407,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0006078098471986417,
      "loss": 2.7291,
      "step": 717
    },
    {
      "epoch": 0.004878340854533464,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0006086587436332768,
      "loss": 2.6841,
      "step": 718
    },
    {
      "epoch": 0.004885135201127521,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0006095076400679117,
      "loss": 2.8075,
      "step": 719
    },
    {
      "epoch": 0.00489192954772158,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0006103565365025468,
      "loss": 2.6959,
      "step": 720
    },
    {
      "epoch": 0.004898723894315637,
      "grad_norm": 0.203125,
      "learning_rate": 0.0006112054329371816,
      "loss": 2.7464,
      "step": 721
    },
    {
      "epoch": 0.004905518240909695,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0006120543293718166,
      "loss": 2.7132,
      "step": 722
    },
    {
      "epoch": 0.0049123125875037525,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0006129032258064516,
      "loss": 2.7256,
      "step": 723
    },
    {
      "epoch": 0.004919106934097811,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0006137521222410866,
      "loss": 2.7689,
      "step": 724
    },
    {
      "epoch": 0.004925901280691868,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0006146010186757216,
      "loss": 2.6467,
      "step": 725
    },
    {
      "epoch": 0.004932695627285926,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0006154499151103565,
      "loss": 2.6178,
      "step": 726
    },
    {
      "epoch": 0.004939489973879984,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0006162988115449915,
      "loss": 2.5453,
      "step": 727
    },
    {
      "epoch": 0.004946284320474042,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0006171477079796265,
      "loss": 2.5951,
      "step": 728
    },
    {
      "epoch": 0.004953078667068099,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0006179966044142614,
      "loss": 2.7329,
      "step": 729
    },
    {
      "epoch": 0.004959873013662157,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0006188455008488965,
      "loss": 2.6395,
      "step": 730
    },
    {
      "epoch": 0.004966667360256215,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0006196943972835314,
      "loss": 2.5432,
      "step": 731
    },
    {
      "epoch": 0.004973461706850273,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0006205432937181664,
      "loss": 2.7148,
      "step": 732
    },
    {
      "epoch": 0.00498025605344433,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0006213921901528014,
      "loss": 2.6614,
      "step": 733
    },
    {
      "epoch": 0.004987050400038388,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0006222410865874364,
      "loss": 2.683,
      "step": 734
    },
    {
      "epoch": 0.004993844746632446,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0006230899830220714,
      "loss": 2.5418,
      "step": 735
    },
    {
      "epoch": 0.005000639093226503,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0006239388794567062,
      "loss": 2.6564,
      "step": 736
    },
    {
      "epoch": 0.005007433439820561,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0006247877758913413,
      "loss": 2.6016,
      "step": 737
    },
    {
      "epoch": 0.005014227786414619,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0006256366723259762,
      "loss": 2.638,
      "step": 738
    },
    {
      "epoch": 0.005021022133008677,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0006264855687606113,
      "loss": 2.5493,
      "step": 739
    },
    {
      "epoch": 0.005027816479602734,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0006273344651952462,
      "loss": 2.5992,
      "step": 740
    },
    {
      "epoch": 0.0050346108261967925,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0006281833616298812,
      "loss": 2.638,
      "step": 741
    },
    {
      "epoch": 0.00504140517279085,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0006290322580645161,
      "loss": 2.7383,
      "step": 742
    },
    {
      "epoch": 0.005048199519384908,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0006298811544991511,
      "loss": 2.5357,
      "step": 743
    },
    {
      "epoch": 0.005054993865978965,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0006307300509337861,
      "loss": 2.5131,
      "step": 744
    },
    {
      "epoch": 0.005061788212573024,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0006315789473684211,
      "loss": 2.6649,
      "step": 745
    },
    {
      "epoch": 0.005068582559167081,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.000632427843803056,
      "loss": 2.6176,
      "step": 746
    },
    {
      "epoch": 0.005075376905761139,
      "grad_norm": 0.234375,
      "learning_rate": 0.000633276740237691,
      "loss": 2.5852,
      "step": 747
    },
    {
      "epoch": 0.0050821712523551965,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.000634125636672326,
      "loss": 2.6613,
      "step": 748
    },
    {
      "epoch": 0.005088965598949254,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.000634974533106961,
      "loss": 2.8307,
      "step": 749
    },
    {
      "epoch": 0.005095759945543312,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0006358234295415959,
      "loss": 2.6423,
      "step": 750
    },
    {
      "epoch": 0.0051025542921373694,
      "grad_norm": 0.330078125,
      "learning_rate": 0.0006366723259762309,
      "loss": 2.7369,
      "step": 751
    },
    {
      "epoch": 0.005109348638731428,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0006375212224108658,
      "loss": 2.6697,
      "step": 752
    },
    {
      "epoch": 0.005116142985325485,
      "grad_norm": 0.2412109375,
      "learning_rate": 0.0006383701188455009,
      "loss": 2.6637,
      "step": 753
    },
    {
      "epoch": 0.005122937331919543,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0006392190152801358,
      "loss": 2.6853,
      "step": 754
    },
    {
      "epoch": 0.0051297316785136006,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0006400679117147709,
      "loss": 2.6808,
      "step": 755
    },
    {
      "epoch": 0.005136526025107659,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0006409168081494057,
      "loss": 2.6624,
      "step": 756
    },
    {
      "epoch": 0.005143320371701716,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0006417657045840407,
      "loss": 2.7953,
      "step": 757
    },
    {
      "epoch": 0.005150114718295774,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0006426146010186757,
      "loss": 2.5435,
      "step": 758
    },
    {
      "epoch": 0.005156909064889832,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0006434634974533107,
      "loss": 2.6264,
      "step": 759
    },
    {
      "epoch": 0.00516370341148389,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0006443123938879457,
      "loss": 2.7749,
      "step": 760
    },
    {
      "epoch": 0.005170497758077947,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0006451612903225806,
      "loss": 2.968,
      "step": 761
    },
    {
      "epoch": 0.0051772921046720054,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0006460101867572156,
      "loss": 2.8078,
      "step": 762
    },
    {
      "epoch": 0.005184086451266063,
      "grad_norm": 0.3125,
      "learning_rate": 0.0006468590831918506,
      "loss": 2.6942,
      "step": 763
    },
    {
      "epoch": 0.00519088079786012,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0006477079796264856,
      "loss": 2.776,
      "step": 764
    },
    {
      "epoch": 0.005197675144454178,
      "grad_norm": 0.296875,
      "learning_rate": 0.0006485568760611206,
      "loss": 2.7269,
      "step": 765
    },
    {
      "epoch": 0.005204469491048236,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0006494057724957555,
      "loss": 2.7935,
      "step": 766
    },
    {
      "epoch": 0.005211263837642294,
      "grad_norm": 0.25,
      "learning_rate": 0.0006502546689303905,
      "loss": 2.7449,
      "step": 767
    },
    {
      "epoch": 0.005218058184236351,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0006511035653650255,
      "loss": 2.5755,
      "step": 768
    },
    {
      "epoch": 0.0052248525308304095,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0006519524617996605,
      "loss": 2.8197,
      "step": 769
    },
    {
      "epoch": 0.005231646877424467,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0006528013582342955,
      "loss": 2.745,
      "step": 770
    },
    {
      "epoch": 0.005238441224018525,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0006536502546689303,
      "loss": 2.8323,
      "step": 771
    },
    {
      "epoch": 0.005245235570612582,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0006544991511035654,
      "loss": 2.7359,
      "step": 772
    },
    {
      "epoch": 0.005252029917206641,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0006553480475382003,
      "loss": 2.5879,
      "step": 773
    },
    {
      "epoch": 0.005258824263800698,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0006561969439728354,
      "loss": 2.696,
      "step": 774
    },
    {
      "epoch": 0.005265618610394756,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0006570458404074703,
      "loss": 2.7964,
      "step": 775
    },
    {
      "epoch": 0.0052724129569888135,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0006578947368421054,
      "loss": 2.7279,
      "step": 776
    },
    {
      "epoch": 0.005279207303582872,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0006587436332767402,
      "loss": 2.7301,
      "step": 777
    },
    {
      "epoch": 0.005286001650176929,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0006595925297113752,
      "loss": 2.7537,
      "step": 778
    },
    {
      "epoch": 0.005292795996770986,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0006604414261460102,
      "loss": 2.6057,
      "step": 779
    },
    {
      "epoch": 0.005299590343365045,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0006612903225806452,
      "loss": 2.6449,
      "step": 780
    },
    {
      "epoch": 0.005306384689959102,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0006621392190152802,
      "loss": 2.7426,
      "step": 781
    },
    {
      "epoch": 0.00531317903655316,
      "grad_norm": 0.25390625,
      "learning_rate": 0.0006629881154499151,
      "loss": 2.8054,
      "step": 782
    },
    {
      "epoch": 0.0053199733831472175,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0006638370118845501,
      "loss": 2.7947,
      "step": 783
    },
    {
      "epoch": 0.005326767729741276,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0006646859083191851,
      "loss": 2.563,
      "step": 784
    },
    {
      "epoch": 0.005333562076335333,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00066553480475382,
      "loss": 2.8006,
      "step": 785
    },
    {
      "epoch": 0.005340356422929391,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0006663837011884551,
      "loss": 2.6889,
      "step": 786
    },
    {
      "epoch": 0.005347150769523449,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0006672325976230899,
      "loss": 2.6311,
      "step": 787
    },
    {
      "epoch": 0.005353945116117507,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000668081494057725,
      "loss": 2.7047,
      "step": 788
    },
    {
      "epoch": 0.005360739462711564,
      "grad_norm": 0.203125,
      "learning_rate": 0.0006689303904923599,
      "loss": 2.7376,
      "step": 789
    },
    {
      "epoch": 0.005367533809305622,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.000669779286926995,
      "loss": 2.819,
      "step": 790
    },
    {
      "epoch": 0.00537432815589968,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0006706281833616299,
      "loss": 2.6387,
      "step": 791
    },
    {
      "epoch": 0.005381122502493738,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0006714770797962648,
      "loss": 2.8,
      "step": 792
    },
    {
      "epoch": 0.005387916849087795,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0006723259762308998,
      "loss": 2.6243,
      "step": 793
    },
    {
      "epoch": 0.005394711195681853,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0006731748726655348,
      "loss": 2.6197,
      "step": 794
    },
    {
      "epoch": 0.005401505542275911,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0006740237691001698,
      "loss": 2.7556,
      "step": 795
    },
    {
      "epoch": 0.005408299888869968,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0006748726655348048,
      "loss": 2.6793,
      "step": 796
    },
    {
      "epoch": 0.005415094235464026,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0006757215619694397,
      "loss": 2.6814,
      "step": 797
    },
    {
      "epoch": 0.005421888582058084,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0006765704584040747,
      "loss": 2.8796,
      "step": 798
    },
    {
      "epoch": 0.005428682928652142,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0006774193548387097,
      "loss": 2.8024,
      "step": 799
    },
    {
      "epoch": 0.005435477275246199,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0006782682512733447,
      "loss": 2.7042,
      "step": 800
    },
    {
      "epoch": 0.0054422716218402575,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0006791171477079797,
      "loss": 2.6571,
      "step": 801
    },
    {
      "epoch": 0.005449065968434315,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0006799660441426146,
      "loss": 2.6964,
      "step": 802
    },
    {
      "epoch": 0.005455860315028373,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0006808149405772496,
      "loss": 2.6889,
      "step": 803
    },
    {
      "epoch": 0.00546265466162243,
      "grad_norm": 0.232421875,
      "learning_rate": 0.0006816638370118846,
      "loss": 2.777,
      "step": 804
    },
    {
      "epoch": 0.005469449008216489,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0006825127334465196,
      "loss": 2.7171,
      "step": 805
    },
    {
      "epoch": 0.005476243354810546,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0006833616298811545,
      "loss": 2.6666,
      "step": 806
    },
    {
      "epoch": 0.005483037701404604,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0006842105263157895,
      "loss": 2.8895,
      "step": 807
    },
    {
      "epoch": 0.0054898320479986615,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0006850594227504244,
      "loss": 2.7788,
      "step": 808
    },
    {
      "epoch": 0.005496626394592719,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0006859083191850595,
      "loss": 2.707,
      "step": 809
    },
    {
      "epoch": 0.005503420741186777,
      "grad_norm": 0.33984375,
      "learning_rate": 0.0006867572156196944,
      "loss": 2.597,
      "step": 810
    },
    {
      "epoch": 0.005510215087780834,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0006876061120543295,
      "loss": 2.837,
      "step": 811
    },
    {
      "epoch": 0.005517009434374893,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0006884550084889643,
      "loss": 2.609,
      "step": 812
    },
    {
      "epoch": 0.00552380378096895,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0006893039049235993,
      "loss": 2.7571,
      "step": 813
    },
    {
      "epoch": 0.005530598127563008,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0006901528013582343,
      "loss": 2.8308,
      "step": 814
    },
    {
      "epoch": 0.0055373924741570655,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0006910016977928693,
      "loss": 2.7423,
      "step": 815
    },
    {
      "epoch": 0.005544186820751124,
      "grad_norm": 0.267578125,
      "learning_rate": 0.0006918505942275043,
      "loss": 2.6147,
      "step": 816
    },
    {
      "epoch": 0.005550981167345181,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0006926994906621392,
      "loss": 2.7139,
      "step": 817
    },
    {
      "epoch": 0.005557775513939239,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0006935483870967742,
      "loss": 2.5881,
      "step": 818
    },
    {
      "epoch": 0.005564569860533297,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0006943972835314092,
      "loss": 2.7379,
      "step": 819
    },
    {
      "epoch": 0.005571364207127355,
      "grad_norm": 0.24609375,
      "learning_rate": 0.0006952461799660441,
      "loss": 2.6698,
      "step": 820
    },
    {
      "epoch": 0.005578158553721412,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0006960950764006792,
      "loss": 2.6775,
      "step": 821
    },
    {
      "epoch": 0.00558495290031547,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.000696943972835314,
      "loss": 2.7965,
      "step": 822
    },
    {
      "epoch": 0.005591747246909528,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0006977928692699491,
      "loss": 2.5805,
      "step": 823
    },
    {
      "epoch": 0.005598541593503585,
      "grad_norm": 0.22265625,
      "learning_rate": 0.000698641765704584,
      "loss": 2.6451,
      "step": 824
    },
    {
      "epoch": 0.005605335940097643,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0006994906621392191,
      "loss": 2.8378,
      "step": 825
    },
    {
      "epoch": 0.005612130286691701,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.000700339558573854,
      "loss": 2.8296,
      "step": 826
    },
    {
      "epoch": 0.005618924633285759,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007011884550084889,
      "loss": 2.6331,
      "step": 827
    },
    {
      "epoch": 0.005625718979879816,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0007020373514431239,
      "loss": 2.6905,
      "step": 828
    },
    {
      "epoch": 0.005632513326473874,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0007028862478777589,
      "loss": 2.716,
      "step": 829
    },
    {
      "epoch": 0.005639307673067932,
      "grad_norm": 0.25,
      "learning_rate": 0.0007037351443123939,
      "loss": 2.7681,
      "step": 830
    },
    {
      "epoch": 0.00564610201966199,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007045840407470289,
      "loss": 2.6439,
      "step": 831
    },
    {
      "epoch": 0.005652896366256047,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.000705432937181664,
      "loss": 2.8458,
      "step": 832
    },
    {
      "epoch": 0.0056596907128501055,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0007062818336162988,
      "loss": 2.8657,
      "step": 833
    },
    {
      "epoch": 0.005666485059444163,
      "grad_norm": 0.21875,
      "learning_rate": 0.0007071307300509338,
      "loss": 2.4533,
      "step": 834
    },
    {
      "epoch": 0.005673279406038221,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0007079796264855688,
      "loss": 2.625,
      "step": 835
    },
    {
      "epoch": 0.0056800737526322784,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0007088285229202038,
      "loss": 2.655,
      "step": 836
    },
    {
      "epoch": 0.005686868099226337,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0007096774193548388,
      "loss": 2.7429,
      "step": 837
    },
    {
      "epoch": 0.005693662445820394,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.0007105263157894737,
      "loss": 2.783,
      "step": 838
    },
    {
      "epoch": 0.005700456792414451,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0007113752122241087,
      "loss": 2.709,
      "step": 839
    },
    {
      "epoch": 0.0057072511390085096,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0007122241086587437,
      "loss": 2.6443,
      "step": 840
    },
    {
      "epoch": 0.005714045485602567,
      "grad_norm": 0.29296875,
      "learning_rate": 0.0007130730050933786,
      "loss": 2.7772,
      "step": 841
    },
    {
      "epoch": 0.005720839832196625,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0007139219015280137,
      "loss": 2.6576,
      "step": 842
    },
    {
      "epoch": 0.0057276341787906825,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0007147707979626485,
      "loss": 2.7802,
      "step": 843
    },
    {
      "epoch": 0.005734428525384741,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0007156196943972836,
      "loss": 2.6801,
      "step": 844
    },
    {
      "epoch": 0.005741222871978798,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0007164685908319185,
      "loss": 2.6338,
      "step": 845
    },
    {
      "epoch": 0.005748017218572856,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0007173174872665536,
      "loss": 2.9029,
      "step": 846
    },
    {
      "epoch": 0.005754811565166914,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0007181663837011885,
      "loss": 2.8254,
      "step": 847
    },
    {
      "epoch": 0.005761605911760972,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0007190152801358234,
      "loss": 2.6741,
      "step": 848
    },
    {
      "epoch": 0.005768400258355029,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0007198641765704584,
      "loss": 2.7627,
      "step": 849
    },
    {
      "epoch": 0.005775194604949087,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0007207130730050934,
      "loss": 2.6955,
      "step": 850
    },
    {
      "epoch": 0.005781988951543145,
      "grad_norm": 0.203125,
      "learning_rate": 0.0007215619694397284,
      "loss": 2.7956,
      "step": 851
    },
    {
      "epoch": 0.005788783298137203,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0007224108658743634,
      "loss": 2.6513,
      "step": 852
    },
    {
      "epoch": 0.00579557764473126,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007232597623089983,
      "loss": 2.5792,
      "step": 853
    },
    {
      "epoch": 0.005802371991325318,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0007241086587436333,
      "loss": 2.6692,
      "step": 854
    },
    {
      "epoch": 0.005809166337919376,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0007249575551782682,
      "loss": 2.7281,
      "step": 855
    },
    {
      "epoch": 0.005815960684513433,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0007258064516129033,
      "loss": 2.6334,
      "step": 856
    },
    {
      "epoch": 0.005822755031107491,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0007266553480475382,
      "loss": 2.83,
      "step": 857
    },
    {
      "epoch": 0.005829549377701549,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0007275042444821732,
      "loss": 2.6795,
      "step": 858
    },
    {
      "epoch": 0.005836343724295607,
      "grad_norm": 0.244140625,
      "learning_rate": 0.0007283531409168081,
      "loss": 2.8291,
      "step": 859
    },
    {
      "epoch": 0.005843138070889664,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0007292020373514432,
      "loss": 2.6627,
      "step": 860
    },
    {
      "epoch": 0.0058499324174837225,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0007300509337860781,
      "loss": 2.6229,
      "step": 861
    },
    {
      "epoch": 0.00585672676407778,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0007308998302207131,
      "loss": 2.7626,
      "step": 862
    },
    {
      "epoch": 0.005863521110671838,
      "grad_norm": 0.1953125,
      "learning_rate": 0.000731748726655348,
      "loss": 2.6186,
      "step": 863
    },
    {
      "epoch": 0.005870315457265895,
      "grad_norm": 0.20703125,
      "learning_rate": 0.000732597623089983,
      "loss": 2.6786,
      "step": 864
    },
    {
      "epoch": 0.005877109803859954,
      "grad_norm": 0.21875,
      "learning_rate": 0.000733446519524618,
      "loss": 2.56,
      "step": 865
    },
    {
      "epoch": 0.005883904150454011,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.000734295415959253,
      "loss": 2.6953,
      "step": 866
    },
    {
      "epoch": 0.005890698497048069,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0007351443123938881,
      "loss": 2.5586,
      "step": 867
    },
    {
      "epoch": 0.0058974928436421265,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007359932088285229,
      "loss": 2.7052,
      "step": 868
    },
    {
      "epoch": 0.005904287190236184,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0007368421052631579,
      "loss": 2.6319,
      "step": 869
    },
    {
      "epoch": 0.005911081536830242,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0007376910016977929,
      "loss": 2.6295,
      "step": 870
    },
    {
      "epoch": 0.005917875883424299,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0007385398981324279,
      "loss": 2.6104,
      "step": 871
    },
    {
      "epoch": 0.005924670230018358,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0007393887945670629,
      "loss": 2.4868,
      "step": 872
    },
    {
      "epoch": 0.005931464576612415,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007402376910016978,
      "loss": 2.8152,
      "step": 873
    },
    {
      "epoch": 0.005938258923206473,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0007410865874363328,
      "loss": 2.5584,
      "step": 874
    },
    {
      "epoch": 0.0059450532698005305,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0007419354838709678,
      "loss": 2.5485,
      "step": 875
    },
    {
      "epoch": 0.005951847616394589,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0007427843803056027,
      "loss": 2.672,
      "step": 876
    },
    {
      "epoch": 0.005958641962988646,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0007436332767402378,
      "loss": 2.5863,
      "step": 877
    },
    {
      "epoch": 0.005965436309582704,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0007444821731748726,
      "loss": 2.7825,
      "step": 878
    },
    {
      "epoch": 0.005972230656176762,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0007453310696095077,
      "loss": 2.8305,
      "step": 879
    },
    {
      "epoch": 0.00597902500277082,
      "grad_norm": 0.21875,
      "learning_rate": 0.0007461799660441426,
      "loss": 2.7859,
      "step": 880
    },
    {
      "epoch": 0.005985819349364877,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007470288624787777,
      "loss": 2.7307,
      "step": 881
    },
    {
      "epoch": 0.005992613695958935,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0007478777589134126,
      "loss": 2.6784,
      "step": 882
    },
    {
      "epoch": 0.005999408042552993,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0007487266553480475,
      "loss": 2.5952,
      "step": 883
    },
    {
      "epoch": 0.00600620238914705,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0007495755517826825,
      "loss": 2.5534,
      "step": 884
    },
    {
      "epoch": 0.006012996735741108,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0007504244482173175,
      "loss": 2.6682,
      "step": 885
    },
    {
      "epoch": 0.006019791082335166,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0007512733446519525,
      "loss": 2.7836,
      "step": 886
    },
    {
      "epoch": 0.006026585428929224,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0007521222410865875,
      "loss": 2.724,
      "step": 887
    },
    {
      "epoch": 0.006033379775523281,
      "grad_norm": 0.25,
      "learning_rate": 0.0007529711375212223,
      "loss": 2.6176,
      "step": 888
    },
    {
      "epoch": 0.006040174122117339,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0007538200339558574,
      "loss": 2.8249,
      "step": 889
    },
    {
      "epoch": 0.006046968468711397,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0007546689303904923,
      "loss": 2.5895,
      "step": 890
    },
    {
      "epoch": 0.006053762815305455,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0007555178268251274,
      "loss": 2.7643,
      "step": 891
    },
    {
      "epoch": 0.006060557161899512,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0007563667232597623,
      "loss": 2.6205,
      "step": 892
    },
    {
      "epoch": 0.0060673515084935705,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007572156196943974,
      "loss": 2.8294,
      "step": 893
    },
    {
      "epoch": 0.006074145855087628,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0007580645161290322,
      "loss": 2.6502,
      "step": 894
    },
    {
      "epoch": 0.006080940201681686,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0007589134125636672,
      "loss": 2.6436,
      "step": 895
    },
    {
      "epoch": 0.006087734548275743,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0007597623089983022,
      "loss": 2.6681,
      "step": 896
    },
    {
      "epoch": 0.006094528894869802,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0007606112054329372,
      "loss": 2.7071,
      "step": 897
    },
    {
      "epoch": 0.006101323241463859,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0007614601018675721,
      "loss": 2.5545,
      "step": 898
    },
    {
      "epoch": 0.006108117588057916,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0007623089983022071,
      "loss": 2.8749,
      "step": 899
    },
    {
      "epoch": 0.0061149119346519745,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0007631578947368421,
      "loss": 2.6774,
      "step": 900
    },
    {
      "epoch": 0.006121706281246032,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0007640067911714771,
      "loss": 2.5476,
      "step": 901
    },
    {
      "epoch": 0.00612850062784009,
      "grad_norm": 0.201171875,
      "learning_rate": 0.000764855687606112,
      "loss": 2.6813,
      "step": 902
    },
    {
      "epoch": 0.006135294974434147,
      "grad_norm": 0.19921875,
      "learning_rate": 0.000765704584040747,
      "loss": 2.7319,
      "step": 903
    },
    {
      "epoch": 0.006142089321028206,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.000766553480475382,
      "loss": 2.7752,
      "step": 904
    },
    {
      "epoch": 0.006148883667622263,
      "grad_norm": 0.220703125,
      "learning_rate": 0.000767402376910017,
      "loss": 2.64,
      "step": 905
    },
    {
      "epoch": 0.006155678014216321,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.000768251273344652,
      "loss": 2.7637,
      "step": 906
    },
    {
      "epoch": 0.0061624723608103785,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.000769100169779287,
      "loss": 2.7751,
      "step": 907
    },
    {
      "epoch": 0.006169266707404437,
      "grad_norm": 0.232421875,
      "learning_rate": 0.000769949066213922,
      "loss": 2.7993,
      "step": 908
    },
    {
      "epoch": 0.006176061053998494,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0007707979626485568,
      "loss": 2.7882,
      "step": 909
    },
    {
      "epoch": 0.006182855400592552,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0007716468590831919,
      "loss": 2.6477,
      "step": 910
    },
    {
      "epoch": 0.00618964974718661,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007724957555178268,
      "loss": 2.7019,
      "step": 911
    },
    {
      "epoch": 0.006196444093780668,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0007733446519524619,
      "loss": 2.6772,
      "step": 912
    },
    {
      "epoch": 0.006203238440374725,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0007741935483870968,
      "loss": 2.5568,
      "step": 913
    },
    {
      "epoch": 0.0062100327869687826,
      "grad_norm": 0.203125,
      "learning_rate": 0.0007750424448217318,
      "loss": 2.6795,
      "step": 914
    },
    {
      "epoch": 0.006216827133562841,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0007758913412563667,
      "loss": 2.7424,
      "step": 915
    },
    {
      "epoch": 0.006223621480156898,
      "grad_norm": 0.2890625,
      "learning_rate": 0.0007767402376910017,
      "loss": 2.9228,
      "step": 916
    },
    {
      "epoch": 0.006230415826750956,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0007775891341256367,
      "loss": 2.7721,
      "step": 917
    },
    {
      "epoch": 0.006237210173345014,
      "grad_norm": 0.21875,
      "learning_rate": 0.0007784380305602717,
      "loss": 2.701,
      "step": 918
    },
    {
      "epoch": 0.006244004519939072,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0007792869269949066,
      "loss": 2.7814,
      "step": 919
    },
    {
      "epoch": 0.006250798866533129,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0007801358234295416,
      "loss": 2.6029,
      "step": 920
    },
    {
      "epoch": 0.0062575932131271874,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0007809847198641766,
      "loss": 2.5953,
      "step": 921
    },
    {
      "epoch": 0.006264387559721245,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0007818336162988116,
      "loss": 2.6348,
      "step": 922
    },
    {
      "epoch": 0.006271181906315303,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0007826825127334465,
      "loss": 2.724,
      "step": 923
    },
    {
      "epoch": 0.00627797625290936,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0007835314091680815,
      "loss": 2.6399,
      "step": 924
    },
    {
      "epoch": 0.0062847705995034185,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.0007843803056027164,
      "loss": 2.6929,
      "step": 925
    },
    {
      "epoch": 0.006291564946097476,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0007852292020373515,
      "loss": 2.6891,
      "step": 926
    },
    {
      "epoch": 0.006298359292691534,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0007860780984719864,
      "loss": 2.5286,
      "step": 927
    },
    {
      "epoch": 0.0063051536392855915,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0007869269949066215,
      "loss": 2.7965,
      "step": 928
    },
    {
      "epoch": 0.006311947985879649,
      "grad_norm": 0.255859375,
      "learning_rate": 0.0007877758913412563,
      "loss": 2.711,
      "step": 929
    },
    {
      "epoch": 0.006318742332473707,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0007886247877758913,
      "loss": 2.7403,
      "step": 930
    },
    {
      "epoch": 0.006325536679067764,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0007894736842105263,
      "loss": 2.6823,
      "step": 931
    },
    {
      "epoch": 0.0063323310256618226,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0007903225806451613,
      "loss": 2.4934,
      "step": 932
    },
    {
      "epoch": 0.00633912537225588,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0007911714770797963,
      "loss": 2.5619,
      "step": 933
    },
    {
      "epoch": 0.006345919718849938,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0007920203735144312,
      "loss": 2.7715,
      "step": 934
    },
    {
      "epoch": 0.0063527140654439955,
      "grad_norm": 0.203125,
      "learning_rate": 0.0007928692699490662,
      "loss": 2.6318,
      "step": 935
    },
    {
      "epoch": 0.006359508412038054,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0007937181663837012,
      "loss": 2.7752,
      "step": 936
    },
    {
      "epoch": 0.006366302758632111,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0007945670628183361,
      "loss": 2.7039,
      "step": 937
    },
    {
      "epoch": 0.006373097105226169,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0007954159592529712,
      "loss": 2.6902,
      "step": 938
    },
    {
      "epoch": 0.006379891451820227,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0007962648556876061,
      "loss": 2.6848,
      "step": 939
    },
    {
      "epoch": 0.006386685798414285,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0007971137521222411,
      "loss": 2.7442,
      "step": 940
    },
    {
      "epoch": 0.006393480145008342,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0007979626485568761,
      "loss": 2.671,
      "step": 941
    },
    {
      "epoch": 0.0064002744916024,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0007988115449915111,
      "loss": 2.6396,
      "step": 942
    },
    {
      "epoch": 0.006407068838196458,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0007996604414261461,
      "loss": 2.5772,
      "step": 943
    },
    {
      "epoch": 0.006413863184790515,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0008005093378607809,
      "loss": 2.6505,
      "step": 944
    },
    {
      "epoch": 0.006420657531384573,
      "grad_norm": 0.216796875,
      "learning_rate": 0.000801358234295416,
      "loss": 2.6935,
      "step": 945
    },
    {
      "epoch": 0.006427451877978631,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0008022071307300509,
      "loss": 2.7836,
      "step": 946
    },
    {
      "epoch": 0.006434246224572689,
      "grad_norm": 0.259765625,
      "learning_rate": 0.000803056027164686,
      "loss": 2.654,
      "step": 947
    },
    {
      "epoch": 0.006441040571166746,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0008039049235993209,
      "loss": 2.6335,
      "step": 948
    },
    {
      "epoch": 0.006447834917760804,
      "grad_norm": 0.2109375,
      "learning_rate": 0.000804753820033956,
      "loss": 2.5837,
      "step": 949
    },
    {
      "epoch": 0.006454629264354862,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0008056027164685908,
      "loss": 2.5326,
      "step": 950
    },
    {
      "epoch": 0.00646142361094892,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0008064516129032258,
      "loss": 2.6926,
      "step": 951
    },
    {
      "epoch": 0.006468217957542977,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0008073005093378608,
      "loss": 2.6126,
      "step": 952
    },
    {
      "epoch": 0.0064750123041370355,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0008081494057724958,
      "loss": 2.688,
      "step": 953
    },
    {
      "epoch": 0.006481806650731093,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0008089983022071307,
      "loss": 2.8501,
      "step": 954
    },
    {
      "epoch": 0.006488600997325151,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0008098471986417657,
      "loss": 2.7232,
      "step": 955
    },
    {
      "epoch": 0.006495395343919208,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0008106960950764007,
      "loss": 2.7858,
      "step": 956
    },
    {
      "epoch": 0.006502189690513267,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0008115449915110357,
      "loss": 2.6091,
      "step": 957
    },
    {
      "epoch": 0.006508984037107324,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0008123938879456706,
      "loss": 2.6804,
      "step": 958
    },
    {
      "epoch": 0.006515778383701381,
      "grad_norm": 0.203125,
      "learning_rate": 0.0008132427843803057,
      "loss": 2.5909,
      "step": 959
    },
    {
      "epoch": 0.0065225727302954395,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0008140916808149405,
      "loss": 2.6389,
      "step": 960
    },
    {
      "epoch": 0.006529367076889497,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0008149405772495756,
      "loss": 2.6291,
      "step": 961
    },
    {
      "epoch": 0.006536161423483555,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0008157894736842105,
      "loss": 2.6857,
      "step": 962
    },
    {
      "epoch": 0.006542955770077612,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0008166383701188456,
      "loss": 2.6645,
      "step": 963
    },
    {
      "epoch": 0.006549750116671671,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0008174872665534804,
      "loss": 2.658,
      "step": 964
    },
    {
      "epoch": 0.006556544463265728,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0008183361629881154,
      "loss": 2.5338,
      "step": 965
    },
    {
      "epoch": 0.006563338809859786,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0008191850594227504,
      "loss": 2.6706,
      "step": 966
    },
    {
      "epoch": 0.0065701331564538435,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0008200339558573854,
      "loss": 2.7558,
      "step": 967
    },
    {
      "epoch": 0.006576927503047902,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0008208828522920204,
      "loss": 2.6726,
      "step": 968
    },
    {
      "epoch": 0.006583721849641959,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0008217317487266554,
      "loss": 2.5188,
      "step": 969
    },
    {
      "epoch": 0.006590516196236017,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0008225806451612903,
      "loss": 2.7744,
      "step": 970
    },
    {
      "epoch": 0.006597310542830075,
      "grad_norm": 0.1875,
      "learning_rate": 0.0008234295415959253,
      "loss": 2.7391,
      "step": 971
    },
    {
      "epoch": 0.006604104889424133,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0008242784380305603,
      "loss": 2.4801,
      "step": 972
    },
    {
      "epoch": 0.00661089923601819,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0008251273344651953,
      "loss": 2.6826,
      "step": 973
    },
    {
      "epoch": 0.0066176935826122475,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0008259762308998303,
      "loss": 2.726,
      "step": 974
    },
    {
      "epoch": 0.006624487929206306,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0008268251273344652,
      "loss": 2.6519,
      "step": 975
    },
    {
      "epoch": 0.006631282275800363,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0008276740237691002,
      "loss": 2.525,
      "step": 976
    },
    {
      "epoch": 0.006638076622394421,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0008285229202037352,
      "loss": 2.7051,
      "step": 977
    },
    {
      "epoch": 0.006644870968988479,
      "grad_norm": 0.236328125,
      "learning_rate": 0.0008293718166383702,
      "loss": 2.6734,
      "step": 978
    },
    {
      "epoch": 0.006651665315582537,
      "grad_norm": 0.22265625,
      "learning_rate": 0.000830220713073005,
      "loss": 2.6123,
      "step": 979
    },
    {
      "epoch": 0.006658459662176594,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0008310696095076401,
      "loss": 2.7887,
      "step": 980
    },
    {
      "epoch": 0.006665254008770652,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000831918505942275,
      "loss": 2.6891,
      "step": 981
    },
    {
      "epoch": 0.00667204835536471,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0008327674023769101,
      "loss": 2.7161,
      "step": 982
    },
    {
      "epoch": 0.006678842701958768,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.000833616298811545,
      "loss": 2.7503,
      "step": 983
    },
    {
      "epoch": 0.006685637048552825,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0008344651952461801,
      "loss": 2.7347,
      "step": 984
    },
    {
      "epoch": 0.0066924313951468835,
      "grad_norm": 0.203125,
      "learning_rate": 0.0008353140916808149,
      "loss": 2.7091,
      "step": 985
    },
    {
      "epoch": 0.006699225741740941,
      "grad_norm": 0.234375,
      "learning_rate": 0.0008361629881154499,
      "loss": 2.7911,
      "step": 986
    },
    {
      "epoch": 0.006706020088334999,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0008370118845500849,
      "loss": 2.6804,
      "step": 987
    },
    {
      "epoch": 0.006712814434929056,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0008378607809847199,
      "loss": 2.7396,
      "step": 988
    },
    {
      "epoch": 0.006719608781523114,
      "grad_norm": 0.251953125,
      "learning_rate": 0.0008387096774193549,
      "loss": 2.7447,
      "step": 989
    },
    {
      "epoch": 0.006726403128117172,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0008395585738539898,
      "loss": 2.6138,
      "step": 990
    },
    {
      "epoch": 0.006733197474711229,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0008404074702886248,
      "loss": 2.6208,
      "step": 991
    },
    {
      "epoch": 0.0067399918213052875,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0008412563667232598,
      "loss": 2.5898,
      "step": 992
    },
    {
      "epoch": 0.006746786167899345,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0008421052631578947,
      "loss": 2.7868,
      "step": 993
    },
    {
      "epoch": 0.006753580514493403,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0008429541595925298,
      "loss": 2.7648,
      "step": 994
    },
    {
      "epoch": 0.0067603748610874604,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0008438030560271646,
      "loss": 2.6452,
      "step": 995
    },
    {
      "epoch": 0.006767169207681519,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0008446519524617997,
      "loss": 2.5667,
      "step": 996
    },
    {
      "epoch": 0.006773963554275576,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0008455008488964346,
      "loss": 2.7462,
      "step": 997
    },
    {
      "epoch": 0.006780757900869634,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0008463497453310697,
      "loss": 2.6757,
      "step": 998
    },
    {
      "epoch": 0.0067875522474636915,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0008471986417657046,
      "loss": 2.6559,
      "step": 999
    },
    {
      "epoch": 0.00679434659405775,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0008480475382003395,
      "loss": 2.6375,
      "step": 1000
    },
    {
      "epoch": 0.006801140940651807,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0008488964346349745,
      "loss": 2.61,
      "step": 1001
    },
    {
      "epoch": 0.006807935287245865,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0008497453310696095,
      "loss": 2.5952,
      "step": 1002
    },
    {
      "epoch": 0.006814729633839923,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.0008505942275042445,
      "loss": 2.6986,
      "step": 1003
    },
    {
      "epoch": 0.00682152398043398,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0008514431239388795,
      "loss": 2.7703,
      "step": 1004
    },
    {
      "epoch": 0.006828318327028038,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0008522920203735144,
      "loss": 2.4923,
      "step": 1005
    },
    {
      "epoch": 0.0068351126736220956,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0008531409168081494,
      "loss": 2.7121,
      "step": 1006
    },
    {
      "epoch": 0.006841907020216154,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0008539898132427844,
      "loss": 2.6048,
      "step": 1007
    },
    {
      "epoch": 0.006848701366810211,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0008548387096774194,
      "loss": 2.6403,
      "step": 1008
    },
    {
      "epoch": 0.006855495713404269,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0008556876061120544,
      "loss": 2.7569,
      "step": 1009
    },
    {
      "epoch": 0.006862290059998327,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0008565365025466893,
      "loss": 2.6565,
      "step": 1010
    },
    {
      "epoch": 0.006869084406592385,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0008573853989813243,
      "loss": 2.6241,
      "step": 1011
    },
    {
      "epoch": 0.006875878753186442,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0008582342954159593,
      "loss": 2.706,
      "step": 1012
    },
    {
      "epoch": 0.0068826730997805004,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0008590831918505943,
      "loss": 2.5579,
      "step": 1013
    },
    {
      "epoch": 0.006889467446374558,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0008599320882852292,
      "loss": 2.5186,
      "step": 1014
    },
    {
      "epoch": 0.006896261792968616,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0008607809847198643,
      "loss": 2.5289,
      "step": 1015
    },
    {
      "epoch": 0.006903056139562673,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0008616298811544991,
      "loss": 2.76,
      "step": 1016
    },
    {
      "epoch": 0.0069098504861567316,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0008624787775891342,
      "loss": 2.7527,
      "step": 1017
    },
    {
      "epoch": 0.006916644832750789,
      "grad_norm": 0.46875,
      "learning_rate": 0.0008633276740237691,
      "loss": 2.6566,
      "step": 1018
    },
    {
      "epoch": 0.006923439179344846,
      "grad_norm": 0.21875,
      "learning_rate": 0.0008641765704584042,
      "loss": 2.5629,
      "step": 1019
    },
    {
      "epoch": 0.0069302335259389045,
      "grad_norm": 0.201171875,
      "learning_rate": 0.000865025466893039,
      "loss": 2.6367,
      "step": 1020
    },
    {
      "epoch": 0.006937027872532962,
      "grad_norm": 0.2373046875,
      "learning_rate": 0.000865874363327674,
      "loss": 2.6543,
      "step": 1021
    },
    {
      "epoch": 0.00694382221912702,
      "grad_norm": 0.203125,
      "learning_rate": 0.000866723259762309,
      "loss": 2.6716,
      "step": 1022
    },
    {
      "epoch": 0.006950616565721077,
      "grad_norm": 0.212890625,
      "learning_rate": 0.000867572156196944,
      "loss": 2.5954,
      "step": 1023
    },
    {
      "epoch": 0.006957410912315136,
      "grad_norm": 0.216796875,
      "learning_rate": 0.000868421052631579,
      "loss": 2.7851,
      "step": 1024
    },
    {
      "epoch": 0.006964205258909193,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.000869269949066214,
      "loss": 2.6956,
      "step": 1025
    },
    {
      "epoch": 0.006970999605503251,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0008701188455008489,
      "loss": 2.7123,
      "step": 1026
    },
    {
      "epoch": 0.0069777939520973085,
      "grad_norm": 0.2177734375,
      "learning_rate": 0.0008709677419354839,
      "loss": 2.7278,
      "step": 1027
    },
    {
      "epoch": 0.006984588298691367,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0008718166383701188,
      "loss": 2.6755,
      "step": 1028
    },
    {
      "epoch": 0.006991382645285424,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0008726655348047539,
      "loss": 2.518,
      "step": 1029
    },
    {
      "epoch": 0.006998176991879482,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0008735144312393887,
      "loss": 2.681,
      "step": 1030
    },
    {
      "epoch": 0.00700497133847354,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0008743633276740238,
      "loss": 2.85,
      "step": 1031
    },
    {
      "epoch": 0.007011765685067598,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0008752122241086587,
      "loss": 2.6702,
      "step": 1032
    },
    {
      "epoch": 0.007018560031661655,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0008760611205432938,
      "loss": 2.6623,
      "step": 1033
    },
    {
      "epoch": 0.0070253543782557125,
      "grad_norm": 0.2490234375,
      "learning_rate": 0.0008769100169779287,
      "loss": 2.6645,
      "step": 1034
    },
    {
      "epoch": 0.007032148724849771,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0008777589134125637,
      "loss": 2.6721,
      "step": 1035
    },
    {
      "epoch": 0.007038943071443828,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0008786078098471986,
      "loss": 2.5493,
      "step": 1036
    },
    {
      "epoch": 0.007045737418037886,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0008794567062818336,
      "loss": 2.7187,
      "step": 1037
    },
    {
      "epoch": 0.007052531764631944,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0008803056027164686,
      "loss": 2.6283,
      "step": 1038
    },
    {
      "epoch": 0.007059326111226002,
      "grad_norm": 0.21875,
      "learning_rate": 0.0008811544991511036,
      "loss": 2.7178,
      "step": 1039
    },
    {
      "epoch": 0.007066120457820059,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0008820033955857387,
      "loss": 2.666,
      "step": 1040
    },
    {
      "epoch": 0.007072914804414117,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0008828522920203735,
      "loss": 2.6156,
      "step": 1041
    },
    {
      "epoch": 0.007079709151008175,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0008837011884550085,
      "loss": 2.7772,
      "step": 1042
    },
    {
      "epoch": 0.007086503497602233,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0008845500848896435,
      "loss": 2.6747,
      "step": 1043
    },
    {
      "epoch": 0.00709329784419629,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0008853989813242785,
      "loss": 2.6119,
      "step": 1044
    },
    {
      "epoch": 0.0071000921907903485,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0008862478777589135,
      "loss": 2.743,
      "step": 1045
    },
    {
      "epoch": 0.007106886537384406,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0008870967741935484,
      "loss": 2.6654,
      "step": 1046
    },
    {
      "epoch": 0.007113680883978464,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0008879456706281834,
      "loss": 2.7059,
      "step": 1047
    },
    {
      "epoch": 0.007120475230572521,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0008887945670628184,
      "loss": 2.6171,
      "step": 1048
    },
    {
      "epoch": 0.007127269577166579,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0008896434634974533,
      "loss": 2.4967,
      "step": 1049
    },
    {
      "epoch": 0.007134063923760637,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0008904923599320884,
      "loss": 2.5316,
      "step": 1050
    },
    {
      "epoch": 0.007140858270354694,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0008913412563667232,
      "loss": 2.6937,
      "step": 1051
    },
    {
      "epoch": 0.0071476526169487525,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0008921901528013583,
      "loss": 2.7245,
      "step": 1052
    },
    {
      "epoch": 0.00715444696354281,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0008930390492359932,
      "loss": 2.7575,
      "step": 1053
    },
    {
      "epoch": 0.007161241310136868,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0008938879456706283,
      "loss": 2.7731,
      "step": 1054
    },
    {
      "epoch": 0.007168035656730925,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0008947368421052632,
      "loss": 2.6912,
      "step": 1055
    },
    {
      "epoch": 0.007174830003324984,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0008955857385398981,
      "loss": 2.6977,
      "step": 1056
    },
    {
      "epoch": 0.007181624349919041,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0008964346349745331,
      "loss": 2.5901,
      "step": 1057
    },
    {
      "epoch": 0.007188418696513099,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0008972835314091681,
      "loss": 2.7131,
      "step": 1058
    },
    {
      "epoch": 0.0071952130431071565,
      "grad_norm": 0.203125,
      "learning_rate": 0.0008981324278438031,
      "loss": 2.6198,
      "step": 1059
    },
    {
      "epoch": 0.007202007389701215,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0008989813242784381,
      "loss": 2.7177,
      "step": 1060
    },
    {
      "epoch": 0.007208801736295272,
      "grad_norm": 0.203125,
      "learning_rate": 0.000899830220713073,
      "loss": 2.8156,
      "step": 1061
    },
    {
      "epoch": 0.00721559608288933,
      "grad_norm": 0.35546875,
      "learning_rate": 0.000900679117147708,
      "loss": 2.5919,
      "step": 1062
    },
    {
      "epoch": 0.007222390429483388,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009015280135823429,
      "loss": 2.7701,
      "step": 1063
    },
    {
      "epoch": 0.007229184776077445,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.000902376910016978,
      "loss": 2.6885,
      "step": 1064
    },
    {
      "epoch": 0.007235979122671503,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0009032258064516129,
      "loss": 2.6694,
      "step": 1065
    },
    {
      "epoch": 0.0072427734692655605,
      "grad_norm": 0.208984375,
      "learning_rate": 0.000904074702886248,
      "loss": 2.6892,
      "step": 1066
    },
    {
      "epoch": 0.007249567815859619,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0009049235993208828,
      "loss": 2.6842,
      "step": 1067
    },
    {
      "epoch": 0.007256362162453676,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009057724957555179,
      "loss": 2.7093,
      "step": 1068
    },
    {
      "epoch": 0.007263156509047734,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0009066213921901528,
      "loss": 2.5624,
      "step": 1069
    },
    {
      "epoch": 0.007269950855641792,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0009074702886247878,
      "loss": 2.6525,
      "step": 1070
    },
    {
      "epoch": 0.00727674520223585,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009083191850594227,
      "loss": 2.653,
      "step": 1071
    },
    {
      "epoch": 0.007283539548829907,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0009091680814940577,
      "loss": 2.8504,
      "step": 1072
    },
    {
      "epoch": 0.007290333895423965,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009100169779286927,
      "loss": 2.588,
      "step": 1073
    },
    {
      "epoch": 0.007297128242018023,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0009108658743633277,
      "loss": 2.5844,
      "step": 1074
    },
    {
      "epoch": 0.007303922588612081,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009117147707979628,
      "loss": 2.8124,
      "step": 1075
    },
    {
      "epoch": 0.007310716935206138,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009125636672325976,
      "loss": 2.6106,
      "step": 1076
    },
    {
      "epoch": 0.0073175112818001965,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009134125636672326,
      "loss": 2.5514,
      "step": 1077
    },
    {
      "epoch": 0.007324305628394254,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009142614601018676,
      "loss": 2.6089,
      "step": 1078
    },
    {
      "epoch": 0.007331099974988311,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0009151103565365026,
      "loss": 2.6225,
      "step": 1079
    },
    {
      "epoch": 0.007337894321582369,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009159592529711376,
      "loss": 2.8088,
      "step": 1080
    },
    {
      "epoch": 0.007344688668176427,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009168081494057726,
      "loss": 2.6727,
      "step": 1081
    },
    {
      "epoch": 0.007351483014770485,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009176570458404074,
      "loss": 2.5649,
      "step": 1082
    },
    {
      "epoch": 0.007358277361364542,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009185059422750425,
      "loss": 2.6911,
      "step": 1083
    },
    {
      "epoch": 0.0073650717079586005,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009193548387096774,
      "loss": 2.5469,
      "step": 1084
    },
    {
      "epoch": 0.007371866054552658,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009202037351443125,
      "loss": 2.6897,
      "step": 1085
    },
    {
      "epoch": 0.007378660401146716,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009210526315789473,
      "loss": 2.6828,
      "step": 1086
    },
    {
      "epoch": 0.0073854547477407734,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009219015280135824,
      "loss": 2.7523,
      "step": 1087
    },
    {
      "epoch": 0.007392249094334832,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009227504244482173,
      "loss": 2.7179,
      "step": 1088
    },
    {
      "epoch": 0.007399043440928889,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009235993208828523,
      "loss": 2.6933,
      "step": 1089
    },
    {
      "epoch": 0.007405837787522947,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009244482173174873,
      "loss": 2.7439,
      "step": 1090
    },
    {
      "epoch": 0.0074126321341170046,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009252971137521223,
      "loss": 2.6226,
      "step": 1091
    },
    {
      "epoch": 0.007419426480711063,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0009261460101867572,
      "loss": 2.7813,
      "step": 1092
    },
    {
      "epoch": 0.00742622082730512,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009269949066213922,
      "loss": 2.677,
      "step": 1093
    },
    {
      "epoch": 0.0074330151738991775,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.0009278438030560272,
      "loss": 2.6709,
      "step": 1094
    },
    {
      "epoch": 0.007439809520493236,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009286926994906622,
      "loss": 2.6805,
      "step": 1095
    },
    {
      "epoch": 0.007446603867087293,
      "grad_norm": 0.1953125,
      "learning_rate": 0.000929541595925297,
      "loss": 2.7433,
      "step": 1096
    },
    {
      "epoch": 0.007453398213681351,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009303904923599321,
      "loss": 2.7638,
      "step": 1097
    },
    {
      "epoch": 0.007460192560275409,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000931239388794567,
      "loss": 2.6954,
      "step": 1098
    },
    {
      "epoch": 0.007466986906869467,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009320882852292021,
      "loss": 2.7476,
      "step": 1099
    },
    {
      "epoch": 0.007473781253463524,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000932937181663837,
      "loss": 2.537,
      "step": 1100
    },
    {
      "epoch": 0.007480575600057582,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009337860780984721,
      "loss": 2.6757,
      "step": 1101
    },
    {
      "epoch": 0.00748736994665164,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009346349745331069,
      "loss": 2.6282,
      "step": 1102
    },
    {
      "epoch": 0.007494164293245698,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009354838709677419,
      "loss": 2.6348,
      "step": 1103
    },
    {
      "epoch": 0.007500958639839755,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009363327674023769,
      "loss": 2.7689,
      "step": 1104
    },
    {
      "epoch": 0.0075077529864338135,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009371816638370119,
      "loss": 2.773,
      "step": 1105
    },
    {
      "epoch": 0.007514547333027871,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009380305602716469,
      "loss": 2.791,
      "step": 1106
    },
    {
      "epoch": 0.007521341679621929,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009388794567062818,
      "loss": 2.5082,
      "step": 1107
    },
    {
      "epoch": 0.007528136026215986,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009397283531409168,
      "loss": 2.6399,
      "step": 1108
    },
    {
      "epoch": 0.007534930372810044,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009405772495755518,
      "loss": 2.8083,
      "step": 1109
    },
    {
      "epoch": 0.007541724719404102,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009414261460101867,
      "loss": 2.634,
      "step": 1110
    },
    {
      "epoch": 0.007548519065998159,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009422750424448218,
      "loss": 2.6872,
      "step": 1111
    },
    {
      "epoch": 0.0075553134125922175,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009431239388794567,
      "loss": 2.52,
      "step": 1112
    },
    {
      "epoch": 0.007562107759186275,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0009439728353140917,
      "loss": 2.6713,
      "step": 1113
    },
    {
      "epoch": 0.007568902105780333,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0009448217317487267,
      "loss": 2.7329,
      "step": 1114
    },
    {
      "epoch": 0.00757569645237439,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009456706281833617,
      "loss": 2.6856,
      "step": 1115
    },
    {
      "epoch": 0.007582490798968449,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0009465195246179967,
      "loss": 2.7138,
      "step": 1116
    },
    {
      "epoch": 0.007589285145562506,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009473684210526315,
      "loss": 2.5676,
      "step": 1117
    },
    {
      "epoch": 0.007596079492156564,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009482173174872666,
      "loss": 2.684,
      "step": 1118
    },
    {
      "epoch": 0.0076028738387506215,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009490662139219015,
      "loss": 2.6605,
      "step": 1119
    },
    {
      "epoch": 0.00760966818534468,
      "grad_norm": 0.376953125,
      "learning_rate": 0.0009499151103565366,
      "loss": 2.5797,
      "step": 1120
    },
    {
      "epoch": 0.007616462531938737,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0009507640067911715,
      "loss": 2.7345,
      "step": 1121
    },
    {
      "epoch": 0.007623256878532795,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0009516129032258065,
      "loss": 2.6278,
      "step": 1122
    },
    {
      "epoch": 0.007630051225126853,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009524617996604414,
      "loss": 2.7689,
      "step": 1123
    },
    {
      "epoch": 0.00763684557172091,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0009533106960950764,
      "loss": 2.7804,
      "step": 1124
    },
    {
      "epoch": 0.007643639918314968,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0009541595925297114,
      "loss": 2.5912,
      "step": 1125
    },
    {
      "epoch": 0.0076504342649090255,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009550084889643464,
      "loss": 2.7612,
      "step": 1126
    },
    {
      "epoch": 0.007657228611503084,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009558573853989813,
      "loss": 2.6197,
      "step": 1127
    },
    {
      "epoch": 0.007664022958097141,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009567062818336163,
      "loss": 2.6149,
      "step": 1128
    },
    {
      "epoch": 0.007670817304691199,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009575551782682513,
      "loss": 2.5466,
      "step": 1129
    },
    {
      "epoch": 0.007677611651285257,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0009584040747028863,
      "loss": 2.6653,
      "step": 1130
    },
    {
      "epoch": 0.007684405997879315,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0009592529711375212,
      "loss": 2.7092,
      "step": 1131
    },
    {
      "epoch": 0.007691200344473372,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009601018675721562,
      "loss": 2.7682,
      "step": 1132
    },
    {
      "epoch": 0.00769799469106743,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009609507640067911,
      "loss": 2.5882,
      "step": 1133
    },
    {
      "epoch": 0.007704789037661488,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009617996604414262,
      "loss": 2.6811,
      "step": 1134
    },
    {
      "epoch": 0.007711583384255546,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009626485568760611,
      "loss": 2.8217,
      "step": 1135
    },
    {
      "epoch": 0.007718377730849603,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009634974533106962,
      "loss": 2.6515,
      "step": 1136
    },
    {
      "epoch": 0.0077251720774436615,
      "grad_norm": 0.20703125,
      "learning_rate": 0.000964346349745331,
      "loss": 2.6942,
      "step": 1137
    },
    {
      "epoch": 0.007731966424037719,
      "grad_norm": 0.21484375,
      "learning_rate": 0.000965195246179966,
      "loss": 2.7653,
      "step": 1138
    },
    {
      "epoch": 0.007738760770631776,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.000966044142614601,
      "loss": 2.5742,
      "step": 1139
    },
    {
      "epoch": 0.007745555117225834,
      "grad_norm": 0.234375,
      "learning_rate": 0.000966893039049236,
      "loss": 2.4985,
      "step": 1140
    },
    {
      "epoch": 0.007752349463819892,
      "grad_norm": 0.208984375,
      "learning_rate": 0.000967741935483871,
      "loss": 2.7573,
      "step": 1141
    },
    {
      "epoch": 0.00775914381041395,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000968590831918506,
      "loss": 2.7641,
      "step": 1142
    },
    {
      "epoch": 0.007765938157008007,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009694397283531409,
      "loss": 2.6567,
      "step": 1143
    },
    {
      "epoch": 0.0077727325036020655,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009702886247877759,
      "loss": 2.6866,
      "step": 1144
    },
    {
      "epoch": 0.007779526850196123,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009711375212224108,
      "loss": 2.6039,
      "step": 1145
    },
    {
      "epoch": 0.007786321196790181,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009719864176570459,
      "loss": 2.7392,
      "step": 1146
    },
    {
      "epoch": 0.007793115543384238,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009728353140916809,
      "loss": 2.6212,
      "step": 1147
    },
    {
      "epoch": 0.007799909889978297,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009736842105263158,
      "loss": 2.6691,
      "step": 1148
    },
    {
      "epoch": 0.007806704236572354,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009745331069609508,
      "loss": 2.7382,
      "step": 1149
    },
    {
      "epoch": 0.007813498583166411,
      "grad_norm": 0.2119140625,
      "learning_rate": 0.0009753820033955858,
      "loss": 2.6853,
      "step": 1150
    },
    {
      "epoch": 0.00782029292976047,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009762308998302208,
      "loss": 2.6993,
      "step": 1151
    },
    {
      "epoch": 0.007827087276354528,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009770797962648558,
      "loss": 2.7139,
      "step": 1152
    },
    {
      "epoch": 0.007833881622948584,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009779286926994907,
      "loss": 2.4639,
      "step": 1153
    },
    {
      "epoch": 0.007840675969542642,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009787775891341257,
      "loss": 2.5455,
      "step": 1154
    },
    {
      "epoch": 0.0078474703161367,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009796264855687607,
      "loss": 2.5654,
      "step": 1155
    },
    {
      "epoch": 0.007854264662730759,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009804753820033957,
      "loss": 2.8418,
      "step": 1156
    },
    {
      "epoch": 0.007861059009324815,
      "grad_norm": 0.2451171875,
      "learning_rate": 0.0009813242784380307,
      "loss": 2.6356,
      "step": 1157
    },
    {
      "epoch": 0.007867853355918874,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009821731748726656,
      "loss": 2.5741,
      "step": 1158
    },
    {
      "epoch": 0.007874647702512932,
      "grad_norm": 0.23046875,
      "learning_rate": 0.0009830220713073004,
      "loss": 2.5625,
      "step": 1159
    },
    {
      "epoch": 0.00788144204910699,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0009838709677419356,
      "loss": 2.6893,
      "step": 1160
    },
    {
      "epoch": 0.007888236395701046,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0009847198641765704,
      "loss": 2.7522,
      "step": 1161
    },
    {
      "epoch": 0.007895030742295105,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009855687606112056,
      "loss": 2.6221,
      "step": 1162
    },
    {
      "epoch": 0.007901825088889163,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0009864176570458403,
      "loss": 2.8088,
      "step": 1163
    },
    {
      "epoch": 0.007908619435483221,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009872665534804755,
      "loss": 2.4991,
      "step": 1164
    },
    {
      "epoch": 0.007915413782077278,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009881154499151103,
      "loss": 2.5738,
      "step": 1165
    },
    {
      "epoch": 0.007922208128671336,
      "grad_norm": 0.25,
      "learning_rate": 0.0009889643463497453,
      "loss": 2.745,
      "step": 1166
    },
    {
      "epoch": 0.007929002475265394,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009898132427843803,
      "loss": 2.6481,
      "step": 1167
    },
    {
      "epoch": 0.00793579682185945,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009906621392190152,
      "loss": 2.6307,
      "step": 1168
    },
    {
      "epoch": 0.007942591168453509,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009915110356536502,
      "loss": 2.7606,
      "step": 1169
    },
    {
      "epoch": 0.007949385515047567,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009923599320882852,
      "loss": 2.5686,
      "step": 1170
    },
    {
      "epoch": 0.007956179861641625,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0009932088285229202,
      "loss": 2.7208,
      "step": 1171
    },
    {
      "epoch": 0.007962974208235682,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009940577249575552,
      "loss": 2.6542,
      "step": 1172
    },
    {
      "epoch": 0.00796976855482974,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009949066213921901,
      "loss": 2.7405,
      "step": 1173
    },
    {
      "epoch": 0.007976562901423798,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009957555178268251,
      "loss": 2.6206,
      "step": 1174
    },
    {
      "epoch": 0.007983357248017856,
      "grad_norm": 0.197265625,
      "learning_rate": 0.00099660441426146,
      "loss": 2.6741,
      "step": 1175
    },
    {
      "epoch": 0.007990151594611913,
      "grad_norm": 0.28515625,
      "learning_rate": 0.000997453310696095,
      "loss": 2.7418,
      "step": 1176
    },
    {
      "epoch": 0.007996945941205971,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.00099830220713073,
      "loss": 2.6398,
      "step": 1177
    },
    {
      "epoch": 0.008003740287800029,
      "grad_norm": 0.205078125,
      "learning_rate": 0.000999151103565365,
      "loss": 2.6666,
      "step": 1178
    },
    {
      "epoch": 0.008010534634394087,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.001,
      "loss": 2.5245,
      "step": 1179
    },
    {
      "epoch": 0.008017328980988144,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009999999998842528,
      "loss": 2.7995,
      "step": 1180
    },
    {
      "epoch": 0.008024123327582202,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009999999995370114,
      "loss": 2.6818,
      "step": 1181
    },
    {
      "epoch": 0.00803091767417626,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009999999989582758,
      "loss": 2.5461,
      "step": 1182
    },
    {
      "epoch": 0.008037712020770317,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009999999981480459,
      "loss": 2.6973,
      "step": 1183
    },
    {
      "epoch": 0.008044506367364375,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0009999999971063217,
      "loss": 2.7354,
      "step": 1184
    },
    {
      "epoch": 0.008051300713958433,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0009999999958331032,
      "loss": 2.7253,
      "step": 1185
    },
    {
      "epoch": 0.008058095060552491,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999999943283904,
      "loss": 2.6418,
      "step": 1186
    },
    {
      "epoch": 0.008064889407146548,
      "grad_norm": 0.25,
      "learning_rate": 0.0009999999925921835,
      "loss": 2.6612,
      "step": 1187
    },
    {
      "epoch": 0.008071683753740606,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.0009999999906244823,
      "loss": 2.736,
      "step": 1188
    },
    {
      "epoch": 0.008078478100334664,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0009999999884252866,
      "loss": 2.6955,
      "step": 1189
    },
    {
      "epoch": 0.008085272446928722,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.000999999985994597,
      "loss": 2.5635,
      "step": 1190
    },
    {
      "epoch": 0.008092066793522779,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009999999833324128,
      "loss": 2.6143,
      "step": 1191
    },
    {
      "epoch": 0.008098861140116837,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999999804387344,
      "loss": 2.6205,
      "step": 1192
    },
    {
      "epoch": 0.008105655486710895,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.000999999977313562,
      "loss": 2.6701,
      "step": 1193
    },
    {
      "epoch": 0.008112449833304954,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.000999999973956895,
      "loss": 2.4727,
      "step": 1194
    },
    {
      "epoch": 0.00811924417989901,
      "grad_norm": 0.201171875,
      "learning_rate": 0.000999999970368734,
      "loss": 2.5972,
      "step": 1195
    },
    {
      "epoch": 0.008126038526493068,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999999665490787,
      "loss": 2.5731,
      "step": 1196
    },
    {
      "epoch": 0.008132832873087126,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009999999624979291,
      "loss": 2.7138,
      "step": 1197
    },
    {
      "epoch": 0.008139627219681183,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009999999582152852,
      "loss": 2.6471,
      "step": 1198
    },
    {
      "epoch": 0.008146421566275241,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999999537011472,
      "loss": 2.6517,
      "step": 1199
    },
    {
      "epoch": 0.0081532159128693,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999999489555148,
      "loss": 2.5884,
      "step": 1200
    },
    {
      "epoch": 0.008160010259463358,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009999999439783882,
      "loss": 2.6182,
      "step": 1201
    },
    {
      "epoch": 0.008166804606057414,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999999387697675,
      "loss": 2.6514,
      "step": 1202
    },
    {
      "epoch": 0.008173598952651472,
      "grad_norm": 0.21875,
      "learning_rate": 0.0009999999333296524,
      "loss": 2.6717,
      "step": 1203
    },
    {
      "epoch": 0.00818039329924553,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999999276580432,
      "loss": 2.6518,
      "step": 1204
    },
    {
      "epoch": 0.008187187645839589,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999999217549396,
      "loss": 2.5004,
      "step": 1205
    },
    {
      "epoch": 0.008193981992433645,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0009999999156203418,
      "loss": 2.7035,
      "step": 1206
    },
    {
      "epoch": 0.008200776339027703,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009999999092542497,
      "loss": 2.6241,
      "step": 1207
    },
    {
      "epoch": 0.008207570685621762,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999999026566635,
      "loss": 2.667,
      "step": 1208
    },
    {
      "epoch": 0.00821436503221582,
      "grad_norm": 0.208984375,
      "learning_rate": 0.000999999895827583,
      "loss": 2.8096,
      "step": 1209
    },
    {
      "epoch": 0.008221159378809876,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999998887670085,
      "loss": 2.6645,
      "step": 1210
    },
    {
      "epoch": 0.008227953725403934,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999998814749396,
      "loss": 2.6407,
      "step": 1211
    },
    {
      "epoch": 0.008234748071997993,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999998739513764,
      "loss": 2.4686,
      "step": 1212
    },
    {
      "epoch": 0.00824154241859205,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009999998661963192,
      "loss": 2.6302,
      "step": 1213
    },
    {
      "epoch": 0.008248336765186107,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.000999999858209768,
      "loss": 2.6096,
      "step": 1214
    },
    {
      "epoch": 0.008255131111780166,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.000999999849991722,
      "loss": 2.7709,
      "step": 1215
    },
    {
      "epoch": 0.008261925458374224,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000999999841542182,
      "loss": 2.5897,
      "step": 1216
    },
    {
      "epoch": 0.00826871980496828,
      "grad_norm": 0.236328125,
      "learning_rate": 0.000999999832861148,
      "loss": 2.6768,
      "step": 1217
    },
    {
      "epoch": 0.008275514151562339,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0009999998239486197,
      "loss": 2.6248,
      "step": 1218
    },
    {
      "epoch": 0.008282308498156397,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009999998148045972,
      "loss": 2.8092,
      "step": 1219
    },
    {
      "epoch": 0.008289102844750455,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0009999998054290807,
      "loss": 2.7682,
      "step": 1220
    },
    {
      "epoch": 0.008295897191344511,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0009999997958220696,
      "loss": 2.711,
      "step": 1221
    },
    {
      "epoch": 0.00830269153793857,
      "grad_norm": 0.2265625,
      "learning_rate": 0.0009999997859835648,
      "loss": 2.7354,
      "step": 1222
    },
    {
      "epoch": 0.008309485884532628,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0009999997759135655,
      "loss": 2.6014,
      "step": 1223
    },
    {
      "epoch": 0.008316280231126686,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999997656120723,
      "loss": 2.6217,
      "step": 1224
    },
    {
      "epoch": 0.008323074577720743,
      "grad_norm": 0.228515625,
      "learning_rate": 0.0009999997550790847,
      "loss": 2.8207,
      "step": 1225
    },
    {
      "epoch": 0.0083298689243148,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999997443146031,
      "loss": 2.6133,
      "step": 1226
    },
    {
      "epoch": 0.008336663270908859,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999997333186273,
      "loss": 2.5856,
      "step": 1227
    },
    {
      "epoch": 0.008343457617502915,
      "grad_norm": 0.216796875,
      "learning_rate": 0.0009999997220911574,
      "loss": 2.7209,
      "step": 1228
    },
    {
      "epoch": 0.008350251964096974,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999997106321932,
      "loss": 2.4977,
      "step": 1229
    },
    {
      "epoch": 0.008357046310691032,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009999996989417351,
      "loss": 2.8688,
      "step": 1230
    },
    {
      "epoch": 0.00836384065728509,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999996870197828,
      "loss": 2.6709,
      "step": 1231
    },
    {
      "epoch": 0.008370635003879147,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0009999996748663363,
      "loss": 2.597,
      "step": 1232
    },
    {
      "epoch": 0.008377429350473205,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999996624813956,
      "loss": 2.8106,
      "step": 1233
    },
    {
      "epoch": 0.008384223697067263,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009999996498649608,
      "loss": 2.7618,
      "step": 1234
    },
    {
      "epoch": 0.008391018043661321,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009999996370170321,
      "loss": 2.601,
      "step": 1235
    },
    {
      "epoch": 0.008397812390255378,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0009999996239376092,
      "loss": 2.5235,
      "step": 1236
    },
    {
      "epoch": 0.008404606736849436,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009999996106266921,
      "loss": 2.743,
      "step": 1237
    },
    {
      "epoch": 0.008411401083443494,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0009999995970842812,
      "loss": 2.7665,
      "step": 1238
    },
    {
      "epoch": 0.008418195430037552,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999995833103758,
      "loss": 2.6228,
      "step": 1239
    },
    {
      "epoch": 0.008424989776631609,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009999995693049768,
      "loss": 2.6866,
      "step": 1240
    },
    {
      "epoch": 0.008431784123225667,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999995550680834,
      "loss": 2.8104,
      "step": 1241
    },
    {
      "epoch": 0.008438578469819725,
      "grad_norm": 0.193359375,
      "learning_rate": 0.000999999540599696,
      "loss": 2.6741,
      "step": 1242
    },
    {
      "epoch": 0.008445372816413782,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999995258998145,
      "loss": 2.6848,
      "step": 1243
    },
    {
      "epoch": 0.00845216716300784,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009999995109684391,
      "loss": 2.6833,
      "step": 1244
    },
    {
      "epoch": 0.008458961509601898,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999994958055697,
      "loss": 2.4429,
      "step": 1245
    },
    {
      "epoch": 0.008465755856195956,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000999999480411206,
      "loss": 2.6409,
      "step": 1246
    },
    {
      "epoch": 0.008472550202790013,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0009999994647853483,
      "loss": 2.6923,
      "step": 1247
    },
    {
      "epoch": 0.008479344549384071,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999994489279969,
      "loss": 2.7318,
      "step": 1248
    },
    {
      "epoch": 0.00848613889597813,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0009999994328391514,
      "loss": 2.8109,
      "step": 1249
    },
    {
      "epoch": 0.008492933242572187,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009999994165188115,
      "loss": 2.6565,
      "step": 1250
    },
    {
      "epoch": 0.008499727589166244,
      "grad_norm": 0.197265625,
      "learning_rate": 0.000999999399966978,
      "loss": 2.7442,
      "step": 1251
    },
    {
      "epoch": 0.008506521935760302,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999993831836505,
      "loss": 2.6301,
      "step": 1252
    },
    {
      "epoch": 0.00851331628235436,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009999993661688289,
      "loss": 2.7996,
      "step": 1253
    },
    {
      "epoch": 0.008520110628948419,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999993489225132,
      "loss": 2.4962,
      "step": 1254
    },
    {
      "epoch": 0.008526904975542475,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999993314447038,
      "loss": 2.6769,
      "step": 1255
    },
    {
      "epoch": 0.008533699322136533,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999993137354002,
      "loss": 2.6614,
      "step": 1256
    },
    {
      "epoch": 0.008540493668730591,
      "grad_norm": 0.1953125,
      "learning_rate": 0.000999999295794603,
      "loss": 2.5062,
      "step": 1257
    },
    {
      "epoch": 0.008547288015324648,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009999992776223115,
      "loss": 2.8139,
      "step": 1258
    },
    {
      "epoch": 0.008554082361918706,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000999999259218526,
      "loss": 2.5848,
      "step": 1259
    },
    {
      "epoch": 0.008560876708512764,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999992405832468,
      "loss": 2.5991,
      "step": 1260
    },
    {
      "epoch": 0.008567671055106823,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999992217164736,
      "loss": 2.5394,
      "step": 1261
    },
    {
      "epoch": 0.008574465401700879,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999992026182066,
      "loss": 2.6085,
      "step": 1262
    },
    {
      "epoch": 0.008581259748294937,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999991832884457,
      "loss": 2.7391,
      "step": 1263
    },
    {
      "epoch": 0.008588054094888995,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009999991637271908,
      "loss": 2.5626,
      "step": 1264
    },
    {
      "epoch": 0.008594848441483054,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999991439344422,
      "loss": 2.7046,
      "step": 1265
    },
    {
      "epoch": 0.00860164278807711,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999991239101995,
      "loss": 2.5995,
      "step": 1266
    },
    {
      "epoch": 0.008608437134671168,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999991036544632,
      "loss": 2.7097,
      "step": 1267
    },
    {
      "epoch": 0.008615231481265227,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999990831672328,
      "loss": 2.7318,
      "step": 1268
    },
    {
      "epoch": 0.008622025827859285,
      "grad_norm": 0.322265625,
      "learning_rate": 0.0009999990624485085,
      "loss": 2.6461,
      "step": 1269
    },
    {
      "epoch": 0.008628820174453341,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009999990414982906,
      "loss": 2.7408,
      "step": 1270
    },
    {
      "epoch": 0.0086356145210474,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0009999990203165788,
      "loss": 2.6393,
      "step": 1271
    },
    {
      "epoch": 0.008642408867641458,
      "grad_norm": 0.20703125,
      "learning_rate": 0.000999998998903373,
      "loss": 2.7123,
      "step": 1272
    },
    {
      "epoch": 0.008649203214235514,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0009999989772586737,
      "loss": 2.7546,
      "step": 1273
    },
    {
      "epoch": 0.008655997560829572,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0009999989553824804,
      "loss": 2.7215,
      "step": 1274
    },
    {
      "epoch": 0.00866279190742363,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009999989332747934,
      "loss": 2.5769,
      "step": 1275
    },
    {
      "epoch": 0.008669586254017689,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009999989109356127,
      "loss": 2.6188,
      "step": 1276
    },
    {
      "epoch": 0.008676380600611745,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000999998888364938,
      "loss": 2.6119,
      "step": 1277
    },
    {
      "epoch": 0.008683174947205803,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999988655627699,
      "loss": 2.6035,
      "step": 1278
    },
    {
      "epoch": 0.008689969293799862,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009999988425291079,
      "loss": 2.6146,
      "step": 1279
    },
    {
      "epoch": 0.00869676364039392,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.000999998819263952,
      "loss": 2.5456,
      "step": 1280
    },
    {
      "epoch": 0.008703557986987976,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999987957673025,
      "loss": 2.4306,
      "step": 1281
    },
    {
      "epoch": 0.008710352333582035,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009999987720391594,
      "loss": 2.6211,
      "step": 1282
    },
    {
      "epoch": 0.008717146680176093,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999987480795224,
      "loss": 2.6405,
      "step": 1283
    },
    {
      "epoch": 0.008723941026770151,
      "grad_norm": 0.1875,
      "learning_rate": 0.000999998723888392,
      "loss": 2.565,
      "step": 1284
    },
    {
      "epoch": 0.008730735373364207,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999986994657674,
      "loss": 2.6301,
      "step": 1285
    },
    {
      "epoch": 0.008737529719958266,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999986748116497,
      "loss": 2.6354,
      "step": 1286
    },
    {
      "epoch": 0.008744324066552324,
      "grad_norm": 0.228515625,
      "learning_rate": 0.000999998649926038,
      "loss": 2.6598,
      "step": 1287
    },
    {
      "epoch": 0.00875111841314638,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999986248089329,
      "loss": 2.488,
      "step": 1288
    },
    {
      "epoch": 0.008757912759740439,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.000999998599460334,
      "loss": 2.5885,
      "step": 1289
    },
    {
      "epoch": 0.008764707106334497,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0009999985738802415,
      "loss": 2.6191,
      "step": 1290
    },
    {
      "epoch": 0.008771501452928555,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999985480686553,
      "loss": 2.6101,
      "step": 1291
    },
    {
      "epoch": 0.008778295799522612,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009999985220255755,
      "loss": 2.6958,
      "step": 1292
    },
    {
      "epoch": 0.00878509014611667,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009999984957510025,
      "loss": 2.7305,
      "step": 1293
    },
    {
      "epoch": 0.008791884492710728,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999984692449354,
      "loss": 2.7062,
      "step": 1294
    },
    {
      "epoch": 0.008798678839304786,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999984425073751,
      "loss": 2.6528,
      "step": 1295
    },
    {
      "epoch": 0.008805473185898843,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009999984155383212,
      "loss": 2.6798,
      "step": 1296
    },
    {
      "epoch": 0.0088122675324929,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999983883377736,
      "loss": 2.6871,
      "step": 1297
    },
    {
      "epoch": 0.008819061879086959,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999983609057328,
      "loss": 2.4891,
      "step": 1298
    },
    {
      "epoch": 0.008825856225681017,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999983332421981,
      "loss": 2.5725,
      "step": 1299
    },
    {
      "epoch": 0.008832650572275074,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999983053471703,
      "loss": 2.7012,
      "step": 1300
    },
    {
      "epoch": 0.008839444918869132,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999982772206486,
      "loss": 2.6216,
      "step": 1301
    },
    {
      "epoch": 0.00884623926546319,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999982488626338,
      "loss": 2.6059,
      "step": 1302
    },
    {
      "epoch": 0.008853033612057247,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999982202731253,
      "loss": 2.5803,
      "step": 1303
    },
    {
      "epoch": 0.008859827958651305,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999981914521235,
      "loss": 2.5786,
      "step": 1304
    },
    {
      "epoch": 0.008866622305245363,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999981623996282,
      "loss": 2.6712,
      "step": 1305
    },
    {
      "epoch": 0.008873416651839421,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999981331156394,
      "loss": 2.6231,
      "step": 1306
    },
    {
      "epoch": 0.008880210998433478,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999981036001573,
      "loss": 2.6122,
      "step": 1307
    },
    {
      "epoch": 0.008887005345027536,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009999980738531818,
      "loss": 2.7224,
      "step": 1308
    },
    {
      "epoch": 0.008893799691621594,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999980438747129,
      "loss": 2.7616,
      "step": 1309
    },
    {
      "epoch": 0.008900594038215652,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999980136647505,
      "loss": 2.6591,
      "step": 1310
    },
    {
      "epoch": 0.008907388384809709,
      "grad_norm": 0.18359375,
      "learning_rate": 0.000999997983223295,
      "loss": 2.6841,
      "step": 1311
    },
    {
      "epoch": 0.008914182731403767,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999979525503462,
      "loss": 2.6657,
      "step": 1312
    },
    {
      "epoch": 0.008920977077997825,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999979216459038,
      "loss": 2.6621,
      "step": 1313
    },
    {
      "epoch": 0.008927771424591883,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999978905099684,
      "loss": 2.6478,
      "step": 1314
    },
    {
      "epoch": 0.00893456577118594,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999978591425395,
      "loss": 2.7414,
      "step": 1315
    },
    {
      "epoch": 0.008941360117779998,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999978275436175,
      "loss": 2.7082,
      "step": 1316
    },
    {
      "epoch": 0.008948154464374056,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.000999997795713202,
      "loss": 2.6316,
      "step": 1317
    },
    {
      "epoch": 0.008954948810968113,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999977636512933,
      "loss": 2.6413,
      "step": 1318
    },
    {
      "epoch": 0.008961743157562171,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999977313578916,
      "loss": 2.758,
      "step": 1319
    },
    {
      "epoch": 0.00896853750415623,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999976988329965,
      "loss": 2.5566,
      "step": 1320
    },
    {
      "epoch": 0.008975331850750288,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999976660766082,
      "loss": 2.661,
      "step": 1321
    },
    {
      "epoch": 0.008982126197344344,
      "grad_norm": 0.201171875,
      "learning_rate": 0.000999997633088727,
      "loss": 2.5687,
      "step": 1322
    },
    {
      "epoch": 0.008988920543938402,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999975998693522,
      "loss": 2.6194,
      "step": 1323
    },
    {
      "epoch": 0.00899571489053246,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999975664184845,
      "loss": 2.7156,
      "step": 1324
    },
    {
      "epoch": 0.009002509237126519,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999975327361233,
      "loss": 2.6819,
      "step": 1325
    },
    {
      "epoch": 0.009009303583720575,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999974988222694,
      "loss": 2.703,
      "step": 1326
    },
    {
      "epoch": 0.009016097930314633,
      "grad_norm": 0.1875,
      "learning_rate": 0.000999997464676922,
      "loss": 2.6602,
      "step": 1327
    },
    {
      "epoch": 0.009022892276908692,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999974303000819,
      "loss": 2.5719,
      "step": 1328
    },
    {
      "epoch": 0.00902968662350275,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999973956917485,
      "loss": 2.4238,
      "step": 1329
    },
    {
      "epoch": 0.009036480970096806,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999973608519222,
      "loss": 2.6276,
      "step": 1330
    },
    {
      "epoch": 0.009043275316690864,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999973257806027,
      "loss": 2.6354,
      "step": 1331
    },
    {
      "epoch": 0.009050069663284923,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999972904777901,
      "loss": 2.558,
      "step": 1332
    },
    {
      "epoch": 0.009056864009878979,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999972549434848,
      "loss": 2.6362,
      "step": 1333
    },
    {
      "epoch": 0.009063658356473037,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999972191776863,
      "loss": 2.5223,
      "step": 1334
    },
    {
      "epoch": 0.009070452703067096,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999971831803948,
      "loss": 2.6644,
      "step": 1335
    },
    {
      "epoch": 0.009077247049661154,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009999971469516103,
      "loss": 2.7577,
      "step": 1336
    },
    {
      "epoch": 0.00908404139625521,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999997110491333,
      "loss": 2.671,
      "step": 1337
    },
    {
      "epoch": 0.009090835742849268,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999970737995627,
      "loss": 2.4956,
      "step": 1338
    },
    {
      "epoch": 0.009097630089443327,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999970368762994,
      "loss": 2.6131,
      "step": 1339
    },
    {
      "epoch": 0.009104424436037385,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999969997215431,
      "loss": 2.5994,
      "step": 1340
    },
    {
      "epoch": 0.009111218782631441,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009999969623352943,
      "loss": 2.6795,
      "step": 1341
    },
    {
      "epoch": 0.0091180131292255,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999969247175525,
      "loss": 2.6131,
      "step": 1342
    },
    {
      "epoch": 0.009124807475819558,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999968868683179,
      "loss": 2.6338,
      "step": 1343
    },
    {
      "epoch": 0.009131601822413616,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999968487875903,
      "loss": 2.6028,
      "step": 1344
    },
    {
      "epoch": 0.009138396169007672,
      "grad_norm": 0.1796875,
      "learning_rate": 0.00099999681047537,
      "loss": 2.5791,
      "step": 1345
    },
    {
      "epoch": 0.00914519051560173,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999967719316568,
      "loss": 2.6012,
      "step": 1346
    },
    {
      "epoch": 0.009151984862195789,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999967331564509,
      "loss": 2.6089,
      "step": 1347
    },
    {
      "epoch": 0.009158779208789845,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999966941497522,
      "loss": 2.6714,
      "step": 1348
    },
    {
      "epoch": 0.009165573555383904,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999966549115607,
      "loss": 2.6465,
      "step": 1349
    },
    {
      "epoch": 0.009172367901977962,
      "grad_norm": 0.2138671875,
      "learning_rate": 0.0009999966154418767,
      "loss": 2.7738,
      "step": 1350
    },
    {
      "epoch": 0.00917916224857202,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999965757406999,
      "loss": 2.532,
      "step": 1351
    },
    {
      "epoch": 0.009185956595166076,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009999965358080303,
      "loss": 2.7543,
      "step": 1352
    },
    {
      "epoch": 0.009192750941760135,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999964956438681,
      "loss": 2.5967,
      "step": 1353
    },
    {
      "epoch": 0.009199545288354193,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0009999964552482134,
      "loss": 2.5745,
      "step": 1354
    },
    {
      "epoch": 0.009206339634948251,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.000999996414621066,
      "loss": 2.5855,
      "step": 1355
    },
    {
      "epoch": 0.009213133981542308,
      "grad_norm": 0.205078125,
      "learning_rate": 0.000999996373762426,
      "loss": 2.6058,
      "step": 1356
    },
    {
      "epoch": 0.009219928328136366,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999963326722933,
      "loss": 2.6957,
      "step": 1357
    },
    {
      "epoch": 0.009226722674730424,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999962913506682,
      "loss": 2.5993,
      "step": 1358
    },
    {
      "epoch": 0.009233517021324482,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0009999962497975502,
      "loss": 2.6196,
      "step": 1359
    },
    {
      "epoch": 0.009240311367918539,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999962080129402,
      "loss": 2.6137,
      "step": 1360
    },
    {
      "epoch": 0.009247105714512597,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999961659968373,
      "loss": 2.6028,
      "step": 1361
    },
    {
      "epoch": 0.009253900061106655,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.000999996123749242,
      "loss": 2.6131,
      "step": 1362
    },
    {
      "epoch": 0.009260694407700712,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.0009999960812701542,
      "loss": 2.6163,
      "step": 1363
    },
    {
      "epoch": 0.00926748875429477,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.000999996038559574,
      "loss": 2.5883,
      "step": 1364
    },
    {
      "epoch": 0.009274283100888828,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999959956175013,
      "loss": 2.5693,
      "step": 1365
    },
    {
      "epoch": 0.009281077447482886,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999959524439363,
      "loss": 2.6366,
      "step": 1366
    },
    {
      "epoch": 0.009287871794076943,
      "grad_norm": 0.212890625,
      "learning_rate": 0.0009999959090388787,
      "loss": 2.838,
      "step": 1367
    },
    {
      "epoch": 0.009294666140671001,
      "grad_norm": 0.185546875,
      "learning_rate": 0.000999995865402329,
      "loss": 2.8007,
      "step": 1368
    },
    {
      "epoch": 0.00930146048726506,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999958215342868,
      "loss": 2.7708,
      "step": 1369
    },
    {
      "epoch": 0.009308254833859117,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009999957774347523,
      "loss": 2.6914,
      "step": 1370
    },
    {
      "epoch": 0.009315049180453174,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999957331037256,
      "loss": 2.5633,
      "step": 1371
    },
    {
      "epoch": 0.009321843527047232,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999956885412063,
      "loss": 2.4429,
      "step": 1372
    },
    {
      "epoch": 0.00932863787364129,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000999995643747195,
      "loss": 2.7808,
      "step": 1373
    },
    {
      "epoch": 0.009335432220235348,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999955987216914,
      "loss": 2.5848,
      "step": 1374
    },
    {
      "epoch": 0.009342226566829405,
      "grad_norm": 0.2421875,
      "learning_rate": 0.0009999955534646956,
      "loss": 2.6134,
      "step": 1375
    },
    {
      "epoch": 0.009349020913423463,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999955079762074,
      "loss": 2.6333,
      "step": 1376
    },
    {
      "epoch": 0.009355815260017521,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999954622562273,
      "loss": 2.7224,
      "step": 1377
    },
    {
      "epoch": 0.009362609606611578,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999954163047546,
      "loss": 2.6098,
      "step": 1378
    },
    {
      "epoch": 0.009369403953205636,
      "grad_norm": 0.19140625,
      "learning_rate": 0.00099999537012179,
      "loss": 2.6959,
      "step": 1379
    },
    {
      "epoch": 0.009376198299799694,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999953237073333,
      "loss": 2.5268,
      "step": 1380
    },
    {
      "epoch": 0.009382992646393752,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999952770613845,
      "loss": 2.6843,
      "step": 1381
    },
    {
      "epoch": 0.009389786992987809,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999952301839436,
      "loss": 2.7754,
      "step": 1382
    },
    {
      "epoch": 0.009396581339581867,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999951830750107,
      "loss": 2.5517,
      "step": 1383
    },
    {
      "epoch": 0.009403375686175925,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999951357345857,
      "loss": 2.5479,
      "step": 1384
    },
    {
      "epoch": 0.009410170032769984,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999950881626688,
      "loss": 2.5469,
      "step": 1385
    },
    {
      "epoch": 0.00941696437936404,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999950403592598,
      "loss": 2.6148,
      "step": 1386
    },
    {
      "epoch": 0.009423758725958098,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009999949923243587,
      "loss": 2.4499,
      "step": 1387
    },
    {
      "epoch": 0.009430553072552156,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999949440579658,
      "loss": 2.5314,
      "step": 1388
    },
    {
      "epoch": 0.009437347419146215,
      "grad_norm": 0.19140625,
      "learning_rate": 0.000999994895560081,
      "loss": 2.5671,
      "step": 1389
    },
    {
      "epoch": 0.009444141765740271,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999948468307042,
      "loss": 2.7821,
      "step": 1390
    },
    {
      "epoch": 0.00945093611233433,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999947978698357,
      "loss": 2.4877,
      "step": 1391
    },
    {
      "epoch": 0.009457730458928388,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999947486774752,
      "loss": 2.5869,
      "step": 1392
    },
    {
      "epoch": 0.009464524805522444,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999946992536228,
      "loss": 2.5932,
      "step": 1393
    },
    {
      "epoch": 0.009471319152116502,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999946495982787,
      "loss": 2.5115,
      "step": 1394
    },
    {
      "epoch": 0.00947811349871056,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999945997114428,
      "loss": 2.5907,
      "step": 1395
    },
    {
      "epoch": 0.009484907845304619,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999994549593115,
      "loss": 2.653,
      "step": 1396
    },
    {
      "epoch": 0.009491702191898675,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999944992432957,
      "loss": 2.5073,
      "step": 1397
    },
    {
      "epoch": 0.009498496538492733,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999944486619844,
      "loss": 2.5504,
      "step": 1398
    },
    {
      "epoch": 0.009505290885086792,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999943978491817,
      "loss": 2.6157,
      "step": 1399
    },
    {
      "epoch": 0.00951208523168085,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000999994346804887,
      "loss": 2.6276,
      "step": 1400
    },
    {
      "epoch": 0.009518879578274906,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999994295529101,
      "loss": 2.6013,
      "step": 1401
    },
    {
      "epoch": 0.009525673924868965,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999994244021823,
      "loss": 2.4835,
      "step": 1402
    },
    {
      "epoch": 0.009532468271463023,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999941922830538,
      "loss": 2.6499,
      "step": 1403
    },
    {
      "epoch": 0.009539262618057081,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999941403127928,
      "loss": 2.5643,
      "step": 1404
    },
    {
      "epoch": 0.009546056964651137,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999940881110403,
      "loss": 2.6376,
      "step": 1405
    },
    {
      "epoch": 0.009552851311245196,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999940356777961,
      "loss": 2.6133,
      "step": 1406
    },
    {
      "epoch": 0.009559645657839254,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999939830130605,
      "loss": 2.6345,
      "step": 1407
    },
    {
      "epoch": 0.00956644000443331,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999939301168334,
      "loss": 2.6395,
      "step": 1408
    },
    {
      "epoch": 0.009573234351027369,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000999993876989115,
      "loss": 2.596,
      "step": 1409
    },
    {
      "epoch": 0.009580028697621427,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999993823629905,
      "loss": 2.5809,
      "step": 1410
    },
    {
      "epoch": 0.009586823044215485,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999937700392036,
      "loss": 2.4626,
      "step": 1411
    },
    {
      "epoch": 0.009593617390809541,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999937162170108,
      "loss": 2.7259,
      "step": 1412
    },
    {
      "epoch": 0.0096004117374036,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999936621633268,
      "loss": 2.6005,
      "step": 1413
    },
    {
      "epoch": 0.009607206083997658,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999936078781513,
      "loss": 2.6218,
      "step": 1414
    },
    {
      "epoch": 0.009614000430591716,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999935533614845,
      "loss": 2.4652,
      "step": 1415
    },
    {
      "epoch": 0.009620794777185773,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999934986133266,
      "loss": 2.6585,
      "step": 1416
    },
    {
      "epoch": 0.00962758912377983,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999934436336773,
      "loss": 2.5986,
      "step": 1417
    },
    {
      "epoch": 0.009634383470373889,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999933884225368,
      "loss": 2.7393,
      "step": 1418
    },
    {
      "epoch": 0.009641177816967947,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999993332979905,
      "loss": 2.5973,
      "step": 1419
    },
    {
      "epoch": 0.009647972163562004,
      "grad_norm": 0.19140625,
      "learning_rate": 0.000999993277305782,
      "loss": 2.7175,
      "step": 1420
    },
    {
      "epoch": 0.009654766510156062,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999932214001682,
      "loss": 2.5846,
      "step": 1421
    },
    {
      "epoch": 0.00966156085675012,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999931652630628,
      "loss": 2.6728,
      "step": 1422
    },
    {
      "epoch": 0.009668355203344177,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999931088944664,
      "loss": 2.5936,
      "step": 1423
    },
    {
      "epoch": 0.009675149549938235,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999930522943791,
      "loss": 2.5828,
      "step": 1424
    },
    {
      "epoch": 0.009681943896532293,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999929954628008,
      "loss": 2.5089,
      "step": 1425
    },
    {
      "epoch": 0.009688738243126351,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999929383997313,
      "loss": 2.5679,
      "step": 1426
    },
    {
      "epoch": 0.009695532589720408,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.000999992881105171,
      "loss": 2.5707,
      "step": 1427
    },
    {
      "epoch": 0.009702326936314466,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999928235791195,
      "loss": 2.4736,
      "step": 1428
    },
    {
      "epoch": 0.009709121282908524,
      "grad_norm": 0.193359375,
      "learning_rate": 0.000999992765821577,
      "loss": 2.679,
      "step": 1429
    },
    {
      "epoch": 0.009715915629502582,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009999927078325438,
      "loss": 2.6136,
      "step": 1430
    },
    {
      "epoch": 0.009722709976096639,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999926496120197,
      "loss": 2.6456,
      "step": 1431
    },
    {
      "epoch": 0.009729504322690697,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009999925911600045,
      "loss": 2.573,
      "step": 1432
    },
    {
      "epoch": 0.009736298669284755,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999925324764988,
      "loss": 2.5251,
      "step": 1433
    },
    {
      "epoch": 0.009743093015878813,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999992473561502,
      "loss": 2.7339,
      "step": 1434
    },
    {
      "epoch": 0.00974988736247287,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999924144150146,
      "loss": 2.6411,
      "step": 1435
    },
    {
      "epoch": 0.009756681709066928,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999923550370363,
      "loss": 2.7328,
      "step": 1436
    },
    {
      "epoch": 0.009763476055660986,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999922954275673,
      "loss": 2.5196,
      "step": 1437
    },
    {
      "epoch": 0.009770270402255043,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999922355866078,
      "loss": 2.5892,
      "step": 1438
    },
    {
      "epoch": 0.009777064748849101,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999921755141574,
      "loss": 2.5445,
      "step": 1439
    },
    {
      "epoch": 0.00978385909544316,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999921152102165,
      "loss": 2.6302,
      "step": 1440
    },
    {
      "epoch": 0.009790653442037217,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999920546747848,
      "loss": 2.5785,
      "step": 1441
    },
    {
      "epoch": 0.009797447788631274,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999919939078627,
      "loss": 2.5942,
      "step": 1442
    },
    {
      "epoch": 0.009804242135225332,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00099999193290945,
      "loss": 2.6712,
      "step": 1443
    },
    {
      "epoch": 0.00981103648181939,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999918716795467,
      "loss": 2.5345,
      "step": 1444
    },
    {
      "epoch": 0.009817830828413449,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999991810218153,
      "loss": 2.4568,
      "step": 1445
    },
    {
      "epoch": 0.009824625175007505,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999917485252687,
      "loss": 2.6127,
      "step": 1446
    },
    {
      "epoch": 0.009831419521601563,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.000999991686600894,
      "loss": 2.6645,
      "step": 1447
    },
    {
      "epoch": 0.009838213868195621,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999916244450289,
      "loss": 2.53,
      "step": 1448
    },
    {
      "epoch": 0.00984500821478968,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999915620576732,
      "loss": 2.6127,
      "step": 1449
    },
    {
      "epoch": 0.009851802561383736,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009999914994388275,
      "loss": 2.5425,
      "step": 1450
    },
    {
      "epoch": 0.009858596907977794,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999914365884913,
      "loss": 2.5596,
      "step": 1451
    },
    {
      "epoch": 0.009865391254571853,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999913735066646,
      "loss": 2.4026,
      "step": 1452
    },
    {
      "epoch": 0.009872185601165909,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000999991310193348,
      "loss": 2.5632,
      "step": 1453
    },
    {
      "epoch": 0.009878979947759967,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999991246648541,
      "loss": 2.5897,
      "step": 1454
    },
    {
      "epoch": 0.009885774294354025,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999911828722436,
      "loss": 2.7136,
      "step": 1455
    },
    {
      "epoch": 0.009892568640948084,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999911188644562,
      "loss": 2.5,
      "step": 1456
    },
    {
      "epoch": 0.00989936298754214,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999910546251787,
      "loss": 2.7419,
      "step": 1457
    },
    {
      "epoch": 0.009906157334136198,
      "grad_norm": 0.189453125,
      "learning_rate": 0.000999990990154411,
      "loss": 2.7433,
      "step": 1458
    },
    {
      "epoch": 0.009912951680730257,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999909254521531,
      "loss": 2.693,
      "step": 1459
    },
    {
      "epoch": 0.009919746027324315,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999908605184053,
      "loss": 2.529,
      "step": 1460
    },
    {
      "epoch": 0.009926540373918371,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999907953531673,
      "loss": 2.6109,
      "step": 1461
    },
    {
      "epoch": 0.00993333472051243,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999907299564396,
      "loss": 2.5627,
      "step": 1462
    },
    {
      "epoch": 0.009940129067106488,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999906643282216,
      "loss": 2.6432,
      "step": 1463
    },
    {
      "epoch": 0.009946923413700546,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999905984685138,
      "loss": 2.6184,
      "step": 1464
    },
    {
      "epoch": 0.009953717760294602,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.000999990532377316,
      "loss": 2.5538,
      "step": 1465
    },
    {
      "epoch": 0.00996051210688866,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999904660546283,
      "loss": 2.5997,
      "step": 1466
    },
    {
      "epoch": 0.009967306453482719,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009999903995004508,
      "loss": 2.4242,
      "step": 1467
    },
    {
      "epoch": 0.009974100800076775,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999903327147835,
      "loss": 2.574,
      "step": 1468
    },
    {
      "epoch": 0.009980895146670834,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999902656976265,
      "loss": 2.338,
      "step": 1469
    },
    {
      "epoch": 0.009987689493264892,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999901984489796,
      "loss": 2.5988,
      "step": 1470
    },
    {
      "epoch": 0.00999448383985895,
      "grad_norm": 0.18359375,
      "learning_rate": 0.000999990130968843,
      "loss": 2.5543,
      "step": 1471
    },
    {
      "epoch": 0.010001278186453006,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999900632572168,
      "loss": 2.6342,
      "step": 1472
    },
    {
      "epoch": 0.010008072533047065,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999899953141006,
      "loss": 2.7288,
      "step": 1473
    },
    {
      "epoch": 0.010014866879641123,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009999899271394951,
      "loss": 2.6749,
      "step": 1474
    },
    {
      "epoch": 0.010021661226235181,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999898587333999,
      "loss": 2.486,
      "step": 1475
    },
    {
      "epoch": 0.010028455572829238,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999897900958153,
      "loss": 2.6731,
      "step": 1476
    },
    {
      "epoch": 0.010035249919423296,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999897212267408,
      "loss": 2.5908,
      "step": 1477
    },
    {
      "epoch": 0.010042044266017354,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999989652126177,
      "loss": 2.5756,
      "step": 1478
    },
    {
      "epoch": 0.010048838612611412,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000999989582794124,
      "loss": 2.6321,
      "step": 1479
    },
    {
      "epoch": 0.010055632959205469,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999989513230581,
      "loss": 2.5853,
      "step": 1480
    },
    {
      "epoch": 0.010062427305799527,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999989443435549,
      "loss": 2.4737,
      "step": 1481
    },
    {
      "epoch": 0.010069221652393585,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999893734090274,
      "loss": 2.5598,
      "step": 1482
    },
    {
      "epoch": 0.010076015998987642,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999893031510166,
      "loss": 2.5388,
      "step": 1483
    },
    {
      "epoch": 0.0100828103455817,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999892326615163,
      "loss": 2.6051,
      "step": 1484
    },
    {
      "epoch": 0.010089604692175758,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999891619405267,
      "loss": 2.4773,
      "step": 1485
    },
    {
      "epoch": 0.010096399038769816,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999890909880482,
      "loss": 2.7008,
      "step": 1486
    },
    {
      "epoch": 0.010103193385363873,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999890198040802,
      "loss": 2.6394,
      "step": 1487
    },
    {
      "epoch": 0.01010998773195793,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.000999988948388623,
      "loss": 2.5066,
      "step": 1488
    },
    {
      "epoch": 0.010116782078551989,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999888767416767,
      "loss": 2.5099,
      "step": 1489
    },
    {
      "epoch": 0.010123576425146047,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999888048632414,
      "loss": 2.5122,
      "step": 1490
    },
    {
      "epoch": 0.010130370771740104,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999887327533168,
      "loss": 2.5982,
      "step": 1491
    },
    {
      "epoch": 0.010137165118334162,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999886604119034,
      "loss": 2.4783,
      "step": 1492
    },
    {
      "epoch": 0.01014395946492822,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999885878390008,
      "loss": 2.664,
      "step": 1493
    },
    {
      "epoch": 0.010150753811522278,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999885150346094,
      "loss": 2.5704,
      "step": 1494
    },
    {
      "epoch": 0.010157548158116335,
      "grad_norm": 0.19921875,
      "learning_rate": 0.000999988441998729,
      "loss": 2.4098,
      "step": 1495
    },
    {
      "epoch": 0.010164342504710393,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999883687313593,
      "loss": 2.6404,
      "step": 1496
    },
    {
      "epoch": 0.010171136851304451,
      "grad_norm": 0.177734375,
      "learning_rate": 0.000999988295232501,
      "loss": 2.657,
      "step": 1497
    },
    {
      "epoch": 0.010177931197898508,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.000999988221502154,
      "loss": 2.6092,
      "step": 1498
    },
    {
      "epoch": 0.010184725544492566,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009999881475403181,
      "loss": 2.4532,
      "step": 1499
    },
    {
      "epoch": 0.010191519891086624,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999880733469932,
      "loss": 2.4432,
      "step": 1500
    },
    {
      "epoch": 0.010198314237680682,
      "grad_norm": 0.208984375,
      "learning_rate": 0.00099998799892218,
      "loss": 2.4648,
      "step": 1501
    },
    {
      "epoch": 0.010205108584274739,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999879242658775,
      "loss": 2.5954,
      "step": 1502
    },
    {
      "epoch": 0.010211902930868797,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999878493780867,
      "loss": 2.6679,
      "step": 1503
    },
    {
      "epoch": 0.010218697277462855,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999877742588072,
      "loss": 2.7564,
      "step": 1504
    },
    {
      "epoch": 0.010225491624056914,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999876989080392,
      "loss": 2.749,
      "step": 1505
    },
    {
      "epoch": 0.01023228597065097,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999876233257826,
      "loss": 2.5956,
      "step": 1506
    },
    {
      "epoch": 0.010239080317245028,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999875475120372,
      "loss": 2.692,
      "step": 1507
    },
    {
      "epoch": 0.010245874663839086,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999874714668037,
      "loss": 2.3688,
      "step": 1508
    },
    {
      "epoch": 0.010252669010433145,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999873951900813,
      "loss": 2.5104,
      "step": 1509
    },
    {
      "epoch": 0.010259463357027201,
      "grad_norm": 0.21875,
      "learning_rate": 0.0009999873186818707,
      "loss": 2.8246,
      "step": 1510
    },
    {
      "epoch": 0.01026625770362126,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999872419421717,
      "loss": 2.5879,
      "step": 1511
    },
    {
      "epoch": 0.010273052050215318,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999871649709845,
      "loss": 2.632,
      "step": 1512
    },
    {
      "epoch": 0.010279846396809374,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999870877683086,
      "loss": 2.6064,
      "step": 1513
    },
    {
      "epoch": 0.010286640743403432,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999870103341447,
      "loss": 2.5541,
      "step": 1514
    },
    {
      "epoch": 0.01029343508999749,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999869326684924,
      "loss": 2.5204,
      "step": 1515
    },
    {
      "epoch": 0.010300229436591549,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000999986854771352,
      "loss": 2.6496,
      "step": 1516
    },
    {
      "epoch": 0.010307023783185605,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999867766427232,
      "loss": 2.5582,
      "step": 1517
    },
    {
      "epoch": 0.010313818129779663,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999866982826065,
      "loss": 2.5626,
      "step": 1518
    },
    {
      "epoch": 0.010320612476373722,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999866196910017,
      "loss": 2.6853,
      "step": 1519
    },
    {
      "epoch": 0.01032740682296778,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999865408679084,
      "loss": 2.6069,
      "step": 1520
    },
    {
      "epoch": 0.010334201169561836,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999864618133276,
      "loss": 2.5395,
      "step": 1521
    },
    {
      "epoch": 0.010340995516155894,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999863825272586,
      "loss": 2.5225,
      "step": 1522
    },
    {
      "epoch": 0.010347789862749953,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999863030097015,
      "loss": 2.5426,
      "step": 1523
    },
    {
      "epoch": 0.010354584209344011,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999862232606567,
      "loss": 2.6119,
      "step": 1524
    },
    {
      "epoch": 0.010361378555938067,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999986143280124,
      "loss": 2.4472,
      "step": 1525
    },
    {
      "epoch": 0.010368172902532126,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999860630681033,
      "loss": 2.6942,
      "step": 1526
    },
    {
      "epoch": 0.010374967249126184,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999859826245949,
      "loss": 2.5684,
      "step": 1527
    },
    {
      "epoch": 0.01038176159572024,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009999859019495987,
      "loss": 2.6466,
      "step": 1528
    },
    {
      "epoch": 0.010388555942314298,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999858210431145,
      "loss": 2.4227,
      "step": 1529
    },
    {
      "epoch": 0.010395350288908357,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000999985739905143,
      "loss": 2.4882,
      "step": 1530
    },
    {
      "epoch": 0.010402144635502415,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999856585356836,
      "loss": 2.5667,
      "step": 1531
    },
    {
      "epoch": 0.010408938982096471,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999855769347366,
      "loss": 2.5914,
      "step": 1532
    },
    {
      "epoch": 0.01041573332869053,
      "grad_norm": 0.19140625,
      "learning_rate": 0.000999985495102302,
      "loss": 2.5453,
      "step": 1533
    },
    {
      "epoch": 0.010422527675284588,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0009999854130383799,
      "loss": 2.6631,
      "step": 1534
    },
    {
      "epoch": 0.010429322021878646,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999853307429702,
      "loss": 2.6499,
      "step": 1535
    },
    {
      "epoch": 0.010436116368472702,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009999852482160732,
      "loss": 2.6192,
      "step": 1536
    },
    {
      "epoch": 0.01044291071506676,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999851654576887,
      "loss": 2.5463,
      "step": 1537
    },
    {
      "epoch": 0.010449705061660819,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999850824678168,
      "loss": 2.7272,
      "step": 1538
    },
    {
      "epoch": 0.010456499408254877,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999849992464573,
      "loss": 2.5266,
      "step": 1539
    },
    {
      "epoch": 0.010463293754848934,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999849157936106,
      "loss": 2.7379,
      "step": 1540
    },
    {
      "epoch": 0.010470088101442992,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999848321092769,
      "loss": 2.6026,
      "step": 1541
    },
    {
      "epoch": 0.01047688244803705,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999847481934557,
      "loss": 2.5857,
      "step": 1542
    },
    {
      "epoch": 0.010483676794631107,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999846640461475,
      "loss": 2.5987,
      "step": 1543
    },
    {
      "epoch": 0.010490471141225165,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999845796673519,
      "loss": 2.5728,
      "step": 1544
    },
    {
      "epoch": 0.010497265487819223,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999844950570691,
      "loss": 2.5499,
      "step": 1545
    },
    {
      "epoch": 0.010504059834413281,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999844102152994,
      "loss": 2.539,
      "step": 1546
    },
    {
      "epoch": 0.010510854181007338,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999843251420426,
      "loss": 2.488,
      "step": 1547
    },
    {
      "epoch": 0.010517648527601396,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999842398372989,
      "loss": 2.5638,
      "step": 1548
    },
    {
      "epoch": 0.010524442874195454,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999984154301068,
      "loss": 2.6054,
      "step": 1549
    },
    {
      "epoch": 0.010531237220789512,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999840685333504,
      "loss": 2.6737,
      "step": 1550
    },
    {
      "epoch": 0.010538031567383569,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999839825341457,
      "loss": 2.6337,
      "step": 1551
    },
    {
      "epoch": 0.010544825913977627,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999838963034543,
      "loss": 2.4496,
      "step": 1552
    },
    {
      "epoch": 0.010551620260571685,
      "grad_norm": 0.1875,
      "learning_rate": 0.000999983809841276,
      "loss": 2.7446,
      "step": 1553
    },
    {
      "epoch": 0.010558414607165743,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000999983723147611,
      "loss": 2.6631,
      "step": 1554
    },
    {
      "epoch": 0.0105652089537598,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999836362224593,
      "loss": 2.6023,
      "step": 1555
    },
    {
      "epoch": 0.010572003300353858,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999835490658209,
      "loss": 2.5839,
      "step": 1556
    },
    {
      "epoch": 0.010578797646947916,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999834616776957,
      "loss": 2.5996,
      "step": 1557
    },
    {
      "epoch": 0.010585591993541973,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999833740580838,
      "loss": 2.6576,
      "step": 1558
    },
    {
      "epoch": 0.010592386340136031,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999832862069857,
      "loss": 2.7733,
      "step": 1559
    },
    {
      "epoch": 0.01059918068673009,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999831981244009,
      "loss": 2.7646,
      "step": 1560
    },
    {
      "epoch": 0.010605975033324147,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999831098103295,
      "loss": 2.6365,
      "step": 1561
    },
    {
      "epoch": 0.010612769379918204,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999830212647717,
      "loss": 2.5735,
      "step": 1562
    },
    {
      "epoch": 0.010619563726512262,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999829324877276,
      "loss": 2.6208,
      "step": 1563
    },
    {
      "epoch": 0.01062635807310632,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999982843479197,
      "loss": 2.6328,
      "step": 1564
    },
    {
      "epoch": 0.010633152419700378,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999827542391802,
      "loss": 2.6072,
      "step": 1565
    },
    {
      "epoch": 0.010639946766294435,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000999982664767677,
      "loss": 2.5925,
      "step": 1566
    },
    {
      "epoch": 0.010646741112888493,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999825750646877,
      "loss": 2.6904,
      "step": 1567
    },
    {
      "epoch": 0.010653535459482551,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999824851302122,
      "loss": 2.481,
      "step": 1568
    },
    {
      "epoch": 0.01066032980607661,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009999823949642503,
      "loss": 2.7012,
      "step": 1569
    },
    {
      "epoch": 0.010667124152670666,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999823045668027,
      "loss": 2.5585,
      "step": 1570
    },
    {
      "epoch": 0.010673918499264724,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999822139378687,
      "loss": 2.4648,
      "step": 1571
    },
    {
      "epoch": 0.010680712845858783,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999821230774486,
      "loss": 2.5637,
      "step": 1572
    },
    {
      "epoch": 0.010687507192452839,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999820319855426,
      "loss": 2.5681,
      "step": 1573
    },
    {
      "epoch": 0.010694301539046897,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999819406621509,
      "loss": 2.4927,
      "step": 1574
    },
    {
      "epoch": 0.010701095885640955,
      "grad_norm": 0.18359375,
      "learning_rate": 0.000999981849107273,
      "loss": 2.7313,
      "step": 1575
    },
    {
      "epoch": 0.010707890232235014,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999817573209094,
      "loss": 2.6455,
      "step": 1576
    },
    {
      "epoch": 0.01071468457882907,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999816653030599,
      "loss": 2.5127,
      "step": 1577
    },
    {
      "epoch": 0.010721478925423128,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009999815730537245,
      "loss": 2.6397,
      "step": 1578
    },
    {
      "epoch": 0.010728273272017187,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999814805729036,
      "loss": 2.5851,
      "step": 1579
    },
    {
      "epoch": 0.010735067618611245,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999813878605968,
      "loss": 2.526,
      "step": 1580
    },
    {
      "epoch": 0.010741861965205301,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999812949168046,
      "loss": 2.5983,
      "step": 1581
    },
    {
      "epoch": 0.01074865631179936,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009999812017415265,
      "loss": 2.5034,
      "step": 1582
    },
    {
      "epoch": 0.010755450658393418,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999811083347629,
      "loss": 2.6934,
      "step": 1583
    },
    {
      "epoch": 0.010762245004987476,
      "grad_norm": 0.181640625,
      "learning_rate": 0.000999981014696514,
      "loss": 2.551,
      "step": 1584
    },
    {
      "epoch": 0.010769039351581532,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999809208267793,
      "loss": 2.5508,
      "step": 1585
    },
    {
      "epoch": 0.01077583369817559,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0009999808267255595,
      "loss": 2.6378,
      "step": 1586
    },
    {
      "epoch": 0.010782628044769649,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999807323928542,
      "loss": 2.6351,
      "step": 1587
    },
    {
      "epoch": 0.010789422391363705,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999806378286635,
      "loss": 2.6326,
      "step": 1588
    },
    {
      "epoch": 0.010796216737957763,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999805430329874,
      "loss": 2.5629,
      "step": 1589
    },
    {
      "epoch": 0.010803011084551822,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009999804480058262,
      "loss": 2.4566,
      "step": 1590
    },
    {
      "epoch": 0.01080980543114588,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999803527471797,
      "loss": 2.5675,
      "step": 1591
    },
    {
      "epoch": 0.010816599777739936,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.000999980257257048,
      "loss": 2.6124,
      "step": 1592
    },
    {
      "epoch": 0.010823394124333995,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999801615354314,
      "loss": 2.6167,
      "step": 1593
    },
    {
      "epoch": 0.010830188470928053,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999800655823294,
      "loss": 2.4413,
      "step": 1594
    },
    {
      "epoch": 0.010836982817522111,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999799693977424,
      "loss": 2.7878,
      "step": 1595
    },
    {
      "epoch": 0.010843777164116167,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999798729816707,
      "loss": 2.6053,
      "step": 1596
    },
    {
      "epoch": 0.010850571510710226,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999797763341138,
      "loss": 2.6456,
      "step": 1597
    },
    {
      "epoch": 0.010857365857304284,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999979679455072,
      "loss": 2.6491,
      "step": 1598
    },
    {
      "epoch": 0.010864160203898342,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999795823445454,
      "loss": 2.5794,
      "step": 1599
    },
    {
      "epoch": 0.010870954550492399,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999979485002534,
      "loss": 2.5805,
      "step": 1600
    },
    {
      "epoch": 0.010877748897086457,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999793874290379,
      "loss": 2.5503,
      "step": 1601
    },
    {
      "epoch": 0.010884543243680515,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.000999979289624057,
      "loss": 2.7102,
      "step": 1602
    },
    {
      "epoch": 0.010891337590274571,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999791915875913,
      "loss": 2.7267,
      "step": 1603
    },
    {
      "epoch": 0.01089813193686863,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999979093319641,
      "loss": 2.5849,
      "step": 1604
    },
    {
      "epoch": 0.010904926283462688,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999789948202064,
      "loss": 2.7263,
      "step": 1605
    },
    {
      "epoch": 0.010911720630056746,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.000999978896089287,
      "loss": 2.5117,
      "step": 1606
    },
    {
      "epoch": 0.010918514976650803,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999787971268833,
      "loss": 2.5215,
      "step": 1607
    },
    {
      "epoch": 0.01092530932324486,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999786979329947,
      "loss": 2.4686,
      "step": 1608
    },
    {
      "epoch": 0.010932103669838919,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999785985076223,
      "loss": 2.4355,
      "step": 1609
    },
    {
      "epoch": 0.010938898016432977,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999978498850765,
      "loss": 2.6854,
      "step": 1610
    },
    {
      "epoch": 0.010945692363027034,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999783989624238,
      "loss": 2.577,
      "step": 1611
    },
    {
      "epoch": 0.010952486709621092,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999782988425981,
      "loss": 2.6009,
      "step": 1612
    },
    {
      "epoch": 0.01095928105621515,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999781984912882,
      "loss": 2.6147,
      "step": 1613
    },
    {
      "epoch": 0.010966075402809208,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999780979084941,
      "loss": 2.5147,
      "step": 1614
    },
    {
      "epoch": 0.010972869749403265,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999977997094216,
      "loss": 2.5162,
      "step": 1615
    },
    {
      "epoch": 0.010979664095997323,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999778960484537,
      "loss": 2.5428,
      "step": 1616
    },
    {
      "epoch": 0.010986458442591381,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999777947712074,
      "loss": 2.664,
      "step": 1617
    },
    {
      "epoch": 0.010993252789185438,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999776932624771,
      "loss": 2.3497,
      "step": 1618
    },
    {
      "epoch": 0.011000047135779496,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999977591522263,
      "loss": 2.6667,
      "step": 1619
    },
    {
      "epoch": 0.011006841482373554,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999774895505648,
      "loss": 2.5757,
      "step": 1620
    },
    {
      "epoch": 0.011013635828967612,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999773873473829,
      "loss": 2.5692,
      "step": 1621
    },
    {
      "epoch": 0.011020430175561669,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999977284912717,
      "loss": 2.5588,
      "step": 1622
    },
    {
      "epoch": 0.011027224522155727,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999771822465676,
      "loss": 2.5599,
      "step": 1623
    },
    {
      "epoch": 0.011034018868749785,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999770793489345,
      "loss": 2.6664,
      "step": 1624
    },
    {
      "epoch": 0.011040813215343843,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999769762198177,
      "loss": 2.5177,
      "step": 1625
    },
    {
      "epoch": 0.0110476075619379,
      "grad_norm": 0.189453125,
      "learning_rate": 0.000999976872859217,
      "loss": 2.4249,
      "step": 1626
    },
    {
      "epoch": 0.011054401908531958,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999767692671331,
      "loss": 2.5081,
      "step": 1627
    },
    {
      "epoch": 0.011061196255126016,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999766654435656,
      "loss": 2.4133,
      "step": 1628
    },
    {
      "epoch": 0.011067990601720075,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999765613885145,
      "loss": 2.6344,
      "step": 1629
    },
    {
      "epoch": 0.011074784948314131,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999764571019803,
      "loss": 2.6265,
      "step": 1630
    },
    {
      "epoch": 0.01108157929490819,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999763525839625,
      "loss": 2.6176,
      "step": 1631
    },
    {
      "epoch": 0.011088373641502247,
      "grad_norm": 0.22265625,
      "learning_rate": 0.0009999762478344614,
      "loss": 2.5747,
      "step": 1632
    },
    {
      "epoch": 0.011095167988096304,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999976142853477,
      "loss": 2.6334,
      "step": 1633
    },
    {
      "epoch": 0.011101962334690362,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999760376410093,
      "loss": 2.5354,
      "step": 1634
    },
    {
      "epoch": 0.01110875668128442,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999759321970584,
      "loss": 2.4379,
      "step": 1635
    },
    {
      "epoch": 0.011115551027878479,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999758265216247,
      "loss": 2.5494,
      "step": 1636
    },
    {
      "epoch": 0.011122345374472535,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999757206147078,
      "loss": 2.4593,
      "step": 1637
    },
    {
      "epoch": 0.011129139721066593,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999756144763077,
      "loss": 2.5653,
      "step": 1638
    },
    {
      "epoch": 0.011135934067660651,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999755081064247,
      "loss": 2.519,
      "step": 1639
    },
    {
      "epoch": 0.01114272841425471,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009999754015050588,
      "loss": 2.7228,
      "step": 1640
    },
    {
      "epoch": 0.011149522760848766,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.00099997529467221,
      "loss": 2.4272,
      "step": 1641
    },
    {
      "epoch": 0.011156317107442824,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999751876078784,
      "loss": 2.4533,
      "step": 1642
    },
    {
      "epoch": 0.011163111454036883,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.000999975080312064,
      "loss": 2.5529,
      "step": 1643
    },
    {
      "epoch": 0.01116990580063094,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999749727847668,
      "loss": 2.572,
      "step": 1644
    },
    {
      "epoch": 0.011176700147224997,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999748650259873,
      "loss": 2.6018,
      "step": 1645
    },
    {
      "epoch": 0.011183494493819056,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999747570357247,
      "loss": 2.4978,
      "step": 1646
    },
    {
      "epoch": 0.011190288840413114,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009999746488139798,
      "loss": 2.4788,
      "step": 1647
    },
    {
      "epoch": 0.01119708318700717,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999745403607523,
      "loss": 2.4732,
      "step": 1648
    },
    {
      "epoch": 0.011203877533601228,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999744316760422,
      "loss": 2.6912,
      "step": 1649
    },
    {
      "epoch": 0.011210671880195287,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999743227598497,
      "loss": 2.5594,
      "step": 1650
    },
    {
      "epoch": 0.011217466226789345,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999742136121749,
      "loss": 2.4957,
      "step": 1651
    },
    {
      "epoch": 0.011224260573383401,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009999741042330177,
      "loss": 2.5261,
      "step": 1652
    },
    {
      "epoch": 0.01123105491997746,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999739946223783,
      "loss": 2.6219,
      "step": 1653
    },
    {
      "epoch": 0.011237849266571518,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999738847802566,
      "loss": 2.5315,
      "step": 1654
    },
    {
      "epoch": 0.011244643613165576,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999737747066527,
      "loss": 2.4062,
      "step": 1655
    },
    {
      "epoch": 0.011251437959759632,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999736644015669,
      "loss": 2.6758,
      "step": 1656
    },
    {
      "epoch": 0.01125823230635369,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999735538649987,
      "loss": 2.5569,
      "step": 1657
    },
    {
      "epoch": 0.011265026652947749,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999734430969488,
      "loss": 2.6008,
      "step": 1658
    },
    {
      "epoch": 0.011271820999541807,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999733320974168,
      "loss": 2.5079,
      "step": 1659
    },
    {
      "epoch": 0.011278615346135864,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999732208664028,
      "loss": 2.5404,
      "step": 1660
    },
    {
      "epoch": 0.011285409692729922,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999731094039069,
      "loss": 2.4362,
      "step": 1661
    },
    {
      "epoch": 0.01129220403932398,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999729977099292,
      "loss": 2.422,
      "step": 1662
    },
    {
      "epoch": 0.011298998385918036,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999728857844699,
      "loss": 2.5864,
      "step": 1663
    },
    {
      "epoch": 0.011305792732512095,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999727736275288,
      "loss": 2.6754,
      "step": 1664
    },
    {
      "epoch": 0.011312587079106153,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999726612391058,
      "loss": 2.5293,
      "step": 1665
    },
    {
      "epoch": 0.011319381425700211,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999725486192015,
      "loss": 2.5667,
      "step": 1666
    },
    {
      "epoch": 0.011326175772294268,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999724357678155,
      "loss": 2.6836,
      "step": 1667
    },
    {
      "epoch": 0.011332970118888326,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999972322684948,
      "loss": 2.562,
      "step": 1668
    },
    {
      "epoch": 0.011339764465482384,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999722093705992,
      "loss": 2.6751,
      "step": 1669
    },
    {
      "epoch": 0.011346558812076442,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999972095824769,
      "loss": 2.6443,
      "step": 1670
    },
    {
      "epoch": 0.011353353158670499,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999719820474571,
      "loss": 2.5217,
      "step": 1671
    },
    {
      "epoch": 0.011360147505264557,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999718680386643,
      "loss": 2.5109,
      "step": 1672
    },
    {
      "epoch": 0.011366941851858615,
      "grad_norm": 0.158203125,
      "learning_rate": 0.00099997175379839,
      "loss": 2.4539,
      "step": 1673
    },
    {
      "epoch": 0.011373736198452673,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999716393266345,
      "loss": 2.5249,
      "step": 1674
    },
    {
      "epoch": 0.01138053054504673,
      "grad_norm": 0.2001953125,
      "learning_rate": 0.000999971524623398,
      "loss": 2.7375,
      "step": 1675
    },
    {
      "epoch": 0.011387324891640788,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999714096886804,
      "loss": 2.6095,
      "step": 1676
    },
    {
      "epoch": 0.011394119238234846,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999712945224818,
      "loss": 2.47,
      "step": 1677
    },
    {
      "epoch": 0.011400913584828903,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999711791248021,
      "loss": 2.3325,
      "step": 1678
    },
    {
      "epoch": 0.011407707931422961,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999710634956416,
      "loss": 2.5726,
      "step": 1679
    },
    {
      "epoch": 0.011414502278017019,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999709476350004,
      "loss": 2.5833,
      "step": 1680
    },
    {
      "epoch": 0.011421296624611077,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999970831542878,
      "loss": 2.4614,
      "step": 1681
    },
    {
      "epoch": 0.011428090971205134,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999970715219275,
      "loss": 2.6559,
      "step": 1682
    },
    {
      "epoch": 0.011434885317799192,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999705986641913,
      "loss": 2.4925,
      "step": 1683
    },
    {
      "epoch": 0.01144167966439325,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999970481877627,
      "loss": 2.6456,
      "step": 1684
    },
    {
      "epoch": 0.011448474010987308,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999970364859582,
      "loss": 2.5079,
      "step": 1685
    },
    {
      "epoch": 0.011455268357581365,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999702476100566,
      "loss": 2.6282,
      "step": 1686
    },
    {
      "epoch": 0.011462062704175423,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999701301290505,
      "loss": 2.4737,
      "step": 1687
    },
    {
      "epoch": 0.011468857050769481,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999970012416564,
      "loss": 2.663,
      "step": 1688
    },
    {
      "epoch": 0.01147565139736354,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999698944725972,
      "loss": 2.5982,
      "step": 1689
    },
    {
      "epoch": 0.011482445743957596,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999697762971501,
      "loss": 2.552,
      "step": 1690
    },
    {
      "epoch": 0.011489240090551654,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999696578902226,
      "loss": 2.4967,
      "step": 1691
    },
    {
      "epoch": 0.011496034437145712,
      "grad_norm": 0.1875,
      "learning_rate": 0.000999969539251815,
      "loss": 2.6006,
      "step": 1692
    },
    {
      "epoch": 0.011502828783739769,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009999694203819273,
      "loss": 2.6194,
      "step": 1693
    },
    {
      "epoch": 0.011509623130333827,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999693012805592,
      "loss": 2.6769,
      "step": 1694
    },
    {
      "epoch": 0.011516417476927885,
      "grad_norm": 0.25,
      "learning_rate": 0.0009999691819477113,
      "loss": 2.5216,
      "step": 1695
    },
    {
      "epoch": 0.011523211823521944,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999690623833833,
      "loss": 2.5914,
      "step": 1696
    },
    {
      "epoch": 0.011530006170116,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009999689425875752,
      "loss": 2.5597,
      "step": 1697
    },
    {
      "epoch": 0.011536800516710058,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999688225602874,
      "loss": 2.653,
      "step": 1698
    },
    {
      "epoch": 0.011543594863304116,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999687023015198,
      "loss": 2.4087,
      "step": 1699
    },
    {
      "epoch": 0.011550389209898175,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999685818112725,
      "loss": 2.4209,
      "step": 1700
    },
    {
      "epoch": 0.011557183556492231,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999684610895454,
      "loss": 2.6576,
      "step": 1701
    },
    {
      "epoch": 0.01156397790308629,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999683401363383,
      "loss": 2.4802,
      "step": 1702
    },
    {
      "epoch": 0.011570772249680348,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999682189516521,
      "loss": 2.6137,
      "step": 1703
    },
    {
      "epoch": 0.011577566596274406,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999680975354861,
      "loss": 2.465,
      "step": 1704
    },
    {
      "epoch": 0.011584360942868462,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999679758878406,
      "loss": 2.5888,
      "step": 1705
    },
    {
      "epoch": 0.01159115528946252,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999678540087158,
      "loss": 2.4507,
      "step": 1706
    },
    {
      "epoch": 0.011597949636056579,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999677318981114,
      "loss": 2.4019,
      "step": 1707
    },
    {
      "epoch": 0.011604743982650635,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999676095560277,
      "loss": 2.5332,
      "step": 1708
    },
    {
      "epoch": 0.011611538329244693,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999674869824648,
      "loss": 2.7745,
      "step": 1709
    },
    {
      "epoch": 0.011618332675838752,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999673641774226,
      "loss": 2.7106,
      "step": 1710
    },
    {
      "epoch": 0.01162512702243281,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0009999672411409016,
      "loss": 2.5791,
      "step": 1711
    },
    {
      "epoch": 0.011631921369026866,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999671178729012,
      "loss": 2.3192,
      "step": 1712
    },
    {
      "epoch": 0.011638715715620924,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999669943734218,
      "loss": 2.4134,
      "step": 1713
    },
    {
      "epoch": 0.011645510062214983,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999668706424634,
      "loss": 2.6582,
      "step": 1714
    },
    {
      "epoch": 0.011652304408809041,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999667466800259,
      "loss": 2.527,
      "step": 1715
    },
    {
      "epoch": 0.011659098755403097,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0009999666224861099,
      "loss": 2.6326,
      "step": 1716
    },
    {
      "epoch": 0.011665893101997156,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999664980607148,
      "loss": 2.389,
      "step": 1717
    },
    {
      "epoch": 0.011672687448591214,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999663734038412,
      "loss": 2.5999,
      "step": 1718
    },
    {
      "epoch": 0.011679481795185272,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999662485154885,
      "loss": 2.5532,
      "step": 1719
    },
    {
      "epoch": 0.011686276141779329,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999661233956575,
      "loss": 2.4686,
      "step": 1720
    },
    {
      "epoch": 0.011693070488373387,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999965998044348,
      "loss": 2.537,
      "step": 1721
    },
    {
      "epoch": 0.011699864834967445,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999658724615596,
      "loss": 2.6487,
      "step": 1722
    },
    {
      "epoch": 0.011706659181561501,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999657466472928,
      "loss": 2.5697,
      "step": 1723
    },
    {
      "epoch": 0.01171345352815556,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999656206015478,
      "loss": 2.6771,
      "step": 1724
    },
    {
      "epoch": 0.011720247874749618,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999654943243243,
      "loss": 2.7259,
      "step": 1725
    },
    {
      "epoch": 0.011727042221343676,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999653678156226,
      "loss": 2.5447,
      "step": 1726
    },
    {
      "epoch": 0.011733836567937733,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009999652410754428,
      "loss": 2.6676,
      "step": 1727
    },
    {
      "epoch": 0.01174063091453179,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999651141037844,
      "loss": 2.5809,
      "step": 1728
    },
    {
      "epoch": 0.011747425261125849,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999649869006481,
      "loss": 2.3757,
      "step": 1729
    },
    {
      "epoch": 0.011754219607719907,
      "grad_norm": 0.193359375,
      "learning_rate": 0.000999964859466034,
      "loss": 2.6336,
      "step": 1730
    },
    {
      "epoch": 0.011761013954313964,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999647317999417,
      "loss": 2.5537,
      "step": 1731
    },
    {
      "epoch": 0.011767808300908022,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999646039023714,
      "loss": 2.5534,
      "step": 1732
    },
    {
      "epoch": 0.01177460264750208,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999644757733232,
      "loss": 2.4963,
      "step": 1733
    },
    {
      "epoch": 0.011781396994096138,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999643474127972,
      "loss": 2.5925,
      "step": 1734
    },
    {
      "epoch": 0.011788191340690195,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999642188207936,
      "loss": 2.6466,
      "step": 1735
    },
    {
      "epoch": 0.011794985687284253,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999640899973122,
      "loss": 2.4465,
      "step": 1736
    },
    {
      "epoch": 0.011801780033878311,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.000999963960942353,
      "loss": 2.7364,
      "step": 1737
    },
    {
      "epoch": 0.011808574380472368,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999638316559164,
      "loss": 2.5476,
      "step": 1738
    },
    {
      "epoch": 0.011815368727066426,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999637021380025,
      "loss": 2.4985,
      "step": 1739
    },
    {
      "epoch": 0.011822163073660484,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999635723886107,
      "loss": 2.6135,
      "step": 1740
    },
    {
      "epoch": 0.011828957420254542,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999634424077417,
      "loss": 2.6509,
      "step": 1741
    },
    {
      "epoch": 0.011835751766848599,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999633121953952,
      "loss": 2.5756,
      "step": 1742
    },
    {
      "epoch": 0.011842546113442657,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999631817515717,
      "loss": 2.5703,
      "step": 1743
    },
    {
      "epoch": 0.011849340460036715,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999630510762708,
      "loss": 2.3488,
      "step": 1744
    },
    {
      "epoch": 0.011856134806630773,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999629201694925,
      "loss": 2.6229,
      "step": 1745
    },
    {
      "epoch": 0.01186292915322483,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999627890312376,
      "loss": 2.5756,
      "step": 1746
    },
    {
      "epoch": 0.011869723499818888,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999626576615054,
      "loss": 2.7397,
      "step": 1747
    },
    {
      "epoch": 0.011876517846412946,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999625260602963,
      "loss": 2.5889,
      "step": 1748
    },
    {
      "epoch": 0.011883312193007005,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.00099996239422761,
      "loss": 2.5931,
      "step": 1749
    },
    {
      "epoch": 0.011890106539601061,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999622621634473,
      "loss": 2.5169,
      "step": 1750
    },
    {
      "epoch": 0.01189690088619512,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999621298678076,
      "loss": 2.7034,
      "step": 1751
    },
    {
      "epoch": 0.011903695232789177,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999961997340691,
      "loss": 2.6796,
      "step": 1752
    },
    {
      "epoch": 0.011910489579383234,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999961864582098,
      "loss": 2.4825,
      "step": 1753
    },
    {
      "epoch": 0.011917283925977292,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999617315920282,
      "loss": 2.5311,
      "step": 1754
    },
    {
      "epoch": 0.01192407827257135,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999961598370482,
      "loss": 2.6391,
      "step": 1755
    },
    {
      "epoch": 0.011930872619165409,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999961464917459,
      "loss": 2.535,
      "step": 1756
    },
    {
      "epoch": 0.011937666965759465,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00099996133123296,
      "loss": 2.5008,
      "step": 1757
    },
    {
      "epoch": 0.011944461312353523,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999611973169845,
      "loss": 2.4768,
      "step": 1758
    },
    {
      "epoch": 0.011951255658947581,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999610631695325,
      "loss": 2.6703,
      "step": 1759
    },
    {
      "epoch": 0.01195805000554164,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999609287906044,
      "loss": 2.441,
      "step": 1760
    },
    {
      "epoch": 0.011964844352135696,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999607941802,
      "loss": 2.509,
      "step": 1761
    },
    {
      "epoch": 0.011971638698729754,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999606593383197,
      "loss": 2.562,
      "step": 1762
    },
    {
      "epoch": 0.011978433045323813,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999605242649633,
      "loss": 2.5297,
      "step": 1763
    },
    {
      "epoch": 0.01198522739191787,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999603889601309,
      "loss": 2.4751,
      "step": 1764
    },
    {
      "epoch": 0.011992021738511927,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999602534238223,
      "loss": 2.3118,
      "step": 1765
    },
    {
      "epoch": 0.011998816085105985,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999601176560382,
      "loss": 2.4771,
      "step": 1766
    },
    {
      "epoch": 0.012005610431700044,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999599816567781,
      "loss": 2.4367,
      "step": 1767
    },
    {
      "epoch": 0.0120124047782941,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999598454260422,
      "loss": 2.4811,
      "step": 1768
    },
    {
      "epoch": 0.012019199124888158,
      "grad_norm": 0.244140625,
      "learning_rate": 0.000999959708963831,
      "loss": 2.5701,
      "step": 1769
    },
    {
      "epoch": 0.012025993471482217,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999595722701437,
      "loss": 2.6701,
      "step": 1770
    },
    {
      "epoch": 0.012032787818076275,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0009999594353449811,
      "loss": 2.5161,
      "step": 1771
    },
    {
      "epoch": 0.012039582164670331,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999959298188343,
      "loss": 2.4415,
      "step": 1772
    },
    {
      "epoch": 0.01204637651126439,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999591608002294,
      "loss": 2.6143,
      "step": 1773
    },
    {
      "epoch": 0.012053170857858448,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999590231806405,
      "loss": 2.4261,
      "step": 1774
    },
    {
      "epoch": 0.012059965204452506,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009999588853295764,
      "loss": 2.7031,
      "step": 1775
    },
    {
      "epoch": 0.012066759551046562,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000999958747247037,
      "loss": 2.5293,
      "step": 1776
    },
    {
      "epoch": 0.01207355389764062,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999586089330223,
      "loss": 2.523,
      "step": 1777
    },
    {
      "epoch": 0.012080348244234679,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009999584703875325,
      "loss": 2.462,
      "step": 1778
    },
    {
      "epoch": 0.012087142590828737,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.000999958331610568,
      "loss": 2.6165,
      "step": 1779
    },
    {
      "epoch": 0.012093936937422793,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999581926021283,
      "loss": 2.5133,
      "step": 1780
    },
    {
      "epoch": 0.012100731284016852,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009999580533622137,
      "loss": 2.7136,
      "step": 1781
    },
    {
      "epoch": 0.01210752563061091,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0009999579138908242,
      "loss": 2.6201,
      "step": 1782
    },
    {
      "epoch": 0.012114319977204966,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00099995777418796,
      "loss": 2.5854,
      "step": 1783
    },
    {
      "epoch": 0.012121114323799025,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999576342536209,
      "loss": 2.4144,
      "step": 1784
    },
    {
      "epoch": 0.012127908670393083,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999574940878074,
      "loss": 2.5248,
      "step": 1785
    },
    {
      "epoch": 0.012134703016987141,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999957353690519,
      "loss": 2.6181,
      "step": 1786
    },
    {
      "epoch": 0.012141497363581197,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999572130617563,
      "loss": 2.565,
      "step": 1787
    },
    {
      "epoch": 0.012148291710175256,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999570722015192,
      "loss": 2.5028,
      "step": 1788
    },
    {
      "epoch": 0.012155086056769314,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009999569311098075,
      "loss": 2.7083,
      "step": 1789
    },
    {
      "epoch": 0.012161880403363372,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999567897866217,
      "loss": 2.6056,
      "step": 1790
    },
    {
      "epoch": 0.012168674749957429,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999566482319616,
      "loss": 2.4986,
      "step": 1791
    },
    {
      "epoch": 0.012175469096551487,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999956506445827,
      "loss": 2.5186,
      "step": 1792
    },
    {
      "epoch": 0.012182263443145545,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999563644282185,
      "loss": 2.6664,
      "step": 1793
    },
    {
      "epoch": 0.012189057789739603,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999956222179136,
      "loss": 2.4187,
      "step": 1794
    },
    {
      "epoch": 0.01219585213633366,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999560796985793,
      "loss": 2.5234,
      "step": 1795
    },
    {
      "epoch": 0.012202646482927718,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999955936986549,
      "loss": 2.6782,
      "step": 1796
    },
    {
      "epoch": 0.012209440829521776,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999557940430445,
      "loss": 2.5716,
      "step": 1797
    },
    {
      "epoch": 0.012216235176115833,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999556508680665,
      "loss": 2.477,
      "step": 1798
    },
    {
      "epoch": 0.01222302952270989,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999555074616145,
      "loss": 2.6077,
      "step": 1799
    },
    {
      "epoch": 0.012229823869303949,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999955363823689,
      "loss": 2.4868,
      "step": 1800
    },
    {
      "epoch": 0.012236618215898007,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999552199542899,
      "loss": 2.5151,
      "step": 1801
    },
    {
      "epoch": 0.012243412562492064,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999550758534172,
      "loss": 2.5448,
      "step": 1802
    },
    {
      "epoch": 0.012250206909086122,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999954931521071,
      "loss": 2.5702,
      "step": 1803
    },
    {
      "epoch": 0.01225700125568018,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999547869572515,
      "loss": 2.5454,
      "step": 1804
    },
    {
      "epoch": 0.012263795602274238,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999546421619583,
      "loss": 2.5509,
      "step": 1805
    },
    {
      "epoch": 0.012270589948868295,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999544971351921,
      "loss": 2.5997,
      "step": 1806
    },
    {
      "epoch": 0.012277384295462353,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999543518769529,
      "loss": 2.6007,
      "step": 1807
    },
    {
      "epoch": 0.012284178642056411,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999542063872404,
      "loss": 2.6915,
      "step": 1808
    },
    {
      "epoch": 0.01229097298865047,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999540606660548,
      "loss": 2.505,
      "step": 1809
    },
    {
      "epoch": 0.012297767335244526,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999539147133962,
      "loss": 2.5841,
      "step": 1810
    },
    {
      "epoch": 0.012304561681838584,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999537685292646,
      "loss": 2.5857,
      "step": 1811
    },
    {
      "epoch": 0.012311356028432642,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999536221136603,
      "loss": 2.672,
      "step": 1812
    },
    {
      "epoch": 0.012318150375026699,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999953475466583,
      "loss": 2.5549,
      "step": 1813
    },
    {
      "epoch": 0.012324944721620757,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999533285880331,
      "loss": 2.5068,
      "step": 1814
    },
    {
      "epoch": 0.012331739068214815,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999531814780108,
      "loss": 2.5573,
      "step": 1815
    },
    {
      "epoch": 0.012338533414808874,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999530341365155,
      "loss": 2.6424,
      "step": 1816
    },
    {
      "epoch": 0.01234532776140293,
      "grad_norm": 0.177734375,
      "learning_rate": 0.000999952886563548,
      "loss": 2.6092,
      "step": 1817
    },
    {
      "epoch": 0.012352122107996988,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999527387591078,
      "loss": 2.3842,
      "step": 1818
    },
    {
      "epoch": 0.012358916454591046,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999525907231953,
      "loss": 2.5992,
      "step": 1819
    },
    {
      "epoch": 0.012365710801185105,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999524424558104,
      "loss": 2.5472,
      "step": 1820
    },
    {
      "epoch": 0.012372505147779161,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999522939569533,
      "loss": 2.6199,
      "step": 1821
    },
    {
      "epoch": 0.01237929949437322,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999952145226624,
      "loss": 2.5493,
      "step": 1822
    },
    {
      "epoch": 0.012386093840967278,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999519962648226,
      "loss": 2.5925,
      "step": 1823
    },
    {
      "epoch": 0.012392888187561336,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009999518470715492,
      "loss": 2.5861,
      "step": 1824
    },
    {
      "epoch": 0.012399682534155392,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009999516976468038,
      "loss": 2.4902,
      "step": 1825
    },
    {
      "epoch": 0.01240647688074945,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999515479905867,
      "loss": 2.5549,
      "step": 1826
    },
    {
      "epoch": 0.012413271227343509,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999513981028974,
      "loss": 2.3807,
      "step": 1827
    },
    {
      "epoch": 0.012420065573937565,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999512479837366,
      "loss": 2.594,
      "step": 1828
    },
    {
      "epoch": 0.012426859920531623,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999951097633104,
      "loss": 2.4214,
      "step": 1829
    },
    {
      "epoch": 0.012433654267125682,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999509470509998,
      "loss": 2.5788,
      "step": 1830
    },
    {
      "epoch": 0.01244044861371974,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999507962374242,
      "loss": 2.5081,
      "step": 1831
    },
    {
      "epoch": 0.012447242960313796,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009999506451923768,
      "loss": 2.3875,
      "step": 1832
    },
    {
      "epoch": 0.012454037306907854,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009999504939158582,
      "loss": 2.5184,
      "step": 1833
    },
    {
      "epoch": 0.012460831653501913,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999950342407868,
      "loss": 2.5247,
      "step": 1834
    },
    {
      "epoch": 0.01246762600009597,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009999501906684067,
      "loss": 2.4899,
      "step": 1835
    },
    {
      "epoch": 0.012474420346690027,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999500386974741,
      "loss": 2.415,
      "step": 1836
    },
    {
      "epoch": 0.012481214693284086,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999498864950707,
      "loss": 2.4722,
      "step": 1837
    },
    {
      "epoch": 0.012488009039878144,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999949734061196,
      "loss": 2.6776,
      "step": 1838
    },
    {
      "epoch": 0.012494803386472202,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009999495813958503,
      "loss": 2.5165,
      "step": 1839
    },
    {
      "epoch": 0.012501597733066258,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999494284990335,
      "loss": 2.5861,
      "step": 1840
    },
    {
      "epoch": 0.012508392079660317,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999492753707461,
      "loss": 2.4657,
      "step": 1841
    },
    {
      "epoch": 0.012515186426254375,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999491220109879,
      "loss": 2.4928,
      "step": 1842
    },
    {
      "epoch": 0.012521980772848431,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999948968419759,
      "loss": 2.5104,
      "step": 1843
    },
    {
      "epoch": 0.01252877511944249,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999488145970594,
      "loss": 2.5537,
      "step": 1844
    },
    {
      "epoch": 0.012535569466036548,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999486605428891,
      "loss": 2.6914,
      "step": 1845
    },
    {
      "epoch": 0.012542363812630606,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009999485062572484,
      "loss": 2.4604,
      "step": 1846
    },
    {
      "epoch": 0.012549158159224662,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999483517401373,
      "loss": 2.4908,
      "step": 1847
    },
    {
      "epoch": 0.01255595250581872,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999481969915557,
      "loss": 2.4677,
      "step": 1848
    },
    {
      "epoch": 0.012562746852412779,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999480420115042,
      "loss": 2.4775,
      "step": 1849
    },
    {
      "epoch": 0.012569541199006837,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009999478867999821,
      "loss": 2.4952,
      "step": 1850
    },
    {
      "epoch": 0.012576335545600894,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00099994773135699,
      "loss": 2.4432,
      "step": 1851
    },
    {
      "epoch": 0.012583129892194952,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999475756825279,
      "loss": 2.5943,
      "step": 1852
    },
    {
      "epoch": 0.01258992423878901,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999474197765958,
      "loss": 2.5353,
      "step": 1853
    },
    {
      "epoch": 0.012596718585383068,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999472636391935,
      "loss": 2.488,
      "step": 1854
    },
    {
      "epoch": 0.012603512931977125,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999471072703217,
      "loss": 2.6558,
      "step": 1855
    },
    {
      "epoch": 0.012610307278571183,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.00099994695066998,
      "loss": 2.5077,
      "step": 1856
    },
    {
      "epoch": 0.012617101625165241,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999467938381686,
      "loss": 2.6264,
      "step": 1857
    },
    {
      "epoch": 0.012623895971759298,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999466367748874,
      "loss": 2.4153,
      "step": 1858
    },
    {
      "epoch": 0.012630690318353356,
      "grad_norm": 0.177734375,
      "learning_rate": 0.000999946479480137,
      "loss": 2.4032,
      "step": 1859
    },
    {
      "epoch": 0.012637484664947414,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009999463219539167,
      "loss": 2.7303,
      "step": 1860
    },
    {
      "epoch": 0.012644279011541472,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999461641962274,
      "loss": 2.6869,
      "step": 1861
    },
    {
      "epoch": 0.012651073358135529,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999460062070685,
      "loss": 2.5647,
      "step": 1862
    },
    {
      "epoch": 0.012657867704729587,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999458479864403,
      "loss": 2.5511,
      "step": 1863
    },
    {
      "epoch": 0.012664662051323645,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999945689534343,
      "loss": 2.5919,
      "step": 1864
    },
    {
      "epoch": 0.012671456397917703,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009999455308507764,
      "loss": 2.5139,
      "step": 1865
    },
    {
      "epoch": 0.01267825074451176,
      "grad_norm": 0.181640625,
      "learning_rate": 0.000999945371935741,
      "loss": 2.5482,
      "step": 1866
    },
    {
      "epoch": 0.012685045091105818,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009999452127892367,
      "loss": 2.5501,
      "step": 1867
    },
    {
      "epoch": 0.012691839437699876,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999450534112632,
      "loss": 2.4847,
      "step": 1868
    },
    {
      "epoch": 0.012698633784293934,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999944893801821,
      "loss": 2.5192,
      "step": 1869
    },
    {
      "epoch": 0.012705428130887991,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00099994473396091,
      "loss": 2.5921,
      "step": 1870
    },
    {
      "epoch": 0.01271222247748205,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999445738885302,
      "loss": 2.5803,
      "step": 1871
    },
    {
      "epoch": 0.012719016824076107,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999944413584682,
      "loss": 2.5696,
      "step": 1872
    },
    {
      "epoch": 0.012725811170670164,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999442530493653,
      "loss": 2.2956,
      "step": 1873
    },
    {
      "epoch": 0.012732605517264222,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.00099994409228258,
      "loss": 2.5193,
      "step": 1874
    },
    {
      "epoch": 0.01273939986385828,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999439312843264,
      "loss": 2.5092,
      "step": 1875
    },
    {
      "epoch": 0.012746194210452338,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999437700546046,
      "loss": 2.5823,
      "step": 1876
    },
    {
      "epoch": 0.012752988557046395,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999436085934145,
      "loss": 2.4379,
      "step": 1877
    },
    {
      "epoch": 0.012759782903640453,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000999943446900756,
      "loss": 2.647,
      "step": 1878
    },
    {
      "epoch": 0.012766577250234511,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999432849766296,
      "loss": 2.5162,
      "step": 1879
    },
    {
      "epoch": 0.01277337159682857,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999431228210351,
      "loss": 2.5588,
      "step": 1880
    },
    {
      "epoch": 0.012780165943422626,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999942960433973,
      "loss": 2.3865,
      "step": 1881
    },
    {
      "epoch": 0.012786960290016684,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999427978154428,
      "loss": 2.5292,
      "step": 1882
    },
    {
      "epoch": 0.012793754636610742,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009999426349654448,
      "loss": 2.5273,
      "step": 1883
    },
    {
      "epoch": 0.0128005489832048,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.000999942471883979,
      "loss": 2.5328,
      "step": 1884
    },
    {
      "epoch": 0.012807343329798857,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999423085710456,
      "loss": 2.5262,
      "step": 1885
    },
    {
      "epoch": 0.012814137676392915,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999421450266447,
      "loss": 2.5235,
      "step": 1886
    },
    {
      "epoch": 0.012820932022986974,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999419812507764,
      "loss": 2.6734,
      "step": 1887
    },
    {
      "epoch": 0.01282772636958103,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999418172434405,
      "loss": 2.6359,
      "step": 1888
    },
    {
      "epoch": 0.012834520716175088,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009999416530046374,
      "loss": 2.4957,
      "step": 1889
    },
    {
      "epoch": 0.012841315062769146,
      "grad_norm": 0.1875,
      "learning_rate": 0.000999941488534367,
      "loss": 2.55,
      "step": 1890
    },
    {
      "epoch": 0.012848109409363205,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999413238326296,
      "loss": 2.6027,
      "step": 1891
    },
    {
      "epoch": 0.012854903755957261,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999941158899425,
      "loss": 2.5101,
      "step": 1892
    },
    {
      "epoch": 0.01286169810255132,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999409937347533,
      "loss": 2.6178,
      "step": 1893
    },
    {
      "epoch": 0.012868492449145378,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999408283386147,
      "loss": 2.4809,
      "step": 1894
    },
    {
      "epoch": 0.012875286795739436,
      "grad_norm": 0.177734375,
      "learning_rate": 0.000999940662711009,
      "loss": 2.3814,
      "step": 1895
    },
    {
      "epoch": 0.012882081142333492,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999940496851937,
      "loss": 2.464,
      "step": 1896
    },
    {
      "epoch": 0.01288887548892755,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009999403307613978,
      "loss": 2.6896,
      "step": 1897
    },
    {
      "epoch": 0.012895669835521609,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999401644393922,
      "loss": 2.5551,
      "step": 1898
    },
    {
      "epoch": 0.012902464182115667,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00099993999788592,
      "loss": 2.6613,
      "step": 1899
    },
    {
      "epoch": 0.012909258528709723,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999398311009813,
      "loss": 2.3909,
      "step": 1900
    },
    {
      "epoch": 0.012916052875303782,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999396640845762,
      "loss": 2.4155,
      "step": 1901
    },
    {
      "epoch": 0.01292284722189784,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009999394968367048,
      "loss": 2.354,
      "step": 1902
    },
    {
      "epoch": 0.012929641568491896,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999393293573672,
      "loss": 2.5014,
      "step": 1903
    },
    {
      "epoch": 0.012936435915085955,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999391616465631,
      "loss": 2.4146,
      "step": 1904
    },
    {
      "epoch": 0.012943230261680013,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999389937042932,
      "loss": 2.6561,
      "step": 1905
    },
    {
      "epoch": 0.012950024608274071,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999388255305572,
      "loss": 2.6039,
      "step": 1906
    },
    {
      "epoch": 0.012956818954868127,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999386571253553,
      "loss": 2.4605,
      "step": 1907
    },
    {
      "epoch": 0.012963613301462186,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999384884886876,
      "loss": 2.5678,
      "step": 1908
    },
    {
      "epoch": 0.012970407648056244,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999938319620554,
      "loss": 2.4771,
      "step": 1909
    },
    {
      "epoch": 0.012977201994650302,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009999381505209546,
      "loss": 2.3222,
      "step": 1910
    },
    {
      "epoch": 0.012983996341244359,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999379811898898,
      "loss": 2.5479,
      "step": 1911
    },
    {
      "epoch": 0.012990790687838417,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999378116273593,
      "loss": 2.4891,
      "step": 1912
    },
    {
      "epoch": 0.012997585034432475,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009999376418333632,
      "loss": 2.4881,
      "step": 1913
    },
    {
      "epoch": 0.013004379381026533,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999937471807902,
      "loss": 2.548,
      "step": 1914
    },
    {
      "epoch": 0.01301117372762059,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999373015509752,
      "loss": 2.5019,
      "step": 1915
    },
    {
      "epoch": 0.013017968074214648,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999371310625833,
      "loss": 2.6007,
      "step": 1916
    },
    {
      "epoch": 0.013024762420808706,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999369603427262,
      "loss": 2.4559,
      "step": 1917
    },
    {
      "epoch": 0.013031556767402763,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000999936789391404,
      "loss": 2.5009,
      "step": 1918
    },
    {
      "epoch": 0.01303835111399682,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.000999936618208617,
      "loss": 2.4684,
      "step": 1919
    },
    {
      "epoch": 0.013045145460590879,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999364467943648,
      "loss": 2.3606,
      "step": 1920
    },
    {
      "epoch": 0.013051939807184937,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999362751486479,
      "loss": 2.6336,
      "step": 1921
    },
    {
      "epoch": 0.013058734153778994,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999361032714661,
      "loss": 2.5496,
      "step": 1922
    },
    {
      "epoch": 0.013065528500373052,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999359311628198,
      "loss": 2.4849,
      "step": 1923
    },
    {
      "epoch": 0.01307232284696711,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999357588227089,
      "loss": 2.4186,
      "step": 1924
    },
    {
      "epoch": 0.013079117193561168,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999355862511333,
      "loss": 2.3496,
      "step": 1925
    },
    {
      "epoch": 0.013085911540155225,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999354134480934,
      "loss": 2.3962,
      "step": 1926
    },
    {
      "epoch": 0.013092705886749283,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999352404135892,
      "loss": 2.5615,
      "step": 1927
    },
    {
      "epoch": 0.013099500233343341,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999350671476203,
      "loss": 2.459,
      "step": 1928
    },
    {
      "epoch": 0.0131062945799374,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999348936501876,
      "loss": 2.4645,
      "step": 1929
    },
    {
      "epoch": 0.013113088926531456,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999347199212906,
      "loss": 2.6362,
      "step": 1930
    },
    {
      "epoch": 0.013119883273125514,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009999345459609297,
      "loss": 2.4604,
      "step": 1931
    },
    {
      "epoch": 0.013126677619719572,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000999934371769105,
      "loss": 2.4992,
      "step": 1932
    },
    {
      "epoch": 0.013133471966313629,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.000999934197345816,
      "loss": 2.4458,
      "step": 1933
    },
    {
      "epoch": 0.013140266312907687,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999340226910633,
      "loss": 2.6709,
      "step": 1934
    },
    {
      "epoch": 0.013147060659501745,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999338478048469,
      "loss": 2.5879,
      "step": 1935
    },
    {
      "epoch": 0.013153855006095803,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.000999933672687167,
      "loss": 2.4976,
      "step": 1936
    },
    {
      "epoch": 0.01316064935268986,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999334973380236,
      "loss": 2.6083,
      "step": 1937
    },
    {
      "epoch": 0.013167443699283918,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999333217574164,
      "loss": 2.5771,
      "step": 1938
    },
    {
      "epoch": 0.013174238045877976,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999933145945346,
      "loss": 2.5892,
      "step": 1939
    },
    {
      "epoch": 0.013181032392472035,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999329699018123,
      "loss": 2.5325,
      "step": 1940
    },
    {
      "epoch": 0.013187826739066091,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999327936268153,
      "loss": 2.4554,
      "step": 1941
    },
    {
      "epoch": 0.01319462108566015,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999326171203553,
      "loss": 2.6694,
      "step": 1942
    },
    {
      "epoch": 0.013201415432254207,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999324403824322,
      "loss": 2.4979,
      "step": 1943
    },
    {
      "epoch": 0.013208209778848266,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999932263413046,
      "loss": 2.5487,
      "step": 1944
    },
    {
      "epoch": 0.013215004125442322,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999320862121967,
      "loss": 2.676,
      "step": 1945
    },
    {
      "epoch": 0.01322179847203638,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999931908779885,
      "loss": 2.5408,
      "step": 1946
    },
    {
      "epoch": 0.013228592818630439,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999317311161103,
      "loss": 2.5576,
      "step": 1947
    },
    {
      "epoch": 0.013235387165224495,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999931553220873,
      "loss": 2.5827,
      "step": 1948
    },
    {
      "epoch": 0.013242181511818553,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999931375094173,
      "loss": 2.553,
      "step": 1949
    },
    {
      "epoch": 0.013248975858412611,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999311967360106,
      "loss": 2.4618,
      "step": 1950
    },
    {
      "epoch": 0.01325577020500667,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999310181463858,
      "loss": 2.4509,
      "step": 1951
    },
    {
      "epoch": 0.013262564551600726,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999308393252987,
      "loss": 2.3837,
      "step": 1952
    },
    {
      "epoch": 0.013269358898194784,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999306602727495,
      "loss": 2.6,
      "step": 1953
    },
    {
      "epoch": 0.013276153244788843,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999930480988738,
      "loss": 2.4766,
      "step": 1954
    },
    {
      "epoch": 0.0132829475913829,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999303014732643,
      "loss": 2.5811,
      "step": 1955
    },
    {
      "epoch": 0.013289741937976957,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999301217263287,
      "loss": 2.4661,
      "step": 1956
    },
    {
      "epoch": 0.013296536284571015,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999929941747931,
      "loss": 2.4696,
      "step": 1957
    },
    {
      "epoch": 0.013303330631165074,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999297615380718,
      "loss": 2.4896,
      "step": 1958
    },
    {
      "epoch": 0.013310124977759132,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999295810967507,
      "loss": 2.6186,
      "step": 1959
    },
    {
      "epoch": 0.013316919324353188,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000999929400423968,
      "loss": 2.6704,
      "step": 1960
    },
    {
      "epoch": 0.013323713670947247,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999292195197237,
      "loss": 2.6577,
      "step": 1961
    },
    {
      "epoch": 0.013330508017541305,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999290383840176,
      "loss": 2.6091,
      "step": 1962
    },
    {
      "epoch": 0.013337302364135361,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999288570168505,
      "loss": 2.4497,
      "step": 1963
    },
    {
      "epoch": 0.01334409671072942,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999286754182218,
      "loss": 2.4829,
      "step": 1964
    },
    {
      "epoch": 0.013350891057323478,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999928493588132,
      "loss": 2.519,
      "step": 1965
    },
    {
      "epoch": 0.013357685403917536,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999928311526581,
      "loss": 2.6587,
      "step": 1966
    },
    {
      "epoch": 0.013364479750511592,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999928129233569,
      "loss": 2.5282,
      "step": 1967
    },
    {
      "epoch": 0.01337127409710565,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999279467090957,
      "loss": 2.5794,
      "step": 1968
    },
    {
      "epoch": 0.013378068443699709,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999277639531619,
      "loss": 2.6389,
      "step": 1969
    },
    {
      "epoch": 0.013384862790293767,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999927580965767,
      "loss": 2.4484,
      "step": 1970
    },
    {
      "epoch": 0.013391657136887824,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999273977469116,
      "loss": 2.4778,
      "step": 1971
    },
    {
      "epoch": 0.013398451483481882,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999272142965953,
      "loss": 2.6706,
      "step": 1972
    },
    {
      "epoch": 0.01340524583007594,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999270306148187,
      "loss": 2.5478,
      "step": 1973
    },
    {
      "epoch": 0.013412040176669998,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999268467015814,
      "loss": 2.4139,
      "step": 1974
    },
    {
      "epoch": 0.013418834523264055,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999266625568836,
      "loss": 2.6096,
      "step": 1975
    },
    {
      "epoch": 0.013425628869858113,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999264781807259,
      "loss": 2.371,
      "step": 1976
    },
    {
      "epoch": 0.013432423216452171,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999262935731074,
      "loss": 2.6461,
      "step": 1977
    },
    {
      "epoch": 0.013439217563046228,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999261087340293,
      "loss": 2.3862,
      "step": 1978
    },
    {
      "epoch": 0.013446011909640286,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999259236634908,
      "loss": 2.5556,
      "step": 1979
    },
    {
      "epoch": 0.013452806256234344,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999257383614924,
      "loss": 2.6546,
      "step": 1980
    },
    {
      "epoch": 0.013459600602828402,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999255528280343,
      "loss": 2.3889,
      "step": 1981
    },
    {
      "epoch": 0.013466394949422459,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999253670631163,
      "loss": 2.6583,
      "step": 1982
    },
    {
      "epoch": 0.013473189296016517,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999251810667385,
      "loss": 2.5902,
      "step": 1983
    },
    {
      "epoch": 0.013479983642610575,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000999924994838901,
      "loss": 2.6605,
      "step": 1984
    },
    {
      "epoch": 0.013486777989204633,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999248083796043,
      "loss": 2.5201,
      "step": 1985
    },
    {
      "epoch": 0.01349357233579869,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999246216888479,
      "loss": 2.4713,
      "step": 1986
    },
    {
      "epoch": 0.013500366682392748,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009999244347666321,
      "loss": 2.5557,
      "step": 1987
    },
    {
      "epoch": 0.013507161028986806,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999924247612957,
      "loss": 2.5386,
      "step": 1988
    },
    {
      "epoch": 0.013513955375580864,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009999240602278229,
      "loss": 2.4708,
      "step": 1989
    },
    {
      "epoch": 0.013520749722174921,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999238726112294,
      "loss": 2.3753,
      "step": 1990
    },
    {
      "epoch": 0.013527544068768979,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999236847631772,
      "loss": 2.4744,
      "step": 1991
    },
    {
      "epoch": 0.013534338415363037,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999234966836658,
      "loss": 2.4617,
      "step": 1992
    },
    {
      "epoch": 0.013541132761957094,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999233083726957,
      "loss": 2.5906,
      "step": 1993
    },
    {
      "epoch": 0.013547927108551152,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999231198302668,
      "loss": 2.556,
      "step": 1994
    },
    {
      "epoch": 0.01355472145514521,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999229310563792,
      "loss": 2.6724,
      "step": 1995
    },
    {
      "epoch": 0.013561515801739268,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009999227420510331,
      "loss": 2.4979,
      "step": 1996
    },
    {
      "epoch": 0.013568310148333325,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009999225528142283,
      "loss": 2.5952,
      "step": 1997
    },
    {
      "epoch": 0.013575104494927383,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999223633459654,
      "loss": 2.5679,
      "step": 1998
    },
    {
      "epoch": 0.013581898841521441,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999922173646244,
      "loss": 2.5653,
      "step": 1999
    },
    {
      "epoch": 0.0135886931881155,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999219837150645,
      "loss": 2.6216,
      "step": 2000
    },
    {
      "epoch": 0.013595487534709556,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999217935524264,
      "loss": 2.5051,
      "step": 2001
    },
    {
      "epoch": 0.013602281881303614,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009999216031583308,
      "loss": 2.7447,
      "step": 2002
    },
    {
      "epoch": 0.013609076227897672,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000999921412532777,
      "loss": 2.6145,
      "step": 2003
    },
    {
      "epoch": 0.01361587057449173,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999212216757652,
      "loss": 2.4823,
      "step": 2004
    },
    {
      "epoch": 0.013622664921085787,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999210305872957,
      "loss": 2.3726,
      "step": 2005
    },
    {
      "epoch": 0.013629459267679845,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999208392673686,
      "loss": 2.4454,
      "step": 2006
    },
    {
      "epoch": 0.013636253614273904,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009999206477159838,
      "loss": 2.5177,
      "step": 2007
    },
    {
      "epoch": 0.01364304796086796,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999204559331413,
      "loss": 2.5067,
      "step": 2008
    },
    {
      "epoch": 0.013649842307462018,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999202639188417,
      "loss": 2.3803,
      "step": 2009
    },
    {
      "epoch": 0.013656636654056076,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999200716730844,
      "loss": 2.476,
      "step": 2010
    },
    {
      "epoch": 0.013663431000650135,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999198791958701,
      "loss": 2.4458,
      "step": 2011
    },
    {
      "epoch": 0.013670225347244191,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009999196864871986,
      "loss": 2.4829,
      "step": 2012
    },
    {
      "epoch": 0.01367701969383825,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999194935470701,
      "loss": 2.5437,
      "step": 2013
    },
    {
      "epoch": 0.013683814040432308,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999193003754842,
      "loss": 2.3675,
      "step": 2014
    },
    {
      "epoch": 0.013690608387026366,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999919106972442,
      "loss": 2.4306,
      "step": 2015
    },
    {
      "epoch": 0.013697402733620422,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999189133379424,
      "loss": 2.423,
      "step": 2016
    },
    {
      "epoch": 0.01370419708021448,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009999187194719864,
      "loss": 2.537,
      "step": 2017
    },
    {
      "epoch": 0.013710991426808539,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009999185253745737,
      "loss": 2.4871,
      "step": 2018
    },
    {
      "epoch": 0.013717785773402597,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999183310457046,
      "loss": 2.6225,
      "step": 2019
    },
    {
      "epoch": 0.013724580119996653,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999181364853786,
      "loss": 2.6145,
      "step": 2020
    },
    {
      "epoch": 0.013731374466590712,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999179416935967,
      "loss": 2.4893,
      "step": 2021
    },
    {
      "epoch": 0.01373816881318477,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999177466703583,
      "loss": 2.5433,
      "step": 2022
    },
    {
      "epoch": 0.013744963159778826,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999175514156637,
      "loss": 2.4249,
      "step": 2023
    },
    {
      "epoch": 0.013751757506372884,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009999173559295132,
      "loss": 2.6069,
      "step": 2024
    },
    {
      "epoch": 0.013758551852966943,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999171602119066,
      "loss": 2.452,
      "step": 2025
    },
    {
      "epoch": 0.013765346199561001,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999169642628438,
      "loss": 2.5681,
      "step": 2026
    },
    {
      "epoch": 0.013772140546155057,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009999167680823256,
      "loss": 2.4483,
      "step": 2027
    },
    {
      "epoch": 0.013778934892749116,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999165716703515,
      "loss": 2.5214,
      "step": 2028
    },
    {
      "epoch": 0.013785729239343174,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009999163750269216,
      "loss": 2.5926,
      "step": 2029
    },
    {
      "epoch": 0.013792523585937232,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009999161781520363,
      "loss": 2.4865,
      "step": 2030
    },
    {
      "epoch": 0.013799317932531288,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999159810456955,
      "loss": 2.4913,
      "step": 2031
    },
    {
      "epoch": 0.013806112279125347,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009999157837078994,
      "loss": 2.5492,
      "step": 2032
    },
    {
      "epoch": 0.013812906625719405,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009999155861386481,
      "loss": 2.5492,
      "step": 2033
    },
    {
      "epoch": 0.013819700972313463,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009999153883379415,
      "loss": 2.5942,
      "step": 2034
    },
    {
      "epoch": 0.01382649531890752,
      "grad_norm": 0.1640625,
      "learning_rate": 0.00099991519030578,
      "loss": 2.4374,
      "step": 2035
    },
    {
      "epoch": 0.013833289665501578,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999914992042163,
      "loss": 2.5787,
      "step": 2036
    },
    {
      "epoch": 0.013840084012095636,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009999147935470915,
      "loss": 2.5233,
      "step": 2037
    },
    {
      "epoch": 0.013846878358689692,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999145948205652,
      "loss": 2.4482,
      "step": 2038
    },
    {
      "epoch": 0.01385367270528375,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999914395862584,
      "loss": 2.5095,
      "step": 2039
    },
    {
      "epoch": 0.013860467051877809,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999141966731482,
      "loss": 2.5839,
      "step": 2040
    },
    {
      "epoch": 0.013867261398471867,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999913997252258,
      "loss": 2.5238,
      "step": 2041
    },
    {
      "epoch": 0.013874055745065924,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999913797599913,
      "loss": 2.4936,
      "step": 2042
    },
    {
      "epoch": 0.013880850091659982,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999913597716114,
      "loss": 2.4948,
      "step": 2043
    },
    {
      "epoch": 0.01388764443825404,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999133976008605,
      "loss": 2.5413,
      "step": 2044
    },
    {
      "epoch": 0.013894438784848098,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009999131972541532,
      "loss": 2.5887,
      "step": 2045
    },
    {
      "epoch": 0.013901233131442155,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009999129966759915,
      "loss": 2.4182,
      "step": 2046
    },
    {
      "epoch": 0.013908027478036213,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999127958663758,
      "loss": 2.4892,
      "step": 2047
    },
    {
      "epoch": 0.013914821824630271,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999125948253064,
      "loss": 2.5611,
      "step": 2048
    },
    {
      "epoch": 0.01392161617122433,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.000999912393552783,
      "loss": 2.4854,
      "step": 2049
    },
    {
      "epoch": 0.013928410517818386,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009999121920488058,
      "loss": 2.452,
      "step": 2050
    },
    {
      "epoch": 0.013935204864412444,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009999119903133754,
      "loss": 2.5697,
      "step": 2051
    },
    {
      "epoch": 0.013941999211006502,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009999117883464912,
      "loss": 2.5267,
      "step": 2052
    },
    {
      "epoch": 0.013948793557600559,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009999115861481535,
      "loss": 2.5699,
      "step": 2053
    },
    {
      "epoch": 0.013955587904194617,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009999113837183627,
      "loss": 2.49,
      "step": 2054
    },
    {
      "epoch": 0.013962382250788675,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999111810571183,
      "loss": 2.4486,
      "step": 2055
    },
    {
      "epoch": 0.013969176597382733,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009999109781644211,
      "loss": 2.447,
      "step": 2056
    },
    {
      "epoch": 0.01397597094397679,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009999107750402708,
      "loss": 2.3994,
      "step": 2057
    },
    {
      "epoch": 0.013982765290570848,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009999105716846674,
      "loss": 2.4314,
      "step": 2058
    },
    {
      "epoch": 0.013989559637164906,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009999103680976113,
      "loss": 2.5847,
      "step": 2059
    },
    {
      "epoch": 0.013996353983758964,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999101642791024,
      "loss": 2.4736,
      "step": 2060
    },
    {
      "epoch": 0.014003148330353021,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999099602291408,
      "loss": 2.5872,
      "step": 2061
    },
    {
      "epoch": 0.01400994267694708,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999097559477265,
      "loss": 2.6536,
      "step": 2062
    },
    {
      "epoch": 0.014016737023541137,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.00099990955143486,
      "loss": 2.6006,
      "step": 2063
    },
    {
      "epoch": 0.014023531370135196,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009999093466905409,
      "loss": 2.4067,
      "step": 2064
    },
    {
      "epoch": 0.014030325716729252,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009999091417147695,
      "loss": 2.7735,
      "step": 2065
    },
    {
      "epoch": 0.01403712006332331,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999089365075459,
      "loss": 2.5009,
      "step": 2066
    },
    {
      "epoch": 0.014043914409917369,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009999087310688702,
      "loss": 2.5844,
      "step": 2067
    },
    {
      "epoch": 0.014050708756511425,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999085253987426,
      "loss": 2.5638,
      "step": 2068
    },
    {
      "epoch": 0.014057503103105483,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999908319497163,
      "loss": 2.579,
      "step": 2069
    },
    {
      "epoch": 0.014064297449699541,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999081133641315,
      "loss": 2.5324,
      "step": 2070
    },
    {
      "epoch": 0.0140710917962936,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999079069996483,
      "loss": 2.4432,
      "step": 2071
    },
    {
      "epoch": 0.014077886142887656,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999077004037135,
      "loss": 2.5123,
      "step": 2072
    },
    {
      "epoch": 0.014084680489481714,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009999074935763273,
      "loss": 2.4771,
      "step": 2073
    },
    {
      "epoch": 0.014091474836075773,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999072865174894,
      "loss": 2.5707,
      "step": 2074
    },
    {
      "epoch": 0.01409826918266983,
      "grad_norm": 0.25,
      "learning_rate": 0.0009999070792272003,
      "loss": 2.6653,
      "step": 2075
    },
    {
      "epoch": 0.014105063529263887,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.00099990687170546,
      "loss": 2.4295,
      "step": 2076
    },
    {
      "epoch": 0.014111857875857945,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009999066639522684,
      "loss": 2.4679,
      "step": 2077
    },
    {
      "epoch": 0.014118652222452004,
      "grad_norm": 0.19140625,
      "learning_rate": 0.000999906455967626,
      "loss": 2.5484,
      "step": 2078
    },
    {
      "epoch": 0.014125446569046062,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999062477515325,
      "loss": 2.4676,
      "step": 2079
    },
    {
      "epoch": 0.014132240915640118,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999906039303988,
      "loss": 2.4617,
      "step": 2080
    },
    {
      "epoch": 0.014139035262234177,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.000999905830624993,
      "loss": 2.5623,
      "step": 2081
    },
    {
      "epoch": 0.014145829608828235,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999056217145472,
      "loss": 2.622,
      "step": 2082
    },
    {
      "epoch": 0.014152623955422291,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009999054125726507,
      "loss": 2.4889,
      "step": 2083
    },
    {
      "epoch": 0.01415941830201635,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999052031993039,
      "loss": 2.5035,
      "step": 2084
    },
    {
      "epoch": 0.014166212648610408,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009999049935945066,
      "loss": 2.6188,
      "step": 2085
    },
    {
      "epoch": 0.014173006995204466,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999904783758259,
      "loss": 2.5124,
      "step": 2086
    },
    {
      "epoch": 0.014179801341798522,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009999045736905614,
      "loss": 2.5937,
      "step": 2087
    },
    {
      "epoch": 0.01418659568839258,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009999043633914138,
      "loss": 2.4566,
      "step": 2088
    },
    {
      "epoch": 0.014193390034986639,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009999041528608158,
      "loss": 2.4399,
      "step": 2089
    },
    {
      "epoch": 0.014200184381580697,
      "grad_norm": 0.23828125,
      "learning_rate": 0.0009999039420987683,
      "loss": 2.7274,
      "step": 2090
    },
    {
      "epoch": 0.014206978728174753,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009999037311052708,
      "loss": 2.5352,
      "step": 2091
    },
    {
      "epoch": 0.014213773074768812,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009999035198803236,
      "loss": 2.4615,
      "step": 2092
    },
    {
      "epoch": 0.01422056742136287,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999903308423927,
      "loss": 2.5524,
      "step": 2093
    },
    {
      "epoch": 0.014227361767956928,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009999030967360806,
      "loss": 2.5317,
      "step": 2094
    },
    {
      "epoch": 0.014234156114550985,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999902884816785,
      "loss": 2.5833,
      "step": 2095
    },
    {
      "epoch": 0.014240950461145043,
      "grad_norm": 0.171875,
      "learning_rate": 0.00099990267266604,
      "loss": 2.8311,
      "step": 2096
    },
    {
      "epoch": 0.014247744807739101,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999902460283846,
      "loss": 2.3695,
      "step": 2097
    },
    {
      "epoch": 0.014254539154333157,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009999022476702027,
      "loss": 2.4098,
      "step": 2098
    },
    {
      "epoch": 0.014261333500927216,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999020348251105,
      "loss": 2.4746,
      "step": 2099
    },
    {
      "epoch": 0.014268127847521274,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999018217485692,
      "loss": 2.568,
      "step": 2100
    },
    {
      "epoch": 0.014274922194115332,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009999016084405793,
      "loss": 2.6766,
      "step": 2101
    },
    {
      "epoch": 0.014281716540709389,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009999013949011403,
      "loss": 2.3248,
      "step": 2102
    },
    {
      "epoch": 0.014288510887303447,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009999011811302532,
      "loss": 2.5323,
      "step": 2103
    },
    {
      "epoch": 0.014295305233897505,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009999009671279174,
      "loss": 2.5146,
      "step": 2104
    },
    {
      "epoch": 0.014302099580491563,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000999900752894133,
      "loss": 2.3099,
      "step": 2105
    },
    {
      "epoch": 0.01430889392708562,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009999005384289004,
      "loss": 2.4054,
      "step": 2106
    },
    {
      "epoch": 0.014315688273679678,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009999003237322196,
      "loss": 2.4931,
      "step": 2107
    },
    {
      "epoch": 0.014322482620273736,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009999001088040907,
      "loss": 2.5905,
      "step": 2108
    },
    {
      "epoch": 0.014329276966867794,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999899893644514,
      "loss": 2.4226,
      "step": 2109
    },
    {
      "epoch": 0.01433607131346185,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999899678253489,
      "loss": 2.537,
      "step": 2110
    },
    {
      "epoch": 0.014342865660055909,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998994626310162,
      "loss": 2.3555,
      "step": 2111
    },
    {
      "epoch": 0.014349660006649967,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998992467770958,
      "loss": 2.6346,
      "step": 2112
    },
    {
      "epoch": 0.014356454353244024,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009998990306917279,
      "loss": 2.5121,
      "step": 2113
    },
    {
      "epoch": 0.014363248699838082,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009998988143749123,
      "loss": 2.4632,
      "step": 2114
    },
    {
      "epoch": 0.01437004304643214,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998985978266492,
      "loss": 2.439,
      "step": 2115
    },
    {
      "epoch": 0.014376837393026198,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998983810469392,
      "loss": 2.462,
      "step": 2116
    },
    {
      "epoch": 0.014383631739620255,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998981640357816,
      "loss": 2.5072,
      "step": 2117
    },
    {
      "epoch": 0.014390426086214313,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009998979467931769,
      "loss": 2.3255,
      "step": 2118
    },
    {
      "epoch": 0.014397220432808371,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009998977293191253,
      "loss": 2.6009,
      "step": 2119
    },
    {
      "epoch": 0.01440401477940243,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009998975116136267,
      "loss": 2.6384,
      "step": 2120
    },
    {
      "epoch": 0.014410809125996486,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998972936766813,
      "loss": 2.6351,
      "step": 2121
    },
    {
      "epoch": 0.014417603472590544,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998970755082893,
      "loss": 2.573,
      "step": 2122
    },
    {
      "epoch": 0.014424397819184602,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998968571084505,
      "loss": 2.5757,
      "step": 2123
    },
    {
      "epoch": 0.01443119216577866,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009998966384771654,
      "loss": 2.4654,
      "step": 2124
    },
    {
      "epoch": 0.014437986512372717,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009998964196144338,
      "loss": 2.5275,
      "step": 2125
    },
    {
      "epoch": 0.014444780858966775,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999896200520256,
      "loss": 2.5507,
      "step": 2126
    },
    {
      "epoch": 0.014451575205560833,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998959811946316,
      "loss": 2.6245,
      "step": 2127
    },
    {
      "epoch": 0.01445836955215489,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998957616375614,
      "loss": 2.6327,
      "step": 2128
    },
    {
      "epoch": 0.014465163898748948,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009998955418490452,
      "loss": 2.5147,
      "step": 2129
    },
    {
      "epoch": 0.014471958245343006,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999895321829083,
      "loss": 2.425,
      "step": 2130
    },
    {
      "epoch": 0.014478752591937065,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999895101577675,
      "loss": 2.5019,
      "step": 2131
    },
    {
      "epoch": 0.014485546938531121,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009998948810948213,
      "loss": 2.5224,
      "step": 2132
    },
    {
      "epoch": 0.01449234128512518,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999894660380522,
      "loss": 2.5463,
      "step": 2133
    },
    {
      "epoch": 0.014499135631719237,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009998944394347773,
      "loss": 2.7299,
      "step": 2134
    },
    {
      "epoch": 0.014505929978313296,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998942182575872,
      "loss": 2.4817,
      "step": 2135
    },
    {
      "epoch": 0.014512724324907352,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998939968489519,
      "loss": 2.5593,
      "step": 2136
    },
    {
      "epoch": 0.01451951867150141,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999893775208871,
      "loss": 2.3293,
      "step": 2137
    },
    {
      "epoch": 0.014526313018095469,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009998935533373453,
      "loss": 2.3836,
      "step": 2138
    },
    {
      "epoch": 0.014533107364689527,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009998933312343747,
      "loss": 2.5754,
      "step": 2139
    },
    {
      "epoch": 0.014539901711283583,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998931088999589,
      "loss": 2.5659,
      "step": 2140
    },
    {
      "epoch": 0.014546696057877642,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998928863340986,
      "loss": 2.4524,
      "step": 2141
    },
    {
      "epoch": 0.0145534904044717,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009998926635367937,
      "loss": 2.4239,
      "step": 2142
    },
    {
      "epoch": 0.014560284751065756,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009998924405080442,
      "loss": 2.455,
      "step": 2143
    },
    {
      "epoch": 0.014567079097659814,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00099989221724785,
      "loss": 2.4561,
      "step": 2144
    },
    {
      "epoch": 0.014573873444253873,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009998919937562114,
      "loss": 2.449,
      "step": 2145
    },
    {
      "epoch": 0.01458066779084793,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998917700331288,
      "loss": 2.4347,
      "step": 2146
    },
    {
      "epoch": 0.014587462137441987,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999891546078602,
      "loss": 2.4218,
      "step": 2147
    },
    {
      "epoch": 0.014594256484036046,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999891321892631,
      "loss": 2.4932,
      "step": 2148
    },
    {
      "epoch": 0.014601050830630104,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999891097475216,
      "loss": 2.4849,
      "step": 2149
    },
    {
      "epoch": 0.014607845177224162,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998908728263574,
      "loss": 2.5091,
      "step": 2150
    },
    {
      "epoch": 0.014614639523818218,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998906479460547,
      "loss": 2.3915,
      "step": 2151
    },
    {
      "epoch": 0.014621433870412277,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009998904228343088,
      "loss": 2.4486,
      "step": 2152
    },
    {
      "epoch": 0.014628228217006335,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009998901974911192,
      "loss": 2.6125,
      "step": 2153
    },
    {
      "epoch": 0.014635022563600393,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999889971916486,
      "loss": 2.5845,
      "step": 2154
    },
    {
      "epoch": 0.01464181691019445,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998897461104097,
      "loss": 2.5164,
      "step": 2155
    },
    {
      "epoch": 0.014648611256788508,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00099988952007289,
      "loss": 2.491,
      "step": 2156
    },
    {
      "epoch": 0.014655405603382566,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998892938039273,
      "loss": 2.5607,
      "step": 2157
    },
    {
      "epoch": 0.014662199949976622,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009998890673035214,
      "loss": 2.6482,
      "step": 2158
    },
    {
      "epoch": 0.01466899429657068,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998888405716728,
      "loss": 2.5503,
      "step": 2159
    },
    {
      "epoch": 0.014675788643164739,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998886136083813,
      "loss": 2.5157,
      "step": 2160
    },
    {
      "epoch": 0.014682582989758797,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009998883864136472,
      "loss": 2.5242,
      "step": 2161
    },
    {
      "epoch": 0.014689377336352854,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998881589874705,
      "loss": 2.5402,
      "step": 2162
    },
    {
      "epoch": 0.014696171682946912,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009998879313298512,
      "loss": 2.639,
      "step": 2163
    },
    {
      "epoch": 0.01470296602954097,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009998877034407896,
      "loss": 2.5286,
      "step": 2164
    },
    {
      "epoch": 0.014709760376135028,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009998874753202858,
      "loss": 2.5999,
      "step": 2165
    },
    {
      "epoch": 0.014716554722729085,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009998872469683397,
      "loss": 2.5262,
      "step": 2166
    },
    {
      "epoch": 0.014723349069323143,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998870183849516,
      "loss": 2.4646,
      "step": 2167
    },
    {
      "epoch": 0.014730143415917201,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998867895701215,
      "loss": 2.3842,
      "step": 2168
    },
    {
      "epoch": 0.01473693776251126,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998865605238495,
      "loss": 2.413,
      "step": 2169
    },
    {
      "epoch": 0.014743732109105316,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998863312461357,
      "loss": 2.3489,
      "step": 2170
    },
    {
      "epoch": 0.014750526455699374,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998861017369806,
      "loss": 2.669,
      "step": 2171
    },
    {
      "epoch": 0.014757320802293432,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998858719963838,
      "loss": 2.4072,
      "step": 2172
    },
    {
      "epoch": 0.014764115148887489,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009998856420243454,
      "loss": 2.396,
      "step": 2173
    },
    {
      "epoch": 0.014770909495481547,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998854118208659,
      "loss": 2.4118,
      "step": 2174
    },
    {
      "epoch": 0.014777703842075605,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999885181385945,
      "loss": 2.4806,
      "step": 2175
    },
    {
      "epoch": 0.014784498188669663,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999884950719583,
      "loss": 2.6117,
      "step": 2176
    },
    {
      "epoch": 0.01479129253526372,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.00099988471982178,
      "loss": 2.4919,
      "step": 2177
    },
    {
      "epoch": 0.014798086881857778,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998844886925362,
      "loss": 2.5345,
      "step": 2178
    },
    {
      "epoch": 0.014804881228451836,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998842573318517,
      "loss": 2.5274,
      "step": 2179
    },
    {
      "epoch": 0.014811675575045894,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998840257397265,
      "loss": 2.5066,
      "step": 2180
    },
    {
      "epoch": 0.014818469921639951,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998837939161605,
      "loss": 2.6064,
      "step": 2181
    },
    {
      "epoch": 0.014825264268234009,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998835618611542,
      "loss": 2.6247,
      "step": 2182
    },
    {
      "epoch": 0.014832058614828067,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009998833295747074,
      "loss": 2.5729,
      "step": 2183
    },
    {
      "epoch": 0.014838852961422126,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998830970568205,
      "loss": 2.4526,
      "step": 2184
    },
    {
      "epoch": 0.014845647308016182,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998828643074935,
      "loss": 2.5505,
      "step": 2185
    },
    {
      "epoch": 0.01485244165461024,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009998826313267263,
      "loss": 2.5454,
      "step": 2186
    },
    {
      "epoch": 0.014859236001204298,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998823981145193,
      "loss": 2.5319,
      "step": 2187
    },
    {
      "epoch": 0.014866030347798355,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998821646708723,
      "loss": 2.3783,
      "step": 2188
    },
    {
      "epoch": 0.014872824694392413,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998819309957859,
      "loss": 2.3592,
      "step": 2189
    },
    {
      "epoch": 0.014879619040986471,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009998816970892596,
      "loss": 2.3519,
      "step": 2190
    },
    {
      "epoch": 0.01488641338758053,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009998814629512938,
      "loss": 2.4069,
      "step": 2191
    },
    {
      "epoch": 0.014893207734174586,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998812285818889,
      "loss": 2.6588,
      "step": 2192
    },
    {
      "epoch": 0.014900002080768644,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998809939810445,
      "loss": 2.3189,
      "step": 2193
    },
    {
      "epoch": 0.014906796427362702,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999880759148761,
      "loss": 2.4787,
      "step": 2194
    },
    {
      "epoch": 0.01491359077395676,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998805240850385,
      "loss": 2.4193,
      "step": 2195
    },
    {
      "epoch": 0.014920385120550817,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009998802887898769,
      "loss": 2.5115,
      "step": 2196
    },
    {
      "epoch": 0.014927179467144875,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998800532632767,
      "loss": 2.566,
      "step": 2197
    },
    {
      "epoch": 0.014933973813738934,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998798175052375,
      "loss": 2.6297,
      "step": 2198
    },
    {
      "epoch": 0.014940768160332992,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998795815157598,
      "loss": 2.5822,
      "step": 2199
    },
    {
      "epoch": 0.014947562506927048,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009998793452948435,
      "loss": 2.4448,
      "step": 2200
    },
    {
      "epoch": 0.014954356853521106,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998791088424888,
      "loss": 2.4946,
      "step": 2201
    },
    {
      "epoch": 0.014961151200115165,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998788721586958,
      "loss": 2.5128,
      "step": 2202
    },
    {
      "epoch": 0.014967945546709221,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009998786352434649,
      "loss": 2.4354,
      "step": 2203
    },
    {
      "epoch": 0.01497473989330328,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009998783980967956,
      "loss": 2.464,
      "step": 2204
    },
    {
      "epoch": 0.014981534239897338,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998781607186884,
      "loss": 2.6124,
      "step": 2205
    },
    {
      "epoch": 0.014988328586491396,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998779231091433,
      "loss": 2.367,
      "step": 2206
    },
    {
      "epoch": 0.014995122933085452,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998776852681605,
      "loss": 2.5448,
      "step": 2207
    },
    {
      "epoch": 0.01500191727967951,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998774471957402,
      "loss": 2.5675,
      "step": 2208
    },
    {
      "epoch": 0.015008711626273569,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009998772088918822,
      "loss": 2.4686,
      "step": 2209
    },
    {
      "epoch": 0.015015505972867627,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998769703565868,
      "loss": 2.5869,
      "step": 2210
    },
    {
      "epoch": 0.015022300319461683,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998767315898543,
      "loss": 2.5916,
      "step": 2211
    },
    {
      "epoch": 0.015029094666055742,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009998764925916843,
      "loss": 2.693,
      "step": 2212
    },
    {
      "epoch": 0.0150358890126498,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009998762533620774,
      "loss": 2.5858,
      "step": 2213
    },
    {
      "epoch": 0.015042683359243858,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998760139010333,
      "loss": 2.4737,
      "step": 2214
    },
    {
      "epoch": 0.015049477705837914,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009998757742085526,
      "loss": 2.4727,
      "step": 2215
    },
    {
      "epoch": 0.015056272052431973,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999875534284635,
      "loss": 2.4003,
      "step": 2216
    },
    {
      "epoch": 0.015063066399026031,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999875294129281,
      "loss": 2.4918,
      "step": 2217
    },
    {
      "epoch": 0.015069860745620087,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998750537424902,
      "loss": 2.595,
      "step": 2218
    },
    {
      "epoch": 0.015076655092214146,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999874813124263,
      "loss": 2.5208,
      "step": 2219
    },
    {
      "epoch": 0.015083449438808204,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998745722745997,
      "loss": 2.5465,
      "step": 2220
    },
    {
      "epoch": 0.015090243785402262,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998743311935,
      "loss": 2.4743,
      "step": 2221
    },
    {
      "epoch": 0.015097038131996319,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009998740898809642,
      "loss": 2.5221,
      "step": 2222
    },
    {
      "epoch": 0.015103832478590377,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009998738483369927,
      "loss": 2.5317,
      "step": 2223
    },
    {
      "epoch": 0.015110626825184435,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999873606561585,
      "loss": 2.4955,
      "step": 2224
    },
    {
      "epoch": 0.015117421171778493,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998733645547417,
      "loss": 2.452,
      "step": 2225
    },
    {
      "epoch": 0.01512421551837255,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009998731223164628,
      "loss": 2.62,
      "step": 2226
    },
    {
      "epoch": 0.015131009864966608,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009998728798467484,
      "loss": 2.501,
      "step": 2227
    },
    {
      "epoch": 0.015137804211560666,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998726371455984,
      "loss": 2.5955,
      "step": 2228
    },
    {
      "epoch": 0.015144598558154724,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998723942130133,
      "loss": 2.6662,
      "step": 2229
    },
    {
      "epoch": 0.01515139290474878,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999872151048993,
      "loss": 2.4452,
      "step": 2230
    },
    {
      "epoch": 0.015158187251342839,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998719076535374,
      "loss": 2.4063,
      "step": 2231
    },
    {
      "epoch": 0.015164981597936897,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999871664026647,
      "loss": 2.6533,
      "step": 2232
    },
    {
      "epoch": 0.015171775944530954,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998714201683218,
      "loss": 2.3874,
      "step": 2233
    },
    {
      "epoch": 0.015178570291125012,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000999871176078562,
      "loss": 2.5929,
      "step": 2234
    },
    {
      "epoch": 0.01518536463771907,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009998709317573674,
      "loss": 2.4718,
      "step": 2235
    },
    {
      "epoch": 0.015192158984313128,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998706872047381,
      "loss": 2.41,
      "step": 2236
    },
    {
      "epoch": 0.015198953330907185,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998704424206747,
      "loss": 2.4818,
      "step": 2237
    },
    {
      "epoch": 0.015205747677501243,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009998701974051768,
      "loss": 2.485,
      "step": 2238
    },
    {
      "epoch": 0.015212542024095301,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998699521582448,
      "loss": 2.5491,
      "step": 2239
    },
    {
      "epoch": 0.01521933637068936,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998697066798787,
      "loss": 2.5018,
      "step": 2240
    },
    {
      "epoch": 0.015226130717283416,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998694609700787,
      "loss": 2.4856,
      "step": 2241
    },
    {
      "epoch": 0.015232925063877474,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999869215028845,
      "loss": 2.4584,
      "step": 2242
    },
    {
      "epoch": 0.015239719410471532,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998689688561775,
      "loss": 2.5488,
      "step": 2243
    },
    {
      "epoch": 0.01524651375706559,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998687224520764,
      "loss": 2.3791,
      "step": 2244
    },
    {
      "epoch": 0.015253308103659647,
      "grad_norm": 0.28125,
      "learning_rate": 0.0009998684758165417,
      "loss": 2.4826,
      "step": 2245
    },
    {
      "epoch": 0.015260102450253705,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998682289495738,
      "loss": 2.4146,
      "step": 2246
    },
    {
      "epoch": 0.015266896796847763,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009998679818511726,
      "loss": 2.5857,
      "step": 2247
    },
    {
      "epoch": 0.01527369114344182,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009998677345213383,
      "loss": 2.5277,
      "step": 2248
    },
    {
      "epoch": 0.015280485490035878,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009998674869600708,
      "loss": 2.5356,
      "step": 2249
    },
    {
      "epoch": 0.015287279836629936,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998672391673704,
      "loss": 2.4612,
      "step": 2250
    },
    {
      "epoch": 0.015294074183223995,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998669911432372,
      "loss": 2.6845,
      "step": 2251
    },
    {
      "epoch": 0.015300868529818051,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009998667428876714,
      "loss": 2.7088,
      "step": 2252
    },
    {
      "epoch": 0.01530766287641211,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999866494400673,
      "loss": 2.2921,
      "step": 2253
    },
    {
      "epoch": 0.015314457223006167,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998662456822422,
      "loss": 2.4353,
      "step": 2254
    },
    {
      "epoch": 0.015321251569600226,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999865996732379,
      "loss": 2.4443,
      "step": 2255
    },
    {
      "epoch": 0.015328045916194282,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998657475510837,
      "loss": 2.5417,
      "step": 2256
    },
    {
      "epoch": 0.01533484026278834,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999865498138356,
      "loss": 2.4215,
      "step": 2257
    },
    {
      "epoch": 0.015341634609382399,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009998652484941965,
      "loss": 2.4794,
      "step": 2258
    },
    {
      "epoch": 0.015348428955976457,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998649986186052,
      "loss": 2.6019,
      "step": 2259
    },
    {
      "epoch": 0.015355223302570513,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999864748511582,
      "loss": 2.6094,
      "step": 2260
    },
    {
      "epoch": 0.015362017649164571,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009998644981731271,
      "loss": 2.3438,
      "step": 2261
    },
    {
      "epoch": 0.01536881199575863,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999864247603241,
      "loss": 2.451,
      "step": 2262
    },
    {
      "epoch": 0.015375606342352686,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999863996801923,
      "loss": 2.5671,
      "step": 2263
    },
    {
      "epoch": 0.015382400688946744,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999863745769174,
      "loss": 2.4355,
      "step": 2264
    },
    {
      "epoch": 0.015389195035540803,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009998634945049937,
      "loss": 2.6717,
      "step": 2265
    },
    {
      "epoch": 0.01539598938213486,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998632430093825,
      "loss": 2.5682,
      "step": 2266
    },
    {
      "epoch": 0.015402783728728917,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009998629912823403,
      "loss": 2.3739,
      "step": 2267
    },
    {
      "epoch": 0.015409578075322975,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998627393238671,
      "loss": 2.5518,
      "step": 2268
    },
    {
      "epoch": 0.015416372421917034,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998624871339632,
      "loss": 2.5159,
      "step": 2269
    },
    {
      "epoch": 0.015423166768511092,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000999862234712629,
      "loss": 2.625,
      "step": 2270
    },
    {
      "epoch": 0.015429961115105148,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999861982059864,
      "loss": 2.4711,
      "step": 2271
    },
    {
      "epoch": 0.015436755461699207,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998617291756687,
      "loss": 2.5488,
      "step": 2272
    },
    {
      "epoch": 0.015443549808293265,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009998614760600433,
      "loss": 2.4987,
      "step": 2273
    },
    {
      "epoch": 0.015450344154887323,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998612227129875,
      "loss": 2.6985,
      "step": 2274
    },
    {
      "epoch": 0.01545713850148138,
      "grad_norm": 0.177734375,
      "learning_rate": 0.000999860969134502,
      "loss": 2.3829,
      "step": 2275
    },
    {
      "epoch": 0.015463932848075438,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009998607153245864,
      "loss": 2.4886,
      "step": 2276
    },
    {
      "epoch": 0.015470727194669496,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999860461283241,
      "loss": 2.5671,
      "step": 2277
    },
    {
      "epoch": 0.015477521541263552,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999860207010466,
      "loss": 2.5857,
      "step": 2278
    },
    {
      "epoch": 0.01548431588785761,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998599525062614,
      "loss": 2.3412,
      "step": 2279
    },
    {
      "epoch": 0.015491110234451669,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998596977706274,
      "loss": 2.5017,
      "step": 2280
    },
    {
      "epoch": 0.015497904581045727,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999859442803564,
      "loss": 2.3989,
      "step": 2281
    },
    {
      "epoch": 0.015504698927639783,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998591876050717,
      "loss": 2.545,
      "step": 2282
    },
    {
      "epoch": 0.015511493274233842,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998589321751502,
      "loss": 2.42,
      "step": 2283
    },
    {
      "epoch": 0.0155182876208279,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009998586765137996,
      "loss": 2.4718,
      "step": 2284
    },
    {
      "epoch": 0.015525081967421958,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998584206210204,
      "loss": 2.5292,
      "step": 2285
    },
    {
      "epoch": 0.015531876314016015,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009998581644968122,
      "loss": 2.502,
      "step": 2286
    },
    {
      "epoch": 0.015538670660610073,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998579081411756,
      "loss": 2.3626,
      "step": 2287
    },
    {
      "epoch": 0.015545465007204131,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998576515541104,
      "loss": 2.5235,
      "step": 2288
    },
    {
      "epoch": 0.01555225935379819,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009998573947356169,
      "loss": 2.4626,
      "step": 2289
    },
    {
      "epoch": 0.015559053700392246,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998571376856951,
      "loss": 2.5169,
      "step": 2290
    },
    {
      "epoch": 0.015565848046986304,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998568804043453,
      "loss": 2.5019,
      "step": 2291
    },
    {
      "epoch": 0.015572642393580362,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009998566228915675,
      "loss": 2.4656,
      "step": 2292
    },
    {
      "epoch": 0.015579436740174419,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998563651473617,
      "loss": 2.5495,
      "step": 2293
    },
    {
      "epoch": 0.015586231086768477,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009998561071717282,
      "loss": 2.4398,
      "step": 2294
    },
    {
      "epoch": 0.015593025433362535,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999855848964667,
      "loss": 2.5587,
      "step": 2295
    },
    {
      "epoch": 0.015599819779956593,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998555905261783,
      "loss": 2.4193,
      "step": 2296
    },
    {
      "epoch": 0.01560661412655065,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999855331856262,
      "loss": 2.3749,
      "step": 2297
    },
    {
      "epoch": 0.015613408473144708,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009998550729549188,
      "loss": 2.4803,
      "step": 2298
    },
    {
      "epoch": 0.015620202819738766,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998548138221482,
      "loss": 2.5001,
      "step": 2299
    },
    {
      "epoch": 0.015626997166332823,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009998545544579506,
      "loss": 2.7227,
      "step": 2300
    },
    {
      "epoch": 0.015633791512926883,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999854294862326,
      "loss": 2.6164,
      "step": 2301
    },
    {
      "epoch": 0.01564058585952094,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998540350352747,
      "loss": 2.4594,
      "step": 2302
    },
    {
      "epoch": 0.015647380206114996,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998537749767967,
      "loss": 2.4624,
      "step": 2303
    },
    {
      "epoch": 0.015654174552709055,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999853514686892,
      "loss": 2.7168,
      "step": 2304
    },
    {
      "epoch": 0.015660968899303112,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999853254165561,
      "loss": 2.4882,
      "step": 2305
    },
    {
      "epoch": 0.01566776324589717,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998529934128035,
      "loss": 2.5107,
      "step": 2306
    },
    {
      "epoch": 0.01567455759249123,
      "grad_norm": 0.166015625,
      "learning_rate": 0.00099985273242862,
      "loss": 2.5698,
      "step": 2307
    },
    {
      "epoch": 0.015681351939085285,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998524712130103,
      "loss": 2.3421,
      "step": 2308
    },
    {
      "epoch": 0.015688146285679345,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009998522097659748,
      "loss": 2.6694,
      "step": 2309
    },
    {
      "epoch": 0.0156949406322734,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009998519480875132,
      "loss": 2.6905,
      "step": 2310
    },
    {
      "epoch": 0.015701734978867458,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998516861776262,
      "loss": 2.5415,
      "step": 2311
    },
    {
      "epoch": 0.015708529325461518,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009998514240363132,
      "loss": 2.4483,
      "step": 2312
    },
    {
      "epoch": 0.015715323672055574,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000999851161663575,
      "loss": 2.6412,
      "step": 2313
    },
    {
      "epoch": 0.01572211801864963,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998508990594114,
      "loss": 2.5525,
      "step": 2314
    },
    {
      "epoch": 0.01572891236524369,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009998506362238224,
      "loss": 2.3833,
      "step": 2315
    },
    {
      "epoch": 0.015735706711837747,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009998503731568084,
      "loss": 2.48,
      "step": 2316
    },
    {
      "epoch": 0.015742501058431807,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998501098583696,
      "loss": 2.5133,
      "step": 2317
    },
    {
      "epoch": 0.015749295405025864,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009998498463285057,
      "loss": 2.4552,
      "step": 2318
    },
    {
      "epoch": 0.01575608975161992,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999849582567217,
      "loss": 2.4368,
      "step": 2319
    },
    {
      "epoch": 0.01576288409821398,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999849318574504,
      "loss": 2.4183,
      "step": 2320
    },
    {
      "epoch": 0.015769678444808036,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998490543503661,
      "loss": 2.5579,
      "step": 2321
    },
    {
      "epoch": 0.015776472791402093,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999848789894804,
      "loss": 2.4901,
      "step": 2322
    },
    {
      "epoch": 0.015783267137996153,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998485252078178,
      "loss": 2.4862,
      "step": 2323
    },
    {
      "epoch": 0.01579006148459021,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998482602894072,
      "loss": 2.5745,
      "step": 2324
    },
    {
      "epoch": 0.015796855831184266,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998479951395726,
      "loss": 2.4131,
      "step": 2325
    },
    {
      "epoch": 0.015803650177778326,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998477297583142,
      "loss": 2.5676,
      "step": 2326
    },
    {
      "epoch": 0.015810444524372382,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009998474641456322,
      "loss": 2.4766,
      "step": 2327
    },
    {
      "epoch": 0.015817238870966442,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998471983015263,
      "loss": 2.5706,
      "step": 2328
    },
    {
      "epoch": 0.0158240332175605,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999846932225997,
      "loss": 2.4713,
      "step": 2329
    },
    {
      "epoch": 0.015830827564154555,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999846665919044,
      "loss": 2.4788,
      "step": 2330
    },
    {
      "epoch": 0.015837621910748615,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999846399380668,
      "loss": 2.6587,
      "step": 2331
    },
    {
      "epoch": 0.01584441625734267,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998461326108689,
      "loss": 2.3546,
      "step": 2332
    },
    {
      "epoch": 0.015851210603936728,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998458656096468,
      "loss": 2.5032,
      "step": 2333
    },
    {
      "epoch": 0.015858004950530788,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009998455983770017,
      "loss": 2.5015,
      "step": 2334
    },
    {
      "epoch": 0.015864799297124844,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998453309129338,
      "loss": 2.5583,
      "step": 2335
    },
    {
      "epoch": 0.0158715936437189,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998450632174433,
      "loss": 2.4191,
      "step": 2336
    },
    {
      "epoch": 0.01587838799031296,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009998447952905302,
      "loss": 2.6782,
      "step": 2337
    },
    {
      "epoch": 0.015885182336907017,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998445271321948,
      "loss": 2.5973,
      "step": 2338
    },
    {
      "epoch": 0.015891976683501077,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999844258742437,
      "loss": 2.5806,
      "step": 2339
    },
    {
      "epoch": 0.015898771030095134,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999843990121257,
      "loss": 2.5036,
      "step": 2340
    },
    {
      "epoch": 0.01590556537668919,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999843721268655,
      "loss": 2.5338,
      "step": 2341
    },
    {
      "epoch": 0.01591235972328325,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009998434521846311,
      "loss": 2.6669,
      "step": 2342
    },
    {
      "epoch": 0.015919154069877307,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998431828691855,
      "loss": 2.5282,
      "step": 2343
    },
    {
      "epoch": 0.015925948416471363,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999842913322318,
      "loss": 2.5125,
      "step": 2344
    },
    {
      "epoch": 0.015932742763065423,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009998426435440294,
      "loss": 2.5165,
      "step": 2345
    },
    {
      "epoch": 0.01593953710965948,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998423735343188,
      "loss": 2.6151,
      "step": 2346
    },
    {
      "epoch": 0.01594633145625354,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0009998421032931875,
      "loss": 2.4584,
      "step": 2347
    },
    {
      "epoch": 0.015953125802847596,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009998418328206347,
      "loss": 2.3959,
      "step": 2348
    },
    {
      "epoch": 0.015959920149441652,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999841562116661,
      "loss": 2.5203,
      "step": 2349
    },
    {
      "epoch": 0.015966714496035712,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009998412911812662,
      "loss": 2.5221,
      "step": 2350
    },
    {
      "epoch": 0.01597350884262977,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998410200144508,
      "loss": 2.573,
      "step": 2351
    },
    {
      "epoch": 0.015980303189223825,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009998407486162147,
      "loss": 2.4689,
      "step": 2352
    },
    {
      "epoch": 0.015987097535817885,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000999840476986558,
      "loss": 2.5316,
      "step": 2353
    },
    {
      "epoch": 0.015993891882411942,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000999840205125481,
      "loss": 2.4173,
      "step": 2354
    },
    {
      "epoch": 0.016000686229006,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998399330329835,
      "loss": 2.4556,
      "step": 2355
    },
    {
      "epoch": 0.016007480575600058,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999839660709066,
      "loss": 2.5463,
      "step": 2356
    },
    {
      "epoch": 0.016014274922194115,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009998393881537285,
      "loss": 2.5003,
      "step": 2357
    },
    {
      "epoch": 0.016021069268788175,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999839115366971,
      "loss": 2.5066,
      "step": 2358
    },
    {
      "epoch": 0.01602786361538223,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998388423487937,
      "loss": 2.5561,
      "step": 2359
    },
    {
      "epoch": 0.016034657961976288,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998385690991969,
      "loss": 2.4723,
      "step": 2360
    },
    {
      "epoch": 0.016041452308570348,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998382956181804,
      "loss": 2.4169,
      "step": 2361
    },
    {
      "epoch": 0.016048246655164404,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998380219057447,
      "loss": 2.4627,
      "step": 2362
    },
    {
      "epoch": 0.01605504100175846,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009998377479618894,
      "loss": 2.6305,
      "step": 2363
    },
    {
      "epoch": 0.01606183534835252,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009998374737866151,
      "loss": 2.4351,
      "step": 2364
    },
    {
      "epoch": 0.016068629694946577,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999837199379922,
      "loss": 2.2586,
      "step": 2365
    },
    {
      "epoch": 0.016075424041540633,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009998369247418099,
      "loss": 2.6484,
      "step": 2366
    },
    {
      "epoch": 0.016082218388134693,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998366498722789,
      "loss": 2.5547,
      "step": 2367
    },
    {
      "epoch": 0.01608901273472875,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998363747713292,
      "loss": 2.428,
      "step": 2368
    },
    {
      "epoch": 0.01609580708132281,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009998360994389612,
      "loss": 2.6323,
      "step": 2369
    },
    {
      "epoch": 0.016102601427916866,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998358238751748,
      "loss": 2.4871,
      "step": 2370
    },
    {
      "epoch": 0.016109395774510923,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00099983554807997,
      "loss": 2.3781,
      "step": 2371
    },
    {
      "epoch": 0.016116190121104983,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009998352720533472,
      "loss": 2.4037,
      "step": 2372
    },
    {
      "epoch": 0.01612298446769904,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009998349957953062,
      "loss": 2.6112,
      "step": 2373
    },
    {
      "epoch": 0.016129778814293096,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998347193058478,
      "loss": 2.4987,
      "step": 2374
    },
    {
      "epoch": 0.016136573160887156,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998344425849713,
      "loss": 2.6261,
      "step": 2375
    },
    {
      "epoch": 0.016143367507481212,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999834165632677,
      "loss": 2.4926,
      "step": 2376
    },
    {
      "epoch": 0.016150161854075272,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998338884489655,
      "loss": 2.4737,
      "step": 2377
    },
    {
      "epoch": 0.01615695620066933,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998336110338367,
      "loss": 2.5199,
      "step": 2378
    },
    {
      "epoch": 0.016163750547263385,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998333333872905,
      "loss": 2.5547,
      "step": 2379
    },
    {
      "epoch": 0.016170544893857445,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.000999833055509327,
      "loss": 2.3919,
      "step": 2380
    },
    {
      "epoch": 0.0161773392404515,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999832777399947,
      "loss": 2.3592,
      "step": 2381
    },
    {
      "epoch": 0.016184133587045558,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998324990591499,
      "loss": 2.4678,
      "step": 2382
    },
    {
      "epoch": 0.016190927933639618,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998322204869362,
      "loss": 2.4459,
      "step": 2383
    },
    {
      "epoch": 0.016197722280233674,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998319416833056,
      "loss": 2.4486,
      "step": 2384
    },
    {
      "epoch": 0.01620451662682773,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998316626482588,
      "loss": 2.5067,
      "step": 2385
    },
    {
      "epoch": 0.01621131097342179,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009998313833817956,
      "loss": 2.4234,
      "step": 2386
    },
    {
      "epoch": 0.016218105320015847,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998311038839163,
      "loss": 2.4653,
      "step": 2387
    },
    {
      "epoch": 0.016224899666609907,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998308241546208,
      "loss": 2.5188,
      "step": 2388
    },
    {
      "epoch": 0.016231694013203964,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998305441939095,
      "loss": 2.6324,
      "step": 2389
    },
    {
      "epoch": 0.01623848835979802,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998302640017822,
      "loss": 2.3684,
      "step": 2390
    },
    {
      "epoch": 0.01624528270639208,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998299835782393,
      "loss": 2.5428,
      "step": 2391
    },
    {
      "epoch": 0.016252077052986137,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999829702923281,
      "loss": 2.4307,
      "step": 2392
    },
    {
      "epoch": 0.016258871399580193,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999829422036907,
      "loss": 2.6083,
      "step": 2393
    },
    {
      "epoch": 0.016265665746174253,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000999829140919118,
      "loss": 2.4129,
      "step": 2394
    },
    {
      "epoch": 0.01627246009276831,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998288595699136,
      "loss": 2.5319,
      "step": 2395
    },
    {
      "epoch": 0.016279254439362366,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998285779892944,
      "loss": 2.4832,
      "step": 2396
    },
    {
      "epoch": 0.016286048785956426,
      "grad_norm": 0.177734375,
      "learning_rate": 0.00099982829617726,
      "loss": 2.4322,
      "step": 2397
    },
    {
      "epoch": 0.016292843132550482,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999828014133811,
      "loss": 2.5358,
      "step": 2398
    },
    {
      "epoch": 0.016299637479144542,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998277318589474,
      "loss": 2.4844,
      "step": 2399
    },
    {
      "epoch": 0.0163064318257386,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009998274493526692,
      "loss": 2.6051,
      "step": 2400
    },
    {
      "epoch": 0.016313226172332655,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998271666149768,
      "loss": 2.3481,
      "step": 2401
    },
    {
      "epoch": 0.016320020518926715,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.00099982688364587,
      "loss": 2.432,
      "step": 2402
    },
    {
      "epoch": 0.01632681486552077,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999826600445349,
      "loss": 2.6073,
      "step": 2403
    },
    {
      "epoch": 0.016333609212114828,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999826317013414,
      "loss": 2.4962,
      "step": 2404
    },
    {
      "epoch": 0.016340403558708888,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009998260333500652,
      "loss": 2.2949,
      "step": 2405
    },
    {
      "epoch": 0.016347197905302945,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999825749455303,
      "loss": 2.4557,
      "step": 2406
    },
    {
      "epoch": 0.016353992251897004,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009998254653291268,
      "loss": 2.5389,
      "step": 2407
    },
    {
      "epoch": 0.01636078659849106,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998251809715372,
      "loss": 2.4961,
      "step": 2408
    },
    {
      "epoch": 0.016367580945085117,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998248963825345,
      "loss": 2.5482,
      "step": 2409
    },
    {
      "epoch": 0.016374375291679177,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009998246115621183,
      "loss": 2.5176,
      "step": 2410
    },
    {
      "epoch": 0.016381169638273234,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998243265102894,
      "loss": 2.4229,
      "step": 2411
    },
    {
      "epoch": 0.01638796398486729,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998240412270472,
      "loss": 2.4165,
      "step": 2412
    },
    {
      "epoch": 0.01639475833146135,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009998237557123923,
      "loss": 2.6035,
      "step": 2413
    },
    {
      "epoch": 0.016401552678055407,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999823469966325,
      "loss": 2.5344,
      "step": 2414
    },
    {
      "epoch": 0.016408347024649463,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009998231839888447,
      "loss": 2.4083,
      "step": 2415
    },
    {
      "epoch": 0.016415141371243523,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998228977799524,
      "loss": 2.5365,
      "step": 2416
    },
    {
      "epoch": 0.01642193571783758,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998226113396476,
      "loss": 2.6202,
      "step": 2417
    },
    {
      "epoch": 0.01642873006443164,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009998223246679308,
      "loss": 2.4663,
      "step": 2418
    },
    {
      "epoch": 0.016435524411025696,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999822037764802,
      "loss": 2.5071,
      "step": 2419
    },
    {
      "epoch": 0.016442318757619753,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998217506302612,
      "loss": 2.5994,
      "step": 2420
    },
    {
      "epoch": 0.016449113104213813,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009998214632643088,
      "loss": 2.4154,
      "step": 2421
    },
    {
      "epoch": 0.01645590745080787,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009998211756669447,
      "loss": 2.4282,
      "step": 2422
    },
    {
      "epoch": 0.016462701797401925,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999820887838169,
      "loss": 2.5947,
      "step": 2423
    },
    {
      "epoch": 0.016469496143995985,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000999820599777982,
      "loss": 2.4104,
      "step": 2424
    },
    {
      "epoch": 0.016476290490590042,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999820311486384,
      "loss": 2.3217,
      "step": 2425
    },
    {
      "epoch": 0.0164830848371841,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999820022963375,
      "loss": 2.4834,
      "step": 2426
    },
    {
      "epoch": 0.01648987918377816,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998197342089548,
      "loss": 2.5018,
      "step": 2427
    },
    {
      "epoch": 0.016496673530372215,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009998194452231237,
      "loss": 2.4363,
      "step": 2428
    },
    {
      "epoch": 0.016503467876966275,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998191560058821,
      "loss": 2.4798,
      "step": 2429
    },
    {
      "epoch": 0.01651026222356033,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.00099981886655723,
      "loss": 2.6328,
      "step": 2430
    },
    {
      "epoch": 0.016517056570154388,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998185768771675,
      "loss": 2.3868,
      "step": 2431
    },
    {
      "epoch": 0.016523850916748448,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009998182869656947,
      "loss": 2.535,
      "step": 2432
    },
    {
      "epoch": 0.016530645263342504,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009998179968228116,
      "loss": 2.6124,
      "step": 2433
    },
    {
      "epoch": 0.01653743960993656,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009998177064485187,
      "loss": 2.5827,
      "step": 2434
    },
    {
      "epoch": 0.01654423395653062,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999817415842816,
      "loss": 2.467,
      "step": 2435
    },
    {
      "epoch": 0.016551028303124677,
      "grad_norm": 0.19140625,
      "learning_rate": 0.0009998171250057033,
      "loss": 2.5894,
      "step": 2436
    },
    {
      "epoch": 0.016557822649718737,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998168339371812,
      "loss": 2.546,
      "step": 2437
    },
    {
      "epoch": 0.016564616996312793,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998165426372495,
      "loss": 2.5907,
      "step": 2438
    },
    {
      "epoch": 0.01657141134290685,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998162511059086,
      "loss": 2.5224,
      "step": 2439
    },
    {
      "epoch": 0.01657820568950091,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998159593431583,
      "loss": 2.5119,
      "step": 2440
    },
    {
      "epoch": 0.016585000036094966,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009998156673489992,
      "loss": 2.3257,
      "step": 2441
    },
    {
      "epoch": 0.016591794382689023,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999815375123431,
      "loss": 2.5717,
      "step": 2442
    },
    {
      "epoch": 0.016598588729283083,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998150826664542,
      "loss": 2.4522,
      "step": 2443
    },
    {
      "epoch": 0.01660538307587714,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009998147899780685,
      "loss": 2.6537,
      "step": 2444
    },
    {
      "epoch": 0.016612177422471196,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998144970582744,
      "loss": 2.5083,
      "step": 2445
    },
    {
      "epoch": 0.016618971769065256,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999814203907072,
      "loss": 2.542,
      "step": 2446
    },
    {
      "epoch": 0.016625766115659312,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009998139105244612,
      "loss": 2.403,
      "step": 2447
    },
    {
      "epoch": 0.016632560462253372,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998136169104424,
      "loss": 2.4667,
      "step": 2448
    },
    {
      "epoch": 0.01663935480884743,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998133230650157,
      "loss": 2.5828,
      "step": 2449
    },
    {
      "epoch": 0.016646149155441485,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999813028988181,
      "loss": 2.5805,
      "step": 2450
    },
    {
      "epoch": 0.016652943502035545,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998127346799385,
      "loss": 2.4176,
      "step": 2451
    },
    {
      "epoch": 0.0166597378486296,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999812440140289,
      "loss": 2.4595,
      "step": 2452
    },
    {
      "epoch": 0.016666532195223658,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998121453692315,
      "loss": 2.6241,
      "step": 2453
    },
    {
      "epoch": 0.016673326541817718,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009998118503667668,
      "loss": 2.5672,
      "step": 2454
    },
    {
      "epoch": 0.016680120888411774,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999811555132895,
      "loss": 2.4223,
      "step": 2455
    },
    {
      "epoch": 0.01668691523500583,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009998112596676162,
      "loss": 2.5456,
      "step": 2456
    },
    {
      "epoch": 0.01669370958159989,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998109639709305,
      "loss": 2.4135,
      "step": 2457
    },
    {
      "epoch": 0.016700503928193947,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009998106680428381,
      "loss": 2.4187,
      "step": 2458
    },
    {
      "epoch": 0.016707298274788007,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999810371883339,
      "loss": 2.6044,
      "step": 2459
    },
    {
      "epoch": 0.016714092621382064,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009998100754924336,
      "loss": 2.5197,
      "step": 2460
    },
    {
      "epoch": 0.01672088696797612,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009998097788701218,
      "loss": 2.3358,
      "step": 2461
    },
    {
      "epoch": 0.01672768131457018,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009998094820164037,
      "loss": 2.5398,
      "step": 2462
    },
    {
      "epoch": 0.016734475661164237,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998091849312794,
      "loss": 2.3728,
      "step": 2463
    },
    {
      "epoch": 0.016741270007758293,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009998088876147493,
      "loss": 2.4691,
      "step": 2464
    },
    {
      "epoch": 0.016748064354352353,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009998085900668136,
      "loss": 2.3579,
      "step": 2465
    },
    {
      "epoch": 0.01675485870094641,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999808292287472,
      "loss": 2.473,
      "step": 2466
    },
    {
      "epoch": 0.01676165304754047,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999807994276725,
      "loss": 2.6675,
      "step": 2467
    },
    {
      "epoch": 0.016768447394134526,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009998076960345728,
      "loss": 2.528,
      "step": 2468
    },
    {
      "epoch": 0.016775241740728582,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998073975610153,
      "loss": 2.5801,
      "step": 2469
    },
    {
      "epoch": 0.016782036087322642,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009998070988560526,
      "loss": 2.517,
      "step": 2470
    },
    {
      "epoch": 0.0167888304339167,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999806799919685,
      "loss": 2.469,
      "step": 2471
    },
    {
      "epoch": 0.016795624780510755,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998065007519124,
      "loss": 2.3229,
      "step": 2472
    },
    {
      "epoch": 0.016802419127104815,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998062013527353,
      "loss": 2.2927,
      "step": 2473
    },
    {
      "epoch": 0.01680921347369887,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009998059017221536,
      "loss": 2.4703,
      "step": 2474
    },
    {
      "epoch": 0.016816007820292928,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009998056018601675,
      "loss": 2.7051,
      "step": 2475
    },
    {
      "epoch": 0.016822802166886988,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009998053017667772,
      "loss": 2.5454,
      "step": 2476
    },
    {
      "epoch": 0.016829596513481045,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999805001441983,
      "loss": 2.5018,
      "step": 2477
    },
    {
      "epoch": 0.016836390860075105,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009998047008857843,
      "loss": 2.5141,
      "step": 2478
    },
    {
      "epoch": 0.01684318520666916,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999804400098182,
      "loss": 2.5439,
      "step": 2479
    },
    {
      "epoch": 0.016849979553263218,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999804099079176,
      "loss": 2.3733,
      "step": 2480
    },
    {
      "epoch": 0.016856773899857277,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009998037978287664,
      "loss": 2.7513,
      "step": 2481
    },
    {
      "epoch": 0.016863568246451334,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009998034963469534,
      "loss": 2.4216,
      "step": 2482
    },
    {
      "epoch": 0.01687036259304539,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009998031946337371,
      "loss": 2.5815,
      "step": 2483
    },
    {
      "epoch": 0.01687715693963945,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009998028926891177,
      "loss": 2.4672,
      "step": 2484
    },
    {
      "epoch": 0.016883951286233507,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009998025905130952,
      "loss": 2.4728,
      "step": 2485
    },
    {
      "epoch": 0.016890745632827563,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00099980228810567,
      "loss": 2.4792,
      "step": 2486
    },
    {
      "epoch": 0.016897539979421623,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009998019854668418,
      "loss": 2.5468,
      "step": 2487
    },
    {
      "epoch": 0.01690433432601568,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009998016825966112,
      "loss": 2.5183,
      "step": 2488
    },
    {
      "epoch": 0.01691112867260974,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009998013794949782,
      "loss": 2.4751,
      "step": 2489
    },
    {
      "epoch": 0.016917923019203796,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009998010761619427,
      "loss": 2.4557,
      "step": 2490
    },
    {
      "epoch": 0.016924717365797853,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009998007725975052,
      "loss": 2.5189,
      "step": 2491
    },
    {
      "epoch": 0.016931511712391913,
      "grad_norm": 0.203125,
      "learning_rate": 0.0009998004688016656,
      "loss": 2.6203,
      "step": 2492
    },
    {
      "epoch": 0.01693830605898597,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999800164774424,
      "loss": 2.442,
      "step": 2493
    },
    {
      "epoch": 0.016945100405580026,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999799860515781,
      "loss": 2.324,
      "step": 2494
    },
    {
      "epoch": 0.016951894752174086,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999799556025736,
      "loss": 2.5834,
      "step": 2495
    },
    {
      "epoch": 0.016958689098768142,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997992513042899,
      "loss": 2.5151,
      "step": 2496
    },
    {
      "epoch": 0.016965483445362202,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997989463514422,
      "loss": 2.4685,
      "step": 2497
    },
    {
      "epoch": 0.01697227779195626,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997986411671932,
      "loss": 2.3866,
      "step": 2498
    },
    {
      "epoch": 0.016979072138550315,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009997983357515436,
      "loss": 2.6748,
      "step": 2499
    },
    {
      "epoch": 0.016985866485144375,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997980301044928,
      "loss": 2.4559,
      "step": 2500
    },
    {
      "epoch": 0.01699266083173843,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997977242260413,
      "loss": 2.4548,
      "step": 2501
    },
    {
      "epoch": 0.016999455178332488,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997974181161892,
      "loss": 2.5115,
      "step": 2502
    },
    {
      "epoch": 0.017006249524926548,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997971117749367,
      "loss": 2.4816,
      "step": 2503
    },
    {
      "epoch": 0.017013043871520604,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009997968052022836,
      "loss": 2.5033,
      "step": 2504
    },
    {
      "epoch": 0.01701983821811466,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009997964983982305,
      "loss": 2.5118,
      "step": 2505
    },
    {
      "epoch": 0.01702663256470872,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997961913627775,
      "loss": 2.561,
      "step": 2506
    },
    {
      "epoch": 0.017033426911302777,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997958840959244,
      "loss": 2.4339,
      "step": 2507
    },
    {
      "epoch": 0.017040221257896837,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009997955765976717,
      "loss": 2.4671,
      "step": 2508
    },
    {
      "epoch": 0.017047015604490894,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009997952688680191,
      "loss": 2.5774,
      "step": 2509
    },
    {
      "epoch": 0.01705380995108495,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009997949609069673,
      "loss": 2.3699,
      "step": 2510
    },
    {
      "epoch": 0.01706060429767901,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999794652714516,
      "loss": 2.4395,
      "step": 2511
    },
    {
      "epoch": 0.017067398644273066,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009997943442906655,
      "loss": 2.5053,
      "step": 2512
    },
    {
      "epoch": 0.017074192990867123,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997940356354159,
      "loss": 2.4523,
      "step": 2513
    },
    {
      "epoch": 0.017080987337461183,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997937267487674,
      "loss": 2.4719,
      "step": 2514
    },
    {
      "epoch": 0.01708778168405524,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997934176307203,
      "loss": 2.3203,
      "step": 2515
    },
    {
      "epoch": 0.017094576030649296,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997931082812745,
      "loss": 2.4849,
      "step": 2516
    },
    {
      "epoch": 0.017101370377243356,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997927987004302,
      "loss": 2.6189,
      "step": 2517
    },
    {
      "epoch": 0.017108164723837412,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997924888881875,
      "loss": 2.4527,
      "step": 2518
    },
    {
      "epoch": 0.017114959070431472,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997921788445468,
      "loss": 2.4674,
      "step": 2519
    },
    {
      "epoch": 0.01712175341702553,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009997918685695079,
      "loss": 2.4571,
      "step": 2520
    },
    {
      "epoch": 0.017128547763619585,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999791558063071,
      "loss": 2.5571,
      "step": 2521
    },
    {
      "epoch": 0.017135342110213645,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997912473252366,
      "loss": 2.5796,
      "step": 2522
    },
    {
      "epoch": 0.0171421364568077,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009997909363560044,
      "loss": 2.4314,
      "step": 2523
    },
    {
      "epoch": 0.017148930803401758,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997906251553747,
      "loss": 2.5096,
      "step": 2524
    },
    {
      "epoch": 0.017155725149995818,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009997903137233478,
      "loss": 2.5231,
      "step": 2525
    },
    {
      "epoch": 0.017162519496589874,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997900020599236,
      "loss": 2.5221,
      "step": 2526
    },
    {
      "epoch": 0.017169313843183934,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009997896901651024,
      "loss": 2.429,
      "step": 2527
    },
    {
      "epoch": 0.01717610818977799,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009997893780388845,
      "loss": 2.3666,
      "step": 2528
    },
    {
      "epoch": 0.017182902536372047,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009997890656812695,
      "loss": 2.5151,
      "step": 2529
    },
    {
      "epoch": 0.017189696882966107,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997887530922582,
      "loss": 2.4454,
      "step": 2530
    },
    {
      "epoch": 0.017196491229560164,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0009997884402718502,
      "loss": 2.5382,
      "step": 2531
    },
    {
      "epoch": 0.01720328557615422,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999788127220046,
      "loss": 2.3812,
      "step": 2532
    },
    {
      "epoch": 0.01721007992274828,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009997878139368457,
      "loss": 2.6795,
      "step": 2533
    },
    {
      "epoch": 0.017216874269342337,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997875004222492,
      "loss": 2.5724,
      "step": 2534
    },
    {
      "epoch": 0.017223668615936393,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.000999787186676257,
      "loss": 2.4736,
      "step": 2535
    },
    {
      "epoch": 0.017230462962530453,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009997868726988688,
      "loss": 2.6088,
      "step": 2536
    },
    {
      "epoch": 0.01723725730912451,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009997865584900852,
      "loss": 2.3709,
      "step": 2537
    },
    {
      "epoch": 0.01724405165571857,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000999786244049906,
      "loss": 2.5635,
      "step": 2538
    },
    {
      "epoch": 0.017250846002312626,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009997859293783319,
      "loss": 2.4398,
      "step": 2539
    },
    {
      "epoch": 0.017257640348906682,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009997856144753623,
      "loss": 2.4564,
      "step": 2540
    },
    {
      "epoch": 0.017264434695500742,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009997852993409977,
      "loss": 2.6115,
      "step": 2541
    },
    {
      "epoch": 0.0172712290420948,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009997849839752382,
      "loss": 2.5235,
      "step": 2542
    },
    {
      "epoch": 0.017278023388688855,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999784668378084,
      "loss": 2.4313,
      "step": 2543
    },
    {
      "epoch": 0.017284817735282915,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997843525495354,
      "loss": 2.528,
      "step": 2544
    },
    {
      "epoch": 0.017291612081876972,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997840364895923,
      "loss": 2.3508,
      "step": 2545
    },
    {
      "epoch": 0.01729840642847103,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999783720198255,
      "loss": 2.451,
      "step": 2546
    },
    {
      "epoch": 0.017305200775065088,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009997834036755233,
      "loss": 2.5005,
      "step": 2547
    },
    {
      "epoch": 0.017311995121659145,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997830869213978,
      "loss": 2.4921,
      "step": 2548
    },
    {
      "epoch": 0.017318789468253205,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009997827699358786,
      "loss": 2.5004,
      "step": 2549
    },
    {
      "epoch": 0.01732558381484726,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997824527189655,
      "loss": 2.4104,
      "step": 2550
    },
    {
      "epoch": 0.017332378161441318,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997821352706589,
      "loss": 2.4243,
      "step": 2551
    },
    {
      "epoch": 0.017339172508035378,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997818175909588,
      "loss": 2.3416,
      "step": 2552
    },
    {
      "epoch": 0.017345966854629434,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997814996798656,
      "loss": 2.5281,
      "step": 2553
    },
    {
      "epoch": 0.01735276120122349,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997811815373792,
      "loss": 2.4966,
      "step": 2554
    },
    {
      "epoch": 0.01735955554781755,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997808631635,
      "loss": 2.6543,
      "step": 2555
    },
    {
      "epoch": 0.017366349894411607,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997805445582277,
      "loss": 2.3479,
      "step": 2556
    },
    {
      "epoch": 0.017373144241005667,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999780225721563,
      "loss": 2.4425,
      "step": 2557
    },
    {
      "epoch": 0.017379938587599723,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997799066535056,
      "loss": 2.7024,
      "step": 2558
    },
    {
      "epoch": 0.01738673293419378,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009997795873540561,
      "loss": 2.4953,
      "step": 2559
    },
    {
      "epoch": 0.01739352728078784,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000999779267823214,
      "loss": 2.3427,
      "step": 2560
    },
    {
      "epoch": 0.017400321627381896,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009997789480609802,
      "loss": 2.4086,
      "step": 2561
    },
    {
      "epoch": 0.017407115973975953,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997786280673542,
      "loss": 2.4208,
      "step": 2562
    },
    {
      "epoch": 0.017413910320570013,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009997783078423367,
      "loss": 2.4323,
      "step": 2563
    },
    {
      "epoch": 0.01742070466716407,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997779873859272,
      "loss": 2.4372,
      "step": 2564
    },
    {
      "epoch": 0.017427499013758126,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009997776666981265,
      "loss": 2.4648,
      "step": 2565
    },
    {
      "epoch": 0.017434293360352186,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009997773457789344,
      "loss": 2.3421,
      "step": 2566
    },
    {
      "epoch": 0.017441087706946242,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999777024628351,
      "loss": 2.5618,
      "step": 2567
    },
    {
      "epoch": 0.017447882053540302,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999776703246377,
      "loss": 2.3814,
      "step": 2568
    },
    {
      "epoch": 0.01745467640013436,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009997763816330116,
      "loss": 2.3804,
      "step": 2569
    },
    {
      "epoch": 0.017461470746728415,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009997760597882557,
      "loss": 2.3212,
      "step": 2570
    },
    {
      "epoch": 0.017468265093322475,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997757377121093,
      "loss": 2.5694,
      "step": 2571
    },
    {
      "epoch": 0.01747505943991653,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997754154045723,
      "loss": 2.3903,
      "step": 2572
    },
    {
      "epoch": 0.017481853786510588,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997750928656451,
      "loss": 2.426,
      "step": 2573
    },
    {
      "epoch": 0.017488648133104648,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997747700953277,
      "loss": 2.2721,
      "step": 2574
    },
    {
      "epoch": 0.017495442479698704,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997744470936203,
      "loss": 2.364,
      "step": 2575
    },
    {
      "epoch": 0.01750223682629276,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009997741238605231,
      "loss": 2.4737,
      "step": 2576
    },
    {
      "epoch": 0.01750903117288682,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009997738003960361,
      "loss": 2.412,
      "step": 2577
    },
    {
      "epoch": 0.017515825519480877,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009997734767001598,
      "loss": 2.5677,
      "step": 2578
    },
    {
      "epoch": 0.017522619866074937,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999773152772894,
      "loss": 2.422,
      "step": 2579
    },
    {
      "epoch": 0.017529414212668994,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999772828614239,
      "loss": 2.4716,
      "step": 2580
    },
    {
      "epoch": 0.01753620855926305,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009997725042241948,
      "loss": 2.4975,
      "step": 2581
    },
    {
      "epoch": 0.01754300290585711,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997721796027616,
      "loss": 2.4214,
      "step": 2582
    },
    {
      "epoch": 0.017549797252451167,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997718547499397,
      "loss": 2.3927,
      "step": 2583
    },
    {
      "epoch": 0.017556591599045223,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009997715296657292,
      "loss": 2.5849,
      "step": 2584
    },
    {
      "epoch": 0.017563385945639283,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009997712043501303,
      "loss": 2.3348,
      "step": 2585
    },
    {
      "epoch": 0.01757018029223334,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999770878803143,
      "loss": 2.3681,
      "step": 2586
    },
    {
      "epoch": 0.0175769746388274,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997705530247673,
      "loss": 2.4576,
      "step": 2587
    },
    {
      "epoch": 0.017583768985421456,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009997702270150037,
      "loss": 2.4716,
      "step": 2588
    },
    {
      "epoch": 0.017590563332015512,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997699007738524,
      "loss": 2.4237,
      "step": 2589
    },
    {
      "epoch": 0.017597357678609572,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999769574301313,
      "loss": 2.3496,
      "step": 2590
    },
    {
      "epoch": 0.01760415202520363,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997692475973863,
      "loss": 2.47,
      "step": 2591
    },
    {
      "epoch": 0.017610946371797685,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009997689206620722,
      "loss": 2.4837,
      "step": 2592
    },
    {
      "epoch": 0.017617740718391745,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009997685934953707,
      "loss": 2.507,
      "step": 2593
    },
    {
      "epoch": 0.0176245350649858,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999768266097282,
      "loss": 2.4609,
      "step": 2594
    },
    {
      "epoch": 0.017631329411579858,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997679384678064,
      "loss": 2.4887,
      "step": 2595
    },
    {
      "epoch": 0.017638123758173918,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999767610606944,
      "loss": 2.2331,
      "step": 2596
    },
    {
      "epoch": 0.017644918104767975,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999767282514695,
      "loss": 2.5446,
      "step": 2597
    },
    {
      "epoch": 0.017651712451362035,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997669541910593,
      "loss": 2.4739,
      "step": 2598
    },
    {
      "epoch": 0.01765850679795609,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997666256360374,
      "loss": 2.5017,
      "step": 2599
    },
    {
      "epoch": 0.017665301144550147,
      "grad_norm": 0.140625,
      "learning_rate": 0.000999766296849629,
      "loss": 2.4647,
      "step": 2600
    },
    {
      "epoch": 0.017672095491144207,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997659678318348,
      "loss": 2.4008,
      "step": 2601
    },
    {
      "epoch": 0.017678889837738264,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009997656385826548,
      "loss": 2.3522,
      "step": 2602
    },
    {
      "epoch": 0.01768568418433232,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997653091020887,
      "loss": 2.5564,
      "step": 2603
    },
    {
      "epoch": 0.01769247853092638,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997649793901371,
      "loss": 2.6107,
      "step": 2604
    },
    {
      "epoch": 0.017699272877520437,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997646494468001,
      "loss": 2.5323,
      "step": 2605
    },
    {
      "epoch": 0.017706067224114493,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997643192720777,
      "loss": 2.5052,
      "step": 2606
    },
    {
      "epoch": 0.017712861570708553,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009997639888659703,
      "loss": 2.4066,
      "step": 2607
    },
    {
      "epoch": 0.01771965591730261,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997636582284779,
      "loss": 2.3594,
      "step": 2608
    },
    {
      "epoch": 0.01772645026389667,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009997633273596004,
      "loss": 2.4311,
      "step": 2609
    },
    {
      "epoch": 0.017733244610490726,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997629962593387,
      "loss": 2.5432,
      "step": 2610
    },
    {
      "epoch": 0.017740038957084783,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999762664927692,
      "loss": 2.5888,
      "step": 2611
    },
    {
      "epoch": 0.017746833303678843,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009997623333646611,
      "loss": 2.4904,
      "step": 2612
    },
    {
      "epoch": 0.0177536276502729,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999762001570246,
      "loss": 2.4277,
      "step": 2613
    },
    {
      "epoch": 0.017760421996866955,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997616695444466,
      "loss": 2.5113,
      "step": 2614
    },
    {
      "epoch": 0.017767216343461015,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009997613372872635,
      "loss": 2.4336,
      "step": 2615
    },
    {
      "epoch": 0.017774010690055072,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997610047986966,
      "loss": 2.3508,
      "step": 2616
    },
    {
      "epoch": 0.017780805036649132,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997606720787463,
      "loss": 2.5019,
      "step": 2617
    },
    {
      "epoch": 0.01778759938324319,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999760339127412,
      "loss": 2.2717,
      "step": 2618
    },
    {
      "epoch": 0.017794393729837245,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009997600059446948,
      "loss": 2.499,
      "step": 2619
    },
    {
      "epoch": 0.017801188076431305,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997596725305944,
      "loss": 2.5688,
      "step": 2620
    },
    {
      "epoch": 0.01780798242302536,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999759338885111,
      "loss": 2.4662,
      "step": 2621
    },
    {
      "epoch": 0.017814776769619418,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997590050082445,
      "loss": 2.4294,
      "step": 2622
    },
    {
      "epoch": 0.017821571116213478,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009997586708999958,
      "loss": 2.5577,
      "step": 2623
    },
    {
      "epoch": 0.017828365462807534,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997583365603641,
      "loss": 2.5366,
      "step": 2624
    },
    {
      "epoch": 0.01783515980940159,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997580019893502,
      "loss": 2.5981,
      "step": 2625
    },
    {
      "epoch": 0.01784195415599565,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997576671869541,
      "loss": 2.5134,
      "step": 2626
    },
    {
      "epoch": 0.017848748502589707,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009997573321531758,
      "loss": 2.4699,
      "step": 2627
    },
    {
      "epoch": 0.017855542849183767,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009997569968880158,
      "loss": 2.4314,
      "step": 2628
    },
    {
      "epoch": 0.017862337195777823,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999756661391474,
      "loss": 2.5965,
      "step": 2629
    },
    {
      "epoch": 0.01786913154237188,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997563256635505,
      "loss": 2.4182,
      "step": 2630
    },
    {
      "epoch": 0.01787592588896594,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997559897042454,
      "loss": 2.3539,
      "step": 2631
    },
    {
      "epoch": 0.017882720235559996,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997556535135592,
      "loss": 2.5258,
      "step": 2632
    },
    {
      "epoch": 0.017889514582154053,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999755317091492,
      "loss": 2.547,
      "step": 2633
    },
    {
      "epoch": 0.017896308928748113,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009997549804380435,
      "loss": 2.5187,
      "step": 2634
    },
    {
      "epoch": 0.01790310327534217,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997546435532142,
      "loss": 2.3842,
      "step": 2635
    },
    {
      "epoch": 0.017909897621936226,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997543064370045,
      "loss": 2.4131,
      "step": 2636
    },
    {
      "epoch": 0.017916691968530286,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997539690894141,
      "loss": 2.3995,
      "step": 2637
    },
    {
      "epoch": 0.017923486315124342,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009997536315104435,
      "loss": 2.5064,
      "step": 2638
    },
    {
      "epoch": 0.017930280661718402,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009997532937000926,
      "loss": 2.4616,
      "step": 2639
    },
    {
      "epoch": 0.01793707500831246,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009997529556583615,
      "loss": 2.4831,
      "step": 2640
    },
    {
      "epoch": 0.017943869354906515,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997526173852508,
      "loss": 2.4566,
      "step": 2641
    },
    {
      "epoch": 0.017950663701500575,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00099975227888076,
      "loss": 2.3007,
      "step": 2642
    },
    {
      "epoch": 0.01795745804809463,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.00099975194014489,
      "loss": 2.5661,
      "step": 2643
    },
    {
      "epoch": 0.017964252394688688,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997516011776403,
      "loss": 2.513,
      "step": 2644
    },
    {
      "epoch": 0.017971046741282748,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009997512619790115,
      "loss": 2.4511,
      "step": 2645
    },
    {
      "epoch": 0.017977841087876804,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009997509225490035,
      "loss": 2.3582,
      "step": 2646
    },
    {
      "epoch": 0.017984635434470864,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997505828876169,
      "loss": 2.3738,
      "step": 2647
    },
    {
      "epoch": 0.01799142978106492,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999750242994851,
      "loss": 2.3998,
      "step": 2648
    },
    {
      "epoch": 0.017998224127658977,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997499028707067,
      "loss": 2.445,
      "step": 2649
    },
    {
      "epoch": 0.018005018474253037,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997495625151838,
      "loss": 2.4012,
      "step": 2650
    },
    {
      "epoch": 0.018011812820847094,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999749221928283,
      "loss": 2.3747,
      "step": 2651
    },
    {
      "epoch": 0.01801860716744115,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997488811100035,
      "loss": 2.4886,
      "step": 2652
    },
    {
      "epoch": 0.01802540151403521,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009997485400603462,
      "loss": 2.4628,
      "step": 2653
    },
    {
      "epoch": 0.018032195860629267,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997481987793111,
      "loss": 2.4834,
      "step": 2654
    },
    {
      "epoch": 0.018038990207223323,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997478572668984,
      "loss": 2.482,
      "step": 2655
    },
    {
      "epoch": 0.018045784553817383,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999747515523108,
      "loss": 2.6241,
      "step": 2656
    },
    {
      "epoch": 0.01805257890041144,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997471735479404,
      "loss": 2.4934,
      "step": 2657
    },
    {
      "epoch": 0.0180593732470055,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997468313413954,
      "loss": 2.4172,
      "step": 2658
    },
    {
      "epoch": 0.018066167593599556,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009997464889034734,
      "loss": 2.4573,
      "step": 2659
    },
    {
      "epoch": 0.018072961940193612,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997461462341747,
      "loss": 2.6059,
      "step": 2660
    },
    {
      "epoch": 0.018079756286787672,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999745803333499,
      "loss": 2.5381,
      "step": 2661
    },
    {
      "epoch": 0.01808655063338173,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997454602014468,
      "loss": 2.4745,
      "step": 2662
    },
    {
      "epoch": 0.018093344979975785,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009997451168380182,
      "loss": 2.5225,
      "step": 2663
    },
    {
      "epoch": 0.018100139326569845,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997447732432133,
      "loss": 2.5472,
      "step": 2664
    },
    {
      "epoch": 0.018106933673163902,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997444294170324,
      "loss": 2.5454,
      "step": 2665
    },
    {
      "epoch": 0.018113728019757958,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009997440853594755,
      "loss": 2.3483,
      "step": 2666
    },
    {
      "epoch": 0.018120522366352018,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997437410705428,
      "loss": 2.5223,
      "step": 2667
    },
    {
      "epoch": 0.018127316712946075,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997433965502345,
      "loss": 2.6283,
      "step": 2668
    },
    {
      "epoch": 0.018134111059540135,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009997430517985507,
      "loss": 2.3703,
      "step": 2669
    },
    {
      "epoch": 0.01814090540613419,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009997427068154916,
      "loss": 2.3679,
      "step": 2670
    },
    {
      "epoch": 0.018147699752728248,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009997423616010576,
      "loss": 2.5124,
      "step": 2671
    },
    {
      "epoch": 0.018154494099322308,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997420161552482,
      "loss": 2.3391,
      "step": 2672
    },
    {
      "epoch": 0.018161288445916364,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009997416704780643,
      "loss": 2.4362,
      "step": 2673
    },
    {
      "epoch": 0.01816808279251042,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009997413245695056,
      "loss": 2.3223,
      "step": 2674
    },
    {
      "epoch": 0.01817487713910448,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009997409784295724,
      "loss": 2.4888,
      "step": 2675
    },
    {
      "epoch": 0.018181671485698537,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009997406320582647,
      "loss": 2.4249,
      "step": 2676
    },
    {
      "epoch": 0.018188465832292597,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999740285455583,
      "loss": 2.5716,
      "step": 2677
    },
    {
      "epoch": 0.018195260178886653,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009997399386215274,
      "loss": 2.3694,
      "step": 2678
    },
    {
      "epoch": 0.01820205452548071,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997395915560976,
      "loss": 2.5932,
      "step": 2679
    },
    {
      "epoch": 0.01820884887207477,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009997392442592944,
      "loss": 2.4342,
      "step": 2680
    },
    {
      "epoch": 0.018215643218668826,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997388967311175,
      "loss": 2.5496,
      "step": 2681
    },
    {
      "epoch": 0.018222437565262883,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997385489715674,
      "loss": 2.4985,
      "step": 2682
    },
    {
      "epoch": 0.018229231911856943,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999738200980644,
      "loss": 2.5167,
      "step": 2683
    },
    {
      "epoch": 0.018236026258451,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009997378527583473,
      "loss": 2.4836,
      "step": 2684
    },
    {
      "epoch": 0.018242820605045056,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999737504304678,
      "loss": 2.4593,
      "step": 2685
    },
    {
      "epoch": 0.018249614951639116,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997371556196358,
      "loss": 2.5066,
      "step": 2686
    },
    {
      "epoch": 0.018256409298233172,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997368067032211,
      "loss": 2.5195,
      "step": 2687
    },
    {
      "epoch": 0.018263203644827232,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999736457555434,
      "loss": 2.5289,
      "step": 2688
    },
    {
      "epoch": 0.01826999799142129,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997361081762746,
      "loss": 2.5242,
      "step": 2689
    },
    {
      "epoch": 0.018276792338015345,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999735758565743,
      "loss": 2.4389,
      "step": 2690
    },
    {
      "epoch": 0.018283586684609405,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997354087238398,
      "loss": 2.408,
      "step": 2691
    },
    {
      "epoch": 0.01829038103120346,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009997350586505647,
      "loss": 2.3964,
      "step": 2692
    },
    {
      "epoch": 0.018297175377797518,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999734708345918,
      "loss": 2.4387,
      "step": 2693
    },
    {
      "epoch": 0.018303969724391578,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997343578098997,
      "loss": 2.4892,
      "step": 2694
    },
    {
      "epoch": 0.018310764070985634,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997340070425102,
      "loss": 2.4185,
      "step": 2695
    },
    {
      "epoch": 0.01831755841757969,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997336560437495,
      "loss": 2.4793,
      "step": 2696
    },
    {
      "epoch": 0.01832435276417375,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997333048136182,
      "loss": 2.4413,
      "step": 2697
    },
    {
      "epoch": 0.018331147110767807,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009997329533521157,
      "loss": 2.5904,
      "step": 2698
    },
    {
      "epoch": 0.018337941457361867,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009997326016592427,
      "loss": 2.5861,
      "step": 2699
    },
    {
      "epoch": 0.018344735803955924,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009997322497349993,
      "loss": 2.5402,
      "step": 2700
    },
    {
      "epoch": 0.01835153015054998,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997318975793856,
      "loss": 2.4984,
      "step": 2701
    },
    {
      "epoch": 0.01835832449714404,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009997315451924017,
      "loss": 2.3966,
      "step": 2702
    },
    {
      "epoch": 0.018365118843738096,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997311925740479,
      "loss": 2.5864,
      "step": 2703
    },
    {
      "epoch": 0.018371913190332153,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009997308397243243,
      "loss": 2.3237,
      "step": 2704
    },
    {
      "epoch": 0.018378707536926213,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009997304866432308,
      "loss": 2.4437,
      "step": 2705
    },
    {
      "epoch": 0.01838550188352027,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997301333307682,
      "loss": 2.4153,
      "step": 2706
    },
    {
      "epoch": 0.01839229623011433,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.000999729779786936,
      "loss": 2.5586,
      "step": 2707
    },
    {
      "epoch": 0.018399090576708386,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997294260117348,
      "loss": 2.515,
      "step": 2708
    },
    {
      "epoch": 0.018405884923302442,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009997290720051644,
      "loss": 2.516,
      "step": 2709
    },
    {
      "epoch": 0.018412679269896502,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009997287177672255,
      "loss": 2.4914,
      "step": 2710
    },
    {
      "epoch": 0.01841947361649056,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997283632979176,
      "loss": 2.5559,
      "step": 2711
    },
    {
      "epoch": 0.018426267963084615,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009997280085972413,
      "loss": 2.5196,
      "step": 2712
    },
    {
      "epoch": 0.018433062309678675,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009997276536651968,
      "loss": 2.3755,
      "step": 2713
    },
    {
      "epoch": 0.01843985665627273,
      "grad_norm": 0.1875,
      "learning_rate": 0.000999727298501784,
      "loss": 2.5405,
      "step": 2714
    },
    {
      "epoch": 0.018446651002866788,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0009997269431070031,
      "loss": 2.4733,
      "step": 2715
    },
    {
      "epoch": 0.018453445349460848,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009997265874808546,
      "loss": 2.4196,
      "step": 2716
    },
    {
      "epoch": 0.018460239696054905,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009997262316233382,
      "loss": 2.5621,
      "step": 2717
    },
    {
      "epoch": 0.018467034042648964,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009997258755344543,
      "loss": 2.4784,
      "step": 2718
    },
    {
      "epoch": 0.01847382838924302,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009997255192142032,
      "loss": 2.3577,
      "step": 2719
    },
    {
      "epoch": 0.018480622735837077,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009997251626625847,
      "loss": 2.3819,
      "step": 2720
    },
    {
      "epoch": 0.018487417082431137,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009997248058795994,
      "loss": 2.4904,
      "step": 2721
    },
    {
      "epoch": 0.018494211429025194,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000999724448865247,
      "loss": 2.3782,
      "step": 2722
    },
    {
      "epoch": 0.01850100577561925,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009997240916195282,
      "loss": 2.549,
      "step": 2723
    },
    {
      "epoch": 0.01850780012221331,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.0009997237341424427,
      "loss": 2.4394,
      "step": 2724
    },
    {
      "epoch": 0.018514594468807367,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997233764339909,
      "loss": 2.4592,
      "step": 2725
    },
    {
      "epoch": 0.018521388815401423,
      "grad_norm": 0.2041015625,
      "learning_rate": 0.0009997230184941729,
      "loss": 2.4683,
      "step": 2726
    },
    {
      "epoch": 0.018528183161995483,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009997226603229888,
      "loss": 2.4468,
      "step": 2727
    },
    {
      "epoch": 0.01853497750858954,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999722301920439,
      "loss": 2.5213,
      "step": 2728
    },
    {
      "epoch": 0.0185417718551836,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009997219432865232,
      "loss": 2.3662,
      "step": 2729
    },
    {
      "epoch": 0.018548566201777656,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009997215844212422,
      "loss": 2.4804,
      "step": 2730
    },
    {
      "epoch": 0.018555360548371713,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009997212253245954,
      "loss": 2.4735,
      "step": 2731
    },
    {
      "epoch": 0.018562154894965772,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009997208659965839,
      "loss": 2.3668,
      "step": 2732
    },
    {
      "epoch": 0.01856894924155983,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009997205064372072,
      "loss": 2.4779,
      "step": 2733
    },
    {
      "epoch": 0.018575743588153885,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009997201466464655,
      "loss": 2.3445,
      "step": 2734
    },
    {
      "epoch": 0.018582537934747945,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0009997197866243592,
      "loss": 2.4825,
      "step": 2735
    },
    {
      "epoch": 0.018589332281342002,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009997194263708885,
      "loss": 2.487,
      "step": 2736
    },
    {
      "epoch": 0.018596126627936062,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009997190658860532,
      "loss": 2.4699,
      "step": 2737
    },
    {
      "epoch": 0.01860292097453012,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997187051698539,
      "loss": 2.4012,
      "step": 2738
    },
    {
      "epoch": 0.018609715321124175,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009997183442222903,
      "loss": 2.2918,
      "step": 2739
    },
    {
      "epoch": 0.018616509667718235,
      "grad_norm": 0.19140625,
      "learning_rate": 0.000999717983043363,
      "loss": 2.5108,
      "step": 2740
    },
    {
      "epoch": 0.01862330401431229,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000999717621633072,
      "loss": 2.494,
      "step": 2741
    },
    {
      "epoch": 0.018630098360906348,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009997172599914175,
      "loss": 2.2998,
      "step": 2742
    },
    {
      "epoch": 0.018636892707500408,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009997168981183996,
      "loss": 2.5329,
      "step": 2743
    },
    {
      "epoch": 0.018643687054094464,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009997165360140186,
      "loss": 2.5127,
      "step": 2744
    },
    {
      "epoch": 0.01865048140068852,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997161736782746,
      "loss": 2.5325,
      "step": 2745
    },
    {
      "epoch": 0.01865727574728258,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009997158111111676,
      "loss": 2.4311,
      "step": 2746
    },
    {
      "epoch": 0.018664070093876637,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009997154483126977,
      "loss": 2.5702,
      "step": 2747
    },
    {
      "epoch": 0.018670864440470697,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997150852828656,
      "loss": 2.5769,
      "step": 2748
    },
    {
      "epoch": 0.018677658787064753,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.000999714722021671,
      "loss": 2.3328,
      "step": 2749
    },
    {
      "epoch": 0.01868445313365881,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009997143585291145,
      "loss": 2.4724,
      "step": 2750
    },
    {
      "epoch": 0.01869124748025287,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009997139948051956,
      "loss": 2.3332,
      "step": 2751
    },
    {
      "epoch": 0.018698041826846926,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999713630849915,
      "loss": 2.4651,
      "step": 2752
    },
    {
      "epoch": 0.018704836173440983,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009997132666632728,
      "loss": 2.5468,
      "step": 2753
    },
    {
      "epoch": 0.018711630520035043,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999712902245269,
      "loss": 2.4986,
      "step": 2754
    },
    {
      "epoch": 0.0187184248666291,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999712537595904,
      "loss": 2.3922,
      "step": 2755
    },
    {
      "epoch": 0.018725219213223156,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009997121727151775,
      "loss": 2.4747,
      "step": 2756
    },
    {
      "epoch": 0.018732013559817216,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009997118076030902,
      "loss": 2.3857,
      "step": 2757
    },
    {
      "epoch": 0.018738807906411272,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999711442259642,
      "loss": 2.3732,
      "step": 2758
    },
    {
      "epoch": 0.018745602253005332,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009997110766848333,
      "loss": 2.3997,
      "step": 2759
    },
    {
      "epoch": 0.01875239659959939,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009997107108786639,
      "loss": 2.6209,
      "step": 2760
    },
    {
      "epoch": 0.018759190946193445,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009997103448411344,
      "loss": 2.3825,
      "step": 2761
    },
    {
      "epoch": 0.018765985292787505,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009997099785722446,
      "loss": 2.4182,
      "step": 2762
    },
    {
      "epoch": 0.01877277963938156,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997096120719945,
      "loss": 2.3838,
      "step": 2763
    },
    {
      "epoch": 0.018779573985975618,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000999709245340385,
      "loss": 2.5259,
      "step": 2764
    },
    {
      "epoch": 0.018786368332569678,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009997088783774157,
      "loss": 2.5314,
      "step": 2765
    },
    {
      "epoch": 0.018793162679163734,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999708511183087,
      "loss": 2.4573,
      "step": 2766
    },
    {
      "epoch": 0.018799957025757794,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.000999708143757399,
      "loss": 2.3369,
      "step": 2767
    },
    {
      "epoch": 0.01880675137235185,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009997077761003516,
      "loss": 2.3488,
      "step": 2768
    },
    {
      "epoch": 0.018813545718945907,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009997074082119454,
      "loss": 2.4254,
      "step": 2769
    },
    {
      "epoch": 0.018820340065539967,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009997070400921805,
      "loss": 2.5697,
      "step": 2770
    },
    {
      "epoch": 0.018827134412134024,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999706671741057,
      "loss": 2.4811,
      "step": 2771
    },
    {
      "epoch": 0.01883392875872808,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999706303158575,
      "loss": 2.4355,
      "step": 2772
    },
    {
      "epoch": 0.01884072310532214,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009997059343447346,
      "loss": 2.4006,
      "step": 2773
    },
    {
      "epoch": 0.018847517451916197,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999705565299536,
      "loss": 2.4383,
      "step": 2774
    },
    {
      "epoch": 0.018854311798510253,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009997051960229796,
      "loss": 2.4321,
      "step": 2775
    },
    {
      "epoch": 0.018861106145104313,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009997048265150655,
      "loss": 2.4691,
      "step": 2776
    },
    {
      "epoch": 0.01886790049169837,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009997044567757935,
      "loss": 2.4589,
      "step": 2777
    },
    {
      "epoch": 0.01887469483829243,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997040868051643,
      "loss": 2.3104,
      "step": 2778
    },
    {
      "epoch": 0.018881489184886486,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009997037166031778,
      "loss": 2.3912,
      "step": 2779
    },
    {
      "epoch": 0.018888283531480542,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009997033461698341,
      "loss": 2.4816,
      "step": 2780
    },
    {
      "epoch": 0.018895077878074602,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009997029755051336,
      "loss": 2.3087,
      "step": 2781
    },
    {
      "epoch": 0.01890187222466866,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009997026046090764,
      "loss": 2.6125,
      "step": 2782
    },
    {
      "epoch": 0.018908666571262715,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997022334816625,
      "loss": 2.4239,
      "step": 2783
    },
    {
      "epoch": 0.018915460917856775,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009997018621228924,
      "loss": 2.4032,
      "step": 2784
    },
    {
      "epoch": 0.01892225526445083,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009997014905327659,
      "loss": 2.4932,
      "step": 2785
    },
    {
      "epoch": 0.018929049611044888,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009997011187112832,
      "loss": 2.3712,
      "step": 2786
    },
    {
      "epoch": 0.018935843957638948,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999700746658445,
      "loss": 2.3408,
      "step": 2787
    },
    {
      "epoch": 0.018942638304233005,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009997003743742506,
      "loss": 2.4577,
      "step": 2788
    },
    {
      "epoch": 0.018949432650827065,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999700001858701,
      "loss": 2.44,
      "step": 2789
    },
    {
      "epoch": 0.01895622699742112,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996996291117958,
      "loss": 2.6004,
      "step": 2790
    },
    {
      "epoch": 0.018963021344015177,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996992561335355,
      "loss": 2.5557,
      "step": 2791
    },
    {
      "epoch": 0.018969815690609237,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996988829239202,
      "loss": 2.5375,
      "step": 2792
    },
    {
      "epoch": 0.018976610037203294,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009996985094829502,
      "loss": 2.3402,
      "step": 2793
    },
    {
      "epoch": 0.01898340438379735,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996981358106254,
      "loss": 2.4705,
      "step": 2794
    },
    {
      "epoch": 0.01899019873039141,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999697761906946,
      "loss": 2.4266,
      "step": 2795
    },
    {
      "epoch": 0.018996993076985467,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009996973877719124,
      "loss": 2.6084,
      "step": 2796
    },
    {
      "epoch": 0.019003787423579527,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996970134055246,
      "loss": 2.5165,
      "step": 2797
    },
    {
      "epoch": 0.019010581770173583,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996966388077827,
      "loss": 2.5522,
      "step": 2798
    },
    {
      "epoch": 0.01901737611676764,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999696263978687,
      "loss": 2.4009,
      "step": 2799
    },
    {
      "epoch": 0.0190241704633617,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996958889182377,
      "loss": 2.556,
      "step": 2800
    },
    {
      "epoch": 0.019030964809955756,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996955136264349,
      "loss": 2.6086,
      "step": 2801
    },
    {
      "epoch": 0.019037759156549813,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999695138103279,
      "loss": 2.3919,
      "step": 2802
    },
    {
      "epoch": 0.019044553503143873,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009996947623487698,
      "loss": 2.5828,
      "step": 2803
    },
    {
      "epoch": 0.01905134784973793,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996943863629076,
      "loss": 2.5043,
      "step": 2804
    },
    {
      "epoch": 0.019058142196331986,
      "grad_norm": 0.1953125,
      "learning_rate": 0.0009996940101456926,
      "loss": 2.4534,
      "step": 2805
    },
    {
      "epoch": 0.019064936542926045,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009996936336971252,
      "loss": 2.4508,
      "step": 2806
    },
    {
      "epoch": 0.019071730889520102,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996932570172053,
      "loss": 2.4836,
      "step": 2807
    },
    {
      "epoch": 0.019078525236114162,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996928801059332,
      "loss": 2.544,
      "step": 2808
    },
    {
      "epoch": 0.01908531958270822,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999692502963309,
      "loss": 2.3996,
      "step": 2809
    },
    {
      "epoch": 0.019092113929302275,
      "grad_norm": 0.20703125,
      "learning_rate": 0.000999692125589333,
      "loss": 2.5451,
      "step": 2810
    },
    {
      "epoch": 0.019098908275896335,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999691747984005,
      "loss": 2.519,
      "step": 2811
    },
    {
      "epoch": 0.01910570262249039,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996913701473256,
      "loss": 2.575,
      "step": 2812
    },
    {
      "epoch": 0.019112496969084448,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996909920792947,
      "loss": 2.4628,
      "step": 2813
    },
    {
      "epoch": 0.019119291315678508,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009996906137799129,
      "loss": 2.5002,
      "step": 2814
    },
    {
      "epoch": 0.019126085662272564,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996902352491799,
      "loss": 2.4684,
      "step": 2815
    },
    {
      "epoch": 0.01913288000886662,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009996898564870959,
      "loss": 2.3917,
      "step": 2816
    },
    {
      "epoch": 0.01913967435546068,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996894774936614,
      "loss": 2.4874,
      "step": 2817
    },
    {
      "epoch": 0.019146468702054737,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996890982688765,
      "loss": 2.4151,
      "step": 2818
    },
    {
      "epoch": 0.019153263048648797,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999688718812741,
      "loss": 2.5139,
      "step": 2819
    },
    {
      "epoch": 0.019160057395242854,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009996883391252555,
      "loss": 2.4965,
      "step": 2820
    },
    {
      "epoch": 0.01916685174183691,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00099968795920642,
      "loss": 2.4725,
      "step": 2821
    },
    {
      "epoch": 0.01917364608843097,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009996875790562348,
      "loss": 2.4804,
      "step": 2822
    },
    {
      "epoch": 0.019180440435025026,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009996871986746997,
      "loss": 2.3918,
      "step": 2823
    },
    {
      "epoch": 0.019187234781619083,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996868180618154,
      "loss": 2.4146,
      "step": 2824
    },
    {
      "epoch": 0.019194029128213143,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009996864372175819,
      "loss": 2.5007,
      "step": 2825
    },
    {
      "epoch": 0.0192008234748072,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999686056141999,
      "loss": 2.4879,
      "step": 2826
    },
    {
      "epoch": 0.01920761782140126,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996856748350675,
      "loss": 2.6018,
      "step": 2827
    },
    {
      "epoch": 0.019214412167995316,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009996852932967872,
      "loss": 2.4195,
      "step": 2828
    },
    {
      "epoch": 0.019221206514589372,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996849115271583,
      "loss": 2.4801,
      "step": 2829
    },
    {
      "epoch": 0.019228000861183432,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999684529526181,
      "loss": 2.3581,
      "step": 2830
    },
    {
      "epoch": 0.01923479520777749,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996841472938555,
      "loss": 2.4501,
      "step": 2831
    },
    {
      "epoch": 0.019241589554371545,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999683764830182,
      "loss": 2.4937,
      "step": 2832
    },
    {
      "epoch": 0.019248383900965605,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996833821351604,
      "loss": 2.5837,
      "step": 2833
    },
    {
      "epoch": 0.01925517824755966,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009996829992087914,
      "loss": 2.4106,
      "step": 2834
    },
    {
      "epoch": 0.019261972594153718,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999682616051075,
      "loss": 2.4615,
      "step": 2835
    },
    {
      "epoch": 0.019268766940747778,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009996822326620112,
      "loss": 2.3409,
      "step": 2836
    },
    {
      "epoch": 0.019275561287341834,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009996818490416,
      "loss": 2.5407,
      "step": 2837
    },
    {
      "epoch": 0.019282355633935894,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999681465189842,
      "loss": 2.3636,
      "step": 2838
    },
    {
      "epoch": 0.01928914998052995,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000999681081106737,
      "loss": 2.511,
      "step": 2839
    },
    {
      "epoch": 0.019295944327124007,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009996806967922857,
      "loss": 2.4697,
      "step": 2840
    },
    {
      "epoch": 0.019302738673718067,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000999680312246488,
      "loss": 2.4437,
      "step": 2841
    },
    {
      "epoch": 0.019309533020312124,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996799274693438,
      "loss": 2.4287,
      "step": 2842
    },
    {
      "epoch": 0.01931632736690618,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009996795424608536,
      "loss": 2.3755,
      "step": 2843
    },
    {
      "epoch": 0.01932312171350024,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996791572210175,
      "loss": 2.4617,
      "step": 2844
    },
    {
      "epoch": 0.019329916060094297,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009996787717498359,
      "loss": 2.4674,
      "step": 2845
    },
    {
      "epoch": 0.019336710406688353,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996783860473084,
      "loss": 2.4879,
      "step": 2846
    },
    {
      "epoch": 0.019343504753282413,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996780001134359,
      "loss": 2.493,
      "step": 2847
    },
    {
      "epoch": 0.01935029909987647,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996776139482178,
      "loss": 2.4575,
      "step": 2848
    },
    {
      "epoch": 0.01935709344647053,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999677227551655,
      "loss": 2.5538,
      "step": 2849
    },
    {
      "epoch": 0.019363887793064586,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009996768409237474,
      "loss": 2.5303,
      "step": 2850
    },
    {
      "epoch": 0.019370682139658642,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999676454064495,
      "loss": 2.3839,
      "step": 2851
    },
    {
      "epoch": 0.019377476486252702,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999676066973898,
      "loss": 2.5273,
      "step": 2852
    },
    {
      "epoch": 0.01938427083284676,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999675679651957,
      "loss": 2.2881,
      "step": 2853
    },
    {
      "epoch": 0.019391065179440815,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996752920986716,
      "loss": 2.418,
      "step": 2854
    },
    {
      "epoch": 0.019397859526034875,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996749043140426,
      "loss": 2.4798,
      "step": 2855
    },
    {
      "epoch": 0.019404653872628932,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996745162980697,
      "loss": 2.4397,
      "step": 2856
    },
    {
      "epoch": 0.019411448219222992,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999674128050753,
      "loss": 2.3508,
      "step": 2857
    },
    {
      "epoch": 0.019418242565817048,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996737395720932,
      "loss": 2.5752,
      "step": 2858
    },
    {
      "epoch": 0.019425036912411105,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.00099967335086209,
      "loss": 2.4306,
      "step": 2859
    },
    {
      "epoch": 0.019431831259005165,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996729619207439,
      "loss": 2.5294,
      "step": 2860
    },
    {
      "epoch": 0.01943862560559922,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000999672572748055,
      "loss": 2.5454,
      "step": 2861
    },
    {
      "epoch": 0.019445419952193278,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996721833440232,
      "loss": 2.4199,
      "step": 2862
    },
    {
      "epoch": 0.019452214298787338,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999671793708649,
      "loss": 2.3963,
      "step": 2863
    },
    {
      "epoch": 0.019459008645381394,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996714038419325,
      "loss": 2.339,
      "step": 2864
    },
    {
      "epoch": 0.01946580299197545,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999671013743874,
      "loss": 2.4187,
      "step": 2865
    },
    {
      "epoch": 0.01947259733856951,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009996706234144735,
      "loss": 2.5611,
      "step": 2866
    },
    {
      "epoch": 0.019479391685163567,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996702328537311,
      "loss": 2.4943,
      "step": 2867
    },
    {
      "epoch": 0.019486186031757627,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996698420616473,
      "loss": 2.2907,
      "step": 2868
    },
    {
      "epoch": 0.019492980378351683,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996694510382219,
      "loss": 2.5559,
      "step": 2869
    },
    {
      "epoch": 0.01949977472494574,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996690597834554,
      "loss": 2.5917,
      "step": 2870
    },
    {
      "epoch": 0.0195065690715398,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996686682973478,
      "loss": 2.4,
      "step": 2871
    },
    {
      "epoch": 0.019513363418133856,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996682765798994,
      "loss": 2.4969,
      "step": 2872
    },
    {
      "epoch": 0.019520157764727913,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996678846311102,
      "loss": 2.5375,
      "step": 2873
    },
    {
      "epoch": 0.019526952111321973,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009996674924509805,
      "loss": 2.5549,
      "step": 2874
    },
    {
      "epoch": 0.01953374645791603,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009996671000395106,
      "loss": 2.524,
      "step": 2875
    },
    {
      "epoch": 0.019540540804510086,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009996667073967004,
      "loss": 2.3229,
      "step": 2876
    },
    {
      "epoch": 0.019547335151104146,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009996663145225503,
      "loss": 2.5615,
      "step": 2877
    },
    {
      "epoch": 0.019554129497698202,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009996659214170605,
      "loss": 2.4216,
      "step": 2878
    },
    {
      "epoch": 0.019560923844292262,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999665528080231,
      "loss": 2.5029,
      "step": 2879
    },
    {
      "epoch": 0.01956771819088632,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009996651345120624,
      "loss": 2.2908,
      "step": 2880
    },
    {
      "epoch": 0.019574512537480375,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009996647407125542,
      "loss": 2.3071,
      "step": 2881
    },
    {
      "epoch": 0.019581306884074435,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009996643466817071,
      "loss": 2.3263,
      "step": 2882
    },
    {
      "epoch": 0.01958810123066849,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999663952419521,
      "loss": 2.6676,
      "step": 2883
    },
    {
      "epoch": 0.019594895577262548,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009996635579259964,
      "loss": 2.6877,
      "step": 2884
    },
    {
      "epoch": 0.019601689923856608,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009996631632011332,
      "loss": 2.4761,
      "step": 2885
    },
    {
      "epoch": 0.019608484270450664,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996627682449318,
      "loss": 2.3036,
      "step": 2886
    },
    {
      "epoch": 0.019615278617044724,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996623730573922,
      "loss": 2.4237,
      "step": 2887
    },
    {
      "epoch": 0.01962207296363878,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009996619776385148,
      "loss": 2.6,
      "step": 2888
    },
    {
      "epoch": 0.019628867310232837,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009996615819882993,
      "loss": 2.378,
      "step": 2889
    },
    {
      "epoch": 0.019635661656826897,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009996611861067463,
      "loss": 2.4893,
      "step": 2890
    },
    {
      "epoch": 0.019642456003420954,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009996607899938562,
      "loss": 2.3934,
      "step": 2891
    },
    {
      "epoch": 0.01964925035001501,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996603936496286,
      "loss": 2.4657,
      "step": 2892
    },
    {
      "epoch": 0.01965604469660907,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996599970740641,
      "loss": 2.5851,
      "step": 2893
    },
    {
      "epoch": 0.019662839043203127,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009996596002671625,
      "loss": 2.5608,
      "step": 2894
    },
    {
      "epoch": 0.019669633389797183,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009996592032289245,
      "loss": 2.432,
      "step": 2895
    },
    {
      "epoch": 0.019676427736391243,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996588059593499,
      "loss": 2.5021,
      "step": 2896
    },
    {
      "epoch": 0.0196832220829853,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999658408458439,
      "loss": 2.5609,
      "step": 2897
    },
    {
      "epoch": 0.01969001642957936,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009996580107261921,
      "loss": 2.51,
      "step": 2898
    },
    {
      "epoch": 0.019696810776173416,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999657612762609,
      "loss": 2.4443,
      "step": 2899
    },
    {
      "epoch": 0.019703605122767472,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009996572145676906,
      "loss": 2.4878,
      "step": 2900
    },
    {
      "epoch": 0.019710399469361532,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009996568161414364,
      "loss": 2.365,
      "step": 2901
    },
    {
      "epoch": 0.01971719381595559,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996564174838468,
      "loss": 2.4103,
      "step": 2902
    },
    {
      "epoch": 0.019723988162549645,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009996560185949219,
      "loss": 2.5635,
      "step": 2903
    },
    {
      "epoch": 0.019730782509143705,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009996556194746622,
      "loss": 2.5507,
      "step": 2904
    },
    {
      "epoch": 0.01973757685573776,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996552201230675,
      "loss": 2.4672,
      "step": 2905
    },
    {
      "epoch": 0.019744371202331818,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996548205401384,
      "loss": 2.356,
      "step": 2906
    },
    {
      "epoch": 0.019751165548925878,
      "grad_norm": 0.1982421875,
      "learning_rate": 0.0009996544207258747,
      "loss": 2.6229,
      "step": 2907
    },
    {
      "epoch": 0.019757959895519935,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996540206802767,
      "loss": 2.3556,
      "step": 2908
    },
    {
      "epoch": 0.019764754242113994,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009996536204033447,
      "loss": 2.4438,
      "step": 2909
    },
    {
      "epoch": 0.01977154858870805,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996532198950786,
      "loss": 2.4369,
      "step": 2910
    },
    {
      "epoch": 0.019778342935302107,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996528191554789,
      "loss": 2.5,
      "step": 2911
    },
    {
      "epoch": 0.019785137281896167,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996524181845458,
      "loss": 2.4572,
      "step": 2912
    },
    {
      "epoch": 0.019791931628490224,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999652016982279,
      "loss": 2.4247,
      "step": 2913
    },
    {
      "epoch": 0.01979872597508428,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009996516155486794,
      "loss": 2.7501,
      "step": 2914
    },
    {
      "epoch": 0.01980552032167834,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996512138837467,
      "loss": 2.4968,
      "step": 2915
    },
    {
      "epoch": 0.019812314668272397,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009996508119874815,
      "loss": 2.4467,
      "step": 2916
    },
    {
      "epoch": 0.019819109014866457,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009996504098598834,
      "loss": 2.4119,
      "step": 2917
    },
    {
      "epoch": 0.019825903361460513,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999650007500953,
      "loss": 2.4465,
      "step": 2918
    },
    {
      "epoch": 0.01983269770805457,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996496049106901,
      "loss": 2.3907,
      "step": 2919
    },
    {
      "epoch": 0.01983949205464863,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996492020890955,
      "loss": 2.3743,
      "step": 2920
    },
    {
      "epoch": 0.019846286401242686,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999648799036169,
      "loss": 2.3908,
      "step": 2921
    },
    {
      "epoch": 0.019853080747836743,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996483957519107,
      "loss": 2.5012,
      "step": 2922
    },
    {
      "epoch": 0.019859875094430803,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009996479922363209,
      "loss": 2.4419,
      "step": 2923
    },
    {
      "epoch": 0.01986666944102486,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996475884893998,
      "loss": 2.4865,
      "step": 2924
    },
    {
      "epoch": 0.019873463787618915,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996471845111477,
      "loss": 2.4979,
      "step": 2925
    },
    {
      "epoch": 0.019880258134212975,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996467803015648,
      "loss": 2.2903,
      "step": 2926
    },
    {
      "epoch": 0.019887052480807032,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009996463758606508,
      "loss": 2.3997,
      "step": 2927
    },
    {
      "epoch": 0.019893846827401092,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996459711884068,
      "loss": 2.3722,
      "step": 2928
    },
    {
      "epoch": 0.01990064117399515,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996455662848321,
      "loss": 2.3011,
      "step": 2929
    },
    {
      "epoch": 0.019907435520589205,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996451611499271,
      "loss": 2.5004,
      "step": 2930
    },
    {
      "epoch": 0.019914229867183265,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009996447557836922,
      "loss": 2.472,
      "step": 2931
    },
    {
      "epoch": 0.01992102421377732,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996443501861275,
      "loss": 2.5268,
      "step": 2932
    },
    {
      "epoch": 0.019927818560371378,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996439443572334,
      "loss": 2.4382,
      "step": 2933
    },
    {
      "epoch": 0.019934612906965438,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.00099964353829701,
      "loss": 2.5192,
      "step": 2934
    },
    {
      "epoch": 0.019941407253559494,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999643132005457,
      "loss": 2.4029,
      "step": 2935
    },
    {
      "epoch": 0.01994820160015355,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999642725482575,
      "loss": 2.4144,
      "step": 2936
    },
    {
      "epoch": 0.01995499594674761,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009996423187283643,
      "loss": 2.417,
      "step": 2937
    },
    {
      "epoch": 0.019961790293341667,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996419117428248,
      "loss": 2.4934,
      "step": 2938
    },
    {
      "epoch": 0.019968584639935727,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999641504525957,
      "loss": 2.3783,
      "step": 2939
    },
    {
      "epoch": 0.019975378986529783,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996410970777607,
      "loss": 2.5651,
      "step": 2940
    },
    {
      "epoch": 0.01998217333312384,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996406893982364,
      "loss": 2.4747,
      "step": 2941
    },
    {
      "epoch": 0.0199889676797179,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996402814873844,
      "loss": 2.5113,
      "step": 2942
    },
    {
      "epoch": 0.019995762026311956,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009996398733452044,
      "loss": 2.4641,
      "step": 2943
    },
    {
      "epoch": 0.020002556372906013,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996394649716969,
      "loss": 2.5805,
      "step": 2944
    },
    {
      "epoch": 0.020009350719500073,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999639056366862,
      "loss": 2.6641,
      "step": 2945
    },
    {
      "epoch": 0.02001614506609413,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009996386475307002,
      "loss": 2.4774,
      "step": 2946
    },
    {
      "epoch": 0.02002293941268819,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996382384632112,
      "loss": 2.4897,
      "step": 2947
    },
    {
      "epoch": 0.020029733759282246,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009996378291643954,
      "loss": 2.5157,
      "step": 2948
    },
    {
      "epoch": 0.020036528105876302,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009996374196342532,
      "loss": 2.4738,
      "step": 2949
    },
    {
      "epoch": 0.020043322452470362,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009996370098727845,
      "loss": 2.401,
      "step": 2950
    },
    {
      "epoch": 0.02005011679906442,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996365998799896,
      "loss": 2.4294,
      "step": 2951
    },
    {
      "epoch": 0.020056911145658475,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009996361896558688,
      "loss": 2.4318,
      "step": 2952
    },
    {
      "epoch": 0.020063705492252535,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999635779200422,
      "loss": 2.3904,
      "step": 2953
    },
    {
      "epoch": 0.02007049983884659,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009996353685136497,
      "loss": 2.5206,
      "step": 2954
    },
    {
      "epoch": 0.020077294185440648,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009996349575955518,
      "loss": 2.4195,
      "step": 2955
    },
    {
      "epoch": 0.020084088532034708,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009996345464461287,
      "loss": 2.4027,
      "step": 2956
    },
    {
      "epoch": 0.020090882878628764,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996341350653807,
      "loss": 2.5435,
      "step": 2957
    },
    {
      "epoch": 0.020097677225222824,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009996337234533074,
      "loss": 2.3819,
      "step": 2958
    },
    {
      "epoch": 0.02010447157181688,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00099963331160991,
      "loss": 2.5042,
      "step": 2959
    },
    {
      "epoch": 0.020111265918410937,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996328995351877,
      "loss": 2.4667,
      "step": 2960
    },
    {
      "epoch": 0.020118060265004997,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996324872291412,
      "loss": 2.4311,
      "step": 2961
    },
    {
      "epoch": 0.020124854611599054,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996320746917704,
      "loss": 2.4898,
      "step": 2962
    },
    {
      "epoch": 0.02013164895819311,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999631661923076,
      "loss": 2.299,
      "step": 2963
    },
    {
      "epoch": 0.02013844330478717,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009996312489230577,
      "loss": 2.3737,
      "step": 2964
    },
    {
      "epoch": 0.020145237651381227,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999630835691716,
      "loss": 2.6426,
      "step": 2965
    },
    {
      "epoch": 0.020152031997975283,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996304222290506,
      "loss": 2.4136,
      "step": 2966
    },
    {
      "epoch": 0.020158826344569343,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996300085350625,
      "loss": 2.2587,
      "step": 2967
    },
    {
      "epoch": 0.0201656206911634,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009996295946097511,
      "loss": 2.4365,
      "step": 2968
    },
    {
      "epoch": 0.02017241503775746,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999629180453117,
      "loss": 2.4065,
      "step": 2969
    },
    {
      "epoch": 0.020179209384351516,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996287660651603,
      "loss": 2.4237,
      "step": 2970
    },
    {
      "epoch": 0.020186003730945572,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996283514458813,
      "loss": 2.4484,
      "step": 2971
    },
    {
      "epoch": 0.020192798077539632,
      "grad_norm": 0.158203125,
      "learning_rate": 0.00099962793659528,
      "loss": 2.6155,
      "step": 2972
    },
    {
      "epoch": 0.02019959242413369,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999627521513357,
      "loss": 2.393,
      "step": 2973
    },
    {
      "epoch": 0.020206386770727745,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999627106200112,
      "loss": 2.5876,
      "step": 2974
    },
    {
      "epoch": 0.020213181117321805,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996266906555451,
      "loss": 2.4862,
      "step": 2975
    },
    {
      "epoch": 0.02021997546391586,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999626274879657,
      "loss": 2.4534,
      "step": 2976
    },
    {
      "epoch": 0.02022676981050992,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009996258588724476,
      "loss": 2.4912,
      "step": 2977
    },
    {
      "epoch": 0.020233564157103978,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009996254426339172,
      "loss": 2.3937,
      "step": 2978
    },
    {
      "epoch": 0.020240358503698035,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999625026164066,
      "loss": 2.5024,
      "step": 2979
    },
    {
      "epoch": 0.020247152850292095,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996246094628942,
      "loss": 2.5041,
      "step": 2980
    },
    {
      "epoch": 0.02025394719688615,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996241925304017,
      "loss": 2.4591,
      "step": 2981
    },
    {
      "epoch": 0.020260741543480208,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009996237753665891,
      "loss": 2.5257,
      "step": 2982
    },
    {
      "epoch": 0.020267535890074267,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996233579714563,
      "loss": 2.2095,
      "step": 2983
    },
    {
      "epoch": 0.020274330236668324,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009996229403450038,
      "loss": 2.5356,
      "step": 2984
    },
    {
      "epoch": 0.02028112458326238,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009996225224872316,
      "loss": 2.3799,
      "step": 2985
    },
    {
      "epoch": 0.02028791892985644,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996221043981398,
      "loss": 2.522,
      "step": 2986
    },
    {
      "epoch": 0.020294713276450497,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009996216860777287,
      "loss": 2.4854,
      "step": 2987
    },
    {
      "epoch": 0.020301507623044557,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009996212675259983,
      "loss": 2.601,
      "step": 2988
    },
    {
      "epoch": 0.020308301969638613,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009996208487429493,
      "loss": 2.5109,
      "step": 2989
    },
    {
      "epoch": 0.02031509631623267,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009996204297285814,
      "loss": 2.4759,
      "step": 2990
    },
    {
      "epoch": 0.02032189066282673,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999620010482895,
      "loss": 2.3828,
      "step": 2991
    },
    {
      "epoch": 0.020328685009420786,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996195910058903,
      "loss": 2.4784,
      "step": 2992
    },
    {
      "epoch": 0.020335479356014843,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009996191712975675,
      "loss": 2.4545,
      "step": 2993
    },
    {
      "epoch": 0.020342273702608903,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009996187513579265,
      "loss": 2.3588,
      "step": 2994
    },
    {
      "epoch": 0.02034906804920296,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999618331186968,
      "loss": 2.4013,
      "step": 2995
    },
    {
      "epoch": 0.020355862395797016,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999617910784692,
      "loss": 2.483,
      "step": 2996
    },
    {
      "epoch": 0.020362656742391076,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996174901510985,
      "loss": 2.3866,
      "step": 2997
    },
    {
      "epoch": 0.020369451088985132,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009996170692861876,
      "loss": 2.4176,
      "step": 2998
    },
    {
      "epoch": 0.020376245435579192,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00099961664818996,
      "loss": 2.4738,
      "step": 2999
    },
    {
      "epoch": 0.02038303978217325,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996162268624156,
      "loss": 2.4069,
      "step": 3000
    },
    {
      "epoch": 0.020389834128767305,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009996158053035547,
      "loss": 2.415,
      "step": 3001
    },
    {
      "epoch": 0.020396628475361365,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996153835133772,
      "loss": 2.5287,
      "step": 3002
    },
    {
      "epoch": 0.02040342282195542,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996149614918835,
      "loss": 2.5324,
      "step": 3003
    },
    {
      "epoch": 0.020410217168549478,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999614539239074,
      "loss": 2.6078,
      "step": 3004
    },
    {
      "epoch": 0.020417011515143538,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996141167549486,
      "loss": 2.543,
      "step": 3005
    },
    {
      "epoch": 0.020423805861737594,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009996136940395076,
      "loss": 2.4613,
      "step": 3006
    },
    {
      "epoch": 0.020430600208331654,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996132710927512,
      "loss": 2.4031,
      "step": 3007
    },
    {
      "epoch": 0.02043739455492571,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009996128479146796,
      "loss": 2.5182,
      "step": 3008
    },
    {
      "epoch": 0.020444188901519767,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009996124245052929,
      "loss": 2.3631,
      "step": 3009
    },
    {
      "epoch": 0.020450983248113827,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009996120008645914,
      "loss": 2.3644,
      "step": 3010
    },
    {
      "epoch": 0.020457777594707884,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009996115769925753,
      "loss": 2.5468,
      "step": 3011
    },
    {
      "epoch": 0.02046457194130194,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996111528892447,
      "loss": 2.6742,
      "step": 3012
    },
    {
      "epoch": 0.020471366287896,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996107285545998,
      "loss": 2.4462,
      "step": 3013
    },
    {
      "epoch": 0.020478160634490056,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999610303988641,
      "loss": 2.4343,
      "step": 3014
    },
    {
      "epoch": 0.020484954981084113,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009996098791913684,
      "loss": 2.4654,
      "step": 3015
    },
    {
      "epoch": 0.020491749327678173,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999609454162782,
      "loss": 2.4354,
      "step": 3016
    },
    {
      "epoch": 0.02049854367427223,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996090289028825,
      "loss": 2.4117,
      "step": 3017
    },
    {
      "epoch": 0.02050533802086629,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009996086034116693,
      "loss": 2.3494,
      "step": 3018
    },
    {
      "epoch": 0.020512132367460346,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009996081776891433,
      "loss": 2.4151,
      "step": 3019
    },
    {
      "epoch": 0.020518926714054402,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996077517353044,
      "loss": 2.4706,
      "step": 3020
    },
    {
      "epoch": 0.020525721060648462,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009996073255501528,
      "loss": 2.5204,
      "step": 3021
    },
    {
      "epoch": 0.02053251540724252,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009996068991336888,
      "loss": 2.3746,
      "step": 3022
    },
    {
      "epoch": 0.020539309753836575,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009996064724859125,
      "loss": 2.4427,
      "step": 3023
    },
    {
      "epoch": 0.020546104100430635,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009996060456068242,
      "loss": 2.3068,
      "step": 3024
    },
    {
      "epoch": 0.02055289844702469,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009996056184964238,
      "loss": 2.4859,
      "step": 3025
    },
    {
      "epoch": 0.020559692793618748,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009996051911547118,
      "loss": 2.3645,
      "step": 3026
    },
    {
      "epoch": 0.020566487140212808,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009996047635816884,
      "loss": 2.259,
      "step": 3027
    },
    {
      "epoch": 0.020573281486806864,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009996043357773537,
      "loss": 2.4302,
      "step": 3028
    },
    {
      "epoch": 0.020580075833400924,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009996039077417081,
      "loss": 2.3731,
      "step": 3029
    },
    {
      "epoch": 0.02058687017999498,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009996034794747513,
      "loss": 2.3384,
      "step": 3030
    },
    {
      "epoch": 0.020593664526589037,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000999603050976484,
      "loss": 2.4174,
      "step": 3031
    },
    {
      "epoch": 0.020600458873183097,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009996026222469062,
      "loss": 2.4301,
      "step": 3032
    },
    {
      "epoch": 0.020607253219777154,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009996021932860183,
      "loss": 2.5429,
      "step": 3033
    },
    {
      "epoch": 0.02061404756637121,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099960176409382,
      "loss": 2.505,
      "step": 3034
    },
    {
      "epoch": 0.02062084191296527,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000999601334670312,
      "loss": 2.4054,
      "step": 3035
    },
    {
      "epoch": 0.020627636259559327,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009996009050154942,
      "loss": 2.2034,
      "step": 3036
    },
    {
      "epoch": 0.020634430606153387,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009996004751293669,
      "loss": 2.4288,
      "step": 3037
    },
    {
      "epoch": 0.020641224952747443,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009996000450119302,
      "loss": 2.2349,
      "step": 3038
    },
    {
      "epoch": 0.0206480192993415,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009995996146631848,
      "loss": 2.3999,
      "step": 3039
    },
    {
      "epoch": 0.02065481364593556,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995991840831301,
      "loss": 2.4232,
      "step": 3040
    },
    {
      "epoch": 0.020661607992529616,
      "grad_norm": 0.2216796875,
      "learning_rate": 0.0009995987532717668,
      "loss": 2.4166,
      "step": 3041
    },
    {
      "epoch": 0.020668402339123673,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0009995983222290952,
      "loss": 2.2559,
      "step": 3042
    },
    {
      "epoch": 0.020675196685717732,
      "grad_norm": 0.2236328125,
      "learning_rate": 0.000999597890955115,
      "loss": 2.5108,
      "step": 3043
    },
    {
      "epoch": 0.02068199103231179,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999597459449827,
      "loss": 2.276,
      "step": 3044
    },
    {
      "epoch": 0.020688785378905845,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009995970277132309,
      "loss": 2.3511,
      "step": 3045
    },
    {
      "epoch": 0.020695579725499905,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0009995965957453271,
      "loss": 2.6177,
      "step": 3046
    },
    {
      "epoch": 0.020702374072093962,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999596163546116,
      "loss": 2.295,
      "step": 3047
    },
    {
      "epoch": 0.020709168418688022,
      "grad_norm": 0.240234375,
      "learning_rate": 0.0009995957311155976,
      "loss": 2.4788,
      "step": 3048
    },
    {
      "epoch": 0.020715962765282078,
      "grad_norm": 0.193359375,
      "learning_rate": 0.000999595298453772,
      "loss": 2.396,
      "step": 3049
    },
    {
      "epoch": 0.020722757111876135,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009995948655606396,
      "loss": 2.4981,
      "step": 3050
    },
    {
      "epoch": 0.020729551458470195,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995944324362003,
      "loss": 2.2561,
      "step": 3051
    },
    {
      "epoch": 0.02073634580506425,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009995939990804547,
      "loss": 2.5097,
      "step": 3052
    },
    {
      "epoch": 0.020743140151658308,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009995935654934025,
      "loss": 2.3896,
      "step": 3053
    },
    {
      "epoch": 0.020749934498252368,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995931316750446,
      "loss": 2.4192,
      "step": 3054
    },
    {
      "epoch": 0.020756728844846424,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009995926976253805,
      "loss": 2.4763,
      "step": 3055
    },
    {
      "epoch": 0.02076352319144048,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009995922633444108,
      "loss": 2.4996,
      "step": 3056
    },
    {
      "epoch": 0.02077031753803454,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009995918288321357,
      "loss": 2.5386,
      "step": 3057
    },
    {
      "epoch": 0.020777111884628597,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009995913940885552,
      "loss": 2.4078,
      "step": 3058
    },
    {
      "epoch": 0.020783906231222657,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995909591136698,
      "loss": 2.561,
      "step": 3059
    },
    {
      "epoch": 0.020790700577816713,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009995905239074793,
      "loss": 2.5868,
      "step": 3060
    },
    {
      "epoch": 0.02079749492441077,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009995900884699842,
      "loss": 2.3592,
      "step": 3061
    },
    {
      "epoch": 0.02080428927100483,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009995896528011848,
      "loss": 2.4827,
      "step": 3062
    },
    {
      "epoch": 0.020811083617598886,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009995892169010808,
      "loss": 2.3762,
      "step": 3063
    },
    {
      "epoch": 0.020817877964192943,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999588780769673,
      "loss": 2.4601,
      "step": 3064
    },
    {
      "epoch": 0.020824672310787003,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999588344406961,
      "loss": 2.3951,
      "step": 3065
    },
    {
      "epoch": 0.02083146665738106,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995879078129456,
      "loss": 2.4774,
      "step": 3066
    },
    {
      "epoch": 0.02083826100397512,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009995874709876266,
      "loss": 2.4392,
      "step": 3067
    },
    {
      "epoch": 0.020845055350569176,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009995870339310044,
      "loss": 2.4777,
      "step": 3068
    },
    {
      "epoch": 0.020851849697163232,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009995865966430791,
      "loss": 2.6157,
      "step": 3069
    },
    {
      "epoch": 0.020858644043757292,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009995861591238508,
      "loss": 2.4311,
      "step": 3070
    },
    {
      "epoch": 0.02086543839035135,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995857213733201,
      "loss": 2.3812,
      "step": 3071
    },
    {
      "epoch": 0.020872232736945405,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995852833914868,
      "loss": 2.4277,
      "step": 3072
    },
    {
      "epoch": 0.020879027083539465,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995848451783513,
      "loss": 2.454,
      "step": 3073
    },
    {
      "epoch": 0.02088582143013352,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009995844067339137,
      "loss": 2.5014,
      "step": 3074
    },
    {
      "epoch": 0.020892615776727578,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995839680581743,
      "loss": 2.4139,
      "step": 3075
    },
    {
      "epoch": 0.020899410123321638,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009995835291511331,
      "loss": 2.2695,
      "step": 3076
    },
    {
      "epoch": 0.020906204469915694,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009995830900127905,
      "loss": 2.4417,
      "step": 3077
    },
    {
      "epoch": 0.020912998816509754,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995826506431467,
      "loss": 2.4898,
      "step": 3078
    },
    {
      "epoch": 0.02091979316310381,
      "grad_norm": 0.2099609375,
      "learning_rate": 0.0009995822110422019,
      "loss": 2.3395,
      "step": 3079
    },
    {
      "epoch": 0.020926587509697867,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995817712099562,
      "loss": 2.329,
      "step": 3080
    },
    {
      "epoch": 0.020933381856291927,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995813311464098,
      "loss": 2.3747,
      "step": 3081
    },
    {
      "epoch": 0.020940176202885984,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009995808908515632,
      "loss": 2.4477,
      "step": 3082
    },
    {
      "epoch": 0.02094697054948004,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999580450325416,
      "loss": 2.2464,
      "step": 3083
    },
    {
      "epoch": 0.0209537648960741,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999580009567969,
      "loss": 2.4537,
      "step": 3084
    },
    {
      "epoch": 0.020960559242668157,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995795685792223,
      "loss": 2.5876,
      "step": 3085
    },
    {
      "epoch": 0.020967353589262213,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000999579127359176,
      "loss": 2.496,
      "step": 3086
    },
    {
      "epoch": 0.020974147935856273,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00099957868590783,
      "loss": 2.6321,
      "step": 3087
    },
    {
      "epoch": 0.02098094228245033,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999578244225185,
      "loss": 2.4394,
      "step": 3088
    },
    {
      "epoch": 0.02098773662904439,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009995778023112409,
      "loss": 2.4678,
      "step": 3089
    },
    {
      "epoch": 0.020994530975638446,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009995773601659981,
      "loss": 2.427,
      "step": 3090
    },
    {
      "epoch": 0.021001325322232502,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009995769177894567,
      "loss": 2.2911,
      "step": 3091
    },
    {
      "epoch": 0.021008119668826562,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999576475181617,
      "loss": 2.4822,
      "step": 3092
    },
    {
      "epoch": 0.02101491401542062,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000999576032342479,
      "loss": 2.4185,
      "step": 3093
    },
    {
      "epoch": 0.021021708362014675,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999575589272043,
      "loss": 2.4611,
      "step": 3094
    },
    {
      "epoch": 0.021028502708608735,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009995751459703094,
      "loss": 2.5382,
      "step": 3095
    },
    {
      "epoch": 0.02103529705520279,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995747024372779,
      "loss": 2.4371,
      "step": 3096
    },
    {
      "epoch": 0.02104209140179685,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009995742586729492,
      "loss": 2.2946,
      "step": 3097
    },
    {
      "epoch": 0.021048885748390908,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995738146773233,
      "loss": 2.439,
      "step": 3098
    },
    {
      "epoch": 0.021055680094984965,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009995733704504007,
      "loss": 2.4503,
      "step": 3099
    },
    {
      "epoch": 0.021062474441579025,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009995729259921812,
      "loss": 2.3131,
      "step": 3100
    },
    {
      "epoch": 0.02106926878817308,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995724813026653,
      "loss": 2.2391,
      "step": 3101
    },
    {
      "epoch": 0.021076063134767137,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995720363818528,
      "loss": 2.3707,
      "step": 3102
    },
    {
      "epoch": 0.021082857481361197,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995715912297443,
      "loss": 2.6043,
      "step": 3103
    },
    {
      "epoch": 0.021089651827955254,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009995711458463399,
      "loss": 2.3935,
      "step": 3104
    },
    {
      "epoch": 0.02109644617454931,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009995707002316398,
      "loss": 2.4348,
      "step": 3105
    },
    {
      "epoch": 0.02110324052114337,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999570254385644,
      "loss": 2.3399,
      "step": 3106
    },
    {
      "epoch": 0.021110034867737427,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999569808308353,
      "loss": 2.2408,
      "step": 3107
    },
    {
      "epoch": 0.021116829214331487,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999569361999767,
      "loss": 2.3891,
      "step": 3108
    },
    {
      "epoch": 0.021123623560925543,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995689154598862,
      "loss": 2.5294,
      "step": 3109
    },
    {
      "epoch": 0.0211304179075196,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009995684686887106,
      "loss": 2.3885,
      "step": 3110
    },
    {
      "epoch": 0.02113721225411366,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009995680216862406,
      "loss": 2.5722,
      "step": 3111
    },
    {
      "epoch": 0.021144006600707716,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999567574452476,
      "loss": 2.2132,
      "step": 3112
    },
    {
      "epoch": 0.021150800947301773,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009995671269874178,
      "loss": 2.3052,
      "step": 3113
    },
    {
      "epoch": 0.021157595293895833,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995666792910656,
      "loss": 2.4704,
      "step": 3114
    },
    {
      "epoch": 0.02116438964048989,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995662313634197,
      "loss": 2.5363,
      "step": 3115
    },
    {
      "epoch": 0.021171183987083945,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009995657832044802,
      "loss": 2.3131,
      "step": 3116
    },
    {
      "epoch": 0.021177978333678005,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009995653348142477,
      "loss": 2.3804,
      "step": 3117
    },
    {
      "epoch": 0.021184772680272062,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995648861927222,
      "loss": 2.4247,
      "step": 3118
    },
    {
      "epoch": 0.021191567026866122,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995644373399038,
      "loss": 2.4581,
      "step": 3119
    },
    {
      "epoch": 0.02119836137346018,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995639882557927,
      "loss": 2.4788,
      "step": 3120
    },
    {
      "epoch": 0.021205155720054235,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995635389403895,
      "loss": 2.3608,
      "step": 3121
    },
    {
      "epoch": 0.021211950066648295,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995630893936938,
      "loss": 2.4989,
      "step": 3122
    },
    {
      "epoch": 0.02121874441324235,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009995626396157062,
      "loss": 2.5609,
      "step": 3123
    },
    {
      "epoch": 0.021225538759836408,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995621896064268,
      "loss": 2.4445,
      "step": 3124
    },
    {
      "epoch": 0.021232333106430468,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995617393658558,
      "loss": 2.3595,
      "step": 3125
    },
    {
      "epoch": 0.021239127453024524,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995612888939935,
      "loss": 2.3688,
      "step": 3126
    },
    {
      "epoch": 0.021245921799618584,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009995608381908401,
      "loss": 2.3871,
      "step": 3127
    },
    {
      "epoch": 0.02125271614621264,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995603872563958,
      "loss": 2.4675,
      "step": 3128
    },
    {
      "epoch": 0.021259510492806697,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995599360906606,
      "loss": 2.3355,
      "step": 3129
    },
    {
      "epoch": 0.021266304839400757,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995594846936352,
      "loss": 2.3847,
      "step": 3130
    },
    {
      "epoch": 0.021273099185994813,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999559033065319,
      "loss": 2.4289,
      "step": 3131
    },
    {
      "epoch": 0.02127989353258887,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009995585812057131,
      "loss": 2.4239,
      "step": 3132
    },
    {
      "epoch": 0.02128668787918293,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995581291148171,
      "loss": 2.5341,
      "step": 3133
    },
    {
      "epoch": 0.021293482225776986,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995576767926316,
      "loss": 2.2495,
      "step": 3134
    },
    {
      "epoch": 0.021300276572371043,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009995572242391564,
      "loss": 2.4353,
      "step": 3135
    },
    {
      "epoch": 0.021307070918965103,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999556771454392,
      "loss": 2.5439,
      "step": 3136
    },
    {
      "epoch": 0.02131386526555916,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009995563184383386,
      "loss": 2.538,
      "step": 3137
    },
    {
      "epoch": 0.02132065961215322,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009995558651909964,
      "loss": 2.5559,
      "step": 3138
    },
    {
      "epoch": 0.021327453958747276,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995554117123652,
      "loss": 2.3428,
      "step": 3139
    },
    {
      "epoch": 0.021334248305341332,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999554958002446,
      "loss": 2.4627,
      "step": 3140
    },
    {
      "epoch": 0.021341042651935392,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009995545040612384,
      "loss": 2.4922,
      "step": 3141
    },
    {
      "epoch": 0.02134783699852945,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995540498887428,
      "loss": 2.3684,
      "step": 3142
    },
    {
      "epoch": 0.021354631345123505,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009995535954849595,
      "loss": 2.356,
      "step": 3143
    },
    {
      "epoch": 0.021361425691717565,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995531408498886,
      "loss": 2.452,
      "step": 3144
    },
    {
      "epoch": 0.02136822003831162,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009995526859835303,
      "loss": 2.4364,
      "step": 3145
    },
    {
      "epoch": 0.021375014384905678,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009995522308858848,
      "loss": 2.4118,
      "step": 3146
    },
    {
      "epoch": 0.021381808731499738,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995517755569522,
      "loss": 2.3852,
      "step": 3147
    },
    {
      "epoch": 0.021388603078093794,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995513199967329,
      "loss": 2.4297,
      "step": 3148
    },
    {
      "epoch": 0.021395397424687854,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009995508642052274,
      "loss": 2.3211,
      "step": 3149
    },
    {
      "epoch": 0.02140219177128191,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009995504081824352,
      "loss": 2.342,
      "step": 3150
    },
    {
      "epoch": 0.021408986117875967,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999549951928357,
      "loss": 2.527,
      "step": 3151
    },
    {
      "epoch": 0.021415780464470027,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999549495442993,
      "loss": 2.5288,
      "step": 3152
    },
    {
      "epoch": 0.021422574811064084,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009995490387263433,
      "loss": 2.453,
      "step": 3153
    },
    {
      "epoch": 0.02142936915765814,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999548581778408,
      "loss": 2.465,
      "step": 3154
    },
    {
      "epoch": 0.0214361635042522,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995481245991876,
      "loss": 2.5109,
      "step": 3155
    },
    {
      "epoch": 0.021442957850846257,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009995476671886818,
      "loss": 2.6053,
      "step": 3156
    },
    {
      "epoch": 0.021449752197440317,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995472095468915,
      "loss": 2.4094,
      "step": 3157
    },
    {
      "epoch": 0.021456546544034373,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009995467516738165,
      "loss": 2.2863,
      "step": 3158
    },
    {
      "epoch": 0.02146334089062843,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999546293569457,
      "loss": 2.5506,
      "step": 3159
    },
    {
      "epoch": 0.02147013523722249,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995458352338134,
      "loss": 2.4332,
      "step": 3160
    },
    {
      "epoch": 0.021476929583816546,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009995453766668855,
      "loss": 2.3204,
      "step": 3161
    },
    {
      "epoch": 0.021483723930410602,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995449178686743,
      "loss": 2.5896,
      "step": 3162
    },
    {
      "epoch": 0.021490518277004662,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995444588391793,
      "loss": 2.4546,
      "step": 3163
    },
    {
      "epoch": 0.02149731262359872,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995439995784008,
      "loss": 2.475,
      "step": 3164
    },
    {
      "epoch": 0.021504106970192775,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995435400863392,
      "loss": 2.4396,
      "step": 3165
    },
    {
      "epoch": 0.021510901316786835,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999543080362995,
      "loss": 2.4448,
      "step": 3166
    },
    {
      "epoch": 0.021517695663380892,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009995426204083677,
      "loss": 2.3863,
      "step": 3167
    },
    {
      "epoch": 0.02152449000997495,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999542160222458,
      "loss": 2.3906,
      "step": 3168
    },
    {
      "epoch": 0.021531284356569008,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999541699805266,
      "loss": 2.5826,
      "step": 3169
    },
    {
      "epoch": 0.021538078703163065,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000999541239156792,
      "loss": 2.5013,
      "step": 3170
    },
    {
      "epoch": 0.021544873049757125,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999540778277036,
      "loss": 2.3891,
      "step": 3171
    },
    {
      "epoch": 0.02155166739635118,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995403171659985,
      "loss": 2.2822,
      "step": 3172
    },
    {
      "epoch": 0.021558461742945238,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995398558236797,
      "loss": 2.6529,
      "step": 3173
    },
    {
      "epoch": 0.021565256089539298,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009995393942500794,
      "loss": 2.4234,
      "step": 3174
    },
    {
      "epoch": 0.021572050436133354,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009995389324451982,
      "loss": 2.2978,
      "step": 3175
    },
    {
      "epoch": 0.02157884478272741,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999538470409036,
      "loss": 2.3929,
      "step": 3176
    },
    {
      "epoch": 0.02158563912932147,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995380081415934,
      "loss": 2.3872,
      "step": 3177
    },
    {
      "epoch": 0.021592433475915527,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009995375456428705,
      "loss": 2.39,
      "step": 3178
    },
    {
      "epoch": 0.021599227822509587,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995370829128673,
      "loss": 2.4904,
      "step": 3179
    },
    {
      "epoch": 0.021606022169103643,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009995366199515843,
      "loss": 2.3239,
      "step": 3180
    },
    {
      "epoch": 0.0216128165156977,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995361567590212,
      "loss": 2.6062,
      "step": 3181
    },
    {
      "epoch": 0.02161961086229176,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995356933351792,
      "loss": 2.5954,
      "step": 3182
    },
    {
      "epoch": 0.021626405208885816,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009995352296800575,
      "loss": 2.4862,
      "step": 3183
    },
    {
      "epoch": 0.021633199555479873,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009995347657936566,
      "loss": 2.4593,
      "step": 3184
    },
    {
      "epoch": 0.021639993902073933,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995343016759772,
      "loss": 2.3931,
      "step": 3185
    },
    {
      "epoch": 0.02164678824866799,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995338373270189,
      "loss": 2.3512,
      "step": 3186
    },
    {
      "epoch": 0.02165358259526205,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995333727467822,
      "loss": 2.4739,
      "step": 3187
    },
    {
      "epoch": 0.021660376941856106,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995329079352674,
      "loss": 2.4244,
      "step": 3188
    },
    {
      "epoch": 0.021667171288450162,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009995324428924745,
      "loss": 2.3861,
      "step": 3189
    },
    {
      "epoch": 0.021673965635044222,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995319776184037,
      "loss": 2.3861,
      "step": 3190
    },
    {
      "epoch": 0.02168075998163828,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009995315121130554,
      "loss": 2.3672,
      "step": 3191
    },
    {
      "epoch": 0.021687554328232335,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009995310463764297,
      "loss": 2.3014,
      "step": 3192
    },
    {
      "epoch": 0.021694348674826395,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009995305804085268,
      "loss": 2.43,
      "step": 3193
    },
    {
      "epoch": 0.02170114302142045,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999530114209347,
      "loss": 2.4799,
      "step": 3194
    },
    {
      "epoch": 0.021707937368014508,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995296477788907,
      "loss": 2.5326,
      "step": 3195
    },
    {
      "epoch": 0.021714731714608568,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995291811171578,
      "loss": 2.2239,
      "step": 3196
    },
    {
      "epoch": 0.021721526061202624,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009995287142241485,
      "loss": 2.4452,
      "step": 3197
    },
    {
      "epoch": 0.021728320407796684,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009995282470998633,
      "loss": 2.4693,
      "step": 3198
    },
    {
      "epoch": 0.02173511475439074,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995277797443022,
      "loss": 2.4463,
      "step": 3199
    },
    {
      "epoch": 0.021741909100984797,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995273121574653,
      "loss": 2.5875,
      "step": 3200
    },
    {
      "epoch": 0.021748703447578857,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009995268443393531,
      "loss": 2.4099,
      "step": 3201
    },
    {
      "epoch": 0.021755497794172914,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995263762899657,
      "loss": 2.5218,
      "step": 3202
    },
    {
      "epoch": 0.02176229214076697,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009995259080093034,
      "loss": 2.4102,
      "step": 3203
    },
    {
      "epoch": 0.02176908648736103,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999525439497366,
      "loss": 2.4135,
      "step": 3204
    },
    {
      "epoch": 0.021775880833955086,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009995249707541545,
      "loss": 2.4362,
      "step": 3205
    },
    {
      "epoch": 0.021782675180549143,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995245017796683,
      "loss": 2.3531,
      "step": 3206
    },
    {
      "epoch": 0.021789469527143203,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999524032573908,
      "loss": 2.4502,
      "step": 3207
    },
    {
      "epoch": 0.02179626387373726,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009995235631368741,
      "loss": 2.4499,
      "step": 3208
    },
    {
      "epoch": 0.02180305822033132,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995230934685664,
      "loss": 2.4866,
      "step": 3209
    },
    {
      "epoch": 0.021809852566925376,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009995226235689852,
      "loss": 2.3279,
      "step": 3210
    },
    {
      "epoch": 0.021816646913519432,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009995221534381308,
      "loss": 2.3148,
      "step": 3211
    },
    {
      "epoch": 0.021823441260113492,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009995216830760033,
      "loss": 2.4939,
      "step": 3212
    },
    {
      "epoch": 0.02183023560670755,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009995212124826029,
      "loss": 2.4154,
      "step": 3213
    },
    {
      "epoch": 0.021837029953301605,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00099952074165793,
      "loss": 2.4425,
      "step": 3214
    },
    {
      "epoch": 0.021843824299895665,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995202706019847,
      "loss": 2.4283,
      "step": 3215
    },
    {
      "epoch": 0.02185061864648972,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009995197993147673,
      "loss": 2.5272,
      "step": 3216
    },
    {
      "epoch": 0.021857412993083778,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999519327796278,
      "loss": 2.3625,
      "step": 3217
    },
    {
      "epoch": 0.021864207339677838,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009995188560465167,
      "loss": 2.4927,
      "step": 3218
    },
    {
      "epoch": 0.021871001686271895,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009995183840654843,
      "loss": 2.633,
      "step": 3219
    },
    {
      "epoch": 0.021877796032865954,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009995179118531803,
      "loss": 2.5058,
      "step": 3220
    },
    {
      "epoch": 0.02188459037946001,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009995174394096054,
      "loss": 2.4463,
      "step": 3221
    },
    {
      "epoch": 0.021891384726054067,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009995169667347595,
      "loss": 2.2234,
      "step": 3222
    },
    {
      "epoch": 0.021898179072648127,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995164938286431,
      "loss": 2.3663,
      "step": 3223
    },
    {
      "epoch": 0.021904973419242184,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009995160206912562,
      "loss": 2.3638,
      "step": 3224
    },
    {
      "epoch": 0.02191176776583624,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009995155473225992,
      "loss": 2.4343,
      "step": 3225
    },
    {
      "epoch": 0.0219185621124303,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995150737226721,
      "loss": 2.4169,
      "step": 3226
    },
    {
      "epoch": 0.021925356459024357,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995145998914754,
      "loss": 2.3515,
      "step": 3227
    },
    {
      "epoch": 0.021932150805618417,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009995141258290093,
      "loss": 2.5428,
      "step": 3228
    },
    {
      "epoch": 0.021938945152212473,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009995136515352737,
      "loss": 2.3124,
      "step": 3229
    },
    {
      "epoch": 0.02194573949880653,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999513177010269,
      "loss": 2.3235,
      "step": 3230
    },
    {
      "epoch": 0.02195253384540059,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995127022539953,
      "loss": 2.409,
      "step": 3231
    },
    {
      "epoch": 0.021959328191994646,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999512227266453,
      "loss": 2.5214,
      "step": 3232
    },
    {
      "epoch": 0.021966122538588703,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009995117520476425,
      "loss": 2.3827,
      "step": 3233
    },
    {
      "epoch": 0.021972916885182762,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995112765975637,
      "loss": 2.4666,
      "step": 3234
    },
    {
      "epoch": 0.02197971123177682,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995108009162167,
      "loss": 2.4536,
      "step": 3235
    },
    {
      "epoch": 0.021986505578370875,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000999510325003602,
      "loss": 2.4905,
      "step": 3236
    },
    {
      "epoch": 0.021993299924964935,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00099950984885972,
      "loss": 2.3354,
      "step": 3237
    },
    {
      "epoch": 0.022000094271558992,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009995093724845703,
      "loss": 2.4,
      "step": 3238
    },
    {
      "epoch": 0.022006888618153052,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995088958781536,
      "loss": 2.3769,
      "step": 3239
    },
    {
      "epoch": 0.02201368296474711,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009995084190404702,
      "loss": 2.4389,
      "step": 3240
    },
    {
      "epoch": 0.022020477311341165,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00099950794197152,
      "loss": 2.3431,
      "step": 3241
    },
    {
      "epoch": 0.022027271657935225,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995074646713033,
      "loss": 2.5381,
      "step": 3242
    },
    {
      "epoch": 0.02203406600452928,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009995069871398205,
      "loss": 2.4762,
      "step": 3243
    },
    {
      "epoch": 0.022040860351123338,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009995065093770714,
      "loss": 2.512,
      "step": 3244
    },
    {
      "epoch": 0.022047654697717398,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995060313830568,
      "loss": 2.5975,
      "step": 3245
    },
    {
      "epoch": 0.022054449044311454,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009995055531577767,
      "loss": 2.4823,
      "step": 3246
    },
    {
      "epoch": 0.02206124339090551,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999505074701231,
      "loss": 2.2531,
      "step": 3247
    },
    {
      "epoch": 0.02206803773749957,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009995045960134203,
      "loss": 2.4759,
      "step": 3248
    },
    {
      "epoch": 0.022074832084093627,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009995041170943447,
      "loss": 2.4585,
      "step": 3249
    },
    {
      "epoch": 0.022081626430687687,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009995036379440044,
      "loss": 2.4821,
      "step": 3250
    },
    {
      "epoch": 0.022088420777281743,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009995031585623996,
      "loss": 2.3861,
      "step": 3251
    },
    {
      "epoch": 0.0220952151238758,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009995026789495307,
      "loss": 2.4074,
      "step": 3252
    },
    {
      "epoch": 0.02210200947046986,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009995021991053977,
      "loss": 2.3508,
      "step": 3253
    },
    {
      "epoch": 0.022108803817063916,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009995017190300008,
      "loss": 2.3845,
      "step": 3254
    },
    {
      "epoch": 0.022115598163657973,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009995012387233404,
      "loss": 2.3476,
      "step": 3255
    },
    {
      "epoch": 0.022122392510252033,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009995007581854167,
      "loss": 2.3374,
      "step": 3256
    },
    {
      "epoch": 0.02212918685684609,
      "grad_norm": 0.15625,
      "learning_rate": 0.00099950027741623,
      "loss": 2.5271,
      "step": 3257
    },
    {
      "epoch": 0.02213598120344015,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009994997964157802,
      "loss": 2.4042,
      "step": 3258
    },
    {
      "epoch": 0.022142775550034206,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994993151840677,
      "loss": 2.4104,
      "step": 3259
    },
    {
      "epoch": 0.022149569896628262,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999498833721093,
      "loss": 2.4446,
      "step": 3260
    },
    {
      "epoch": 0.022156364243222322,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994983520268557,
      "loss": 2.2272,
      "step": 3261
    },
    {
      "epoch": 0.02216315858981638,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009994978701013566,
      "loss": 2.3125,
      "step": 3262
    },
    {
      "epoch": 0.022169952936410435,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009994973879445957,
      "loss": 2.3914,
      "step": 3263
    },
    {
      "epoch": 0.022176747283004495,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994969055565732,
      "loss": 2.3496,
      "step": 3264
    },
    {
      "epoch": 0.02218354162959855,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994964229372895,
      "loss": 2.4269,
      "step": 3265
    },
    {
      "epoch": 0.022190335976192608,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994959400867445,
      "loss": 2.5989,
      "step": 3266
    },
    {
      "epoch": 0.022197130322786668,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009994954570049387,
      "loss": 2.5649,
      "step": 3267
    },
    {
      "epoch": 0.022203924669380724,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009994949736918722,
      "loss": 2.3653,
      "step": 3268
    },
    {
      "epoch": 0.022210719015974784,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009994944901475452,
      "loss": 2.616,
      "step": 3269
    },
    {
      "epoch": 0.02221751336256884,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994940063719582,
      "loss": 2.4942,
      "step": 3270
    },
    {
      "epoch": 0.022224307709162897,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999493522365111,
      "loss": 2.3427,
      "step": 3271
    },
    {
      "epoch": 0.022231102055756957,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999493038127004,
      "loss": 2.4837,
      "step": 3272
    },
    {
      "epoch": 0.022237896402351014,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994925536576376,
      "loss": 2.3529,
      "step": 3273
    },
    {
      "epoch": 0.02224469074894507,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994920689570118,
      "loss": 2.3372,
      "step": 3274
    },
    {
      "epoch": 0.02225148509553913,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009994915840251268,
      "loss": 2.4451,
      "step": 3275
    },
    {
      "epoch": 0.022258279442133187,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999491098861983,
      "loss": 2.3739,
      "step": 3276
    },
    {
      "epoch": 0.022265073788727243,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009994906134675806,
      "loss": 2.3004,
      "step": 3277
    },
    {
      "epoch": 0.022271868135321303,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994901278419198,
      "loss": 2.3904,
      "step": 3278
    },
    {
      "epoch": 0.02227866248191536,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009994896419850006,
      "loss": 2.2322,
      "step": 3279
    },
    {
      "epoch": 0.02228545682850942,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009994891558968236,
      "loss": 2.4159,
      "step": 3280
    },
    {
      "epoch": 0.022292251175103476,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009994886695773887,
      "loss": 2.3327,
      "step": 3281
    },
    {
      "epoch": 0.022299045521697532,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994881830266964,
      "loss": 2.4537,
      "step": 3282
    },
    {
      "epoch": 0.022305839868291592,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009994876962447468,
      "loss": 2.5439,
      "step": 3283
    },
    {
      "epoch": 0.02231263421488565,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009994872092315402,
      "loss": 2.5541,
      "step": 3284
    },
    {
      "epoch": 0.022319428561479705,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994867219870766,
      "loss": 2.4133,
      "step": 3285
    },
    {
      "epoch": 0.022326222908073765,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009994862345113565,
      "loss": 2.4599,
      "step": 3286
    },
    {
      "epoch": 0.02233301725466782,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994857468043797,
      "loss": 2.3354,
      "step": 3287
    },
    {
      "epoch": 0.02233981160126188,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999485258866147,
      "loss": 2.4169,
      "step": 3288
    },
    {
      "epoch": 0.022346605947855938,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994847706966583,
      "loss": 2.5094,
      "step": 3289
    },
    {
      "epoch": 0.022353400294449995,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009994842822959138,
      "loss": 2.4807,
      "step": 3290
    },
    {
      "epoch": 0.022360194641044055,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000999483793663914,
      "loss": 2.4435,
      "step": 3291
    },
    {
      "epoch": 0.02236698898763811,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994833048006587,
      "loss": 2.4645,
      "step": 3292
    },
    {
      "epoch": 0.022373783334232168,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009994828157061486,
      "loss": 2.5756,
      "step": 3293
    },
    {
      "epoch": 0.022380577680826227,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009994823263803833,
      "loss": 2.3472,
      "step": 3294
    },
    {
      "epoch": 0.022387372027420284,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009994818368233638,
      "loss": 2.3669,
      "step": 3295
    },
    {
      "epoch": 0.02239416637401434,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009994813470350899,
      "loss": 2.4975,
      "step": 3296
    },
    {
      "epoch": 0.0224009607206084,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009994808570155617,
      "loss": 2.3148,
      "step": 3297
    },
    {
      "epoch": 0.022407755067202457,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994803667647796,
      "loss": 2.4363,
      "step": 3298
    },
    {
      "epoch": 0.022414549413796517,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994798762827438,
      "loss": 2.4504,
      "step": 3299
    },
    {
      "epoch": 0.022421343760390573,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009994793855694546,
      "loss": 2.4499,
      "step": 3300
    },
    {
      "epoch": 0.02242813810698463,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994788946249122,
      "loss": 2.29,
      "step": 3301
    },
    {
      "epoch": 0.02243493245357869,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994784034491168,
      "loss": 2.4097,
      "step": 3302
    },
    {
      "epoch": 0.022441726800172746,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994779120420685,
      "loss": 2.3858,
      "step": 3303
    },
    {
      "epoch": 0.022448521146766803,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009994774204037677,
      "loss": 2.2607,
      "step": 3304
    },
    {
      "epoch": 0.022455315493360863,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009994769285342146,
      "loss": 2.5215,
      "step": 3305
    },
    {
      "epoch": 0.02246210983995492,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994764364334096,
      "loss": 2.2591,
      "step": 3306
    },
    {
      "epoch": 0.022468904186548976,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009994759441013525,
      "loss": 2.4681,
      "step": 3307
    },
    {
      "epoch": 0.022475698533143035,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009994754515380438,
      "loss": 2.3916,
      "step": 3308
    },
    {
      "epoch": 0.022482492879737092,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009994749587434839,
      "loss": 2.4038,
      "step": 3309
    },
    {
      "epoch": 0.022489287226331152,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009994744657176725,
      "loss": 2.4495,
      "step": 3310
    },
    {
      "epoch": 0.02249608157292521,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994739724606104,
      "loss": 2.4735,
      "step": 3311
    },
    {
      "epoch": 0.022502875919519265,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009994734789722974,
      "loss": 2.3282,
      "step": 3312
    },
    {
      "epoch": 0.022509670266113325,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999472985252734,
      "loss": 2.3068,
      "step": 3313
    },
    {
      "epoch": 0.02251646461270738,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009994724913019203,
      "loss": 2.3243,
      "step": 3314
    },
    {
      "epoch": 0.022523258959301438,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994719971198566,
      "loss": 2.3816,
      "step": 3315
    },
    {
      "epoch": 0.022530053305895498,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999471502706543,
      "loss": 2.5088,
      "step": 3316
    },
    {
      "epoch": 0.022536847652489554,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00099947100806198,
      "loss": 2.4777,
      "step": 3317
    },
    {
      "epoch": 0.022543641999083614,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994705131861675,
      "loss": 2.2578,
      "step": 3318
    },
    {
      "epoch": 0.02255043634567767,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994700180791058,
      "loss": 2.2792,
      "step": 3319
    },
    {
      "epoch": 0.022557230692271727,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009994695227407953,
      "loss": 2.2148,
      "step": 3320
    },
    {
      "epoch": 0.022564025038865787,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009994690271712364,
      "loss": 2.5845,
      "step": 3321
    },
    {
      "epoch": 0.022570819385459844,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009994685313704287,
      "loss": 2.3605,
      "step": 3322
    },
    {
      "epoch": 0.0225776137320539,
      "grad_norm": 0.1884765625,
      "learning_rate": 0.000999468035338373,
      "loss": 2.3707,
      "step": 3323
    },
    {
      "epoch": 0.02258440807864796,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999467539075069,
      "loss": 2.3497,
      "step": 3324
    },
    {
      "epoch": 0.022591202425242016,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994670425805176,
      "loss": 2.4813,
      "step": 3325
    },
    {
      "epoch": 0.022597996771836073,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009994665458547187,
      "loss": 2.3863,
      "step": 3326
    },
    {
      "epoch": 0.022604791118430133,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009994660488976724,
      "loss": 2.4569,
      "step": 3327
    },
    {
      "epoch": 0.02261158546502419,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999465551709379,
      "loss": 2.3244,
      "step": 3328
    },
    {
      "epoch": 0.02261837981161825,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009994650542898388,
      "loss": 2.3818,
      "step": 3329
    },
    {
      "epoch": 0.022625174158212306,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009994645566390522,
      "loss": 2.4253,
      "step": 3330
    },
    {
      "epoch": 0.022631968504806362,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999464058757019,
      "loss": 2.4104,
      "step": 3331
    },
    {
      "epoch": 0.022638762851400422,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994635606437397,
      "loss": 2.5486,
      "step": 3332
    },
    {
      "epoch": 0.02264555719799448,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994630622992145,
      "loss": 2.4499,
      "step": 3333
    },
    {
      "epoch": 0.022652351544588535,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009994625637234436,
      "loss": 2.4573,
      "step": 3334
    },
    {
      "epoch": 0.022659145891182595,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994620649164275,
      "loss": 2.457,
      "step": 3335
    },
    {
      "epoch": 0.02266594023777665,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994615658781658,
      "loss": 2.5092,
      "step": 3336
    },
    {
      "epoch": 0.022672734584370708,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994610666086594,
      "loss": 2.4675,
      "step": 3337
    },
    {
      "epoch": 0.022679528930964768,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994605671079081,
      "loss": 2.4499,
      "step": 3338
    },
    {
      "epoch": 0.022686323277558824,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994600673759123,
      "loss": 2.5343,
      "step": 3339
    },
    {
      "epoch": 0.022693117624152884,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009994595674126725,
      "loss": 2.3223,
      "step": 3340
    },
    {
      "epoch": 0.02269991197074694,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994590672181883,
      "loss": 2.4246,
      "step": 3341
    },
    {
      "epoch": 0.022706706317340997,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994585667924606,
      "loss": 2.4638,
      "step": 3342
    },
    {
      "epoch": 0.022713500663935057,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999458066135489,
      "loss": 2.413,
      "step": 3343
    },
    {
      "epoch": 0.022720295010529114,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009994575652472742,
      "loss": 2.5025,
      "step": 3344
    },
    {
      "epoch": 0.02272708935712317,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009994570641278162,
      "loss": 2.4384,
      "step": 3345
    },
    {
      "epoch": 0.02273388370371723,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994565627771152,
      "loss": 2.3851,
      "step": 3346
    },
    {
      "epoch": 0.022740678050311287,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994560611951717,
      "loss": 2.4369,
      "step": 3347
    },
    {
      "epoch": 0.022747472396905347,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009994555593819858,
      "loss": 2.4479,
      "step": 3348
    },
    {
      "epoch": 0.022754266743499403,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009994550573375577,
      "loss": 2.4724,
      "step": 3349
    },
    {
      "epoch": 0.02276106109009346,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994545550618876,
      "loss": 2.4393,
      "step": 3350
    },
    {
      "epoch": 0.02276785543668752,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994540525549759,
      "loss": 2.4023,
      "step": 3351
    },
    {
      "epoch": 0.022774649783281576,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009994535498168223,
      "loss": 2.484,
      "step": 3352
    },
    {
      "epoch": 0.022781444129875632,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994530468474276,
      "loss": 2.4568,
      "step": 3353
    },
    {
      "epoch": 0.022788238476469692,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994525436467921,
      "loss": 2.3654,
      "step": 3354
    },
    {
      "epoch": 0.02279503282306375,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994520402149157,
      "loss": 2.2499,
      "step": 3355
    },
    {
      "epoch": 0.022801827169657805,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009994515365517986,
      "loss": 2.5263,
      "step": 3356
    },
    {
      "epoch": 0.022808621516251865,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009994510326574414,
      "loss": 2.3732,
      "step": 3357
    },
    {
      "epoch": 0.022815415862845922,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009994505285318438,
      "loss": 2.3426,
      "step": 3358
    },
    {
      "epoch": 0.022822210209439982,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009994500241750064,
      "loss": 2.3663,
      "step": 3359
    },
    {
      "epoch": 0.022829004556034038,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009994495195869296,
      "loss": 2.4733,
      "step": 3360
    },
    {
      "epoch": 0.022835798902628095,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999449014767613,
      "loss": 2.4364,
      "step": 3361
    },
    {
      "epoch": 0.022842593249222155,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009994485097170576,
      "loss": 2.4082,
      "step": 3362
    },
    {
      "epoch": 0.02284938759581621,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009994480044352631,
      "loss": 2.4618,
      "step": 3363
    },
    {
      "epoch": 0.022856181942410268,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00099944749892223,
      "loss": 2.5682,
      "step": 3364
    },
    {
      "epoch": 0.022862976289004328,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009994469931779582,
      "loss": 2.4613,
      "step": 3365
    },
    {
      "epoch": 0.022869770635598384,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009994464872024485,
      "loss": 2.4978,
      "step": 3366
    },
    {
      "epoch": 0.02287656498219244,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994459809957005,
      "loss": 2.3497,
      "step": 3367
    },
    {
      "epoch": 0.0228833593287865,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009994454745577148,
      "loss": 2.3011,
      "step": 3368
    },
    {
      "epoch": 0.022890153675380557,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994449678884917,
      "loss": 2.4396,
      "step": 3369
    },
    {
      "epoch": 0.022896948021974617,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009994444609880313,
      "loss": 2.3953,
      "step": 3370
    },
    {
      "epoch": 0.022903742368568673,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994439538563338,
      "loss": 2.3953,
      "step": 3371
    },
    {
      "epoch": 0.02291053671516273,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009994434464933994,
      "loss": 2.2639,
      "step": 3372
    },
    {
      "epoch": 0.02291733106175679,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994429388992285,
      "loss": 2.3237,
      "step": 3373
    },
    {
      "epoch": 0.022924125408350846,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994424310738213,
      "loss": 2.4919,
      "step": 3374
    },
    {
      "epoch": 0.022930919754944903,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994419230171779,
      "loss": 2.3713,
      "step": 3375
    },
    {
      "epoch": 0.022937714101538963,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994414147292984,
      "loss": 2.4648,
      "step": 3376
    },
    {
      "epoch": 0.02294450844813302,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009994409062101836,
      "loss": 2.3666,
      "step": 3377
    },
    {
      "epoch": 0.02295130279472708,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009994403974598332,
      "loss": 2.4781,
      "step": 3378
    },
    {
      "epoch": 0.022958097141321136,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994398884782478,
      "loss": 2.4773,
      "step": 3379
    },
    {
      "epoch": 0.022964891487915192,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009994393792654272,
      "loss": 2.406,
      "step": 3380
    },
    {
      "epoch": 0.022971685834509252,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009994388698213721,
      "loss": 2.4046,
      "step": 3381
    },
    {
      "epoch": 0.02297848018110331,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994383601460823,
      "loss": 2.3729,
      "step": 3382
    },
    {
      "epoch": 0.022985274527697365,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009994378502395584,
      "loss": 2.4275,
      "step": 3383
    },
    {
      "epoch": 0.022992068874291425,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009994373401018006,
      "loss": 2.4489,
      "step": 3384
    },
    {
      "epoch": 0.02299886322088548,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999436829732809,
      "loss": 2.5017,
      "step": 3385
    },
    {
      "epoch": 0.023005657567479538,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994363191325839,
      "loss": 2.5122,
      "step": 3386
    },
    {
      "epoch": 0.023012451914073598,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009994358083011254,
      "loss": 2.3815,
      "step": 3387
    },
    {
      "epoch": 0.023019246260667654,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009994352972384339,
      "loss": 2.4422,
      "step": 3388
    },
    {
      "epoch": 0.023026040607261714,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009994347859445096,
      "loss": 2.4203,
      "step": 3389
    },
    {
      "epoch": 0.02303283495385577,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994342744193527,
      "loss": 2.4111,
      "step": 3390
    },
    {
      "epoch": 0.023039629300449827,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009994337626629635,
      "loss": 2.4681,
      "step": 3391
    },
    {
      "epoch": 0.023046423647043887,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009994332506753421,
      "loss": 2.5608,
      "step": 3392
    },
    {
      "epoch": 0.023053217993637944,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994327384564889,
      "loss": 2.3289,
      "step": 3393
    },
    {
      "epoch": 0.023060012340232,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009994322260064039,
      "loss": 2.4365,
      "step": 3394
    },
    {
      "epoch": 0.02306680668682606,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009994317133250876,
      "loss": 2.2873,
      "step": 3395
    },
    {
      "epoch": 0.023073601033420117,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009994312004125401,
      "loss": 2.5152,
      "step": 3396
    },
    {
      "epoch": 0.023080395380014173,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999430687268762,
      "loss": 2.3426,
      "step": 3397
    },
    {
      "epoch": 0.023087189726608233,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994301738937528,
      "loss": 2.4268,
      "step": 3398
    },
    {
      "epoch": 0.02309398407320229,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009994296602875135,
      "loss": 2.479,
      "step": 3399
    },
    {
      "epoch": 0.02310077841979635,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009994291464500437,
      "loss": 2.3896,
      "step": 3400
    },
    {
      "epoch": 0.023107572766390406,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009994286323813442,
      "loss": 2.3093,
      "step": 3401
    },
    {
      "epoch": 0.023114367112984462,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009994281180814148,
      "loss": 2.4125,
      "step": 3402
    },
    {
      "epoch": 0.023121161459578522,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009994276035502558,
      "loss": 2.3775,
      "step": 3403
    },
    {
      "epoch": 0.02312795580617258,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999427088787868,
      "loss": 2.5307,
      "step": 3404
    },
    {
      "epoch": 0.023134750152766635,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994265737942507,
      "loss": 2.2613,
      "step": 3405
    },
    {
      "epoch": 0.023141544499360695,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994260585694046,
      "loss": 2.4469,
      "step": 3406
    },
    {
      "epoch": 0.02314833884595475,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994255431133303,
      "loss": 2.3644,
      "step": 3407
    },
    {
      "epoch": 0.02315513319254881,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994250274260276,
      "loss": 2.5031,
      "step": 3408
    },
    {
      "epoch": 0.023161927539142868,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994245115074968,
      "loss": 2.5911,
      "step": 3409
    },
    {
      "epoch": 0.023168721885736925,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000999423995357738,
      "loss": 2.478,
      "step": 3410
    },
    {
      "epoch": 0.023175516232330984,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994234789767519,
      "loss": 2.3938,
      "step": 3411
    },
    {
      "epoch": 0.02318231057892504,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994229623645383,
      "loss": 2.3292,
      "step": 3412
    },
    {
      "epoch": 0.023189104925519097,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009994224455210977,
      "loss": 2.4987,
      "step": 3413
    },
    {
      "epoch": 0.023195899272113157,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00099942192844643,
      "loss": 2.4802,
      "step": 3414
    },
    {
      "epoch": 0.023202693618707214,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994214111405358,
      "loss": 2.45,
      "step": 3415
    },
    {
      "epoch": 0.02320948796530127,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009994208936034152,
      "loss": 2.4633,
      "step": 3416
    },
    {
      "epoch": 0.02321628231189533,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009994203758350686,
      "loss": 2.5009,
      "step": 3417
    },
    {
      "epoch": 0.023223076658489387,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999419857835496,
      "loss": 2.5097,
      "step": 3418
    },
    {
      "epoch": 0.023229871005083447,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994193396046978,
      "loss": 2.4642,
      "step": 3419
    },
    {
      "epoch": 0.023236665351677503,
      "grad_norm": 0.177734375,
      "learning_rate": 0.000999418821142674,
      "loss": 2.7502,
      "step": 3420
    },
    {
      "epoch": 0.02324345969827156,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999418302449425,
      "loss": 2.4042,
      "step": 3421
    },
    {
      "epoch": 0.02325025404486562,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994177835249514,
      "loss": 2.3748,
      "step": 3422
    },
    {
      "epoch": 0.023257048391459676,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994172643692527,
      "loss": 2.4496,
      "step": 3423
    },
    {
      "epoch": 0.023263842738053733,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009994167449823297,
      "loss": 2.3453,
      "step": 3424
    },
    {
      "epoch": 0.023270637084647793,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009994162253641825,
      "loss": 2.4475,
      "step": 3425
    },
    {
      "epoch": 0.02327743143124185,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999415705514811,
      "loss": 2.3865,
      "step": 3426
    },
    {
      "epoch": 0.023284225777835905,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009994151854342162,
      "loss": 2.3834,
      "step": 3427
    },
    {
      "epoch": 0.023291020124429965,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994146651223978,
      "loss": 2.4943,
      "step": 3428
    },
    {
      "epoch": 0.023297814471024022,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000999414144579356,
      "loss": 2.3178,
      "step": 3429
    },
    {
      "epoch": 0.023304608817618082,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009994136238050913,
      "loss": 2.3616,
      "step": 3430
    },
    {
      "epoch": 0.02331140316421214,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009994131027996037,
      "loss": 2.4804,
      "step": 3431
    },
    {
      "epoch": 0.023318197510806195,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009994125815628936,
      "loss": 2.2993,
      "step": 3432
    },
    {
      "epoch": 0.023324991857400255,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994120600949612,
      "loss": 2.4242,
      "step": 3433
    },
    {
      "epoch": 0.02333178620399431,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009994115383958066,
      "loss": 2.2197,
      "step": 3434
    },
    {
      "epoch": 0.023338580550588368,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994110164654303,
      "loss": 2.3899,
      "step": 3435
    },
    {
      "epoch": 0.023345374897182428,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009994104943038325,
      "loss": 2.5773,
      "step": 3436
    },
    {
      "epoch": 0.023352169243776484,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009994099719110134,
      "loss": 2.39,
      "step": 3437
    },
    {
      "epoch": 0.023358963590370544,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009994094492869732,
      "loss": 2.5486,
      "step": 3438
    },
    {
      "epoch": 0.0233657579369646,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999408926431712,
      "loss": 2.3594,
      "step": 3439
    },
    {
      "epoch": 0.023372552283558657,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994084033452303,
      "loss": 2.3487,
      "step": 3440
    },
    {
      "epoch": 0.023379346630152717,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009994078800275282,
      "loss": 2.3826,
      "step": 3441
    },
    {
      "epoch": 0.023386140976746773,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994073564786059,
      "loss": 2.2884,
      "step": 3442
    },
    {
      "epoch": 0.02339293532334083,
      "grad_norm": 0.17578125,
      "learning_rate": 0.000999406832698464,
      "loss": 2.5568,
      "step": 3443
    },
    {
      "epoch": 0.02339972966993489,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994063086871021,
      "loss": 2.4594,
      "step": 3444
    },
    {
      "epoch": 0.023406524016528946,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999405784444521,
      "loss": 2.5233,
      "step": 3445
    },
    {
      "epoch": 0.023413318363123003,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009994052599707208,
      "loss": 2.5457,
      "step": 3446
    },
    {
      "epoch": 0.023420112709717063,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994047352657016,
      "loss": 2.2284,
      "step": 3447
    },
    {
      "epoch": 0.02342690705631112,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994042103294637,
      "loss": 2.4118,
      "step": 3448
    },
    {
      "epoch": 0.02343370140290518,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009994036851620075,
      "loss": 2.5134,
      "step": 3449
    },
    {
      "epoch": 0.023440495749499236,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009994031597633328,
      "loss": 2.3231,
      "step": 3450
    },
    {
      "epoch": 0.023447290096093292,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009994026341334406,
      "loss": 2.3792,
      "step": 3451
    },
    {
      "epoch": 0.023454084442687352,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009994021082723303,
      "loss": 2.2861,
      "step": 3452
    },
    {
      "epoch": 0.02346087878928141,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009994015821800028,
      "loss": 2.2263,
      "step": 3453
    },
    {
      "epoch": 0.023467673135875465,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009994010558564582,
      "loss": 2.4556,
      "step": 3454
    },
    {
      "epoch": 0.023474467482469525,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009994005293016963,
      "loss": 2.4231,
      "step": 3455
    },
    {
      "epoch": 0.02348126182906358,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999400002515718,
      "loss": 2.3648,
      "step": 3456
    },
    {
      "epoch": 0.023488056175657638,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999399475498523,
      "loss": 2.419,
      "step": 3457
    },
    {
      "epoch": 0.023494850522251698,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999398948250112,
      "loss": 2.3835,
      "step": 3458
    },
    {
      "epoch": 0.023501644868845754,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009993984207704847,
      "loss": 2.4615,
      "step": 3459
    },
    {
      "epoch": 0.023508439215439814,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009993978930596418,
      "loss": 2.2966,
      "step": 3460
    },
    {
      "epoch": 0.02351523356203387,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993973651175835,
      "loss": 2.4048,
      "step": 3461
    },
    {
      "epoch": 0.023522027908627927,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00099939683694431,
      "loss": 2.411,
      "step": 3462
    },
    {
      "epoch": 0.023528822255221987,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993963085398211,
      "loss": 2.271,
      "step": 3463
    },
    {
      "epoch": 0.023535616601816044,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009993957799041178,
      "loss": 2.5385,
      "step": 3464
    },
    {
      "epoch": 0.0235424109484101,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993952510371999,
      "loss": 2.4208,
      "step": 3465
    },
    {
      "epoch": 0.02354920529500416,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993947219390676,
      "loss": 2.4941,
      "step": 3466
    },
    {
      "epoch": 0.023555999641598217,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009993941926097214,
      "loss": 2.3883,
      "step": 3467
    },
    {
      "epoch": 0.023562793988192277,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009993936630491613,
      "loss": 2.5484,
      "step": 3468
    },
    {
      "epoch": 0.023569588334786333,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009993931332573879,
      "loss": 2.3874,
      "step": 3469
    },
    {
      "epoch": 0.02357638268138039,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999392603234401,
      "loss": 2.3703,
      "step": 3470
    },
    {
      "epoch": 0.02358317702797445,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999392072980201,
      "loss": 2.5808,
      "step": 3471
    },
    {
      "epoch": 0.023589971374568506,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993915424947883,
      "loss": 2.4078,
      "step": 3472
    },
    {
      "epoch": 0.023596765721162562,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999391011778163,
      "loss": 2.4423,
      "step": 3473
    },
    {
      "epoch": 0.023603560067756622,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009993904808303256,
      "loss": 2.4689,
      "step": 3474
    },
    {
      "epoch": 0.02361035441435068,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009993899496512759,
      "loss": 2.4564,
      "step": 3475
    },
    {
      "epoch": 0.023617148760944735,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009993894182410144,
      "loss": 2.4385,
      "step": 3476
    },
    {
      "epoch": 0.023623943107538795,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009993888865995413,
      "loss": 2.314,
      "step": 3477
    },
    {
      "epoch": 0.02363073745413285,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999388354726857,
      "loss": 2.4151,
      "step": 3478
    },
    {
      "epoch": 0.02363753180072691,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009993878226229615,
      "loss": 2.3276,
      "step": 3479
    },
    {
      "epoch": 0.023644326147320968,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009993872902878552,
      "loss": 2.5109,
      "step": 3480
    },
    {
      "epoch": 0.023651120493915025,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009993867577215384,
      "loss": 2.4389,
      "step": 3481
    },
    {
      "epoch": 0.023657914840509085,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009993862249240112,
      "loss": 2.3864,
      "step": 3482
    },
    {
      "epoch": 0.02366470918710314,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993856918952737,
      "loss": 2.2946,
      "step": 3483
    },
    {
      "epoch": 0.023671503533697198,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993851586353266,
      "loss": 2.5283,
      "step": 3484
    },
    {
      "epoch": 0.023678297880291257,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993846251441698,
      "loss": 2.5001,
      "step": 3485
    },
    {
      "epoch": 0.023685092226885314,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993840914218036,
      "loss": 2.4024,
      "step": 3486
    },
    {
      "epoch": 0.02369188657347937,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993835574682284,
      "loss": 2.3926,
      "step": 3487
    },
    {
      "epoch": 0.02369868092007343,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009993830232834444,
      "loss": 2.6459,
      "step": 3488
    },
    {
      "epoch": 0.023705475266667487,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009993824888674516,
      "loss": 2.3526,
      "step": 3489
    },
    {
      "epoch": 0.023712269613261547,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009993819542202505,
      "loss": 2.5812,
      "step": 3490
    },
    {
      "epoch": 0.023719063959855603,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993814193418413,
      "loss": 2.3718,
      "step": 3491
    },
    {
      "epoch": 0.02372585830644966,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993808842322243,
      "loss": 2.3893,
      "step": 3492
    },
    {
      "epoch": 0.02373265265304372,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993803488913995,
      "loss": 2.3716,
      "step": 3493
    },
    {
      "epoch": 0.023739446999637776,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009993798133193675,
      "loss": 2.3995,
      "step": 3494
    },
    {
      "epoch": 0.023746241346231833,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009993792775161282,
      "loss": 2.4132,
      "step": 3495
    },
    {
      "epoch": 0.023753035692825893,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009993787414816822,
      "loss": 2.2082,
      "step": 3496
    },
    {
      "epoch": 0.02375983003941995,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993782052160295,
      "loss": 2.4268,
      "step": 3497
    },
    {
      "epoch": 0.02376662438601401,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009993776687191702,
      "loss": 2.4091,
      "step": 3498
    },
    {
      "epoch": 0.023773418732608066,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999377131991105,
      "loss": 2.3688,
      "step": 3499
    },
    {
      "epoch": 0.023780213079202122,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009993765950318339,
      "loss": 2.5345,
      "step": 3500
    },
    {
      "epoch": 0.023787007425796182,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999376057841357,
      "loss": 2.2418,
      "step": 3501
    },
    {
      "epoch": 0.02379380177239024,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999375520419675,
      "loss": 2.4551,
      "step": 3502
    },
    {
      "epoch": 0.023800596118984295,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009993749827667878,
      "loss": 2.3707,
      "step": 3503
    },
    {
      "epoch": 0.023807390465578355,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993744448826955,
      "loss": 2.4362,
      "step": 3504
    },
    {
      "epoch": 0.02381418481217241,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993739067673986,
      "loss": 2.3381,
      "step": 3505
    },
    {
      "epoch": 0.023820979158766468,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009993733684208975,
      "loss": 2.294,
      "step": 3506
    },
    {
      "epoch": 0.023827773505360528,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999372829843192,
      "loss": 2.5303,
      "step": 3507
    },
    {
      "epoch": 0.023834567851954584,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999372291034283,
      "loss": 2.4944,
      "step": 3508
    },
    {
      "epoch": 0.023841362198548644,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009993717519941698,
      "loss": 2.4874,
      "step": 3509
    },
    {
      "epoch": 0.0238481565451427,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009993712127228536,
      "loss": 2.4747,
      "step": 3510
    },
    {
      "epoch": 0.023854950891736757,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999370673220334,
      "loss": 2.5821,
      "step": 3511
    },
    {
      "epoch": 0.023861745238330817,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009993701334866118,
      "loss": 2.2702,
      "step": 3512
    },
    {
      "epoch": 0.023868539584924874,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009993695935216868,
      "loss": 2.5371,
      "step": 3513
    },
    {
      "epoch": 0.02387533393151893,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009993690533255594,
      "loss": 2.4127,
      "step": 3514
    },
    {
      "epoch": 0.02388212827811299,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993685128982298,
      "loss": 2.4418,
      "step": 3515
    },
    {
      "epoch": 0.023888922624707046,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993679722396984,
      "loss": 2.3697,
      "step": 3516
    },
    {
      "epoch": 0.023895716971301103,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009993674313499654,
      "loss": 2.3441,
      "step": 3517
    },
    {
      "epoch": 0.023902511317895163,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999366890229031,
      "loss": 2.5918,
      "step": 3518
    },
    {
      "epoch": 0.02390930566448922,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999366348876895,
      "loss": 2.3245,
      "step": 3519
    },
    {
      "epoch": 0.02391610001108328,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009993658072935587,
      "loss": 2.5127,
      "step": 3520
    },
    {
      "epoch": 0.023922894357677336,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009993652654790215,
      "loss": 2.4358,
      "step": 3521
    },
    {
      "epoch": 0.023929688704271392,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999364723433284,
      "loss": 2.4169,
      "step": 3522
    },
    {
      "epoch": 0.023936483050865452,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009993641811563464,
      "loss": 2.3476,
      "step": 3523
    },
    {
      "epoch": 0.02394327739745951,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993636386482086,
      "loss": 2.2598,
      "step": 3524
    },
    {
      "epoch": 0.023950071744053565,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009993630959088714,
      "loss": 2.4607,
      "step": 3525
    },
    {
      "epoch": 0.023956866090647625,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009993625529383348,
      "loss": 2.3989,
      "step": 3526
    },
    {
      "epoch": 0.02396366043724168,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000999362009736599,
      "loss": 2.4401,
      "step": 3527
    },
    {
      "epoch": 0.02397045478383574,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993614663036643,
      "loss": 2.3961,
      "step": 3528
    },
    {
      "epoch": 0.023977249130429798,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999360922639531,
      "loss": 2.4145,
      "step": 3529
    },
    {
      "epoch": 0.023984043477023854,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009993603787441995,
      "loss": 2.2411,
      "step": 3530
    },
    {
      "epoch": 0.023990837823617914,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993598346176697,
      "loss": 2.4591,
      "step": 3531
    },
    {
      "epoch": 0.02399763217021197,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999359290259942,
      "loss": 2.3657,
      "step": 3532
    },
    {
      "epoch": 0.024004426516806027,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009993587456710168,
      "loss": 2.3342,
      "step": 3533
    },
    {
      "epoch": 0.024011220863400087,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009993582008508942,
      "loss": 2.4326,
      "step": 3534
    },
    {
      "epoch": 0.024018015209994144,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009993576557995743,
      "loss": 2.4212,
      "step": 3535
    },
    {
      "epoch": 0.0240248095565882,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009993571105170575,
      "loss": 2.4517,
      "step": 3536
    },
    {
      "epoch": 0.02403160390318226,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009993565650033443,
      "loss": 2.4839,
      "step": 3537
    },
    {
      "epoch": 0.024038398249776317,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993560192584348,
      "loss": 2.3318,
      "step": 3538
    },
    {
      "epoch": 0.024045192596370377,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0009993554732823289,
      "loss": 2.3727,
      "step": 3539
    },
    {
      "epoch": 0.024051986942964433,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993549270750272,
      "loss": 2.2587,
      "step": 3540
    },
    {
      "epoch": 0.02405878128955849,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993543806365301,
      "loss": 2.416,
      "step": 3541
    },
    {
      "epoch": 0.02406557563615255,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009993538339668376,
      "loss": 2.3507,
      "step": 3542
    },
    {
      "epoch": 0.024072369982746606,
      "grad_norm": 0.15625,
      "learning_rate": 0.00099935328706595,
      "loss": 2.5922,
      "step": 3543
    },
    {
      "epoch": 0.024079164329340663,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009993527399338672,
      "loss": 2.3288,
      "step": 3544
    },
    {
      "epoch": 0.024085958675934722,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993521925705901,
      "loss": 2.2547,
      "step": 3545
    },
    {
      "epoch": 0.02409275302252878,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009993516449761188,
      "loss": 2.4543,
      "step": 3546
    },
    {
      "epoch": 0.024099547369122835,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993510971504531,
      "loss": 2.5145,
      "step": 3547
    },
    {
      "epoch": 0.024106341715716895,
      "grad_norm": 0.177734375,
      "learning_rate": 0.000999350549093594,
      "loss": 2.4829,
      "step": 3548
    },
    {
      "epoch": 0.024113136062310952,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999350000805541,
      "loss": 2.3907,
      "step": 3549
    },
    {
      "epoch": 0.024119930408905012,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009993494522862946,
      "loss": 2.5419,
      "step": 3550
    },
    {
      "epoch": 0.024126724755499068,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009993489035358552,
      "loss": 2.4118,
      "step": 3551
    },
    {
      "epoch": 0.024133519102093125,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993483545542232,
      "loss": 2.3165,
      "step": 3552
    },
    {
      "epoch": 0.024140313448687185,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009993478053413984,
      "loss": 2.5887,
      "step": 3553
    },
    {
      "epoch": 0.02414710779528124,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009993472558973812,
      "loss": 2.5729,
      "step": 3554
    },
    {
      "epoch": 0.024153902141875298,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009993467062221722,
      "loss": 2.4539,
      "step": 3555
    },
    {
      "epoch": 0.024160696488469358,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009993461563157712,
      "loss": 2.4037,
      "step": 3556
    },
    {
      "epoch": 0.024167490835063414,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009993456061781788,
      "loss": 2.4923,
      "step": 3557
    },
    {
      "epoch": 0.024174285181657474,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009993450558093953,
      "loss": 2.3124,
      "step": 3558
    },
    {
      "epoch": 0.02418107952825153,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009993445052094204,
      "loss": 2.3632,
      "step": 3559
    },
    {
      "epoch": 0.024187873874845587,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009993439543782548,
      "loss": 2.4844,
      "step": 3560
    },
    {
      "epoch": 0.024194668221439647,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009993434033158989,
      "loss": 2.3284,
      "step": 3561
    },
    {
      "epoch": 0.024201462568033703,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993428520223525,
      "loss": 2.3906,
      "step": 3562
    },
    {
      "epoch": 0.02420825691462776,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999342300497616,
      "loss": 2.4195,
      "step": 3563
    },
    {
      "epoch": 0.02421505126122182,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009993417487416901,
      "loss": 2.4907,
      "step": 3564
    },
    {
      "epoch": 0.024221845607815876,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009993411967545744,
      "loss": 2.5468,
      "step": 3565
    },
    {
      "epoch": 0.024228639954409933,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009993406445362694,
      "loss": 2.3897,
      "step": 3566
    },
    {
      "epoch": 0.024235434301003993,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009993400920867757,
      "loss": 2.3448,
      "step": 3567
    },
    {
      "epoch": 0.02424222864759805,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000999339539406093,
      "loss": 2.4035,
      "step": 3568
    },
    {
      "epoch": 0.02424902299419211,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999338986494222,
      "loss": 2.3965,
      "step": 3569
    },
    {
      "epoch": 0.024255817340786166,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009993384333511627,
      "loss": 2.5589,
      "step": 3570
    },
    {
      "epoch": 0.024262611687380222,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009993378799769152,
      "loss": 2.3596,
      "step": 3571
    },
    {
      "epoch": 0.024269406033974282,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009993373263714803,
      "loss": 2.342,
      "step": 3572
    },
    {
      "epoch": 0.02427620038056834,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009993367725348579,
      "loss": 2.4603,
      "step": 3573
    },
    {
      "epoch": 0.024282994727162395,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009993362184670482,
      "loss": 2.5075,
      "step": 3574
    },
    {
      "epoch": 0.024289789073756455,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009993356641680516,
      "loss": 2.5042,
      "step": 3575
    },
    {
      "epoch": 0.02429658342035051,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009993351096378683,
      "loss": 2.3027,
      "step": 3576
    },
    {
      "epoch": 0.024303377766944568,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009993345548764984,
      "loss": 2.3405,
      "step": 3577
    },
    {
      "epoch": 0.024310172113538628,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009993339998839426,
      "loss": 2.4144,
      "step": 3578
    },
    {
      "epoch": 0.024316966460132684,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993334446602008,
      "loss": 2.4317,
      "step": 3579
    },
    {
      "epoch": 0.024323760806726744,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993328892052733,
      "loss": 2.3764,
      "step": 3580
    },
    {
      "epoch": 0.0243305551533208,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009993323335191606,
      "loss": 2.3932,
      "step": 3581
    },
    {
      "epoch": 0.024337349499914857,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009993317776018625,
      "loss": 2.58,
      "step": 3582
    },
    {
      "epoch": 0.024344143846508917,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009993312214533797,
      "loss": 2.383,
      "step": 3583
    },
    {
      "epoch": 0.024350938193102974,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999330665073712,
      "loss": 2.3366,
      "step": 3584
    },
    {
      "epoch": 0.02435773253969703,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.00099933010846286,
      "loss": 2.3442,
      "step": 3585
    },
    {
      "epoch": 0.02436452688629109,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999329551620824,
      "loss": 2.4182,
      "step": 3586
    },
    {
      "epoch": 0.024371321232885147,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009993289945476043,
      "loss": 2.3842,
      "step": 3587
    },
    {
      "epoch": 0.024378115579479206,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009993284372432007,
      "loss": 2.4261,
      "step": 3588
    },
    {
      "epoch": 0.024384909926073263,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009993278797076138,
      "loss": 2.3931,
      "step": 3589
    },
    {
      "epoch": 0.02439170427266732,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.000999327321940844,
      "loss": 2.493,
      "step": 3590
    },
    {
      "epoch": 0.02439849861926138,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009993267639428912,
      "loss": 2.3951,
      "step": 3591
    },
    {
      "epoch": 0.024405292965855436,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999326205713756,
      "loss": 2.5227,
      "step": 3592
    },
    {
      "epoch": 0.024412087312449492,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993256472534383,
      "loss": 2.4207,
      "step": 3593
    },
    {
      "epoch": 0.024418881659043552,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993250885619387,
      "loss": 2.5179,
      "step": 3594
    },
    {
      "epoch": 0.02442567600563761,
      "grad_norm": 0.26953125,
      "learning_rate": 0.0009993245296392572,
      "loss": 2.5485,
      "step": 3595
    },
    {
      "epoch": 0.024432470352231665,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009993239704853942,
      "loss": 2.4811,
      "step": 3596
    },
    {
      "epoch": 0.024439264698825725,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099932341110035,
      "loss": 2.4341,
      "step": 3597
    },
    {
      "epoch": 0.02444605904541978,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993228514841245,
      "loss": 2.3973,
      "step": 3598
    },
    {
      "epoch": 0.02445285339201384,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009993222916367187,
      "loss": 2.3785,
      "step": 3599
    },
    {
      "epoch": 0.024459647738607898,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999321731558132,
      "loss": 2.5044,
      "step": 3600
    },
    {
      "epoch": 0.024466442085201955,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009993211712483654,
      "loss": 2.3107,
      "step": 3601
    },
    {
      "epoch": 0.024473236431796015,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009993206107074186,
      "loss": 2.3306,
      "step": 3602
    },
    {
      "epoch": 0.02448003077839007,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009993200499352922,
      "loss": 2.5648,
      "step": 3603
    },
    {
      "epoch": 0.024486825124984127,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999319488931986,
      "loss": 2.3521,
      "step": 3604
    },
    {
      "epoch": 0.024493619471578187,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999318927697501,
      "loss": 2.5101,
      "step": 3605
    },
    {
      "epoch": 0.024500413818172244,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000999318366231837,
      "loss": 2.5077,
      "step": 3606
    },
    {
      "epoch": 0.0245072081647663,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009993178045349942,
      "loss": 2.4512,
      "step": 3607
    },
    {
      "epoch": 0.02451400251136036,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009993172426069731,
      "loss": 2.4911,
      "step": 3608
    },
    {
      "epoch": 0.024520796857954417,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009993166804477737,
      "loss": 2.5177,
      "step": 3609
    },
    {
      "epoch": 0.024527591204548477,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009993161180573966,
      "loss": 2.2744,
      "step": 3610
    },
    {
      "epoch": 0.024534385551142533,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009993155554358418,
      "loss": 2.3152,
      "step": 3611
    },
    {
      "epoch": 0.02454117989773659,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993149925831095,
      "loss": 2.3693,
      "step": 3612
    },
    {
      "epoch": 0.02454797424433065,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009993144294992,
      "loss": 2.5867,
      "step": 3613
    },
    {
      "epoch": 0.024554768590924706,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999313866184114,
      "loss": 2.4963,
      "step": 3614
    },
    {
      "epoch": 0.024561562937518763,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009993133026378511,
      "loss": 2.5707,
      "step": 3615
    },
    {
      "epoch": 0.024568357284112823,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999312738860412,
      "loss": 2.4253,
      "step": 3616
    },
    {
      "epoch": 0.02457515163070688,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999312174851797,
      "loss": 2.4472,
      "step": 3617
    },
    {
      "epoch": 0.02458194597730094,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009993116106120057,
      "loss": 2.3917,
      "step": 3618
    },
    {
      "epoch": 0.024588740323894995,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999311046141039,
      "loss": 2.5251,
      "step": 3619
    },
    {
      "epoch": 0.024595534670489052,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009993104814388974,
      "loss": 2.5061,
      "step": 3620
    },
    {
      "epoch": 0.024602329017083112,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009993099165055803,
      "loss": 2.4397,
      "step": 3621
    },
    {
      "epoch": 0.02460912336367717,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009993093513410887,
      "loss": 2.261,
      "step": 3622
    },
    {
      "epoch": 0.024615917710271225,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009993087859454226,
      "loss": 2.3991,
      "step": 3623
    },
    {
      "epoch": 0.024622712056865285,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009993082203185821,
      "loss": 2.4202,
      "step": 3624
    },
    {
      "epoch": 0.02462950640345934,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009993076544605676,
      "loss": 2.5286,
      "step": 3625
    },
    {
      "epoch": 0.024636300750053398,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009993070883713797,
      "loss": 2.2925,
      "step": 3626
    },
    {
      "epoch": 0.024643095096647458,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999306522051018,
      "loss": 2.36,
      "step": 3627
    },
    {
      "epoch": 0.024649889443241514,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009993059554994834,
      "loss": 2.59,
      "step": 3628
    },
    {
      "epoch": 0.024656683789835574,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009993053887167756,
      "loss": 2.3654,
      "step": 3629
    },
    {
      "epoch": 0.02466347813642963,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009993048217028953,
      "loss": 2.326,
      "step": 3630
    },
    {
      "epoch": 0.024670272483023687,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993042544578425,
      "loss": 2.3455,
      "step": 3631
    },
    {
      "epoch": 0.024677066829617747,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009993036869816176,
      "loss": 2.4848,
      "step": 3632
    },
    {
      "epoch": 0.024683861176211803,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009993031192742207,
      "loss": 2.388,
      "step": 3633
    },
    {
      "epoch": 0.02469065552280586,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009993025513356524,
      "loss": 2.5178,
      "step": 3634
    },
    {
      "epoch": 0.02469744986939992,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009993019831659126,
      "loss": 2.598,
      "step": 3635
    },
    {
      "epoch": 0.024704244215993976,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009993014147650016,
      "loss": 2.3988,
      "step": 3636
    },
    {
      "epoch": 0.024711038562588033,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099930084613292,
      "loss": 2.3714,
      "step": 3637
    },
    {
      "epoch": 0.024717832909182093,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009993002772696678,
      "loss": 2.4754,
      "step": 3638
    },
    {
      "epoch": 0.02472462725577615,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992997081752452,
      "loss": 2.4496,
      "step": 3639
    },
    {
      "epoch": 0.02473142160237021,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992991388496524,
      "loss": 2.4866,
      "step": 3640
    },
    {
      "epoch": 0.024738215948964266,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009992985692928902,
      "loss": 2.446,
      "step": 3641
    },
    {
      "epoch": 0.024745010295558322,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009992979995049583,
      "loss": 2.4564,
      "step": 3642
    },
    {
      "epoch": 0.024751804642152382,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999297429485857,
      "loss": 2.3741,
      "step": 3643
    },
    {
      "epoch": 0.02475859898874644,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999296859235587,
      "loss": 2.4371,
      "step": 3644
    },
    {
      "epoch": 0.024765393335340495,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999296288754148,
      "loss": 2.1984,
      "step": 3645
    },
    {
      "epoch": 0.024772187681934555,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009992957180415408,
      "loss": 2.6462,
      "step": 3646
    },
    {
      "epoch": 0.02477898202852861,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999295147097765,
      "loss": 2.3342,
      "step": 3647
    },
    {
      "epoch": 0.02478577637512267,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009992945759228218,
      "loss": 2.3111,
      "step": 3648
    },
    {
      "epoch": 0.024792570721716728,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992940045167108,
      "loss": 2.3127,
      "step": 3649
    },
    {
      "epoch": 0.024799365068310784,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999293432879432,
      "loss": 2.2871,
      "step": 3650
    },
    {
      "epoch": 0.024806159414904844,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992928610109864,
      "loss": 2.4497,
      "step": 3651
    },
    {
      "epoch": 0.0248129537614989,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999292288911374,
      "loss": 2.4373,
      "step": 3652
    },
    {
      "epoch": 0.024819748108092957,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009992917165805948,
      "loss": 2.4577,
      "step": 3653
    },
    {
      "epoch": 0.024826542454687017,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009992911440186495,
      "loss": 2.275,
      "step": 3654
    },
    {
      "epoch": 0.024833336801281074,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992905712255377,
      "loss": 2.4712,
      "step": 3655
    },
    {
      "epoch": 0.02484013114787513,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992899982012604,
      "loss": 2.4729,
      "step": 3656
    },
    {
      "epoch": 0.02484692549446919,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992894249458175,
      "loss": 2.4642,
      "step": 3657
    },
    {
      "epoch": 0.024853719841063247,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009992888514592094,
      "loss": 2.401,
      "step": 3658
    },
    {
      "epoch": 0.024860514187657307,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009992882777414363,
      "loss": 2.4176,
      "step": 3659
    },
    {
      "epoch": 0.024867308534251363,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009992877037924984,
      "loss": 2.4009,
      "step": 3660
    },
    {
      "epoch": 0.02487410288084542,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009992871296123959,
      "loss": 2.2949,
      "step": 3661
    },
    {
      "epoch": 0.02488089722743948,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009992865552011293,
      "loss": 2.3736,
      "step": 3662
    },
    {
      "epoch": 0.024887691574033536,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009992859805586986,
      "loss": 2.3807,
      "step": 3663
    },
    {
      "epoch": 0.024894485920627592,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009992854056851044,
      "loss": 2.4353,
      "step": 3664
    },
    {
      "epoch": 0.024901280267221652,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009992848305803467,
      "loss": 2.4272,
      "step": 3665
    },
    {
      "epoch": 0.02490807461381571,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992842552444259,
      "loss": 2.4262,
      "step": 3666
    },
    {
      "epoch": 0.024914868960409765,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000999283679677342,
      "loss": 2.4894,
      "step": 3667
    },
    {
      "epoch": 0.024921663307003825,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992831038790958,
      "loss": 2.3237,
      "step": 3668
    },
    {
      "epoch": 0.024928457653597882,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009992825278496871,
      "loss": 2.3198,
      "step": 3669
    },
    {
      "epoch": 0.02493525200019194,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009992819515891162,
      "loss": 2.5967,
      "step": 3670
    },
    {
      "epoch": 0.024942046346785998,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009992813750973838,
      "loss": 2.3865,
      "step": 3671
    },
    {
      "epoch": 0.024948840693380055,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009992807983744895,
      "loss": 2.4127,
      "step": 3672
    },
    {
      "epoch": 0.024955635039974115,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009992802214204342,
      "loss": 2.339,
      "step": 3673
    },
    {
      "epoch": 0.02496242938656817,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009992796442352177,
      "loss": 2.4177,
      "step": 3674
    },
    {
      "epoch": 0.024969223733162228,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009992790668188404,
      "loss": 2.5013,
      "step": 3675
    },
    {
      "epoch": 0.024976018079756288,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992784891713028,
      "loss": 2.3777,
      "step": 3676
    },
    {
      "epoch": 0.024982812426350344,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992779112926048,
      "loss": 2.4925,
      "step": 3677
    },
    {
      "epoch": 0.024989606772944404,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999277333182747,
      "loss": 2.4211,
      "step": 3678
    },
    {
      "epoch": 0.02499640111953846,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009992767548417294,
      "loss": 2.2178,
      "step": 3679
    },
    {
      "epoch": 0.025003195466132517,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009992761762695525,
      "loss": 2.3934,
      "step": 3680
    },
    {
      "epoch": 0.025009989812726577,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009992755974662162,
      "loss": 2.2983,
      "step": 3681
    },
    {
      "epoch": 0.025016784159320633,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009992750184317214,
      "loss": 2.3221,
      "step": 3682
    },
    {
      "epoch": 0.02502357850591469,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009992744391660677,
      "loss": 2.4716,
      "step": 3683
    },
    {
      "epoch": 0.02503037285250875,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009992738596692559,
      "loss": 2.449,
      "step": 3684
    },
    {
      "epoch": 0.025037167199102806,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009992732799412858,
      "loss": 2.6452,
      "step": 3685
    },
    {
      "epoch": 0.025043961545696863,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000999272699982158,
      "loss": 2.342,
      "step": 3686
    },
    {
      "epoch": 0.025050755892290923,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992721197918727,
      "loss": 2.3317,
      "step": 3687
    },
    {
      "epoch": 0.02505755023888498,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00099927153937043,
      "loss": 2.4437,
      "step": 3688
    },
    {
      "epoch": 0.02506434458547904,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009992709587178304,
      "loss": 2.3386,
      "step": 3689
    },
    {
      "epoch": 0.025071138932073096,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009992703778340742,
      "loss": 2.4306,
      "step": 3690
    },
    {
      "epoch": 0.025077933278667152,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009992697967191612,
      "loss": 2.392,
      "step": 3691
    },
    {
      "epoch": 0.025084727625261212,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009992692153730923,
      "loss": 2.3497,
      "step": 3692
    },
    {
      "epoch": 0.02509152197185527,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009992686337958673,
      "loss": 2.3489,
      "step": 3693
    },
    {
      "epoch": 0.025098316318449325,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009992680519874868,
      "loss": 2.4656,
      "step": 3694
    },
    {
      "epoch": 0.025105110665043385,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009992674699479509,
      "loss": 2.3825,
      "step": 3695
    },
    {
      "epoch": 0.02511190501163744,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009992668876772598,
      "loss": 2.3982,
      "step": 3696
    },
    {
      "epoch": 0.025118699358231498,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992663051754138,
      "loss": 2.537,
      "step": 3697
    },
    {
      "epoch": 0.025125493704825558,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009992657224424132,
      "loss": 2.4263,
      "step": 3698
    },
    {
      "epoch": 0.025132288051419614,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009992651394782586,
      "loss": 2.4101,
      "step": 3699
    },
    {
      "epoch": 0.025139082398013674,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009992645562829498,
      "loss": 2.3979,
      "step": 3700
    },
    {
      "epoch": 0.02514587674460773,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009992639728564872,
      "loss": 2.4071,
      "step": 3701
    },
    {
      "epoch": 0.025152671091201787,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999263389198871,
      "loss": 2.5756,
      "step": 3702
    },
    {
      "epoch": 0.025159465437795847,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009992628053101017,
      "loss": 2.4365,
      "step": 3703
    },
    {
      "epoch": 0.025166259784389904,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009992622211901793,
      "loss": 2.4557,
      "step": 3704
    },
    {
      "epoch": 0.02517305413098396,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009992616368391044,
      "loss": 2.5375,
      "step": 3705
    },
    {
      "epoch": 0.02517984847757802,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999261052256877,
      "loss": 2.3409,
      "step": 3706
    },
    {
      "epoch": 0.025186642824172076,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992604674434976,
      "loss": 2.3652,
      "step": 3707
    },
    {
      "epoch": 0.025193437170766136,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009992598823989663,
      "loss": 2.4748,
      "step": 3708
    },
    {
      "epoch": 0.025200231517360193,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009992592971232832,
      "loss": 2.3395,
      "step": 3709
    },
    {
      "epoch": 0.02520702586395425,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999258711616449,
      "loss": 2.448,
      "step": 3710
    },
    {
      "epoch": 0.02521382021054831,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992581258784635,
      "loss": 2.3758,
      "step": 3711
    },
    {
      "epoch": 0.025220614557142366,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009992575399093272,
      "loss": 2.4032,
      "step": 3712
    },
    {
      "epoch": 0.025227408903736422,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009992569537090406,
      "loss": 2.4639,
      "step": 3713
    },
    {
      "epoch": 0.025234203250330482,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009992563672776037,
      "loss": 2.392,
      "step": 3714
    },
    {
      "epoch": 0.02524099759692454,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992557806150167,
      "loss": 2.4127,
      "step": 3715
    },
    {
      "epoch": 0.025247791943518595,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009992551937212802,
      "loss": 2.3858,
      "step": 3716
    },
    {
      "epoch": 0.025254586290112655,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992546065963942,
      "loss": 2.3309,
      "step": 3717
    },
    {
      "epoch": 0.02526138063670671,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999254019240359,
      "loss": 2.4048,
      "step": 3718
    },
    {
      "epoch": 0.02526817498330077,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992534316531749,
      "loss": 2.3866,
      "step": 3719
    },
    {
      "epoch": 0.025274969329894828,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999252843834842,
      "loss": 2.3178,
      "step": 3720
    },
    {
      "epoch": 0.025281763676488885,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992522557853612,
      "loss": 2.39,
      "step": 3721
    },
    {
      "epoch": 0.025288558023082944,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992516675047319,
      "loss": 2.551,
      "step": 3722
    },
    {
      "epoch": 0.025295352369677,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000999251078992955,
      "loss": 2.3245,
      "step": 3723
    },
    {
      "epoch": 0.025302146716271057,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992504902500306,
      "loss": 2.3661,
      "step": 3724
    },
    {
      "epoch": 0.025308941062865117,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999249901275959,
      "loss": 2.4152,
      "step": 3725
    },
    {
      "epoch": 0.025315735409459174,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009992493120707401,
      "loss": 2.4622,
      "step": 3726
    },
    {
      "epoch": 0.02532252975605323,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009992487226343748,
      "loss": 2.4331,
      "step": 3727
    },
    {
      "epoch": 0.02532932410264729,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999248132966863,
      "loss": 2.4051,
      "step": 3728
    },
    {
      "epoch": 0.025336118449241347,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009992475430682052,
      "loss": 2.3853,
      "step": 3729
    },
    {
      "epoch": 0.025342912795835407,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992469529384012,
      "loss": 2.197,
      "step": 3730
    },
    {
      "epoch": 0.025349707142429463,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009992463625774516,
      "loss": 2.4664,
      "step": 3731
    },
    {
      "epoch": 0.02535650148902352,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009992457719853569,
      "loss": 2.3677,
      "step": 3732
    },
    {
      "epoch": 0.02536329583561758,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000999245181162117,
      "loss": 2.3703,
      "step": 3733
    },
    {
      "epoch": 0.025370090182211636,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992445901077323,
      "loss": 2.1607,
      "step": 3734
    },
    {
      "epoch": 0.025376884528805693,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000999243998822203,
      "loss": 2.2942,
      "step": 3735
    },
    {
      "epoch": 0.025383678875399752,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009992434073055297,
      "loss": 2.33,
      "step": 3736
    },
    {
      "epoch": 0.02539047322199381,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992428155577123,
      "loss": 2.4325,
      "step": 3737
    },
    {
      "epoch": 0.02539726756858787,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999242223578751,
      "loss": 2.1761,
      "step": 3738
    },
    {
      "epoch": 0.025404061915181925,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009992416313686465,
      "loss": 2.252,
      "step": 3739
    },
    {
      "epoch": 0.025410856261775982,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009992410389273987,
      "loss": 2.4307,
      "step": 3740
    },
    {
      "epoch": 0.025417650608370042,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009992404462550082,
      "loss": 2.4302,
      "step": 3741
    },
    {
      "epoch": 0.0254244449549641,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999239853351475,
      "loss": 2.392,
      "step": 3742
    },
    {
      "epoch": 0.025431239301558155,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009992392602167994,
      "loss": 2.3795,
      "step": 3743
    },
    {
      "epoch": 0.025438033648152215,
      "grad_norm": 0.140625,
      "learning_rate": 0.000999238666850982,
      "loss": 2.3977,
      "step": 3744
    },
    {
      "epoch": 0.02544482799474627,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009992380732540226,
      "loss": 2.4269,
      "step": 3745
    },
    {
      "epoch": 0.025451622341340328,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009992374794259217,
      "loss": 2.2891,
      "step": 3746
    },
    {
      "epoch": 0.025458416687934388,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992368853666796,
      "loss": 2.4185,
      "step": 3747
    },
    {
      "epoch": 0.025465211034528444,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992362910762964,
      "loss": 2.4857,
      "step": 3748
    },
    {
      "epoch": 0.025472005381122504,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009992356965547729,
      "loss": 2.2533,
      "step": 3749
    },
    {
      "epoch": 0.02547879972771656,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009992351018021086,
      "loss": 2.3499,
      "step": 3750
    },
    {
      "epoch": 0.025485594074310617,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992345068183044,
      "loss": 2.3708,
      "step": 3751
    },
    {
      "epoch": 0.025492388420904677,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992339116033603,
      "loss": 2.4144,
      "step": 3752
    },
    {
      "epoch": 0.025499182767498733,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009992333161572766,
      "loss": 2.5588,
      "step": 3753
    },
    {
      "epoch": 0.02550597711409279,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992327204800537,
      "loss": 2.3578,
      "step": 3754
    },
    {
      "epoch": 0.02551277146068685,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009992321245716916,
      "loss": 2.4859,
      "step": 3755
    },
    {
      "epoch": 0.025519565807280906,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000999231528432191,
      "loss": 2.5135,
      "step": 3756
    },
    {
      "epoch": 0.025526360153874963,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992309320615517,
      "loss": 2.5097,
      "step": 3757
    },
    {
      "epoch": 0.025533154500469023,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009992303354597744,
      "loss": 2.3989,
      "step": 3758
    },
    {
      "epoch": 0.02553994884706308,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009992297386268588,
      "loss": 2.6617,
      "step": 3759
    },
    {
      "epoch": 0.02554674319365714,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009992291415628059,
      "loss": 2.6091,
      "step": 3760
    },
    {
      "epoch": 0.025553537540251196,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009992285442676154,
      "loss": 2.4299,
      "step": 3761
    },
    {
      "epoch": 0.025560331886845252,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999227946741288,
      "loss": 2.427,
      "step": 3762
    },
    {
      "epoch": 0.025567126233439312,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009992273489838237,
      "loss": 2.4867,
      "step": 3763
    },
    {
      "epoch": 0.02557392058003337,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009992267509952228,
      "loss": 2.3665,
      "step": 3764
    },
    {
      "epoch": 0.025580714926627425,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009992261527754858,
      "loss": 2.4334,
      "step": 3765
    },
    {
      "epoch": 0.025587509273221485,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009992255543246128,
      "loss": 2.4258,
      "step": 3766
    },
    {
      "epoch": 0.02559430361981554,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000999224955642604,
      "loss": 2.4753,
      "step": 3767
    },
    {
      "epoch": 0.0256010979664096,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009992243567294597,
      "loss": 2.3527,
      "step": 3768
    },
    {
      "epoch": 0.025607892313003658,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009992237575851804,
      "loss": 2.3767,
      "step": 3769
    },
    {
      "epoch": 0.025614686659597714,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999223158209766,
      "loss": 2.3533,
      "step": 3770
    },
    {
      "epoch": 0.025621481006191774,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009992225586032174,
      "loss": 2.3929,
      "step": 3771
    },
    {
      "epoch": 0.02562827535278583,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009992219587655342,
      "loss": 2.4257,
      "step": 3772
    },
    {
      "epoch": 0.025635069699379887,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992213586967169,
      "loss": 2.3839,
      "step": 3773
    },
    {
      "epoch": 0.025641864045973947,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999220758396766,
      "loss": 2.4369,
      "step": 3774
    },
    {
      "epoch": 0.025648658392568004,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009992201578656816,
      "loss": 2.4366,
      "step": 3775
    },
    {
      "epoch": 0.02565545273916206,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009992195571034638,
      "loss": 2.4217,
      "step": 3776
    },
    {
      "epoch": 0.02566224708575612,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992189561101132,
      "loss": 2.4166,
      "step": 3777
    },
    {
      "epoch": 0.025669041432350177,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009992183548856298,
      "loss": 2.3494,
      "step": 3778
    },
    {
      "epoch": 0.025675835778944237,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009992177534300145,
      "loss": 2.2236,
      "step": 3779
    },
    {
      "epoch": 0.025682630125538293,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009992171517432668,
      "loss": 2.4525,
      "step": 3780
    },
    {
      "epoch": 0.02568942447213235,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000999216549825387,
      "loss": 2.3527,
      "step": 3781
    },
    {
      "epoch": 0.02569621881872641,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999215947676376,
      "loss": 2.3402,
      "step": 3782
    },
    {
      "epoch": 0.025703013165320466,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009992153452962338,
      "loss": 2.3308,
      "step": 3783
    },
    {
      "epoch": 0.025709807511914522,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992147426849604,
      "loss": 2.4379,
      "step": 3784
    },
    {
      "epoch": 0.025716601858508582,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009992141398425564,
      "loss": 2.2634,
      "step": 3785
    },
    {
      "epoch": 0.02572339620510264,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000999213536769022,
      "loss": 2.289,
      "step": 3786
    },
    {
      "epoch": 0.025730190551696695,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009992129334643574,
      "loss": 2.3721,
      "step": 3787
    },
    {
      "epoch": 0.025736984898290755,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000999212329928563,
      "loss": 2.493,
      "step": 3788
    },
    {
      "epoch": 0.02574377924488481,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009992117261616387,
      "loss": 2.2969,
      "step": 3789
    },
    {
      "epoch": 0.02575057359147887,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009992111221635855,
      "loss": 2.3694,
      "step": 3790
    },
    {
      "epoch": 0.025757367938072928,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000999210517934403,
      "loss": 2.4331,
      "step": 3791
    },
    {
      "epoch": 0.025764162284666985,
      "grad_norm": 0.140625,
      "learning_rate": 0.000999209913474092,
      "loss": 2.3493,
      "step": 3792
    },
    {
      "epoch": 0.025770956631261045,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009992093087826525,
      "loss": 2.2862,
      "step": 3793
    },
    {
      "epoch": 0.0257777509778551,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009992087038600845,
      "loss": 2.476,
      "step": 3794
    },
    {
      "epoch": 0.025784545324449158,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000999208098706389,
      "loss": 2.5016,
      "step": 3795
    },
    {
      "epoch": 0.025791339671043217,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009992074933215657,
      "loss": 2.4056,
      "step": 3796
    },
    {
      "epoch": 0.025798134017637274,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999206887705615,
      "loss": 2.2131,
      "step": 3797
    },
    {
      "epoch": 0.025804928364231334,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009992062818585373,
      "loss": 2.3235,
      "step": 3798
    },
    {
      "epoch": 0.02581172271082539,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009992056757803327,
      "loss": 2.4447,
      "step": 3799
    },
    {
      "epoch": 0.025818517057419447,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009992050694710018,
      "loss": 2.4402,
      "step": 3800
    },
    {
      "epoch": 0.025825311404013507,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009992044629305447,
      "loss": 2.2882,
      "step": 3801
    },
    {
      "epoch": 0.025832105750607563,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009992038561589614,
      "loss": 2.6744,
      "step": 3802
    },
    {
      "epoch": 0.02583890009720162,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009992032491562524,
      "loss": 2.3576,
      "step": 3803
    },
    {
      "epoch": 0.02584569444379568,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009992026419224182,
      "loss": 2.4663,
      "step": 3804
    },
    {
      "epoch": 0.025852488790389736,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009992020344574588,
      "loss": 2.3358,
      "step": 3805
    },
    {
      "epoch": 0.025859283136983793,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009992014267613748,
      "loss": 2.3615,
      "step": 3806
    },
    {
      "epoch": 0.025866077483577853,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000999200818834166,
      "loss": 2.3952,
      "step": 3807
    },
    {
      "epoch": 0.02587287183017191,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009992002106758331,
      "loss": 2.4999,
      "step": 3808
    },
    {
      "epoch": 0.02587966617676597,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991996022863762,
      "loss": 2.378,
      "step": 3809
    },
    {
      "epoch": 0.025886460523360025,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009991989936657957,
      "loss": 2.3857,
      "step": 3810
    },
    {
      "epoch": 0.025893254869954082,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991983848140916,
      "loss": 2.3862,
      "step": 3811
    },
    {
      "epoch": 0.025900049216548142,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009991977757312644,
      "loss": 2.2939,
      "step": 3812
    },
    {
      "epoch": 0.0259068435631422,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009991971664173145,
      "loss": 2.2595,
      "step": 3813
    },
    {
      "epoch": 0.025913637909736255,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999196556872242,
      "loss": 2.2895,
      "step": 3814
    },
    {
      "epoch": 0.025920432256330315,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991959470960472,
      "loss": 2.3751,
      "step": 3815
    },
    {
      "epoch": 0.02592722660292437,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009991953370887304,
      "loss": 2.4019,
      "step": 3816
    },
    {
      "epoch": 0.025934020949518428,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009991947268502919,
      "loss": 2.386,
      "step": 3817
    },
    {
      "epoch": 0.025940815296112488,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009991941163807318,
      "loss": 2.4782,
      "step": 3818
    },
    {
      "epoch": 0.025947609642706544,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009991935056800506,
      "loss": 2.503,
      "step": 3819
    },
    {
      "epoch": 0.025954403989300604,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009991928947482487,
      "loss": 2.3406,
      "step": 3820
    },
    {
      "epoch": 0.02596119833589466,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009991922835853259,
      "loss": 2.4332,
      "step": 3821
    },
    {
      "epoch": 0.025967992682488717,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999191672191283,
      "loss": 2.3708,
      "step": 3822
    },
    {
      "epoch": 0.025974787029082777,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009991910605661202,
      "loss": 2.4045,
      "step": 3823
    },
    {
      "epoch": 0.025981581375676834,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009991904487098376,
      "loss": 2.3267,
      "step": 3824
    },
    {
      "epoch": 0.02598837572227089,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009991898366224355,
      "loss": 2.4521,
      "step": 3825
    },
    {
      "epoch": 0.02599517006886495,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009991892243039141,
      "loss": 2.4618,
      "step": 3826
    },
    {
      "epoch": 0.026001964415459006,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009991886117542738,
      "loss": 2.4812,
      "step": 3827
    },
    {
      "epoch": 0.026008758762053066,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009991879989735151,
      "loss": 2.3222,
      "step": 3828
    },
    {
      "epoch": 0.026015553108647123,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999187385961638,
      "loss": 2.3297,
      "step": 3829
    },
    {
      "epoch": 0.02602234745524118,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999186772718643,
      "loss": 2.3789,
      "step": 3830
    },
    {
      "epoch": 0.02602914180183524,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009991861592445299,
      "loss": 2.4983,
      "step": 3831
    },
    {
      "epoch": 0.026035936148429296,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009991855455392996,
      "loss": 2.4607,
      "step": 3832
    },
    {
      "epoch": 0.026042730495023352,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000999184931602952,
      "loss": 2.4206,
      "step": 3833
    },
    {
      "epoch": 0.026049524841617412,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991843174354876,
      "loss": 2.4545,
      "step": 3834
    },
    {
      "epoch": 0.02605631918821147,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009991837030369067,
      "loss": 2.4922,
      "step": 3835
    },
    {
      "epoch": 0.026063113534805525,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991830884072093,
      "loss": 2.2677,
      "step": 3836
    },
    {
      "epoch": 0.026069907881399585,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009991824735463956,
      "loss": 2.4838,
      "step": 3837
    },
    {
      "epoch": 0.02607670222799364,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009991818584544666,
      "loss": 2.4209,
      "step": 3838
    },
    {
      "epoch": 0.0260834965745877,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999181243131422,
      "loss": 2.3836,
      "step": 3839
    },
    {
      "epoch": 0.026090290921181758,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999180627577262,
      "loss": 2.4728,
      "step": 3840
    },
    {
      "epoch": 0.026097085267775814,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009991800117919872,
      "loss": 2.4827,
      "step": 3841
    },
    {
      "epoch": 0.026103879614369874,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009991793957755978,
      "loss": 2.4003,
      "step": 3842
    },
    {
      "epoch": 0.02611067396096393,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999178779528094,
      "loss": 2.3634,
      "step": 3843
    },
    {
      "epoch": 0.026117468307557987,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009991781630494764,
      "loss": 2.3893,
      "step": 3844
    },
    {
      "epoch": 0.026124262654152047,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009991775463397447,
      "loss": 2.2592,
      "step": 3845
    },
    {
      "epoch": 0.026131057000746104,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009991769293988997,
      "loss": 2.3365,
      "step": 3846
    },
    {
      "epoch": 0.02613785134734016,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009991763122269414,
      "loss": 2.3099,
      "step": 3847
    },
    {
      "epoch": 0.02614464569393422,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009991756948238702,
      "loss": 2.3124,
      "step": 3848
    },
    {
      "epoch": 0.026151440040528277,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009991750771896867,
      "loss": 2.4332,
      "step": 3849
    },
    {
      "epoch": 0.026158234387122337,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009991744593243904,
      "loss": 2.3133,
      "step": 3850
    },
    {
      "epoch": 0.026165028733716393,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009991738412279824,
      "loss": 2.4002,
      "step": 3851
    },
    {
      "epoch": 0.02617182308031045,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009991732229004623,
      "loss": 2.5349,
      "step": 3852
    },
    {
      "epoch": 0.02617861742690451,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999172604341831,
      "loss": 2.3897,
      "step": 3853
    },
    {
      "epoch": 0.026185411773498566,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009991719855520884,
      "loss": 2.3824,
      "step": 3854
    },
    {
      "epoch": 0.026192206120092622,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991713665312347,
      "loss": 2.2649,
      "step": 3855
    },
    {
      "epoch": 0.026199000466686682,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009991707472792708,
      "loss": 2.3053,
      "step": 3856
    },
    {
      "epoch": 0.02620579481328074,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009991701277961961,
      "loss": 2.4064,
      "step": 3857
    },
    {
      "epoch": 0.0262125891598748,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009991695080820118,
      "loss": 2.3747,
      "step": 3858
    },
    {
      "epoch": 0.026219383506468855,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009991688881367175,
      "loss": 2.5981,
      "step": 3859
    },
    {
      "epoch": 0.026226177853062912,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009991682679603138,
      "loss": 2.343,
      "step": 3860
    },
    {
      "epoch": 0.026232972199656972,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999167647552801,
      "loss": 2.4668,
      "step": 3861
    },
    {
      "epoch": 0.026239766546251028,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000999167026914179,
      "loss": 2.2764,
      "step": 3862
    },
    {
      "epoch": 0.026246560892845085,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009991664060444486,
      "loss": 2.4709,
      "step": 3863
    },
    {
      "epoch": 0.026253355239439145,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009991657849436097,
      "loss": 2.4199,
      "step": 3864
    },
    {
      "epoch": 0.0262601495860332,
      "grad_norm": 0.171875,
      "learning_rate": 0.000999165163611663,
      "loss": 2.4872,
      "step": 3865
    },
    {
      "epoch": 0.026266943932627258,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009991645420486083,
      "loss": 2.5156,
      "step": 3866
    },
    {
      "epoch": 0.026273738279221318,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009991639202544464,
      "loss": 2.3007,
      "step": 3867
    },
    {
      "epoch": 0.026280532625815374,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009991632982291772,
      "loss": 2.342,
      "step": 3868
    },
    {
      "epoch": 0.026287326972409434,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009991626759728012,
      "loss": 2.4276,
      "step": 3869
    },
    {
      "epoch": 0.02629412131900349,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991620534853184,
      "loss": 2.3603,
      "step": 3870
    },
    {
      "epoch": 0.026300915665597547,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009991614307667295,
      "loss": 2.3286,
      "step": 3871
    },
    {
      "epoch": 0.026307710012191607,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009991608078170346,
      "loss": 2.4672,
      "step": 3872
    },
    {
      "epoch": 0.026314504358785663,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009991601846362338,
      "loss": 2.3729,
      "step": 3873
    },
    {
      "epoch": 0.02632129870537972,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009991595612243278,
      "loss": 2.5924,
      "step": 3874
    },
    {
      "epoch": 0.02632809305197378,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009991589375813164,
      "loss": 2.3836,
      "step": 3875
    },
    {
      "epoch": 0.026334887398567836,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009991583137072004,
      "loss": 2.3228,
      "step": 3876
    },
    {
      "epoch": 0.026341681745161893,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009991576896019796,
      "loss": 2.3568,
      "step": 3877
    },
    {
      "epoch": 0.026348476091755953,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009991570652656545,
      "loss": 2.4845,
      "step": 3878
    },
    {
      "epoch": 0.02635527043835001,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009991564406982255,
      "loss": 2.3911,
      "step": 3879
    },
    {
      "epoch": 0.02636206478494407,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009991558158996927,
      "loss": 2.3017,
      "step": 3880
    },
    {
      "epoch": 0.026368859131538126,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009991551908700566,
      "loss": 2.4116,
      "step": 3881
    },
    {
      "epoch": 0.026375653478132182,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009991545656093173,
      "loss": 2.3249,
      "step": 3882
    },
    {
      "epoch": 0.026382447824726242,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009991539401174752,
      "loss": 2.5624,
      "step": 3883
    },
    {
      "epoch": 0.0263892421713203,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009991533143945304,
      "loss": 2.4907,
      "step": 3884
    },
    {
      "epoch": 0.026396036517914355,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009991526884404836,
      "loss": 2.4252,
      "step": 3885
    },
    {
      "epoch": 0.026402830864508415,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009991520622553345,
      "loss": 2.3555,
      "step": 3886
    },
    {
      "epoch": 0.02640962521110247,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999151435839084,
      "loss": 2.3586,
      "step": 3887
    },
    {
      "epoch": 0.02641641955769653,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999150809191732,
      "loss": 2.3663,
      "step": 3888
    },
    {
      "epoch": 0.026423213904290588,
      "grad_norm": 0.15625,
      "learning_rate": 0.000999150182313279,
      "loss": 2.5799,
      "step": 3889
    },
    {
      "epoch": 0.026430008250884644,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000999149555203725,
      "loss": 2.404,
      "step": 3890
    },
    {
      "epoch": 0.026436802597478704,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009991489278630705,
      "loss": 2.4543,
      "step": 3891
    },
    {
      "epoch": 0.02644359694407276,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000999148300291316,
      "loss": 2.3569,
      "step": 3892
    },
    {
      "epoch": 0.026450391290666817,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009991476724884613,
      "loss": 2.2223,
      "step": 3893
    },
    {
      "epoch": 0.026457185637260877,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000999147044454507,
      "loss": 2.3854,
      "step": 3894
    },
    {
      "epoch": 0.026463979983854934,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009991464161894534,
      "loss": 2.4354,
      "step": 3895
    },
    {
      "epoch": 0.02647077433044899,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009991457876933007,
      "loss": 2.4229,
      "step": 3896
    },
    {
      "epoch": 0.02647756867704305,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009991451589660492,
      "loss": 2.3335,
      "step": 3897
    },
    {
      "epoch": 0.026484363023637107,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009991445300076994,
      "loss": 2.5508,
      "step": 3898
    },
    {
      "epoch": 0.026491157370231166,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009991439008182512,
      "loss": 2.3221,
      "step": 3899
    },
    {
      "epoch": 0.026497951716825223,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009991432713977052,
      "loss": 2.3824,
      "step": 3900
    },
    {
      "epoch": 0.02650474606341928,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009991426417460614,
      "loss": 2.3779,
      "step": 3901
    },
    {
      "epoch": 0.02651154041001334,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009991420118633206,
      "loss": 2.3518,
      "step": 3902
    },
    {
      "epoch": 0.026518334756607396,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009991413817494826,
      "loss": 2.44,
      "step": 3903
    },
    {
      "epoch": 0.026525129103201452,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009991407514045477,
      "loss": 2.3534,
      "step": 3904
    },
    {
      "epoch": 0.026531923449795512,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009991401208285164,
      "loss": 2.2819,
      "step": 3905
    },
    {
      "epoch": 0.02653871779638957,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009991394900213894,
      "loss": 2.621,
      "step": 3906
    },
    {
      "epoch": 0.026545512142983625,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000999138858983166,
      "loss": 2.5591,
      "step": 3907
    },
    {
      "epoch": 0.026552306489577685,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009991382277138472,
      "loss": 2.5049,
      "step": 3908
    },
    {
      "epoch": 0.02655910083617174,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009991375962134332,
      "loss": 2.346,
      "step": 3909
    },
    {
      "epoch": 0.0265658951827658,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000999136964481924,
      "loss": 2.3856,
      "step": 3910
    },
    {
      "epoch": 0.026572689529359858,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009991363325193203,
      "loss": 2.32,
      "step": 3911
    },
    {
      "epoch": 0.026579483875953915,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009991357003256223,
      "loss": 2.3749,
      "step": 3912
    },
    {
      "epoch": 0.026586278222547974,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009991350679008299,
      "loss": 2.3135,
      "step": 3913
    },
    {
      "epoch": 0.02659307256914203,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009991344352449439,
      "loss": 2.356,
      "step": 3914
    },
    {
      "epoch": 0.026599866915736087,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009991338023579643,
      "loss": 2.3895,
      "step": 3915
    },
    {
      "epoch": 0.026606661262330147,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009991331692398916,
      "loss": 2.3401,
      "step": 3916
    },
    {
      "epoch": 0.026613455608924204,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999132535890726,
      "loss": 2.4406,
      "step": 3917
    },
    {
      "epoch": 0.026620249955518264,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009991319023104676,
      "loss": 2.4128,
      "step": 3918
    },
    {
      "epoch": 0.02662704430211232,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991312684991167,
      "loss": 2.4317,
      "step": 3919
    },
    {
      "epoch": 0.026633838648706377,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999130634456674,
      "loss": 2.3351,
      "step": 3920
    },
    {
      "epoch": 0.026640632995300437,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009991300001831395,
      "loss": 2.3631,
      "step": 3921
    },
    {
      "epoch": 0.026647427341894493,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009991293656785134,
      "loss": 2.5036,
      "step": 3922
    },
    {
      "epoch": 0.02665422168848855,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009991287309427963,
      "loss": 2.4779,
      "step": 3923
    },
    {
      "epoch": 0.02666101603508261,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009991280959759882,
      "loss": 2.5733,
      "step": 3924
    },
    {
      "epoch": 0.026667810381676666,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009991274607780896,
      "loss": 2.5398,
      "step": 3925
    },
    {
      "epoch": 0.026674604728270723,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009991268253491006,
      "loss": 2.3972,
      "step": 3926
    },
    {
      "epoch": 0.026681399074864783,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009991261896890218,
      "loss": 2.3179,
      "step": 3927
    },
    {
      "epoch": 0.02668819342145884,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000999125553797853,
      "loss": 2.3389,
      "step": 3928
    },
    {
      "epoch": 0.0266949877680529,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999124917675595,
      "loss": 2.4492,
      "step": 3929
    },
    {
      "epoch": 0.026701782114646955,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999124281322248,
      "loss": 2.5058,
      "step": 3930
    },
    {
      "epoch": 0.026708576461241012,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009991236447378117,
      "loss": 2.4813,
      "step": 3931
    },
    {
      "epoch": 0.026715370807835072,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009991230079222874,
      "loss": 2.4574,
      "step": 3932
    },
    {
      "epoch": 0.02672216515442913,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009991223708756746,
      "loss": 2.4204,
      "step": 3933
    },
    {
      "epoch": 0.026728959501023185,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009991217335979738,
      "loss": 2.3775,
      "step": 3934
    },
    {
      "epoch": 0.026735753847617245,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009991210960891856,
      "loss": 2.1956,
      "step": 3935
    },
    {
      "epoch": 0.0267425481942113,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099912045834931,
      "loss": 2.3214,
      "step": 3936
    },
    {
      "epoch": 0.026749342540805358,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009991198203783472,
      "loss": 2.4834,
      "step": 3937
    },
    {
      "epoch": 0.026756136887399418,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009991191821762976,
      "loss": 2.451,
      "step": 3938
    },
    {
      "epoch": 0.026762931233993474,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009991185437431618,
      "loss": 2.4149,
      "step": 3939
    },
    {
      "epoch": 0.026769725580587534,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009991179050789396,
      "loss": 2.3837,
      "step": 3940
    },
    {
      "epoch": 0.02677651992718159,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009991172661836319,
      "loss": 2.3954,
      "step": 3941
    },
    {
      "epoch": 0.026783314273775647,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009991166270572384,
      "loss": 2.3696,
      "step": 3942
    },
    {
      "epoch": 0.026790108620369707,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009991159876997595,
      "loss": 2.4483,
      "step": 3943
    },
    {
      "epoch": 0.026796902966963763,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009991153481111955,
      "loss": 2.4587,
      "step": 3944
    },
    {
      "epoch": 0.02680369731355782,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000999114708291547,
      "loss": 2.3766,
      "step": 3945
    },
    {
      "epoch": 0.02681049166015188,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009991140682408142,
      "loss": 2.407,
      "step": 3946
    },
    {
      "epoch": 0.026817286006745936,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009991134279589973,
      "loss": 2.3156,
      "step": 3947
    },
    {
      "epoch": 0.026824080353339996,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009991127874460965,
      "loss": 2.4019,
      "step": 3948
    },
    {
      "epoch": 0.026830874699934053,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009991121467021124,
      "loss": 2.3773,
      "step": 3949
    },
    {
      "epoch": 0.02683766904652811,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999111505727045,
      "loss": 2.5884,
      "step": 3950
    },
    {
      "epoch": 0.02684446339312217,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009991108645208946,
      "loss": 2.3697,
      "step": 3951
    },
    {
      "epoch": 0.026851257739716226,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009991102230836617,
      "loss": 2.4533,
      "step": 3952
    },
    {
      "epoch": 0.026858052086310282,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009991095814153464,
      "loss": 2.3621,
      "step": 3953
    },
    {
      "epoch": 0.026864846432904342,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999108939515949,
      "loss": 2.4841,
      "step": 3954
    },
    {
      "epoch": 0.0268716407794984,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00099910829738547,
      "loss": 2.3307,
      "step": 3955
    },
    {
      "epoch": 0.026878435126092455,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009991076550239096,
      "loss": 2.318,
      "step": 3956
    },
    {
      "epoch": 0.026885229472686515,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009991070124312682,
      "loss": 2.2856,
      "step": 3957
    },
    {
      "epoch": 0.02689202381928057,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009991063696075458,
      "loss": 2.2525,
      "step": 3958
    },
    {
      "epoch": 0.02689881816587463,
      "grad_norm": 0.2197265625,
      "learning_rate": 0.0009991057265527428,
      "loss": 2.5022,
      "step": 3959
    },
    {
      "epoch": 0.026905612512468688,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009991050832668597,
      "loss": 2.4784,
      "step": 3960
    },
    {
      "epoch": 0.026912406859062744,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009991044397498967,
      "loss": 2.4436,
      "step": 3961
    },
    {
      "epoch": 0.026919201205656804,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999103796001854,
      "loss": 2.3055,
      "step": 3962
    },
    {
      "epoch": 0.02692599555225086,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999103152022732,
      "loss": 2.3737,
      "step": 3963
    },
    {
      "epoch": 0.026932789898844917,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999102507812531,
      "loss": 2.4611,
      "step": 3964
    },
    {
      "epoch": 0.026939584245438977,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999101863371251,
      "loss": 2.4221,
      "step": 3965
    },
    {
      "epoch": 0.026946378592033034,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009991012186988928,
      "loss": 2.3293,
      "step": 3966
    },
    {
      "epoch": 0.02695317293862709,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009991005737954564,
      "loss": 2.3272,
      "step": 3967
    },
    {
      "epoch": 0.02695996728522115,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009990999286609422,
      "loss": 2.4702,
      "step": 3968
    },
    {
      "epoch": 0.026966761631815207,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009990992832953505,
      "loss": 2.3614,
      "step": 3969
    },
    {
      "epoch": 0.026973555978409267,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009990986376986815,
      "loss": 2.5865,
      "step": 3970
    },
    {
      "epoch": 0.026980350325003323,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990979918709354,
      "loss": 2.5128,
      "step": 3971
    },
    {
      "epoch": 0.02698714467159738,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999097345812113,
      "loss": 2.5861,
      "step": 3972
    },
    {
      "epoch": 0.02699393901819144,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999096699522214,
      "loss": 2.4155,
      "step": 3973
    },
    {
      "epoch": 0.027000733364785496,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000999096053001239,
      "loss": 2.4111,
      "step": 3974
    },
    {
      "epoch": 0.027007527711379552,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009990954062491881,
      "loss": 2.4887,
      "step": 3975
    },
    {
      "epoch": 0.027014322057973612,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999094759266062,
      "loss": 2.4275,
      "step": 3976
    },
    {
      "epoch": 0.02702111640456767,
      "grad_norm": 0.37890625,
      "learning_rate": 0.0009990941120518605,
      "loss": 2.4323,
      "step": 3977
    },
    {
      "epoch": 0.02702791075116173,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009990934646065843,
      "loss": 2.3226,
      "step": 3978
    },
    {
      "epoch": 0.027034705097755785,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009990928169302336,
      "loss": 2.5035,
      "step": 3979
    },
    {
      "epoch": 0.027041499444349842,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009990921690228086,
      "loss": 2.2919,
      "step": 3980
    },
    {
      "epoch": 0.0270482937909439,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009990915208843096,
      "loss": 2.403,
      "step": 3981
    },
    {
      "epoch": 0.027055088137537958,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999090872514737,
      "loss": 2.3277,
      "step": 3982
    },
    {
      "epoch": 0.027061882484132015,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009990902239140908,
      "loss": 2.4238,
      "step": 3983
    },
    {
      "epoch": 0.027068676830726075,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009990895750823718,
      "loss": 2.39,
      "step": 3984
    },
    {
      "epoch": 0.02707547117732013,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009990889260195802,
      "loss": 2.3975,
      "step": 3985
    },
    {
      "epoch": 0.027082265523914188,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009990882767257159,
      "loss": 2.3928,
      "step": 3986
    },
    {
      "epoch": 0.027089059870508247,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009990876272007794,
      "loss": 2.4534,
      "step": 3987
    },
    {
      "epoch": 0.027095854217102304,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009990869774447713,
      "loss": 2.3724,
      "step": 3988
    },
    {
      "epoch": 0.027102648563696364,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009990863274576915,
      "loss": 2.5535,
      "step": 3989
    },
    {
      "epoch": 0.02710944291029042,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009990856772395405,
      "loss": 2.5338,
      "step": 3990
    },
    {
      "epoch": 0.027116237256884477,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009990850267903185,
      "loss": 2.2883,
      "step": 3991
    },
    {
      "epoch": 0.027123031603478537,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009990843761100258,
      "loss": 2.4039,
      "step": 3992
    },
    {
      "epoch": 0.027129825950072593,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009990837251986628,
      "loss": 2.3876,
      "step": 3993
    },
    {
      "epoch": 0.02713662029666665,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009990830740562297,
      "loss": 2.4954,
      "step": 3994
    },
    {
      "epoch": 0.02714341464326071,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000999082422682727,
      "loss": 2.3815,
      "step": 3995
    },
    {
      "epoch": 0.027150208989854766,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009990817710781549,
      "loss": 2.3837,
      "step": 3996
    },
    {
      "epoch": 0.027157003336448823,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009990811192425136,
      "loss": 2.2734,
      "step": 3997
    },
    {
      "epoch": 0.027163797683042883,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009990804671758033,
      "loss": 2.4005,
      "step": 3998
    },
    {
      "epoch": 0.02717059202963694,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009990798148780246,
      "loss": 2.4587,
      "step": 3999
    },
    {
      "epoch": 0.027177386376231,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009990791623491777,
      "loss": 2.5083,
      "step": 4000
    },
    {
      "epoch": 0.027184180722825056,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999078509589263,
      "loss": 2.4625,
      "step": 4001
    },
    {
      "epoch": 0.027190975069419112,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990778565982803,
      "loss": 2.3446,
      "step": 4002
    },
    {
      "epoch": 0.027197769416013172,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009990772033762305,
      "loss": 2.3159,
      "step": 4003
    },
    {
      "epoch": 0.02720456376260723,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009990765499231138,
      "loss": 2.3523,
      "step": 4004
    },
    {
      "epoch": 0.027211358109201285,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009990758962389302,
      "loss": 2.3221,
      "step": 4005
    },
    {
      "epoch": 0.027218152455795345,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00099907524232368,
      "loss": 2.3499,
      "step": 4006
    },
    {
      "epoch": 0.0272249468023894,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999074588177364,
      "loss": 2.5571,
      "step": 4007
    },
    {
      "epoch": 0.02723174114898346,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000999073933799982,
      "loss": 2.3128,
      "step": 4008
    },
    {
      "epoch": 0.027238535495577518,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009990732791915348,
      "loss": 2.3364,
      "step": 4009
    },
    {
      "epoch": 0.027245329842171574,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999072624352022,
      "loss": 2.3365,
      "step": 4010
    },
    {
      "epoch": 0.027252124188765634,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009990719692814444,
      "loss": 2.5334,
      "step": 4011
    },
    {
      "epoch": 0.02725891853535969,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009990713139798025,
      "loss": 2.4318,
      "step": 4012
    },
    {
      "epoch": 0.027265712881953747,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999070658447096,
      "loss": 2.5258,
      "step": 4013
    },
    {
      "epoch": 0.027272507228547807,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990700026833257,
      "loss": 2.4664,
      "step": 4014
    },
    {
      "epoch": 0.027279301575141864,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009990693466884915,
      "loss": 2.6004,
      "step": 4015
    },
    {
      "epoch": 0.02728609592173592,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009990686904625939,
      "loss": 2.3528,
      "step": 4016
    },
    {
      "epoch": 0.02729289026832998,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009990680340056335,
      "loss": 2.391,
      "step": 4017
    },
    {
      "epoch": 0.027299684614924036,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00099906737731761,
      "loss": 2.4289,
      "step": 4018
    },
    {
      "epoch": 0.027306478961518096,
      "grad_norm": 0.140625,
      "learning_rate": 0.000999066720398524,
      "loss": 2.346,
      "step": 4019
    },
    {
      "epoch": 0.027313273308112153,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009990660632483762,
      "loss": 2.3258,
      "step": 4020
    },
    {
      "epoch": 0.02732006765470621,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009990654058671664,
      "loss": 2.4803,
      "step": 4021
    },
    {
      "epoch": 0.02732686200130027,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999064748254895,
      "loss": 2.4617,
      "step": 4022
    },
    {
      "epoch": 0.027333656347894326,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009990640904115622,
      "loss": 2.3696,
      "step": 4023
    },
    {
      "epoch": 0.027340450694488382,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009990634323371686,
      "loss": 2.3766,
      "step": 4024
    },
    {
      "epoch": 0.027347245041082442,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009990627740317144,
      "loss": 2.4306,
      "step": 4025
    },
    {
      "epoch": 0.0273540393876765,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009990621154951998,
      "loss": 2.4543,
      "step": 4026
    },
    {
      "epoch": 0.027360833734270555,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000999061456727625,
      "loss": 2.3477,
      "step": 4027
    },
    {
      "epoch": 0.027367628080864615,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009990607977289907,
      "loss": 2.3725,
      "step": 4028
    },
    {
      "epoch": 0.02737442242745867,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009990601384992968,
      "loss": 2.5457,
      "step": 4029
    },
    {
      "epoch": 0.02738121677405273,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000999059479038544,
      "loss": 2.4854,
      "step": 4030
    },
    {
      "epoch": 0.027388011120646788,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000999058819346732,
      "loss": 2.425,
      "step": 4031
    },
    {
      "epoch": 0.027394805467240844,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009990581594238618,
      "loss": 2.3747,
      "step": 4032
    },
    {
      "epoch": 0.027401599813834904,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009990574992699334,
      "loss": 2.3621,
      "step": 4033
    },
    {
      "epoch": 0.02740839416042896,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009990568388849469,
      "loss": 2.3284,
      "step": 4034
    },
    {
      "epoch": 0.027415188507023017,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999056178268903,
      "loss": 2.3058,
      "step": 4035
    },
    {
      "epoch": 0.027421982853617077,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009990555174218016,
      "loss": 2.528,
      "step": 4036
    },
    {
      "epoch": 0.027428777200211134,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009990548563436434,
      "loss": 2.3745,
      "step": 4037
    },
    {
      "epoch": 0.027435571546805194,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009990541950344286,
      "loss": 2.4916,
      "step": 4038
    },
    {
      "epoch": 0.02744236589339925,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009990535334941572,
      "loss": 2.5653,
      "step": 4039
    },
    {
      "epoch": 0.027449160239993307,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009990528717228298,
      "loss": 2.353,
      "step": 4040
    },
    {
      "epoch": 0.027455954586587367,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009990522097204466,
      "loss": 2.4072,
      "step": 4041
    },
    {
      "epoch": 0.027462748933181423,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009990515474870081,
      "loss": 2.3337,
      "step": 4042
    },
    {
      "epoch": 0.02746954327977548,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009990508850225145,
      "loss": 2.6013,
      "step": 4043
    },
    {
      "epoch": 0.02747633762636954,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009990502223269659,
      "loss": 2.3503,
      "step": 4044
    },
    {
      "epoch": 0.027483131972963596,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009990495594003627,
      "loss": 2.3208,
      "step": 4045
    },
    {
      "epoch": 0.027489926319557653,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009990488962427055,
      "loss": 2.3223,
      "step": 4046
    },
    {
      "epoch": 0.027496720666151712,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009990482328539943,
      "loss": 2.3629,
      "step": 4047
    },
    {
      "epoch": 0.02750351501274577,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990475692342293,
      "loss": 2.2599,
      "step": 4048
    },
    {
      "epoch": 0.02751030935933983,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009990469053834112,
      "loss": 2.4522,
      "step": 4049
    },
    {
      "epoch": 0.027517103705933885,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009990462413015401,
      "loss": 2.4669,
      "step": 4050
    },
    {
      "epoch": 0.027523898052527942,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009990455769886162,
      "loss": 2.4176,
      "step": 4051
    },
    {
      "epoch": 0.027530692399122002,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00099904491244464,
      "loss": 2.3914,
      "step": 4052
    },
    {
      "epoch": 0.027537486745716058,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009990442476696117,
      "loss": 2.2437,
      "step": 4053
    },
    {
      "epoch": 0.027544281092310115,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009990435826635316,
      "loss": 2.2236,
      "step": 4054
    },
    {
      "epoch": 0.027551075438904175,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009990429174264002,
      "loss": 2.5398,
      "step": 4055
    },
    {
      "epoch": 0.02755786978549823,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009990422519582174,
      "loss": 2.4756,
      "step": 4056
    },
    {
      "epoch": 0.027564664132092288,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000999041586258984,
      "loss": 2.3431,
      "step": 4057
    },
    {
      "epoch": 0.027571458478686348,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009990409203287,
      "loss": 2.3761,
      "step": 4058
    },
    {
      "epoch": 0.027578252825280404,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009990402541673658,
      "loss": 2.4594,
      "step": 4059
    },
    {
      "epoch": 0.027585047171874464,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009990395877749816,
      "loss": 2.4574,
      "step": 4060
    },
    {
      "epoch": 0.02759184151846852,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990389211515479,
      "loss": 2.4027,
      "step": 4061
    },
    {
      "epoch": 0.027598635865062577,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009990382542970646,
      "loss": 2.3755,
      "step": 4062
    },
    {
      "epoch": 0.027605430211656637,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990375872115328,
      "loss": 2.3455,
      "step": 4063
    },
    {
      "epoch": 0.027612224558250693,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000999036919894952,
      "loss": 2.4079,
      "step": 4064
    },
    {
      "epoch": 0.02761901890484475,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009990362523473227,
      "loss": 2.4308,
      "step": 4065
    },
    {
      "epoch": 0.02762581325143881,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009990355845686456,
      "loss": 2.3657,
      "step": 4066
    },
    {
      "epoch": 0.027632607598032866,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009990349165589205,
      "loss": 2.2608,
      "step": 4067
    },
    {
      "epoch": 0.027639401944626926,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990342483181484,
      "loss": 2.3137,
      "step": 4068
    },
    {
      "epoch": 0.027646196291220983,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009990335798463288,
      "loss": 2.4014,
      "step": 4069
    },
    {
      "epoch": 0.02765299063781504,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009990329111434625,
      "loss": 2.3074,
      "step": 4070
    },
    {
      "epoch": 0.0276597849844091,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009990322422095495,
      "loss": 2.4581,
      "step": 4071
    },
    {
      "epoch": 0.027666579331003156,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009990315730445906,
      "loss": 2.4109,
      "step": 4072
    },
    {
      "epoch": 0.027673373677597212,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990309036485854,
      "loss": 2.3866,
      "step": 4073
    },
    {
      "epoch": 0.027680168024191272,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000999030234021535,
      "loss": 2.3066,
      "step": 4074
    },
    {
      "epoch": 0.02768696237078533,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000999029564163439,
      "loss": 2.2533,
      "step": 4075
    },
    {
      "epoch": 0.027693756717379385,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009990288940742982,
      "loss": 2.4344,
      "step": 4076
    },
    {
      "epoch": 0.027700551063973445,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009990282237541128,
      "loss": 2.36,
      "step": 4077
    },
    {
      "epoch": 0.0277073454105675,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000999027553202883,
      "loss": 2.4342,
      "step": 4078
    },
    {
      "epoch": 0.02771413975716156,
      "grad_norm": 0.140625,
      "learning_rate": 0.000999026882420609,
      "loss": 2.2709,
      "step": 4079
    },
    {
      "epoch": 0.027720934103755618,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009990262114072915,
      "loss": 2.3076,
      "step": 4080
    },
    {
      "epoch": 0.027727728450349674,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009990255401629306,
      "loss": 2.4129,
      "step": 4081
    },
    {
      "epoch": 0.027734522796943734,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009990248686875265,
      "loss": 2.4794,
      "step": 4082
    },
    {
      "epoch": 0.02774131714353779,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990241969810795,
      "loss": 2.2957,
      "step": 4083
    },
    {
      "epoch": 0.027748111490131847,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009990235250435901,
      "loss": 2.2728,
      "step": 4084
    },
    {
      "epoch": 0.027754905836725907,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009990228528750586,
      "loss": 2.3861,
      "step": 4085
    },
    {
      "epoch": 0.027761700183319964,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009990221804754853,
      "loss": 2.2069,
      "step": 4086
    },
    {
      "epoch": 0.02776849452991402,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009990215078448704,
      "loss": 2.5017,
      "step": 4087
    },
    {
      "epoch": 0.02777528887650808,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000999020834983214,
      "loss": 2.2624,
      "step": 4088
    },
    {
      "epoch": 0.027782083223102137,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000999020161890517,
      "loss": 2.4797,
      "step": 4089
    },
    {
      "epoch": 0.027788877569696196,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009990194885667793,
      "loss": 2.3159,
      "step": 4090
    },
    {
      "epoch": 0.027795671916290253,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009990188150120014,
      "loss": 2.3246,
      "step": 4091
    },
    {
      "epoch": 0.02780246626288431,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009990181412261835,
      "loss": 2.5318,
      "step": 4092
    },
    {
      "epoch": 0.02780926060947837,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009990174672093257,
      "loss": 2.3394,
      "step": 4093
    },
    {
      "epoch": 0.027816054956072426,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990167929614286,
      "loss": 2.4388,
      "step": 4094
    },
    {
      "epoch": 0.027822849302666482,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009990161184824926,
      "loss": 2.1848,
      "step": 4095
    },
    {
      "epoch": 0.027829643649260542,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009990154437725177,
      "loss": 2.4407,
      "step": 4096
    },
    {
      "epoch": 0.0278364379958546,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009990147688315044,
      "loss": 2.397,
      "step": 4097
    },
    {
      "epoch": 0.02784323234244866,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999014093659453,
      "loss": 2.4204,
      "step": 4098
    },
    {
      "epoch": 0.027850026689042715,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009990134182563638,
      "loss": 2.4696,
      "step": 4099
    },
    {
      "epoch": 0.02785682103563677,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999012742622237,
      "loss": 2.2499,
      "step": 4100
    },
    {
      "epoch": 0.02786361538223083,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009990120667570733,
      "loss": 2.5469,
      "step": 4101
    },
    {
      "epoch": 0.027870409728824888,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009990113906608725,
      "loss": 2.3053,
      "step": 4102
    },
    {
      "epoch": 0.027877204075418945,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000999010714333635,
      "loss": 2.4854,
      "step": 4103
    },
    {
      "epoch": 0.027883998422013005,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009990100377753617,
      "loss": 2.4096,
      "step": 4104
    },
    {
      "epoch": 0.02789079276860706,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009990093609860522,
      "loss": 2.3495,
      "step": 4105
    },
    {
      "epoch": 0.027897587115201117,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000999008683965707,
      "loss": 2.4163,
      "step": 4106
    },
    {
      "epoch": 0.027904381461795177,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009990080067143268,
      "loss": 2.3751,
      "step": 4107
    },
    {
      "epoch": 0.027911175808389234,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009990073292319114,
      "loss": 2.2812,
      "step": 4108
    },
    {
      "epoch": 0.027917970154983294,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990066515184613,
      "loss": 2.3696,
      "step": 4109
    },
    {
      "epoch": 0.02792476450157735,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000999005973573977,
      "loss": 2.5503,
      "step": 4110
    },
    {
      "epoch": 0.027931558848171407,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009990052953984586,
      "loss": 2.4201,
      "step": 4111
    },
    {
      "epoch": 0.027938353194765467,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009990046169919063,
      "loss": 2.4305,
      "step": 4112
    },
    {
      "epoch": 0.027945147541359523,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009990039383543207,
      "loss": 2.3426,
      "step": 4113
    },
    {
      "epoch": 0.02795194188795358,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000999003259485702,
      "loss": 2.2792,
      "step": 4114
    },
    {
      "epoch": 0.02795873623454764,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009990025803860506,
      "loss": 2.2808,
      "step": 4115
    },
    {
      "epoch": 0.027965530581141696,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009990019010553666,
      "loss": 2.4142,
      "step": 4116
    },
    {
      "epoch": 0.027972324927735753,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009990012214936505,
      "loss": 2.3716,
      "step": 4117
    },
    {
      "epoch": 0.027979119274329813,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009990005417009026,
      "loss": 2.2353,
      "step": 4118
    },
    {
      "epoch": 0.02798591362092387,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009989998616771233,
      "loss": 2.2731,
      "step": 4119
    },
    {
      "epoch": 0.02799270796751793,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989991814223125,
      "loss": 2.4531,
      "step": 4120
    },
    {
      "epoch": 0.027999502314111985,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989985009364708,
      "loss": 2.3213,
      "step": 4121
    },
    {
      "epoch": 0.028006296660706042,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009989978202195986,
      "loss": 2.4191,
      "step": 4122
    },
    {
      "epoch": 0.028013091007300102,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009989971392716961,
      "loss": 2.3176,
      "step": 4123
    },
    {
      "epoch": 0.02801988535389416,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009989964580927636,
      "loss": 2.3804,
      "step": 4124
    },
    {
      "epoch": 0.028026679700488215,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009989957766828018,
      "loss": 2.2223,
      "step": 4125
    },
    {
      "epoch": 0.028033474047082275,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009989950950418105,
      "loss": 2.3495,
      "step": 4126
    },
    {
      "epoch": 0.02804026839367633,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099899441316979,
      "loss": 2.2198,
      "step": 4127
    },
    {
      "epoch": 0.02804706274027039,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009989937310667409,
      "loss": 2.4776,
      "step": 4128
    },
    {
      "epoch": 0.028053857086864448,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009989930487326634,
      "loss": 2.3153,
      "step": 4129
    },
    {
      "epoch": 0.028060651433458504,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998992366167558,
      "loss": 2.2622,
      "step": 4130
    },
    {
      "epoch": 0.028067445780052564,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009989916833714248,
      "loss": 2.3784,
      "step": 4131
    },
    {
      "epoch": 0.02807424012664662,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000998991000344264,
      "loss": 2.4407,
      "step": 4132
    },
    {
      "epoch": 0.028081034473240677,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989903170860763,
      "loss": 2.379,
      "step": 4133
    },
    {
      "epoch": 0.028087828819834737,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009989896335968616,
      "loss": 2.4803,
      "step": 4134
    },
    {
      "epoch": 0.028094623166428793,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009989889498766206,
      "loss": 2.3651,
      "step": 4135
    },
    {
      "epoch": 0.02810141751302285,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009989882659253533,
      "loss": 2.3054,
      "step": 4136
    },
    {
      "epoch": 0.02810821185961691,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009989875817430602,
      "loss": 2.4376,
      "step": 4137
    },
    {
      "epoch": 0.028115006206210966,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989868973297416,
      "loss": 2.2905,
      "step": 4138
    },
    {
      "epoch": 0.028121800552805026,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009989862126853978,
      "loss": 2.4218,
      "step": 4139
    },
    {
      "epoch": 0.028128594899399083,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009989855278100292,
      "loss": 2.4347,
      "step": 4140
    },
    {
      "epoch": 0.02813538924599314,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998984842703636,
      "loss": 2.5677,
      "step": 4141
    },
    {
      "epoch": 0.0281421835925872,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009989841573662182,
      "loss": 2.4337,
      "step": 4142
    },
    {
      "epoch": 0.028148977939181256,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009989834717977768,
      "loss": 2.1888,
      "step": 4143
    },
    {
      "epoch": 0.028155772285775312,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009989827859983116,
      "loss": 2.3893,
      "step": 4144
    },
    {
      "epoch": 0.028162566632369372,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009989820999678231,
      "loss": 2.4274,
      "step": 4145
    },
    {
      "epoch": 0.02816936097896343,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009989814137063117,
      "loss": 2.3925,
      "step": 4146
    },
    {
      "epoch": 0.028176155325557485,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009989807272137777,
      "loss": 2.2111,
      "step": 4147
    },
    {
      "epoch": 0.028182949672151545,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009989800404902212,
      "loss": 2.1943,
      "step": 4148
    },
    {
      "epoch": 0.0281897440187456,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989793535356427,
      "loss": 2.2952,
      "step": 4149
    },
    {
      "epoch": 0.02819653836533966,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009989786663500424,
      "loss": 2.4205,
      "step": 4150
    },
    {
      "epoch": 0.028203332711933718,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000998977978933421,
      "loss": 2.3982,
      "step": 4151
    },
    {
      "epoch": 0.028210127058527774,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989772912857781,
      "loss": 2.2933,
      "step": 4152
    },
    {
      "epoch": 0.028216921405121834,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009989766034071148,
      "loss": 2.5844,
      "step": 4153
    },
    {
      "epoch": 0.02822371575171589,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009989759152974308,
      "loss": 2.4433,
      "step": 4154
    },
    {
      "epoch": 0.028230510098309947,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009989752269567267,
      "loss": 2.2772,
      "step": 4155
    },
    {
      "epoch": 0.028237304444904007,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000998974538385003,
      "loss": 2.5011,
      "step": 4156
    },
    {
      "epoch": 0.028244098791498064,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009989738495822596,
      "loss": 2.4402,
      "step": 4157
    },
    {
      "epoch": 0.028250893138092124,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000998973160548497,
      "loss": 2.4557,
      "step": 4158
    },
    {
      "epoch": 0.02825768748468618,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989724712837155,
      "loss": 2.2805,
      "step": 4159
    },
    {
      "epoch": 0.028264481831280237,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009989717817879158,
      "loss": 2.2503,
      "step": 4160
    },
    {
      "epoch": 0.028271276177874297,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009989710920610977,
      "loss": 2.4276,
      "step": 4161
    },
    {
      "epoch": 0.028278070524468353,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009989704021032616,
      "loss": 2.3735,
      "step": 4162
    },
    {
      "epoch": 0.02828486487106241,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998969711914408,
      "loss": 2.2905,
      "step": 4163
    },
    {
      "epoch": 0.02829165921765647,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009989690214945372,
      "loss": 2.4976,
      "step": 4164
    },
    {
      "epoch": 0.028298453564250526,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009989683308436494,
      "loss": 2.4446,
      "step": 4165
    },
    {
      "epoch": 0.028305247910844582,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998967639961745,
      "loss": 2.4838,
      "step": 4166
    },
    {
      "epoch": 0.028312042257438642,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009989669488488244,
      "loss": 2.3987,
      "step": 4167
    },
    {
      "epoch": 0.0283188366040327,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009989662575048876,
      "loss": 2.4915,
      "step": 4168
    },
    {
      "epoch": 0.02832563095062676,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009989655659299352,
      "loss": 2.3742,
      "step": 4169
    },
    {
      "epoch": 0.028332425297220815,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009989648741239676,
      "loss": 2.3329,
      "step": 4170
    },
    {
      "epoch": 0.028339219643814872,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998964182086985,
      "loss": 2.468,
      "step": 4171
    },
    {
      "epoch": 0.02834601399040893,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989634898189877,
      "loss": 2.4435,
      "step": 4172
    },
    {
      "epoch": 0.028352808337002988,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009989627973199758,
      "loss": 2.4055,
      "step": 4173
    },
    {
      "epoch": 0.028359602683597045,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00099896210458995,
      "loss": 2.3495,
      "step": 4174
    },
    {
      "epoch": 0.028366397030191105,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009989614116289105,
      "loss": 2.4974,
      "step": 4175
    },
    {
      "epoch": 0.02837319137678516,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009989607184368574,
      "loss": 2.2341,
      "step": 4176
    },
    {
      "epoch": 0.028379985723379218,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009989600250137915,
      "loss": 2.3807,
      "step": 4177
    },
    {
      "epoch": 0.028386780069973278,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009989593313597127,
      "loss": 2.4005,
      "step": 4178
    },
    {
      "epoch": 0.028393574416567334,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009989586374746214,
      "loss": 2.3771,
      "step": 4179
    },
    {
      "epoch": 0.028400368763161394,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998957943358518,
      "loss": 2.3093,
      "step": 4180
    },
    {
      "epoch": 0.02840716310975545,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009989572490114028,
      "loss": 2.2918,
      "step": 4181
    },
    {
      "epoch": 0.028413957456349507,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009989565544332763,
      "loss": 2.204,
      "step": 4182
    },
    {
      "epoch": 0.028420751802943567,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989558596241383,
      "loss": 2.4398,
      "step": 4183
    },
    {
      "epoch": 0.028427546149537623,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009989551645839896,
      "loss": 2.3104,
      "step": 4184
    },
    {
      "epoch": 0.02843434049613168,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009989544693128304,
      "loss": 2.431,
      "step": 4185
    },
    {
      "epoch": 0.02844113484272574,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009989537738106609,
      "loss": 2.4848,
      "step": 4186
    },
    {
      "epoch": 0.028447929189319796,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009989530780774817,
      "loss": 2.537,
      "step": 4187
    },
    {
      "epoch": 0.028454723535913856,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009989523821132928,
      "loss": 2.4356,
      "step": 4188
    },
    {
      "epoch": 0.028461517882507913,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009989516859180948,
      "loss": 2.4275,
      "step": 4189
    },
    {
      "epoch": 0.02846831222910197,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009989509894918877,
      "loss": 2.247,
      "step": 4190
    },
    {
      "epoch": 0.02847510657569603,
      "grad_norm": 0.15625,
      "learning_rate": 0.000998950292834672,
      "loss": 2.3453,
      "step": 4191
    },
    {
      "epoch": 0.028481900922290086,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009989495959464483,
      "loss": 2.288,
      "step": 4192
    },
    {
      "epoch": 0.028488695268884142,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009989488988272164,
      "loss": 2.3217,
      "step": 4193
    },
    {
      "epoch": 0.028495489615478202,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000998948201476977,
      "loss": 2.4146,
      "step": 4194
    },
    {
      "epoch": 0.02850228396207226,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009989475038957303,
      "loss": 2.4035,
      "step": 4195
    },
    {
      "epoch": 0.028509078308666315,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989468060834766,
      "loss": 2.3478,
      "step": 4196
    },
    {
      "epoch": 0.028515872655260375,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009989461080402164,
      "loss": 2.4092,
      "step": 4197
    },
    {
      "epoch": 0.02852266700185443,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009989454097659497,
      "loss": 2.3402,
      "step": 4198
    },
    {
      "epoch": 0.02852946134844849,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009989447112606772,
      "loss": 2.4193,
      "step": 4199
    },
    {
      "epoch": 0.028536255695042548,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009989440125243987,
      "loss": 2.3594,
      "step": 4200
    },
    {
      "epoch": 0.028543050041636604,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998943313557115,
      "loss": 2.3508,
      "step": 4201
    },
    {
      "epoch": 0.028549844388230664,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009989426143588262,
      "loss": 2.4251,
      "step": 4202
    },
    {
      "epoch": 0.02855663873482472,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998941914929533,
      "loss": 2.3783,
      "step": 4203
    },
    {
      "epoch": 0.028563433081418777,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009989412152692349,
      "loss": 2.2697,
      "step": 4204
    },
    {
      "epoch": 0.028570227428012837,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998940515377933,
      "loss": 2.3647,
      "step": 4205
    },
    {
      "epoch": 0.028577021774606894,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009989398152556274,
      "loss": 2.3154,
      "step": 4206
    },
    {
      "epoch": 0.02858381612120095,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009989391149023185,
      "loss": 2.3296,
      "step": 4207
    },
    {
      "epoch": 0.02859061046779501,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009989384143180063,
      "loss": 2.449,
      "step": 4208
    },
    {
      "epoch": 0.028597404814389066,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989377135026913,
      "loss": 2.3663,
      "step": 4209
    },
    {
      "epoch": 0.028604199160983126,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998937012456374,
      "loss": 2.3111,
      "step": 4210
    },
    {
      "epoch": 0.028610993507577183,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989363111790545,
      "loss": 2.3889,
      "step": 4211
    },
    {
      "epoch": 0.02861778785417124,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009989356096707332,
      "loss": 2.4702,
      "step": 4212
    },
    {
      "epoch": 0.0286245822007653,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009989349079314105,
      "loss": 2.5346,
      "step": 4213
    },
    {
      "epoch": 0.028631376547359356,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009989342059610867,
      "loss": 2.329,
      "step": 4214
    },
    {
      "epoch": 0.028638170893953412,
      "grad_norm": 0.140625,
      "learning_rate": 0.000998933503759762,
      "loss": 2.3195,
      "step": 4215
    },
    {
      "epoch": 0.028644965240547472,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009989328013274368,
      "loss": 2.4723,
      "step": 4216
    },
    {
      "epoch": 0.02865175958714153,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009989320986641115,
      "loss": 2.3275,
      "step": 4217
    },
    {
      "epoch": 0.02865855393373559,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009989313957697863,
      "loss": 2.3058,
      "step": 4218
    },
    {
      "epoch": 0.028665348280329645,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009989306926444616,
      "loss": 2.3338,
      "step": 4219
    },
    {
      "epoch": 0.0286721426269237,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989299892881377,
      "loss": 2.33,
      "step": 4220
    },
    {
      "epoch": 0.02867893697351776,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998929285700815,
      "loss": 2.4768,
      "step": 4221
    },
    {
      "epoch": 0.028685731320111818,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009989285818824936,
      "loss": 2.3516,
      "step": 4222
    },
    {
      "epoch": 0.028692525666705875,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009989278778331741,
      "loss": 2.5082,
      "step": 4223
    },
    {
      "epoch": 0.028699320013299934,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009989271735528567,
      "loss": 2.3134,
      "step": 4224
    },
    {
      "epoch": 0.02870611435989399,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009989264690415417,
      "loss": 2.3442,
      "step": 4225
    },
    {
      "epoch": 0.028712908706488047,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009989257642992295,
      "loss": 2.4777,
      "step": 4226
    },
    {
      "epoch": 0.028719703053082107,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009989250593259204,
      "loss": 2.3637,
      "step": 4227
    },
    {
      "epoch": 0.028726497399676164,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009989243541216148,
      "loss": 2.3755,
      "step": 4228
    },
    {
      "epoch": 0.028733291746270224,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009989236486863129,
      "loss": 2.4615,
      "step": 4229
    },
    {
      "epoch": 0.02874008609286428,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009989229430200151,
      "loss": 2.5625,
      "step": 4230
    },
    {
      "epoch": 0.028746880439458337,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009989222371227216,
      "loss": 2.3724,
      "step": 4231
    },
    {
      "epoch": 0.028753674786052397,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998921530994433,
      "loss": 2.2767,
      "step": 4232
    },
    {
      "epoch": 0.028760469132646453,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009989208246351492,
      "loss": 2.4207,
      "step": 4233
    },
    {
      "epoch": 0.02876726347924051,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998920118044871,
      "loss": 2.2794,
      "step": 4234
    },
    {
      "epoch": 0.02877405782583457,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009989194112235982,
      "loss": 2.2627,
      "step": 4235
    },
    {
      "epoch": 0.028780852172428626,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009989187041713316,
      "loss": 2.4825,
      "step": 4236
    },
    {
      "epoch": 0.028787646519022683,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989179968880714,
      "loss": 2.3742,
      "step": 4237
    },
    {
      "epoch": 0.028794440865616742,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009989172893738179,
      "loss": 2.3691,
      "step": 4238
    },
    {
      "epoch": 0.0288012352122108,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009989165816285713,
      "loss": 2.2403,
      "step": 4239
    },
    {
      "epoch": 0.02880802955880486,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000998915873652332,
      "loss": 2.4619,
      "step": 4240
    },
    {
      "epoch": 0.028814823905398915,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989151654451007,
      "loss": 2.2581,
      "step": 4241
    },
    {
      "epoch": 0.028821618251992972,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998914457006877,
      "loss": 2.4119,
      "step": 4242
    },
    {
      "epoch": 0.028828412598587032,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998913748337662,
      "loss": 2.4317,
      "step": 4243
    },
    {
      "epoch": 0.02883520694518109,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009989130394374554,
      "loss": 2.3904,
      "step": 4244
    },
    {
      "epoch": 0.028842001291775145,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009989123303062578,
      "loss": 2.4172,
      "step": 4245
    },
    {
      "epoch": 0.028848795638369205,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009989116209440697,
      "loss": 2.5516,
      "step": 4246
    },
    {
      "epoch": 0.02885558998496326,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998910911350891,
      "loss": 2.4435,
      "step": 4247
    },
    {
      "epoch": 0.02886238433155732,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009989102015267224,
      "loss": 2.3178,
      "step": 4248
    },
    {
      "epoch": 0.028869178678151378,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009989094914715643,
      "loss": 2.247,
      "step": 4249
    },
    {
      "epoch": 0.028875973024745434,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009989087811854164,
      "loss": 2.3317,
      "step": 4250
    },
    {
      "epoch": 0.028882767371339494,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009989080706682796,
      "loss": 2.3197,
      "step": 4251
    },
    {
      "epoch": 0.02888956171793355,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998907359920154,
      "loss": 2.5493,
      "step": 4252
    },
    {
      "epoch": 0.028896356064527607,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009989066489410404,
      "loss": 2.3031,
      "step": 4253
    },
    {
      "epoch": 0.028903150411121667,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009989059377309385,
      "loss": 2.4316,
      "step": 4254
    },
    {
      "epoch": 0.028909944757715723,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009989052262898489,
      "loss": 2.26,
      "step": 4255
    },
    {
      "epoch": 0.02891673910430978,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009989045146177717,
      "loss": 2.2683,
      "step": 4256
    },
    {
      "epoch": 0.02892353345090384,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009989038027147076,
      "loss": 2.3107,
      "step": 4257
    },
    {
      "epoch": 0.028930327797497896,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009989030905806567,
      "loss": 2.3924,
      "step": 4258
    },
    {
      "epoch": 0.028937122144091956,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009989023782156195,
      "loss": 2.4589,
      "step": 4259
    },
    {
      "epoch": 0.028943916490686013,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998901665619596,
      "loss": 2.2555,
      "step": 4260
    },
    {
      "epoch": 0.02895071083728007,
      "grad_norm": 0.140625,
      "learning_rate": 0.000998900952792587,
      "loss": 2.3845,
      "step": 4261
    },
    {
      "epoch": 0.02895750518387413,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009989002397345925,
      "loss": 2.3296,
      "step": 4262
    },
    {
      "epoch": 0.028964299530468186,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998899526445613,
      "loss": 2.2555,
      "step": 4263
    },
    {
      "epoch": 0.028971093877062242,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009988988129256485,
      "loss": 2.4709,
      "step": 4264
    },
    {
      "epoch": 0.028977888223656302,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009988980991746997,
      "loss": 2.3553,
      "step": 4265
    },
    {
      "epoch": 0.02898468257025036,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998897385192767,
      "loss": 2.2928,
      "step": 4266
    },
    {
      "epoch": 0.028991476916844415,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009988966709798502,
      "loss": 2.2093,
      "step": 4267
    },
    {
      "epoch": 0.028998271263438475,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009988959565359501,
      "loss": 2.4069,
      "step": 4268
    },
    {
      "epoch": 0.02900506561003253,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009988952418610668,
      "loss": 2.3075,
      "step": 4269
    },
    {
      "epoch": 0.02901185995662659,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009988945269552008,
      "loss": 2.369,
      "step": 4270
    },
    {
      "epoch": 0.029018654303220648,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009988938118183524,
      "loss": 2.4403,
      "step": 4271
    },
    {
      "epoch": 0.029025448649814704,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009988930964505219,
      "loss": 2.4466,
      "step": 4272
    },
    {
      "epoch": 0.029032242996408764,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009988923808517095,
      "loss": 2.3996,
      "step": 4273
    },
    {
      "epoch": 0.02903903734300282,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009988916650219156,
      "loss": 2.3962,
      "step": 4274
    },
    {
      "epoch": 0.029045831689596877,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009988909489611407,
      "loss": 2.3253,
      "step": 4275
    },
    {
      "epoch": 0.029052626036190937,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998890232669385,
      "loss": 2.376,
      "step": 4276
    },
    {
      "epoch": 0.029059420382784994,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009988895161466488,
      "loss": 2.3812,
      "step": 4277
    },
    {
      "epoch": 0.029066214729379054,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009988887993929326,
      "loss": 2.6421,
      "step": 4278
    },
    {
      "epoch": 0.02907300907597311,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009988880824082366,
      "loss": 2.2871,
      "step": 4279
    },
    {
      "epoch": 0.029079803422567167,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000998887365192561,
      "loss": 2.5899,
      "step": 4280
    },
    {
      "epoch": 0.029086597769161227,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009988866477459062,
      "loss": 2.4369,
      "step": 4281
    },
    {
      "epoch": 0.029093392115755283,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009988859300682728,
      "loss": 2.3381,
      "step": 4282
    },
    {
      "epoch": 0.02910018646234934,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009988852121596609,
      "loss": 2.2045,
      "step": 4283
    },
    {
      "epoch": 0.0291069808089434,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009988844940200707,
      "loss": 2.5416,
      "step": 4284
    },
    {
      "epoch": 0.029113775155537456,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009988837756495028,
      "loss": 2.3053,
      "step": 4285
    },
    {
      "epoch": 0.029120569502131512,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009988830570479574,
      "loss": 2.3303,
      "step": 4286
    },
    {
      "epoch": 0.029127363848725572,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998882338215435,
      "loss": 2.345,
      "step": 4287
    },
    {
      "epoch": 0.02913415819531963,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009988816191519357,
      "loss": 2.3383,
      "step": 4288
    },
    {
      "epoch": 0.02914095254191369,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00099888089985746,
      "loss": 2.4091,
      "step": 4289
    },
    {
      "epoch": 0.029147746888507745,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009988801803320082,
      "loss": 2.266,
      "step": 4290
    },
    {
      "epoch": 0.0291545412351018,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009988794605755805,
      "loss": 2.2986,
      "step": 4291
    },
    {
      "epoch": 0.02916133558169586,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998878740588177,
      "loss": 2.322,
      "step": 4292
    },
    {
      "epoch": 0.029168129928289918,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009988780203697989,
      "loss": 2.5177,
      "step": 4293
    },
    {
      "epoch": 0.029174924274883975,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000998877299920446,
      "loss": 2.2618,
      "step": 4294
    },
    {
      "epoch": 0.029181718621478035,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009988765792401182,
      "loss": 2.4016,
      "step": 4295
    },
    {
      "epoch": 0.02918851296807209,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009988758583288164,
      "loss": 2.41,
      "step": 4296
    },
    {
      "epoch": 0.029195307314666148,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000998875137186541,
      "loss": 2.2566,
      "step": 4297
    },
    {
      "epoch": 0.029202101661260207,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000998874415813292,
      "loss": 2.4438,
      "step": 4298
    },
    {
      "epoch": 0.029208896007854264,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00099887369420907,
      "loss": 2.4913,
      "step": 4299
    },
    {
      "epoch": 0.029215690354448324,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998872972373875,
      "loss": 2.2494,
      "step": 4300
    },
    {
      "epoch": 0.02922248470104238,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009988722503077075,
      "loss": 2.3279,
      "step": 4301
    },
    {
      "epoch": 0.029229279047636437,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009988715280105681,
      "loss": 2.3173,
      "step": 4302
    },
    {
      "epoch": 0.029236073394230497,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009988708054824568,
      "loss": 2.3692,
      "step": 4303
    },
    {
      "epoch": 0.029242867740824553,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998870082723374,
      "loss": 2.4822,
      "step": 4304
    },
    {
      "epoch": 0.02924966208741861,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.00099886935973332,
      "loss": 2.3661,
      "step": 4305
    },
    {
      "epoch": 0.02925645643401267,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998868636512295,
      "loss": 2.3665,
      "step": 4306
    },
    {
      "epoch": 0.029263250780606726,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009988679130603,
      "loss": 2.3476,
      "step": 4307
    },
    {
      "epoch": 0.029270045127200786,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009988671893773348,
      "loss": 2.356,
      "step": 4308
    },
    {
      "epoch": 0.029276839473794843,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009988664654633998,
      "loss": 2.4203,
      "step": 4309
    },
    {
      "epoch": 0.0292836338203889,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009988657413184951,
      "loss": 2.3834,
      "step": 4310
    },
    {
      "epoch": 0.02929042816698296,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009988650169426214,
      "loss": 2.3519,
      "step": 4311
    },
    {
      "epoch": 0.029297222513577015,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009988642923357792,
      "loss": 2.4562,
      "step": 4312
    },
    {
      "epoch": 0.029304016860171072,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009988635674979681,
      "loss": 2.431,
      "step": 4313
    },
    {
      "epoch": 0.029310811206765132,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009988628424291892,
      "loss": 2.4696,
      "step": 4314
    },
    {
      "epoch": 0.02931760555335919,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009988621171294423,
      "loss": 2.2336,
      "step": 4315
    },
    {
      "epoch": 0.029324399899953245,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998861391598728,
      "loss": 2.4147,
      "step": 4316
    },
    {
      "epoch": 0.029331194246547305,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009988606658370468,
      "loss": 2.3882,
      "step": 4317
    },
    {
      "epoch": 0.02933798859314136,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009988599398443986,
      "loss": 2.3239,
      "step": 4318
    },
    {
      "epoch": 0.02934478293973542,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009988592136207842,
      "loss": 2.4032,
      "step": 4319
    },
    {
      "epoch": 0.029351577286329478,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009988584871662037,
      "loss": 2.3956,
      "step": 4320
    },
    {
      "epoch": 0.029358371632923534,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009988577604806573,
      "loss": 2.3802,
      "step": 4321
    },
    {
      "epoch": 0.029365165979517594,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009988570335641453,
      "loss": 2.2477,
      "step": 4322
    },
    {
      "epoch": 0.02937196032611165,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009988563064166685,
      "loss": 2.4265,
      "step": 4323
    },
    {
      "epoch": 0.029378754672705707,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009988555790382268,
      "loss": 2.3181,
      "step": 4324
    },
    {
      "epoch": 0.029385549019299767,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009988548514288207,
      "loss": 2.5175,
      "step": 4325
    },
    {
      "epoch": 0.029392343365893824,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009988541235884506,
      "loss": 2.2771,
      "step": 4326
    },
    {
      "epoch": 0.02939913771248788,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009988533955171167,
      "loss": 2.5043,
      "step": 4327
    },
    {
      "epoch": 0.02940593205908194,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009988526672148196,
      "loss": 2.314,
      "step": 4328
    },
    {
      "epoch": 0.029412726405675996,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009988519386815592,
      "loss": 2.6101,
      "step": 4329
    },
    {
      "epoch": 0.029419520752270056,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009988512099173362,
      "loss": 2.2591,
      "step": 4330
    },
    {
      "epoch": 0.029426315098864113,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009988504809221505,
      "loss": 2.3667,
      "step": 4331
    },
    {
      "epoch": 0.02943310944545817,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009988497516960032,
      "loss": 2.2576,
      "step": 4332
    },
    {
      "epoch": 0.02943990379205223,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009988490222388939,
      "loss": 2.3315,
      "step": 4333
    },
    {
      "epoch": 0.029446698138646286,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009988482925508233,
      "loss": 2.303,
      "step": 4334
    },
    {
      "epoch": 0.029453492485240342,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009988475626317916,
      "loss": 2.3521,
      "step": 4335
    },
    {
      "epoch": 0.029460286831834402,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009988468324817993,
      "loss": 2.4824,
      "step": 4336
    },
    {
      "epoch": 0.02946708117842846,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009988461021008465,
      "loss": 2.4016,
      "step": 4337
    },
    {
      "epoch": 0.02947387552502252,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009988453714889337,
      "loss": 2.3951,
      "step": 4338
    },
    {
      "epoch": 0.029480669871616575,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009988446406460612,
      "loss": 2.318,
      "step": 4339
    },
    {
      "epoch": 0.02948746421821063,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009988439095722295,
      "loss": 2.5206,
      "step": 4340
    },
    {
      "epoch": 0.02949425856480469,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009988431782674387,
      "loss": 2.3827,
      "step": 4341
    },
    {
      "epoch": 0.029501052911398748,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009988424467316892,
      "loss": 2.3384,
      "step": 4342
    },
    {
      "epoch": 0.029507847257992804,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009988417149649814,
      "loss": 2.3521,
      "step": 4343
    },
    {
      "epoch": 0.029514641604586864,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009988409829673156,
      "loss": 2.3247,
      "step": 4344
    },
    {
      "epoch": 0.02952143595118092,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009988402507386921,
      "loss": 2.2722,
      "step": 4345
    },
    {
      "epoch": 0.029528230297774977,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009988395182791115,
      "loss": 2.428,
      "step": 4346
    },
    {
      "epoch": 0.029535024644369037,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009988387855885737,
      "loss": 2.3886,
      "step": 4347
    },
    {
      "epoch": 0.029541818990963094,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009988380526670792,
      "loss": 2.3856,
      "step": 4348
    },
    {
      "epoch": 0.029548613337557154,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009988373195146285,
      "loss": 2.3971,
      "step": 4349
    },
    {
      "epoch": 0.02955540768415121,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009988365861312218,
      "loss": 2.3919,
      "step": 4350
    },
    {
      "epoch": 0.029562202030745267,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009988358525168596,
      "loss": 2.5236,
      "step": 4351
    },
    {
      "epoch": 0.029568996377339327,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000998835118671542,
      "loss": 2.2331,
      "step": 4352
    },
    {
      "epoch": 0.029575790723933383,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009988343845952696,
      "loss": 2.419,
      "step": 4353
    },
    {
      "epoch": 0.02958258507052744,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009988336502880426,
      "loss": 2.4199,
      "step": 4354
    },
    {
      "epoch": 0.0295893794171215,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009988329157498612,
      "loss": 2.5259,
      "step": 4355
    },
    {
      "epoch": 0.029596173763715556,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000998832180980726,
      "loss": 2.3814,
      "step": 4356
    },
    {
      "epoch": 0.029602968110309612,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009988314459806371,
      "loss": 2.3422,
      "step": 4357
    },
    {
      "epoch": 0.029609762456903672,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998830710749595,
      "loss": 2.4172,
      "step": 4358
    },
    {
      "epoch": 0.02961655680349773,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009988299752876,
      "loss": 2.3776,
      "step": 4359
    },
    {
      "epoch": 0.02962335115009179,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009988292395946524,
      "loss": 2.2635,
      "step": 4360
    },
    {
      "epoch": 0.029630145496685845,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009988285036707527,
      "loss": 2.3604,
      "step": 4361
    },
    {
      "epoch": 0.029636939843279902,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998827767515901,
      "loss": 2.3913,
      "step": 4362
    },
    {
      "epoch": 0.029643734189873962,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009988270311300977,
      "loss": 2.3063,
      "step": 4363
    },
    {
      "epoch": 0.029650528536468018,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009988262945133434,
      "loss": 2.4262,
      "step": 4364
    },
    {
      "epoch": 0.029657322883062075,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009988255576656381,
      "loss": 2.2683,
      "step": 4365
    },
    {
      "epoch": 0.029664117229656135,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009988248205869824,
      "loss": 2.3929,
      "step": 4366
    },
    {
      "epoch": 0.02967091157625019,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009988240832773765,
      "loss": 2.3631,
      "step": 4367
    },
    {
      "epoch": 0.02967770592284425,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009988233457368207,
      "loss": 2.4677,
      "step": 4368
    },
    {
      "epoch": 0.029684500269438308,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009988226079653154,
      "loss": 2.5071,
      "step": 4369
    },
    {
      "epoch": 0.029691294616032364,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998821869962861,
      "loss": 2.3658,
      "step": 4370
    },
    {
      "epoch": 0.029698088962626424,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009988211317294579,
      "loss": 2.4823,
      "step": 4371
    },
    {
      "epoch": 0.02970488330922048,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009988203932651061,
      "loss": 2.3474,
      "step": 4372
    },
    {
      "epoch": 0.029711677655814537,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009988196545698063,
      "loss": 2.372,
      "step": 4373
    },
    {
      "epoch": 0.029718472002408597,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009988189156435589,
      "loss": 2.4614,
      "step": 4374
    },
    {
      "epoch": 0.029725266349002653,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009988181764863638,
      "loss": 2.3504,
      "step": 4375
    },
    {
      "epoch": 0.02973206069559671,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009988174370982217,
      "loss": 2.4075,
      "step": 4376
    },
    {
      "epoch": 0.02973885504219077,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009988166974791329,
      "loss": 2.1562,
      "step": 4377
    },
    {
      "epoch": 0.029745649388784826,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009988159576290975,
      "loss": 2.4392,
      "step": 4378
    },
    {
      "epoch": 0.029752443735378886,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009988152175481161,
      "loss": 2.1739,
      "step": 4379
    },
    {
      "epoch": 0.029759238081972943,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009988144772361891,
      "loss": 2.3379,
      "step": 4380
    },
    {
      "epoch": 0.029766032428567,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009988137366933165,
      "loss": 2.4686,
      "step": 4381
    },
    {
      "epoch": 0.02977282677516106,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009988129959194992,
      "loss": 2.3706,
      "step": 4382
    },
    {
      "epoch": 0.029779621121755116,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009988122549147369,
      "loss": 2.2537,
      "step": 4383
    },
    {
      "epoch": 0.029786415468349172,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009988115136790304,
      "loss": 2.3084,
      "step": 4384
    },
    {
      "epoch": 0.029793209814943232,
      "grad_norm": 0.1640625,
      "learning_rate": 0.00099881077221238,
      "loss": 2.4943,
      "step": 4385
    },
    {
      "epoch": 0.02980000416153729,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009988100305147855,
      "loss": 2.3917,
      "step": 4386
    },
    {
      "epoch": 0.029806798508131345,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998809288586248,
      "loss": 2.3452,
      "step": 4387
    },
    {
      "epoch": 0.029813592854725405,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009988085464267674,
      "loss": 2.3935,
      "step": 4388
    },
    {
      "epoch": 0.02982038720131946,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009988078040363442,
      "loss": 2.2634,
      "step": 4389
    },
    {
      "epoch": 0.02982718154791352,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009988070614149788,
      "loss": 2.4443,
      "step": 4390
    },
    {
      "epoch": 0.029833975894507578,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009988063185626713,
      "loss": 2.3084,
      "step": 4391
    },
    {
      "epoch": 0.029840770241101634,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009988055754794223,
      "loss": 2.2857,
      "step": 4392
    },
    {
      "epoch": 0.029847564587695694,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998804832165232,
      "loss": 2.3236,
      "step": 4393
    },
    {
      "epoch": 0.02985435893428975,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009988040886201007,
      "loss": 2.4432,
      "step": 4394
    },
    {
      "epoch": 0.029861153280883807,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998803344844029,
      "loss": 2.1525,
      "step": 4395
    },
    {
      "epoch": 0.029867947627477867,
      "grad_norm": 0.439453125,
      "learning_rate": 0.000998802600837017,
      "loss": 2.5483,
      "step": 4396
    },
    {
      "epoch": 0.029874741974071924,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000998801856599065,
      "loss": 2.2723,
      "step": 4397
    },
    {
      "epoch": 0.029881536320665984,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009988011121301738,
      "loss": 2.2667,
      "step": 4398
    },
    {
      "epoch": 0.02988833066726004,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009988003674303429,
      "loss": 2.356,
      "step": 4399
    },
    {
      "epoch": 0.029895125013854097,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009987996224995735,
      "loss": 2.4309,
      "step": 4400
    },
    {
      "epoch": 0.029901919360448156,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009987988773378655,
      "loss": 2.2957,
      "step": 4401
    },
    {
      "epoch": 0.029908713707042213,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009987981319452193,
      "loss": 2.4743,
      "step": 4402
    },
    {
      "epoch": 0.02991550805363627,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009987973863216353,
      "loss": 2.3529,
      "step": 4403
    },
    {
      "epoch": 0.02992230240023033,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987966404671137,
      "loss": 2.4987,
      "step": 4404
    },
    {
      "epoch": 0.029929096746824386,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009987958943816553,
      "loss": 2.2033,
      "step": 4405
    },
    {
      "epoch": 0.029935891093418442,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099879514806526,
      "loss": 2.3176,
      "step": 4406
    },
    {
      "epoch": 0.029942685440012502,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987944015179282,
      "loss": 2.374,
      "step": 4407
    },
    {
      "epoch": 0.02994947978660656,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009987936547396603,
      "loss": 2.2863,
      "step": 4408
    },
    {
      "epoch": 0.02995627413320062,
      "grad_norm": 0.1796875,
      "learning_rate": 0.0009987929077304564,
      "loss": 2.3364,
      "step": 4409
    },
    {
      "epoch": 0.029963068479794675,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009987921604903177,
      "loss": 2.4919,
      "step": 4410
    },
    {
      "epoch": 0.02996986282638873,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009987914130192435,
      "loss": 2.4153,
      "step": 4411
    },
    {
      "epoch": 0.02997665717298279,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987906653172347,
      "loss": 2.3486,
      "step": 4412
    },
    {
      "epoch": 0.029983451519576848,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987899173842915,
      "loss": 2.3043,
      "step": 4413
    },
    {
      "epoch": 0.029990245866170905,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009987891692204143,
      "loss": 2.2389,
      "step": 4414
    },
    {
      "epoch": 0.029997040212764964,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009987884208256036,
      "loss": 2.451,
      "step": 4415
    },
    {
      "epoch": 0.03000383455935902,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009987876721998595,
      "loss": 2.5056,
      "step": 4416
    },
    {
      "epoch": 0.030010628905953077,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009987869233431823,
      "loss": 2.5341,
      "step": 4417
    },
    {
      "epoch": 0.030017423252547137,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009987861742555726,
      "loss": 2.2936,
      "step": 4418
    },
    {
      "epoch": 0.030024217599141194,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009987854249370305,
      "loss": 2.3675,
      "step": 4419
    },
    {
      "epoch": 0.030031011945735254,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987846753875564,
      "loss": 2.5402,
      "step": 4420
    },
    {
      "epoch": 0.03003780629232931,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009987839256071509,
      "loss": 2.5657,
      "step": 4421
    },
    {
      "epoch": 0.030044600638923367,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009987831755958142,
      "loss": 2.374,
      "step": 4422
    },
    {
      "epoch": 0.030051394985517427,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987824253535464,
      "loss": 2.3525,
      "step": 4423
    },
    {
      "epoch": 0.030058189332111483,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009987816748803482,
      "loss": 2.2554,
      "step": 4424
    },
    {
      "epoch": 0.03006498367870554,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987809241762198,
      "loss": 2.3858,
      "step": 4425
    },
    {
      "epoch": 0.0300717780252996,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009987801732411618,
      "loss": 2.3042,
      "step": 4426
    },
    {
      "epoch": 0.030078572371893656,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009987794220751738,
      "loss": 2.2606,
      "step": 4427
    },
    {
      "epoch": 0.030085366718487716,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998778670678257,
      "loss": 2.4176,
      "step": 4428
    },
    {
      "epoch": 0.030092161065081773,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009987779190504112,
      "loss": 2.3368,
      "step": 4429
    },
    {
      "epoch": 0.03009895541167583,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000998777167191637,
      "loss": 2.3894,
      "step": 4430
    },
    {
      "epoch": 0.03010574975826989,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009987764151019346,
      "loss": 2.4881,
      "step": 4431
    },
    {
      "epoch": 0.030112544104863945,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009987756627813046,
      "loss": 2.3996,
      "step": 4432
    },
    {
      "epoch": 0.030119338451458002,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009987749102297473,
      "loss": 2.4263,
      "step": 4433
    },
    {
      "epoch": 0.030126132798052062,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009987741574472627,
      "loss": 2.3963,
      "step": 4434
    },
    {
      "epoch": 0.03013292714464612,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987734044338513,
      "loss": 2.4047,
      "step": 4435
    },
    {
      "epoch": 0.030139721491240175,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987726511895138,
      "loss": 2.2383,
      "step": 4436
    },
    {
      "epoch": 0.030146515837834235,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009987718977142502,
      "loss": 2.2155,
      "step": 4437
    },
    {
      "epoch": 0.03015331018442829,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009987711440080608,
      "loss": 2.2859,
      "step": 4438
    },
    {
      "epoch": 0.03016010453102235,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009987703900709462,
      "loss": 2.2006,
      "step": 4439
    },
    {
      "epoch": 0.030166898877616408,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009987696359029065,
      "loss": 2.5297,
      "step": 4440
    },
    {
      "epoch": 0.030173693224210464,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009987688815039422,
      "loss": 2.4404,
      "step": 4441
    },
    {
      "epoch": 0.030180487570804524,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998768126874054,
      "loss": 2.3777,
      "step": 4442
    },
    {
      "epoch": 0.03018728191739858,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987673720132414,
      "loss": 2.505,
      "step": 4443
    },
    {
      "epoch": 0.030194076263992637,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009987666169215053,
      "loss": 2.4813,
      "step": 4444
    },
    {
      "epoch": 0.030200870610586697,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000998765861598846,
      "loss": 2.3518,
      "step": 4445
    },
    {
      "epoch": 0.030207664957180753,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009987651060452641,
      "loss": 2.2391,
      "step": 4446
    },
    {
      "epoch": 0.03021445930377481,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009987643502607593,
      "loss": 2.3769,
      "step": 4447
    },
    {
      "epoch": 0.03022125365036887,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009987635942453326,
      "loss": 2.3715,
      "step": 4448
    },
    {
      "epoch": 0.030228047996962926,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987628379989838,
      "loss": 2.4271,
      "step": 4449
    },
    {
      "epoch": 0.030234842343556986,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009987620815217137,
      "loss": 2.3758,
      "step": 4450
    },
    {
      "epoch": 0.030241636690151043,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009987613248135226,
      "loss": 2.3484,
      "step": 4451
    },
    {
      "epoch": 0.0302484310367451,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009987605678744105,
      "loss": 2.3498,
      "step": 4452
    },
    {
      "epoch": 0.03025522538333916,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998759810704378,
      "loss": 2.4591,
      "step": 4453
    },
    {
      "epoch": 0.030262019729933216,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009987590533034256,
      "loss": 2.3362,
      "step": 4454
    },
    {
      "epoch": 0.030268814076527272,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009987582956715532,
      "loss": 2.2889,
      "step": 4455
    },
    {
      "epoch": 0.030275608423121332,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987575378087616,
      "loss": 2.4715,
      "step": 4456
    },
    {
      "epoch": 0.03028240276971539,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987567797150512,
      "loss": 2.3756,
      "step": 4457
    },
    {
      "epoch": 0.03028919711630945,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000998756021390422,
      "loss": 2.4601,
      "step": 4458
    },
    {
      "epoch": 0.030295991462903505,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009987552628348741,
      "loss": 2.35,
      "step": 4459
    },
    {
      "epoch": 0.03030278580949756,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009987545040484086,
      "loss": 2.5493,
      "step": 4460
    },
    {
      "epoch": 0.03030958015609162,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009987537450310254,
      "loss": 2.4181,
      "step": 4461
    },
    {
      "epoch": 0.030316374502685678,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009987529857827249,
      "loss": 2.4239,
      "step": 4462
    },
    {
      "epoch": 0.030323168849279734,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009987522263035075,
      "loss": 2.4972,
      "step": 4463
    },
    {
      "epoch": 0.030329963195873794,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009987514665933735,
      "loss": 2.3705,
      "step": 4464
    },
    {
      "epoch": 0.03033675754246785,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009987507066523233,
      "loss": 2.4227,
      "step": 4465
    },
    {
      "epoch": 0.030343551889061907,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987499464803574,
      "loss": 2.3452,
      "step": 4466
    },
    {
      "epoch": 0.030350346235655967,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009987491860774759,
      "loss": 2.4116,
      "step": 4467
    },
    {
      "epoch": 0.030357140582250024,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009987484254436792,
      "loss": 2.2221,
      "step": 4468
    },
    {
      "epoch": 0.030363934928844084,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009987476645789677,
      "loss": 2.3337,
      "step": 4469
    },
    {
      "epoch": 0.03037072927543814,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009987469034833418,
      "loss": 2.3749,
      "step": 4470
    },
    {
      "epoch": 0.030377523622032197,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009987461421568017,
      "loss": 2.321,
      "step": 4471
    },
    {
      "epoch": 0.030384317968626257,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998745380599348,
      "loss": 2.3386,
      "step": 4472
    },
    {
      "epoch": 0.030391112315220313,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987446188109807,
      "loss": 2.365,
      "step": 4473
    },
    {
      "epoch": 0.03039790666181437,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009987438567917006,
      "loss": 2.4357,
      "step": 4474
    },
    {
      "epoch": 0.03040470100840843,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009987430945415077,
      "loss": 2.3998,
      "step": 4475
    },
    {
      "epoch": 0.030411495355002486,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009987423320604023,
      "loss": 2.2105,
      "step": 4476
    },
    {
      "epoch": 0.030418289701596542,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009987415693483853,
      "loss": 2.3948,
      "step": 4477
    },
    {
      "epoch": 0.030425084048190602,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009987408064054564,
      "loss": 2.2823,
      "step": 4478
    },
    {
      "epoch": 0.03043187839478466,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009987400432316161,
      "loss": 2.4394,
      "step": 4479
    },
    {
      "epoch": 0.03043867274137872,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000998739279826865,
      "loss": 2.2598,
      "step": 4480
    },
    {
      "epoch": 0.030445467087972775,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009987385161912034,
      "loss": 2.3442,
      "step": 4481
    },
    {
      "epoch": 0.030452261434566832,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009987377523246316,
      "loss": 2.3769,
      "step": 4482
    },
    {
      "epoch": 0.03045905578116089,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00099873698822715,
      "loss": 2.2869,
      "step": 4483
    },
    {
      "epoch": 0.030465850127754948,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009987362238987588,
      "loss": 2.2735,
      "step": 4484
    },
    {
      "epoch": 0.030472644474349005,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987354593394583,
      "loss": 2.3251,
      "step": 4485
    },
    {
      "epoch": 0.030479438820943065,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009987346945492493,
      "loss": 2.3446,
      "step": 4486
    },
    {
      "epoch": 0.03048623316753712,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987339295281317,
      "loss": 2.525,
      "step": 4487
    },
    {
      "epoch": 0.03049302751413118,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987331642761059,
      "loss": 2.3848,
      "step": 4488
    },
    {
      "epoch": 0.030499821860725237,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009987323987931725,
      "loss": 2.3122,
      "step": 4489
    },
    {
      "epoch": 0.030506616207319294,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987316330793317,
      "loss": 2.3809,
      "step": 4490
    },
    {
      "epoch": 0.030513410553913354,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009987308671345837,
      "loss": 2.2525,
      "step": 4491
    },
    {
      "epoch": 0.03052020490050741,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009987301009589293,
      "loss": 2.4207,
      "step": 4492
    },
    {
      "epoch": 0.030526999247101467,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009987293345523685,
      "loss": 2.2898,
      "step": 4493
    },
    {
      "epoch": 0.030533793593695527,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009987285679149017,
      "loss": 2.4213,
      "step": 4494
    },
    {
      "epoch": 0.030540587940289583,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009987278010465293,
      "loss": 2.2832,
      "step": 4495
    },
    {
      "epoch": 0.03054738228688364,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009987270339472515,
      "loss": 2.4406,
      "step": 4496
    },
    {
      "epoch": 0.0305541766334777,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998726266617069,
      "loss": 2.3532,
      "step": 4497
    },
    {
      "epoch": 0.030560970980071756,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987254990559819,
      "loss": 2.3982,
      "step": 4498
    },
    {
      "epoch": 0.030567765326665816,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009987247312639906,
      "loss": 2.2443,
      "step": 4499
    },
    {
      "epoch": 0.030574559673259873,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009987239632410956,
      "loss": 2.4094,
      "step": 4500
    },
    {
      "epoch": 0.03058135401985393,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009987231949872971,
      "loss": 2.5173,
      "step": 4501
    },
    {
      "epoch": 0.03058814836644799,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009987224265025952,
      "loss": 2.3305,
      "step": 4502
    },
    {
      "epoch": 0.030594942713042046,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009987216577869908,
      "loss": 2.3763,
      "step": 4503
    },
    {
      "epoch": 0.030601737059636102,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009987208888404839,
      "loss": 2.3332,
      "step": 4504
    },
    {
      "epoch": 0.030608531406230162,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009987201196630748,
      "loss": 2.2635,
      "step": 4505
    },
    {
      "epoch": 0.03061532575282422,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009987193502547643,
      "loss": 2.4001,
      "step": 4506
    },
    {
      "epoch": 0.030622120099418275,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009987185806155524,
      "loss": 2.4079,
      "step": 4507
    },
    {
      "epoch": 0.030628914446012335,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009987178107454394,
      "loss": 2.2697,
      "step": 4508
    },
    {
      "epoch": 0.03063570879260639,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009987170406444256,
      "loss": 2.3979,
      "step": 4509
    },
    {
      "epoch": 0.03064250313920045,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009987162703125117,
      "loss": 2.4661,
      "step": 4510
    },
    {
      "epoch": 0.030649297485794508,
      "grad_norm": 0.181640625,
      "learning_rate": 0.000998715499749698,
      "loss": 2.3608,
      "step": 4511
    },
    {
      "epoch": 0.030656091832388564,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009987147289559845,
      "loss": 2.3767,
      "step": 4512
    },
    {
      "epoch": 0.030662886178982624,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998713957931372,
      "loss": 2.2012,
      "step": 4513
    },
    {
      "epoch": 0.03066968052557668,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009987131866758608,
      "loss": 2.3769,
      "step": 4514
    },
    {
      "epoch": 0.030676474872170737,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009987124151894509,
      "loss": 2.3625,
      "step": 4515
    },
    {
      "epoch": 0.030683269218764797,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009987116434721428,
      "loss": 2.37,
      "step": 4516
    },
    {
      "epoch": 0.030690063565358854,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998710871523937,
      "loss": 2.1952,
      "step": 4517
    },
    {
      "epoch": 0.030696857911952913,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009987100993448337,
      "loss": 2.2333,
      "step": 4518
    },
    {
      "epoch": 0.03070365225854697,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009987093269348336,
      "loss": 2.3002,
      "step": 4519
    },
    {
      "epoch": 0.030710446605141026,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009987085542939365,
      "loss": 2.2231,
      "step": 4520
    },
    {
      "epoch": 0.030717240951735086,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009987077814221433,
      "loss": 2.3633,
      "step": 4521
    },
    {
      "epoch": 0.030724035298329143,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998707008319454,
      "loss": 2.3967,
      "step": 4522
    },
    {
      "epoch": 0.0307308296449232,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000998706234985869,
      "loss": 2.2638,
      "step": 4523
    },
    {
      "epoch": 0.03073762399151726,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009987054614213889,
      "loss": 2.3888,
      "step": 4524
    },
    {
      "epoch": 0.030744418338111316,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009987046876260136,
      "loss": 2.2699,
      "step": 4525
    },
    {
      "epoch": 0.030751212684705372,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998703913599744,
      "loss": 2.3236,
      "step": 4526
    },
    {
      "epoch": 0.030758007031299432,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00099870313934258,
      "loss": 2.4053,
      "step": 4527
    },
    {
      "epoch": 0.03076480137789349,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987023648545224,
      "loss": 2.3729,
      "step": 4528
    },
    {
      "epoch": 0.03077159572448755,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009987015901355713,
      "loss": 2.47,
      "step": 4529
    },
    {
      "epoch": 0.030778390071081605,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998700815185727,
      "loss": 2.3505,
      "step": 4530
    },
    {
      "epoch": 0.03078518441767566,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00099870004000499,
      "loss": 2.2444,
      "step": 4531
    },
    {
      "epoch": 0.03079197876426972,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009986992645933605,
      "loss": 2.3767,
      "step": 4532
    },
    {
      "epoch": 0.030798773110863778,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009986984889508391,
      "loss": 2.3891,
      "step": 4533
    },
    {
      "epoch": 0.030805567457457834,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009986977130774259,
      "loss": 2.2975,
      "step": 4534
    },
    {
      "epoch": 0.030812361804051894,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009986969369731214,
      "loss": 2.3123,
      "step": 4535
    },
    {
      "epoch": 0.03081915615064595,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009986961606379258,
      "loss": 2.4087,
      "step": 4536
    },
    {
      "epoch": 0.030825950497240007,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009986953840718398,
      "loss": 2.4848,
      "step": 4537
    },
    {
      "epoch": 0.030832744843834067,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009986946072748635,
      "loss": 2.44,
      "step": 4538
    },
    {
      "epoch": 0.030839539190428124,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009986938302469974,
      "loss": 2.4246,
      "step": 4539
    },
    {
      "epoch": 0.030846333537022184,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009986930529882415,
      "loss": 2.3622,
      "step": 4540
    },
    {
      "epoch": 0.03085312788361624,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009986922754985968,
      "loss": 2.2181,
      "step": 4541
    },
    {
      "epoch": 0.030859922230210297,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998691497778063,
      "loss": 2.4357,
      "step": 4542
    },
    {
      "epoch": 0.030866716576804357,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009986907198266408,
      "loss": 2.4238,
      "step": 4543
    },
    {
      "epoch": 0.030873510923398413,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009986899416443307,
      "loss": 2.2067,
      "step": 4544
    },
    {
      "epoch": 0.03088030526999247,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009986891632311327,
      "loss": 2.3862,
      "step": 4545
    },
    {
      "epoch": 0.03088709961658653,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009986883845870473,
      "loss": 2.2923,
      "step": 4546
    },
    {
      "epoch": 0.030893893963180586,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009986876057120752,
      "loss": 2.2789,
      "step": 4547
    },
    {
      "epoch": 0.030900688309774646,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009986868266062161,
      "loss": 2.5179,
      "step": 4548
    },
    {
      "epoch": 0.030907482656368702,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009986860472694708,
      "loss": 2.41,
      "step": 4549
    },
    {
      "epoch": 0.03091427700296276,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009986852677018397,
      "loss": 2.4496,
      "step": 4550
    },
    {
      "epoch": 0.03092107134955682,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000998684487903323,
      "loss": 2.4938,
      "step": 4551
    },
    {
      "epoch": 0.030927865696150875,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009986837078739211,
      "loss": 2.4501,
      "step": 4552
    },
    {
      "epoch": 0.030934660042744932,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009986829276136342,
      "loss": 2.3876,
      "step": 4553
    },
    {
      "epoch": 0.030941454389338992,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998682147122463,
      "loss": 2.4294,
      "step": 4554
    },
    {
      "epoch": 0.030948248735933048,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009986813664004076,
      "loss": 2.1455,
      "step": 4555
    },
    {
      "epoch": 0.030955043082527105,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009986805854474683,
      "loss": 2.194,
      "step": 4556
    },
    {
      "epoch": 0.030961837429121165,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009986798042636458,
      "loss": 2.3819,
      "step": 4557
    },
    {
      "epoch": 0.03096863177571522,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009986790228489403,
      "loss": 2.3464,
      "step": 4558
    },
    {
      "epoch": 0.03097542612230928,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009986782412033522,
      "loss": 2.4459,
      "step": 4559
    },
    {
      "epoch": 0.030982220468903338,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009986774593268815,
      "loss": 2.3214,
      "step": 4560
    },
    {
      "epoch": 0.030989014815497394,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009986766772195291,
      "loss": 2.301,
      "step": 4561
    },
    {
      "epoch": 0.030995809162091454,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998675894881295,
      "loss": 2.363,
      "step": 4562
    },
    {
      "epoch": 0.03100260350868551,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009986751123121798,
      "loss": 2.5105,
      "step": 4563
    },
    {
      "epoch": 0.031009397855279567,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009986743295121836,
      "loss": 2.3947,
      "step": 4564
    },
    {
      "epoch": 0.031016192201873627,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998673546481307,
      "loss": 2.2529,
      "step": 4565
    },
    {
      "epoch": 0.031022986548467683,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009986727632195502,
      "loss": 2.4977,
      "step": 4566
    },
    {
      "epoch": 0.03102978089506174,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009986719797269138,
      "loss": 2.4358,
      "step": 4567
    },
    {
      "epoch": 0.0310365752416558,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009986711960033976,
      "loss": 2.3065,
      "step": 4568
    },
    {
      "epoch": 0.031043369588249856,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009986704120490027,
      "loss": 2.3,
      "step": 4569
    },
    {
      "epoch": 0.031050163934843916,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998669627863729,
      "loss": 2.4657,
      "step": 4570
    },
    {
      "epoch": 0.031056958281437973,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000998668843447577,
      "loss": 2.4694,
      "step": 4571
    },
    {
      "epoch": 0.03106375262803203,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000998668058800547,
      "loss": 2.3617,
      "step": 4572
    },
    {
      "epoch": 0.03107054697462609,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009986672739226395,
      "loss": 2.5132,
      "step": 4573
    },
    {
      "epoch": 0.031077341321220146,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009986664888138547,
      "loss": 2.3706,
      "step": 4574
    },
    {
      "epoch": 0.031084135667814202,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998665703474193,
      "loss": 2.2068,
      "step": 4575
    },
    {
      "epoch": 0.031090930014408262,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009986649179036548,
      "loss": 2.2957,
      "step": 4576
    },
    {
      "epoch": 0.03109772436100232,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009986641321022404,
      "loss": 2.4135,
      "step": 4577
    },
    {
      "epoch": 0.03110451870759638,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009986633460699504,
      "loss": 2.2949,
      "step": 4578
    },
    {
      "epoch": 0.031111313054190435,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000998662559806785,
      "loss": 2.4617,
      "step": 4579
    },
    {
      "epoch": 0.03111810740078449,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009986617733127444,
      "loss": 2.2276,
      "step": 4580
    },
    {
      "epoch": 0.03112490174737855,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009986609865878292,
      "loss": 2.4223,
      "step": 4581
    },
    {
      "epoch": 0.031131696093972608,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009986601996320396,
      "loss": 2.378,
      "step": 4582
    },
    {
      "epoch": 0.031138490440566664,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998659412445376,
      "loss": 2.3657,
      "step": 4583
    },
    {
      "epoch": 0.031145284787160724,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000998658625027839,
      "loss": 2.3522,
      "step": 4584
    },
    {
      "epoch": 0.03115207913375478,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009986578373794286,
      "loss": 2.4221,
      "step": 4585
    },
    {
      "epoch": 0.031158873480348837,
      "grad_norm": 0.205078125,
      "learning_rate": 0.0009986570495001455,
      "loss": 2.5577,
      "step": 4586
    },
    {
      "epoch": 0.031165667826942897,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009986562613899898,
      "loss": 2.2033,
      "step": 4587
    },
    {
      "epoch": 0.031172462173536954,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009986554730489622,
      "loss": 2.2735,
      "step": 4588
    },
    {
      "epoch": 0.031179256520131014,
      "grad_norm": 0.1904296875,
      "learning_rate": 0.0009986546844770626,
      "loss": 2.2566,
      "step": 4589
    },
    {
      "epoch": 0.03118605086672507,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009986538956742914,
      "loss": 2.3273,
      "step": 4590
    },
    {
      "epoch": 0.031192845213319127,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009986531066406494,
      "loss": 2.3575,
      "step": 4591
    },
    {
      "epoch": 0.031199639559913186,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998652317376137,
      "loss": 2.2874,
      "step": 4592
    },
    {
      "epoch": 0.031206433906507243,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009986515278807538,
      "loss": 2.3374,
      "step": 4593
    },
    {
      "epoch": 0.0312132282531013,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009986507381545009,
      "loss": 2.5236,
      "step": 4594
    },
    {
      "epoch": 0.03122002259969536,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009986499481973784,
      "loss": 2.4295,
      "step": 4595
    },
    {
      "epoch": 0.031226816946289416,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009986491580093865,
      "loss": 2.3587,
      "step": 4596
    },
    {
      "epoch": 0.031233611292883472,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000998648367590526,
      "loss": 2.5166,
      "step": 4597
    },
    {
      "epoch": 0.031240405639477532,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000998647576940797,
      "loss": 2.2641,
      "step": 4598
    },
    {
      "epoch": 0.03124719998607159,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009986467860602,
      "loss": 2.2178,
      "step": 4599
    },
    {
      "epoch": 0.031253994332665645,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998645994948735,
      "loss": 2.3643,
      "step": 4600
    },
    {
      "epoch": 0.0312607886792597,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009986452036064026,
      "loss": 2.2942,
      "step": 4601
    },
    {
      "epoch": 0.031267583025853765,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009986444120332035,
      "loss": 2.3564,
      "step": 4602
    },
    {
      "epoch": 0.03127437737244782,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009986436202291374,
      "loss": 2.1914,
      "step": 4603
    },
    {
      "epoch": 0.03128117171904188,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009986428281942052,
      "loss": 2.3407,
      "step": 4604
    },
    {
      "epoch": 0.031287966065635935,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009986420359284073,
      "loss": 2.3981,
      "step": 4605
    },
    {
      "epoch": 0.03129476041222999,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009986412434317435,
      "loss": 2.3877,
      "step": 4606
    },
    {
      "epoch": 0.031301554758824054,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009986404507042147,
      "loss": 2.4201,
      "step": 4607
    },
    {
      "epoch": 0.03130834910541811,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000998639657745821,
      "loss": 2.4729,
      "step": 4608
    },
    {
      "epoch": 0.03131514345201217,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009986388645565629,
      "loss": 2.2514,
      "step": 4609
    },
    {
      "epoch": 0.031321937798606224,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009986380711364407,
      "loss": 2.3965,
      "step": 4610
    },
    {
      "epoch": 0.03132873214520028,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009986372774854547,
      "loss": 2.2665,
      "step": 4611
    },
    {
      "epoch": 0.03133552649179434,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009986364836036055,
      "loss": 2.3323,
      "step": 4612
    },
    {
      "epoch": 0.0313423208383884,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009986356894908933,
      "loss": 2.3346,
      "step": 4613
    },
    {
      "epoch": 0.03134911518498246,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009986348951473184,
      "loss": 2.5207,
      "step": 4614
    },
    {
      "epoch": 0.03135590953157651,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009986341005728814,
      "loss": 2.2602,
      "step": 4615
    },
    {
      "epoch": 0.03136270387817057,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009986333057675826,
      "loss": 2.3868,
      "step": 4616
    },
    {
      "epoch": 0.031369498224764626,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998632510731422,
      "loss": 2.3736,
      "step": 4617
    },
    {
      "epoch": 0.03137629257135869,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009986317154644004,
      "loss": 2.4387,
      "step": 4618
    },
    {
      "epoch": 0.031383086917952746,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998630919966518,
      "loss": 2.464,
      "step": 4619
    },
    {
      "epoch": 0.0313898812645468,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009986301242377754,
      "loss": 2.2,
      "step": 4620
    },
    {
      "epoch": 0.03139667561114086,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009986293282781727,
      "loss": 2.1498,
      "step": 4621
    },
    {
      "epoch": 0.031403469957734916,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009986285320877101,
      "loss": 2.2964,
      "step": 4622
    },
    {
      "epoch": 0.03141026430432898,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009986277356663885,
      "loss": 2.3381,
      "step": 4623
    },
    {
      "epoch": 0.031417058650923035,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009986269390142078,
      "loss": 2.2843,
      "step": 4624
    },
    {
      "epoch": 0.03142385299751709,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009986261421311685,
      "loss": 2.4177,
      "step": 4625
    },
    {
      "epoch": 0.03143064734411115,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998625345017271,
      "loss": 2.5433,
      "step": 4626
    },
    {
      "epoch": 0.031437441690705205,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009986245476725159,
      "loss": 2.4976,
      "step": 4627
    },
    {
      "epoch": 0.03144423603729926,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009986237500969033,
      "loss": 2.2274,
      "step": 4628
    },
    {
      "epoch": 0.031451030383893325,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009986229522904336,
      "loss": 2.3996,
      "step": 4629
    },
    {
      "epoch": 0.03145782473048738,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000998622154253107,
      "loss": 2.414,
      "step": 4630
    },
    {
      "epoch": 0.03146461907708144,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009986213559849241,
      "loss": 2.4731,
      "step": 4631
    },
    {
      "epoch": 0.031471413423675494,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009986205574858856,
      "loss": 2.3352,
      "step": 4632
    },
    {
      "epoch": 0.03147820777026955,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998619758755991,
      "loss": 2.4096,
      "step": 4633
    },
    {
      "epoch": 0.031485002116863614,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009986189597952415,
      "loss": 2.2859,
      "step": 4634
    },
    {
      "epoch": 0.03149179646345767,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000998618160603637,
      "loss": 2.4107,
      "step": 4635
    },
    {
      "epoch": 0.03149859081005173,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998617361181178,
      "loss": 2.3111,
      "step": 4636
    },
    {
      "epoch": 0.031505385156645783,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009986165615278651,
      "loss": 2.4443,
      "step": 4637
    },
    {
      "epoch": 0.03151217950323984,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009986157616436982,
      "loss": 2.4207,
      "step": 4638
    },
    {
      "epoch": 0.031518973849833896,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998614961528678,
      "loss": 2.3893,
      "step": 4639
    },
    {
      "epoch": 0.03152576819642796,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009986141611828048,
      "loss": 2.4356,
      "step": 4640
    },
    {
      "epoch": 0.031532562543022016,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998613360606079,
      "loss": 2.424,
      "step": 4641
    },
    {
      "epoch": 0.03153935688961607,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009986125597985008,
      "loss": 2.2943,
      "step": 4642
    },
    {
      "epoch": 0.03154615123621013,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009986117587600708,
      "loss": 2.4244,
      "step": 4643
    },
    {
      "epoch": 0.031552945582804186,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009986109574907892,
      "loss": 2.6313,
      "step": 4644
    },
    {
      "epoch": 0.03155973992939825,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009986101559906566,
      "loss": 2.2925,
      "step": 4645
    },
    {
      "epoch": 0.031566534275992306,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998609354259673,
      "loss": 2.3568,
      "step": 4646
    },
    {
      "epoch": 0.03157332862258636,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009986085522978392,
      "loss": 2.4792,
      "step": 4647
    },
    {
      "epoch": 0.03158012296918042,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009986077501051552,
      "loss": 2.4491,
      "step": 4648
    },
    {
      "epoch": 0.031586917315774475,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009986069476816215,
      "loss": 2.4121,
      "step": 4649
    },
    {
      "epoch": 0.03159371166236853,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009986061450272387,
      "loss": 2.5174,
      "step": 4650
    },
    {
      "epoch": 0.031600506008962595,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009986053421420066,
      "loss": 2.3691,
      "step": 4651
    },
    {
      "epoch": 0.03160730035555665,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009986045390259263,
      "loss": 2.3558,
      "step": 4652
    },
    {
      "epoch": 0.03161409470215071,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009986037356789976,
      "loss": 2.3427,
      "step": 4653
    },
    {
      "epoch": 0.031620889048744764,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009986029321012214,
      "loss": 2.3601,
      "step": 4654
    },
    {
      "epoch": 0.03162768339533882,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009986021282925974,
      "loss": 2.4838,
      "step": 4655
    },
    {
      "epoch": 0.031634477741932884,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009986013242531265,
      "loss": 2.358,
      "step": 4656
    },
    {
      "epoch": 0.03164127208852694,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009986005199828089,
      "loss": 2.3595,
      "step": 4657
    },
    {
      "epoch": 0.031648066435121,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009985997154816448,
      "loss": 2.3036,
      "step": 4658
    },
    {
      "epoch": 0.031654860781715054,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009985989107496348,
      "loss": 2.4245,
      "step": 4659
    },
    {
      "epoch": 0.03166165512830911,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009985981057867794,
      "loss": 2.2414,
      "step": 4660
    },
    {
      "epoch": 0.03166844947490317,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985973005930788,
      "loss": 2.5643,
      "step": 4661
    },
    {
      "epoch": 0.03167524382149723,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009985964951685334,
      "loss": 2.334,
      "step": 4662
    },
    {
      "epoch": 0.03168203816809129,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009985956895131433,
      "loss": 2.4159,
      "step": 4663
    },
    {
      "epoch": 0.03168883251468534,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985948836269093,
      "loss": 2.4688,
      "step": 4664
    },
    {
      "epoch": 0.0316956268612794,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009985940775098315,
      "loss": 2.4203,
      "step": 4665
    },
    {
      "epoch": 0.031702421207873456,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009985932711619103,
      "loss": 2.2873,
      "step": 4666
    },
    {
      "epoch": 0.03170921555446752,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009985924645831464,
      "loss": 2.3749,
      "step": 4667
    },
    {
      "epoch": 0.031716009901061576,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009985916577735397,
      "loss": 2.5636,
      "step": 4668
    },
    {
      "epoch": 0.03172280424765563,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985908507330907,
      "loss": 2.3457,
      "step": 4669
    },
    {
      "epoch": 0.03172959859424969,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009985900434617999,
      "loss": 2.3497,
      "step": 4670
    },
    {
      "epoch": 0.031736392940843745,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009985892359596676,
      "loss": 2.3361,
      "step": 4671
    },
    {
      "epoch": 0.0317431872874378,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009985884282266945,
      "loss": 2.2168,
      "step": 4672
    },
    {
      "epoch": 0.031749981634031865,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985876202628803,
      "loss": 2.4229,
      "step": 4673
    },
    {
      "epoch": 0.03175677598062592,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998586812068226,
      "loss": 2.1961,
      "step": 4674
    },
    {
      "epoch": 0.03176357032721998,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009985860036427314,
      "loss": 2.279,
      "step": 4675
    },
    {
      "epoch": 0.031770364673814035,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009985851949863977,
      "loss": 2.3657,
      "step": 4676
    },
    {
      "epoch": 0.03177715902040809,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009985843860992243,
      "loss": 2.4736,
      "step": 4677
    },
    {
      "epoch": 0.031783953367002155,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998583576981212,
      "loss": 2.3684,
      "step": 4678
    },
    {
      "epoch": 0.03179074771359621,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009985827676323616,
      "loss": 2.2786,
      "step": 4679
    },
    {
      "epoch": 0.03179754206019027,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998581958052673,
      "loss": 2.3708,
      "step": 4680
    },
    {
      "epoch": 0.031804336406784324,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009985811482421465,
      "loss": 2.3251,
      "step": 4681
    },
    {
      "epoch": 0.03181113075337838,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000998580338200783,
      "loss": 2.3075,
      "step": 4682
    },
    {
      "epoch": 0.031817925099972444,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000998579527928582,
      "loss": 2.5029,
      "step": 4683
    },
    {
      "epoch": 0.0318247194465665,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009985787174255447,
      "loss": 2.3952,
      "step": 4684
    },
    {
      "epoch": 0.03183151379316056,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009985779066916712,
      "loss": 2.4211,
      "step": 4685
    },
    {
      "epoch": 0.03183830813975461,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985770957269618,
      "loss": 2.2979,
      "step": 4686
    },
    {
      "epoch": 0.03184510248634867,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009985762845314168,
      "loss": 2.3075,
      "step": 4687
    },
    {
      "epoch": 0.031851896832942726,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000998575473105037,
      "loss": 2.3945,
      "step": 4688
    },
    {
      "epoch": 0.03185869117953679,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009985746614478222,
      "loss": 2.4103,
      "step": 4689
    },
    {
      "epoch": 0.031865485526130846,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009985738495597732,
      "loss": 2.2649,
      "step": 4690
    },
    {
      "epoch": 0.0318722798727249,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00099857303744089,
      "loss": 2.4617,
      "step": 4691
    },
    {
      "epoch": 0.03187907421931896,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009985722250911732,
      "loss": 2.3151,
      "step": 4692
    },
    {
      "epoch": 0.031885868565913016,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009985714125106235,
      "loss": 2.4231,
      "step": 4693
    },
    {
      "epoch": 0.03189266291250708,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009985705996992406,
      "loss": 2.4487,
      "step": 4694
    },
    {
      "epoch": 0.031899457259101135,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009985697866570255,
      "loss": 2.2519,
      "step": 4695
    },
    {
      "epoch": 0.03190625160569519,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009985689733839783,
      "loss": 2.4015,
      "step": 4696
    },
    {
      "epoch": 0.03191304595228925,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009985681598800992,
      "loss": 2.389,
      "step": 4697
    },
    {
      "epoch": 0.031919840298883305,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009985673461453887,
      "loss": 2.3761,
      "step": 4698
    },
    {
      "epoch": 0.03192663464547736,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009985665321798477,
      "loss": 2.3458,
      "step": 4699
    },
    {
      "epoch": 0.031933428992071425,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009985657179834757,
      "loss": 2.2543,
      "step": 4700
    },
    {
      "epoch": 0.03194022333866548,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009985649035562736,
      "loss": 2.2833,
      "step": 4701
    },
    {
      "epoch": 0.03194701768525954,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985640888982414,
      "loss": 2.3216,
      "step": 4702
    },
    {
      "epoch": 0.031953812031853594,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009985632740093801,
      "loss": 2.383,
      "step": 4703
    },
    {
      "epoch": 0.03196060637844765,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009985624588896896,
      "loss": 2.3498,
      "step": 4704
    },
    {
      "epoch": 0.031967400725041714,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009985616435391705,
      "loss": 2.3443,
      "step": 4705
    },
    {
      "epoch": 0.03197419507163577,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009985608279578229,
      "loss": 2.3597,
      "step": 4706
    },
    {
      "epoch": 0.03198098941822983,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009985600121456475,
      "loss": 2.3413,
      "step": 4707
    },
    {
      "epoch": 0.031987783764823884,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009985591961026443,
      "loss": 2.3976,
      "step": 4708
    },
    {
      "epoch": 0.03199457811141794,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998558379828814,
      "loss": 2.2388,
      "step": 4709
    },
    {
      "epoch": 0.032001372458012,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998557563324157,
      "loss": 2.3462,
      "step": 4710
    },
    {
      "epoch": 0.03200816680460606,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009985567465886735,
      "loss": 2.5033,
      "step": 4711
    },
    {
      "epoch": 0.032014961151200116,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000998555929622364,
      "loss": 2.2384,
      "step": 4712
    },
    {
      "epoch": 0.03202175549779417,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009985551124252288,
      "loss": 2.2994,
      "step": 4713
    },
    {
      "epoch": 0.03202854984438823,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009985542949972684,
      "loss": 2.3751,
      "step": 4714
    },
    {
      "epoch": 0.032035344190982286,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009985534773384829,
      "loss": 2.4677,
      "step": 4715
    },
    {
      "epoch": 0.03204213853757635,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009985526594488729,
      "loss": 2.3725,
      "step": 4716
    },
    {
      "epoch": 0.032048932884170406,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985518413284386,
      "loss": 2.277,
      "step": 4717
    },
    {
      "epoch": 0.03205572723076446,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009985510229771808,
      "loss": 2.4071,
      "step": 4718
    },
    {
      "epoch": 0.03206252157735852,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009985502043950993,
      "loss": 2.3621,
      "step": 4719
    },
    {
      "epoch": 0.032069315923952575,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000998549385582195,
      "loss": 2.3974,
      "step": 4720
    },
    {
      "epoch": 0.03207611027054663,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998548566538468,
      "loss": 2.4261,
      "step": 4721
    },
    {
      "epoch": 0.032082904617140695,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009985477472639187,
      "loss": 2.3353,
      "step": 4722
    },
    {
      "epoch": 0.03208969896373475,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009985469277585474,
      "loss": 2.5169,
      "step": 4723
    },
    {
      "epoch": 0.03209649331032881,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009985461080223548,
      "loss": 2.348,
      "step": 4724
    },
    {
      "epoch": 0.032103287656922865,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009985452880553409,
      "loss": 2.1864,
      "step": 4725
    },
    {
      "epoch": 0.03211008200351692,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0009985444678575065,
      "loss": 2.3764,
      "step": 4726
    },
    {
      "epoch": 0.032116876350110984,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009985436474288514,
      "loss": 2.354,
      "step": 4727
    },
    {
      "epoch": 0.03212367069670504,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985428267693765,
      "loss": 2.4404,
      "step": 4728
    },
    {
      "epoch": 0.0321304650432991,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998542005879082,
      "loss": 2.4717,
      "step": 4729
    },
    {
      "epoch": 0.032137259389893154,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000998541184757968,
      "loss": 2.4363,
      "step": 4730
    },
    {
      "epoch": 0.03214405373648721,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985403634060355,
      "loss": 2.303,
      "step": 4731
    },
    {
      "epoch": 0.03215084808308127,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009985395418232843,
      "loss": 2.43,
      "step": 4732
    },
    {
      "epoch": 0.03215764242967533,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000998538720009715,
      "loss": 2.3252,
      "step": 4733
    },
    {
      "epoch": 0.03216443677626939,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998537897965328,
      "loss": 2.3197,
      "step": 4734
    },
    {
      "epoch": 0.03217123112286344,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998537075690124,
      "loss": 2.146,
      "step": 4735
    },
    {
      "epoch": 0.0321780254694575,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009985362531841028,
      "loss": 2.2906,
      "step": 4736
    },
    {
      "epoch": 0.032184819816051556,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998535430447265,
      "loss": 2.3073,
      "step": 4737
    },
    {
      "epoch": 0.03219161416264562,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009985346074796108,
      "loss": 2.2592,
      "step": 4738
    },
    {
      "epoch": 0.032198408509239676,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009985337842811411,
      "loss": 2.3805,
      "step": 4739
    },
    {
      "epoch": 0.03220520285583373,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009985329608518559,
      "loss": 2.4741,
      "step": 4740
    },
    {
      "epoch": 0.03221199720242779,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009985321371917557,
      "loss": 2.2612,
      "step": 4741
    },
    {
      "epoch": 0.032218791549021845,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009985313133008408,
      "loss": 2.4366,
      "step": 4742
    },
    {
      "epoch": 0.03222558589561591,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009985304891791116,
      "loss": 2.4179,
      "step": 4743
    },
    {
      "epoch": 0.032232380242209965,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009985296648265685,
      "loss": 2.461,
      "step": 4744
    },
    {
      "epoch": 0.03223917458880402,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009985288402432119,
      "loss": 2.5309,
      "step": 4745
    },
    {
      "epoch": 0.03224596893539808,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998528015429042,
      "loss": 2.3093,
      "step": 4746
    },
    {
      "epoch": 0.032252763281992135,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009985271903840596,
      "loss": 2.3775,
      "step": 4747
    },
    {
      "epoch": 0.03225955762858619,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985263651082646,
      "loss": 2.5124,
      "step": 4748
    },
    {
      "epoch": 0.032266351975180255,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009985255396016577,
      "loss": 2.369,
      "step": 4749
    },
    {
      "epoch": 0.03227314632177431,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985247138642392,
      "loss": 2.3735,
      "step": 4750
    },
    {
      "epoch": 0.03227994066836837,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009985238878960096,
      "loss": 2.2847,
      "step": 4751
    },
    {
      "epoch": 0.032286735014962424,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998523061696969,
      "loss": 2.4431,
      "step": 4752
    },
    {
      "epoch": 0.03229352936155648,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000998522235267118,
      "loss": 2.3369,
      "step": 4753
    },
    {
      "epoch": 0.032300323708150544,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009985214086064569,
      "loss": 2.4306,
      "step": 4754
    },
    {
      "epoch": 0.0323071180547446,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998520581714986,
      "loss": 2.381,
      "step": 4755
    },
    {
      "epoch": 0.03231391240133866,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998519754592706,
      "loss": 2.4926,
      "step": 4756
    },
    {
      "epoch": 0.03232070674793271,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998518927239617,
      "loss": 2.3298,
      "step": 4757
    },
    {
      "epoch": 0.03232750109452677,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009985180996557193,
      "loss": 2.3645,
      "step": 4758
    },
    {
      "epoch": 0.032334295441120826,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985172718410137,
      "loss": 2.4101,
      "step": 4759
    },
    {
      "epoch": 0.03234108978771489,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009985164437955,
      "loss": 2.4936,
      "step": 4760
    },
    {
      "epoch": 0.032347884134308946,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998515615519179,
      "loss": 2.3169,
      "step": 4761
    },
    {
      "epoch": 0.032354678480903,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000998514787012051,
      "loss": 2.3951,
      "step": 4762
    },
    {
      "epoch": 0.03236147282749706,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985139582741164,
      "loss": 2.4065,
      "step": 4763
    },
    {
      "epoch": 0.032368267174091116,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985131293053758,
      "loss": 2.3192,
      "step": 4764
    },
    {
      "epoch": 0.03237506152068518,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009985123001058289,
      "loss": 2.3148,
      "step": 4765
    },
    {
      "epoch": 0.032381855867279236,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009985114706754768,
      "loss": 2.2722,
      "step": 4766
    },
    {
      "epoch": 0.03238865021387329,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009985106410143197,
      "loss": 2.2412,
      "step": 4767
    },
    {
      "epoch": 0.03239544456046735,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985098111223577,
      "loss": 2.2915,
      "step": 4768
    },
    {
      "epoch": 0.032402238907061405,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009985089809995914,
      "loss": 2.3586,
      "step": 4769
    },
    {
      "epoch": 0.03240903325365546,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998508150646021,
      "loss": 2.4187,
      "step": 4770
    },
    {
      "epoch": 0.032415827600249525,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009985073200616471,
      "loss": 2.3607,
      "step": 4771
    },
    {
      "epoch": 0.03242262194684358,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009985064892464703,
      "loss": 2.282,
      "step": 4772
    },
    {
      "epoch": 0.03242941629343764,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009985056582004904,
      "loss": 2.3546,
      "step": 4773
    },
    {
      "epoch": 0.032436210640031694,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009985048269237082,
      "loss": 2.3135,
      "step": 4774
    },
    {
      "epoch": 0.03244300498662575,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998503995416124,
      "loss": 2.3399,
      "step": 4775
    },
    {
      "epoch": 0.032449799333219814,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009985031636777382,
      "loss": 2.3193,
      "step": 4776
    },
    {
      "epoch": 0.03245659367981387,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000998502331708551,
      "loss": 2.5315,
      "step": 4777
    },
    {
      "epoch": 0.03246338802640793,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009985014995085634,
      "loss": 2.4023,
      "step": 4778
    },
    {
      "epoch": 0.032470182373001984,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009985006670777748,
      "loss": 2.3766,
      "step": 4779
    },
    {
      "epoch": 0.03247697671959604,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009984998344161862,
      "loss": 2.3079,
      "step": 4780
    },
    {
      "epoch": 0.0324837710661901,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998499001523798,
      "loss": 2.3118,
      "step": 4781
    },
    {
      "epoch": 0.03249056541278416,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009984981684006105,
      "loss": 2.3183,
      "step": 4782
    },
    {
      "epoch": 0.032497359759378217,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998497335046624,
      "loss": 2.4124,
      "step": 4783
    },
    {
      "epoch": 0.03250415410597227,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998496501461839,
      "loss": 2.4138,
      "step": 4784
    },
    {
      "epoch": 0.03251094845256633,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009984956676462558,
      "loss": 2.5,
      "step": 4785
    },
    {
      "epoch": 0.032517742799160386,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009984948335998746,
      "loss": 2.4588,
      "step": 4786
    },
    {
      "epoch": 0.03252453714575445,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009984939993226965,
      "loss": 2.2714,
      "step": 4787
    },
    {
      "epoch": 0.032531331492348506,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009984931648147212,
      "loss": 2.4189,
      "step": 4788
    },
    {
      "epoch": 0.03253812583894256,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998492330075949,
      "loss": 2.3525,
      "step": 4789
    },
    {
      "epoch": 0.03254492018553662,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009984914951063808,
      "loss": 2.1699,
      "step": 4790
    },
    {
      "epoch": 0.032551714532130675,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009984906599060166,
      "loss": 2.6053,
      "step": 4791
    },
    {
      "epoch": 0.03255850887872473,
      "grad_norm": 0.140625,
      "learning_rate": 0.000998489824474857,
      "loss": 2.3716,
      "step": 4792
    },
    {
      "epoch": 0.032565303225318795,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009984889888129025,
      "loss": 2.3209,
      "step": 4793
    },
    {
      "epoch": 0.03257209757191285,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009984881529201532,
      "loss": 2.3034,
      "step": 4794
    },
    {
      "epoch": 0.03257889191850691,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009984873167966095,
      "loss": 2.5428,
      "step": 4795
    },
    {
      "epoch": 0.032585686265100965,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009984864804422722,
      "loss": 2.415,
      "step": 4796
    },
    {
      "epoch": 0.03259248061169502,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009984856438571411,
      "loss": 2.4641,
      "step": 4797
    },
    {
      "epoch": 0.032599274958289085,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009984848070412168,
      "loss": 2.4112,
      "step": 4798
    },
    {
      "epoch": 0.03260606930488314,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009984839699945,
      "loss": 2.478,
      "step": 4799
    },
    {
      "epoch": 0.0326128636514772,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009984831327169905,
      "loss": 2.1256,
      "step": 4800
    },
    {
      "epoch": 0.032619657998071254,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009984822952086893,
      "loss": 2.4344,
      "step": 4801
    },
    {
      "epoch": 0.03262645234466531,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009984814574695963,
      "loss": 2.3415,
      "step": 4802
    },
    {
      "epoch": 0.032633246691259374,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009984806194997125,
      "loss": 2.1906,
      "step": 4803
    },
    {
      "epoch": 0.03264004103785343,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009984797812990377,
      "loss": 2.2405,
      "step": 4804
    },
    {
      "epoch": 0.03264683538444749,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009984789428675723,
      "loss": 2.3324,
      "step": 4805
    },
    {
      "epoch": 0.03265362973104154,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998478104205317,
      "loss": 2.3054,
      "step": 4806
    },
    {
      "epoch": 0.0326604240776356,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000998477265312272,
      "loss": 2.4286,
      "step": 4807
    },
    {
      "epoch": 0.032667218424229656,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009984764261884377,
      "loss": 2.3953,
      "step": 4808
    },
    {
      "epoch": 0.03267401277082372,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009984755868338146,
      "loss": 2.404,
      "step": 4809
    },
    {
      "epoch": 0.032680807117417776,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998474747248403,
      "loss": 2.2847,
      "step": 4810
    },
    {
      "epoch": 0.03268760146401183,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009984739074322033,
      "loss": 2.4653,
      "step": 4811
    },
    {
      "epoch": 0.03269439581060589,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009984730673852158,
      "loss": 2.5537,
      "step": 4812
    },
    {
      "epoch": 0.032701190157199946,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009984722271074412,
      "loss": 2.369,
      "step": 4813
    },
    {
      "epoch": 0.03270798450379401,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009984713865988796,
      "loss": 2.46,
      "step": 4814
    },
    {
      "epoch": 0.032714778850388065,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009984705458595314,
      "loss": 2.5173,
      "step": 4815
    },
    {
      "epoch": 0.03272157319698212,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009984697048893972,
      "loss": 2.4356,
      "step": 4816
    },
    {
      "epoch": 0.03272836754357618,
      "grad_norm": 0.140625,
      "learning_rate": 0.000998468863688477,
      "loss": 2.3443,
      "step": 4817
    },
    {
      "epoch": 0.032735161890170235,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009984680222567715,
      "loss": 2.3447,
      "step": 4818
    },
    {
      "epoch": 0.03274195623676429,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009984671805942811,
      "loss": 2.4191,
      "step": 4819
    },
    {
      "epoch": 0.032748750583358355,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009984663387010062,
      "loss": 2.207,
      "step": 4820
    },
    {
      "epoch": 0.03275554492995241,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000998465496576947,
      "loss": 2.3271,
      "step": 4821
    },
    {
      "epoch": 0.03276233927654647,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998464654222104,
      "loss": 2.4303,
      "step": 4822
    },
    {
      "epoch": 0.032769133623140524,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009984638116364774,
      "loss": 2.3752,
      "step": 4823
    },
    {
      "epoch": 0.03277592796973458,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009984629688200679,
      "loss": 2.3104,
      "step": 4824
    },
    {
      "epoch": 0.032782722316328644,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998462125772876,
      "loss": 2.3721,
      "step": 4825
    },
    {
      "epoch": 0.0327895166629227,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009984612824949016,
      "loss": 2.2201,
      "step": 4826
    },
    {
      "epoch": 0.03279631100951676,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009984604389861455,
      "loss": 2.3084,
      "step": 4827
    },
    {
      "epoch": 0.032803105356110814,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009984595952466076,
      "loss": 2.2925,
      "step": 4828
    },
    {
      "epoch": 0.03280989970270487,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009984587512762888,
      "loss": 2.3484,
      "step": 4829
    },
    {
      "epoch": 0.032816694049298926,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009984579070751894,
      "loss": 2.3508,
      "step": 4830
    },
    {
      "epoch": 0.03282348839589299,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009984570626433095,
      "loss": 2.389,
      "step": 4831
    },
    {
      "epoch": 0.032830282742487046,
      "grad_norm": 0.15625,
      "learning_rate": 0.00099845621798065,
      "loss": 2.3783,
      "step": 4832
    },
    {
      "epoch": 0.0328370770890811,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009984553730872107,
      "loss": 2.4241,
      "step": 4833
    },
    {
      "epoch": 0.03284387143567516,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009984545279629925,
      "loss": 2.4194,
      "step": 4834
    },
    {
      "epoch": 0.032850665782269216,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009984536826079956,
      "loss": 2.3988,
      "step": 4835
    },
    {
      "epoch": 0.03285746012886328,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009984528370222201,
      "loss": 2.3937,
      "step": 4836
    },
    {
      "epoch": 0.032864254475457336,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009984519912056668,
      "loss": 2.2602,
      "step": 4837
    },
    {
      "epoch": 0.03287104882205139,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998451145158336,
      "loss": 2.3078,
      "step": 4838
    },
    {
      "epoch": 0.03287784316864545,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998450298880228,
      "loss": 2.2996,
      "step": 4839
    },
    {
      "epoch": 0.032884637515239505,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000998449452371343,
      "loss": 2.4442,
      "step": 4840
    },
    {
      "epoch": 0.03289143186183356,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009984486056316818,
      "loss": 2.5499,
      "step": 4841
    },
    {
      "epoch": 0.032898226208427625,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009984477586612447,
      "loss": 2.4197,
      "step": 4842
    },
    {
      "epoch": 0.03290502055502168,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009984469114600319,
      "loss": 2.5433,
      "step": 4843
    },
    {
      "epoch": 0.03291181490161574,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000998446064028044,
      "loss": 2.2805,
      "step": 4844
    },
    {
      "epoch": 0.032918609248209794,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000998445216365281,
      "loss": 2.3755,
      "step": 4845
    },
    {
      "epoch": 0.03292540359480385,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009984443684717438,
      "loss": 2.2633,
      "step": 4846
    },
    {
      "epoch": 0.032932197941397914,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009984435203474328,
      "loss": 2.2126,
      "step": 4847
    },
    {
      "epoch": 0.03293899228799197,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009984426719923478,
      "loss": 2.2566,
      "step": 4848
    },
    {
      "epoch": 0.03294578663458603,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009984418234064896,
      "loss": 2.4922,
      "step": 4849
    },
    {
      "epoch": 0.032952580981180084,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009984409745898586,
      "loss": 2.4251,
      "step": 4850
    },
    {
      "epoch": 0.03295937532777414,
      "grad_norm": 0.15625,
      "learning_rate": 0.000998440125542455,
      "loss": 2.3877,
      "step": 4851
    },
    {
      "epoch": 0.0329661696743682,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009984392762642797,
      "loss": 2.3063,
      "step": 4852
    },
    {
      "epoch": 0.03297296402096226,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009984384267553324,
      "loss": 2.4385,
      "step": 4853
    },
    {
      "epoch": 0.03297975836755632,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998437577015614,
      "loss": 2.4313,
      "step": 4854
    },
    {
      "epoch": 0.03298655271415037,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009984367270451247,
      "loss": 2.4058,
      "step": 4855
    },
    {
      "epoch": 0.03299334706074443,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009984358768438649,
      "loss": 2.4081,
      "step": 4856
    },
    {
      "epoch": 0.033000141407338486,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998435026411835,
      "loss": 2.3398,
      "step": 4857
    },
    {
      "epoch": 0.03300693575393255,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009984341757490354,
      "loss": 2.5382,
      "step": 4858
    },
    {
      "epoch": 0.033013730100526606,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009984333248554665,
      "loss": 2.4922,
      "step": 4859
    },
    {
      "epoch": 0.03302052444712066,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009984324737311286,
      "loss": 2.3589,
      "step": 4860
    },
    {
      "epoch": 0.03302731879371472,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009984316223760222,
      "loss": 2.2722,
      "step": 4861
    },
    {
      "epoch": 0.033034113140308775,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009984307707901477,
      "loss": 2.4039,
      "step": 4862
    },
    {
      "epoch": 0.03304090748690283,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009984299189735055,
      "loss": 2.2954,
      "step": 4863
    },
    {
      "epoch": 0.033047701833496895,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009984290669260959,
      "loss": 2.3061,
      "step": 4864
    },
    {
      "epoch": 0.03305449618009095,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009984282146479194,
      "loss": 2.2526,
      "step": 4865
    },
    {
      "epoch": 0.03306129052668501,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009984273621389764,
      "loss": 2.3842,
      "step": 4866
    },
    {
      "epoch": 0.033068084873279065,
      "grad_norm": 0.171875,
      "learning_rate": 0.000998426509399267,
      "loss": 2.4683,
      "step": 4867
    },
    {
      "epoch": 0.03307487921987312,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009984256564287922,
      "loss": 2.4604,
      "step": 4868
    },
    {
      "epoch": 0.033081673566467185,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009984248032275519,
      "loss": 2.4273,
      "step": 4869
    },
    {
      "epoch": 0.03308846791306124,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009984239497955465,
      "loss": 2.5327,
      "step": 4870
    },
    {
      "epoch": 0.0330952622596553,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009984230961327767,
      "loss": 2.4187,
      "step": 4871
    },
    {
      "epoch": 0.033102056606249354,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009984222422392427,
      "loss": 2.1873,
      "step": 4872
    },
    {
      "epoch": 0.03310885095284341,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009984213881149447,
      "loss": 2.5241,
      "step": 4873
    },
    {
      "epoch": 0.033115645299437474,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009984205337598836,
      "loss": 2.4318,
      "step": 4874
    },
    {
      "epoch": 0.03312243964603153,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009984196791740594,
      "loss": 2.4458,
      "step": 4875
    },
    {
      "epoch": 0.03312923399262559,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009984188243574725,
      "loss": 2.5391,
      "step": 4876
    },
    {
      "epoch": 0.03313602833921964,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009984179693101234,
      "loss": 2.2585,
      "step": 4877
    },
    {
      "epoch": 0.0331428226858137,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009984171140320127,
      "loss": 2.4909,
      "step": 4878
    },
    {
      "epoch": 0.033149617032407756,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009984162585231404,
      "loss": 2.3818,
      "step": 4879
    },
    {
      "epoch": 0.03315641137900182,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000998415402783507,
      "loss": 2.2207,
      "step": 4880
    },
    {
      "epoch": 0.033163205725595876,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009984145468131132,
      "loss": 2.5547,
      "step": 4881
    },
    {
      "epoch": 0.03317000007218993,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009984136906119592,
      "loss": 2.1929,
      "step": 4882
    },
    {
      "epoch": 0.03317679441878399,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000998412834180045,
      "loss": 2.2959,
      "step": 4883
    },
    {
      "epoch": 0.033183588765378046,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998411977517372,
      "loss": 2.379,
      "step": 4884
    },
    {
      "epoch": 0.03319038311197211,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009984111206239395,
      "loss": 2.2743,
      "step": 4885
    },
    {
      "epoch": 0.033197177458566166,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009984102634997484,
      "loss": 2.4431,
      "step": 4886
    },
    {
      "epoch": 0.03320397180516022,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009984094061447992,
      "loss": 2.4485,
      "step": 4887
    },
    {
      "epoch": 0.03321076615175428,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009984085485590921,
      "loss": 2.3931,
      "step": 4888
    },
    {
      "epoch": 0.033217560498348335,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009984076907426276,
      "loss": 2.3221,
      "step": 4889
    },
    {
      "epoch": 0.03322435484494239,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009984068326954062,
      "loss": 2.225,
      "step": 4890
    },
    {
      "epoch": 0.033231149191536455,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009984059744174278,
      "loss": 2.3899,
      "step": 4891
    },
    {
      "epoch": 0.03323794353813051,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009984051159086934,
      "loss": 2.2156,
      "step": 4892
    },
    {
      "epoch": 0.03324473788472457,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009984042571692029,
      "loss": 2.4105,
      "step": 4893
    },
    {
      "epoch": 0.033251532231318624,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009984033981989572,
      "loss": 2.4967,
      "step": 4894
    },
    {
      "epoch": 0.03325832657791268,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009984025389979563,
      "loss": 2.3624,
      "step": 4895
    },
    {
      "epoch": 0.033265120924506744,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009984016795662007,
      "loss": 2.291,
      "step": 4896
    },
    {
      "epoch": 0.0332719152711008,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000998400819903691,
      "loss": 2.2533,
      "step": 4897
    },
    {
      "epoch": 0.03327870961769486,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983999600104273,
      "loss": 2.2826,
      "step": 4898
    },
    {
      "epoch": 0.033285503964288914,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009983990998864104,
      "loss": 2.3793,
      "step": 4899
    },
    {
      "epoch": 0.03329229831088297,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009983982395316401,
      "loss": 2.2743,
      "step": 4900
    },
    {
      "epoch": 0.03329909265747703,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983973789461172,
      "loss": 2.4226,
      "step": 4901
    },
    {
      "epoch": 0.03330588700407109,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983965181298421,
      "loss": 2.3693,
      "step": 4902
    },
    {
      "epoch": 0.033312681350665146,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000998395657082815,
      "loss": 2.3379,
      "step": 4903
    },
    {
      "epoch": 0.0333194756972592,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009983947958050366,
      "loss": 2.3826,
      "step": 4904
    },
    {
      "epoch": 0.03332627004385326,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983939342965071,
      "loss": 2.502,
      "step": 4905
    },
    {
      "epoch": 0.033333064390447316,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009983930725572267,
      "loss": 2.4133,
      "step": 4906
    },
    {
      "epoch": 0.03333985873704138,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009983922105871964,
      "loss": 2.3112,
      "step": 4907
    },
    {
      "epoch": 0.033346653083635436,
      "grad_norm": 0.15625,
      "learning_rate": 0.000998391348386416,
      "loss": 2.4414,
      "step": 4908
    },
    {
      "epoch": 0.03335344743022949,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998390485954886,
      "loss": 2.4605,
      "step": 4909
    },
    {
      "epoch": 0.03336024177682355,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983896232926071,
      "loss": 2.5057,
      "step": 4910
    },
    {
      "epoch": 0.033367036123417605,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983887603995795,
      "loss": 2.5591,
      "step": 4911
    },
    {
      "epoch": 0.03337383047001166,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009983878972758037,
      "loss": 2.4039,
      "step": 4912
    },
    {
      "epoch": 0.033380624816605725,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009983870339212797,
      "loss": 2.3196,
      "step": 4913
    },
    {
      "epoch": 0.03338741916319978,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009983861703360085,
      "loss": 2.2821,
      "step": 4914
    },
    {
      "epoch": 0.03339421350979384,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009983853065199902,
      "loss": 2.3242,
      "step": 4915
    },
    {
      "epoch": 0.033401007856387895,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000998384442473225,
      "loss": 2.3385,
      "step": 4916
    },
    {
      "epoch": 0.03340780220298195,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009983835781957138,
      "loss": 2.2713,
      "step": 4917
    },
    {
      "epoch": 0.033414596549576014,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009983827136874565,
      "loss": 2.3238,
      "step": 4918
    },
    {
      "epoch": 0.03342139089617007,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009983818489484539,
      "loss": 2.5152,
      "step": 4919
    },
    {
      "epoch": 0.03342818524276413,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009983809839787061,
      "loss": 2.3614,
      "step": 4920
    },
    {
      "epoch": 0.033434979589358184,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009983801187782135,
      "loss": 2.2787,
      "step": 4921
    },
    {
      "epoch": 0.03344177393595224,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009983792533469768,
      "loss": 2.3493,
      "step": 4922
    },
    {
      "epoch": 0.0334485682825463,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009983783876849964,
      "loss": 2.4739,
      "step": 4923
    },
    {
      "epoch": 0.03345536262914036,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998377521792272,
      "loss": 2.3182,
      "step": 4924
    },
    {
      "epoch": 0.03346215697573442,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009983766556688049,
      "loss": 2.3251,
      "step": 4925
    },
    {
      "epoch": 0.03346895132232847,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009983757893145952,
      "loss": 2.3788,
      "step": 4926
    },
    {
      "epoch": 0.03347574566892253,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009983749227296431,
      "loss": 2.3413,
      "step": 4927
    },
    {
      "epoch": 0.033482540015516586,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009983740559139488,
      "loss": 2.2686,
      "step": 4928
    },
    {
      "epoch": 0.03348933436211065,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009983731888675135,
      "loss": 2.2171,
      "step": 4929
    },
    {
      "epoch": 0.033496128708704706,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009983723215903368,
      "loss": 2.1155,
      "step": 4930
    },
    {
      "epoch": 0.03350292305529876,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009983714540824197,
      "loss": 2.2955,
      "step": 4931
    },
    {
      "epoch": 0.03350971740189282,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009983705863437621,
      "loss": 2.5001,
      "step": 4932
    },
    {
      "epoch": 0.033516511748486875,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009983697183743648,
      "loss": 2.2959,
      "step": 4933
    },
    {
      "epoch": 0.03352330609508094,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998368850174228,
      "loss": 2.3886,
      "step": 4934
    },
    {
      "epoch": 0.033530100441674995,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009983679817433522,
      "loss": 2.3243,
      "step": 4935
    },
    {
      "epoch": 0.03353689478826905,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009983671130817376,
      "loss": 2.3405,
      "step": 4936
    },
    {
      "epoch": 0.03354368913486311,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009983662441893847,
      "loss": 2.2886,
      "step": 4937
    },
    {
      "epoch": 0.033550483481457165,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000998365375066294,
      "loss": 2.1721,
      "step": 4938
    },
    {
      "epoch": 0.03355727782805122,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998364505712466,
      "loss": 2.3685,
      "step": 4939
    },
    {
      "epoch": 0.033564072174645285,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009983636361279007,
      "loss": 2.4204,
      "step": 4940
    },
    {
      "epoch": 0.03357086652123934,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998362766312599,
      "loss": 2.364,
      "step": 4941
    },
    {
      "epoch": 0.0335776608678334,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983618962665609,
      "loss": 2.3785,
      "step": 4942
    },
    {
      "epoch": 0.033584455214427454,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983610259897868,
      "loss": 2.417,
      "step": 4943
    },
    {
      "epoch": 0.03359124956102151,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009983601554822774,
      "loss": 2.4308,
      "step": 4944
    },
    {
      "epoch": 0.033598043907615574,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009983592847440331,
      "loss": 2.3875,
      "step": 4945
    },
    {
      "epoch": 0.03360483825420963,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983584137750541,
      "loss": 2.3007,
      "step": 4946
    },
    {
      "epoch": 0.03361163260080369,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009983575425753408,
      "loss": 2.3194,
      "step": 4947
    },
    {
      "epoch": 0.03361842694739774,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009983566711448937,
      "loss": 2.3132,
      "step": 4948
    },
    {
      "epoch": 0.0336252212939918,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009983557994837132,
      "loss": 2.3371,
      "step": 4949
    },
    {
      "epoch": 0.033632015640585856,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009983549275917995,
      "loss": 2.2831,
      "step": 4950
    },
    {
      "epoch": 0.03363880998717992,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009983540554691535,
      "loss": 2.2748,
      "step": 4951
    },
    {
      "epoch": 0.033645604333773976,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009983531831157751,
      "loss": 2.4455,
      "step": 4952
    },
    {
      "epoch": 0.03365239868036803,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998352310531665,
      "loss": 2.4159,
      "step": 4953
    },
    {
      "epoch": 0.03365919302696209,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009983514377168233,
      "loss": 2.1776,
      "step": 4954
    },
    {
      "epoch": 0.033665987373556146,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009983505646712508,
      "loss": 2.3187,
      "step": 4955
    },
    {
      "epoch": 0.03367278172015021,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009983496913949475,
      "loss": 2.3282,
      "step": 4956
    },
    {
      "epoch": 0.033679576066744266,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009983488178879141,
      "loss": 2.2746,
      "step": 4957
    },
    {
      "epoch": 0.03368637041333832,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998347944150151,
      "loss": 2.2949,
      "step": 4958
    },
    {
      "epoch": 0.03369316475993238,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983470701816584,
      "loss": 2.311,
      "step": 4959
    },
    {
      "epoch": 0.033699959106526435,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009983461959824369,
      "loss": 2.3261,
      "step": 4960
    },
    {
      "epoch": 0.03370675345312049,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983453215524868,
      "loss": 2.3361,
      "step": 4961
    },
    {
      "epoch": 0.033713547799714555,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009983444468918085,
      "loss": 2.3371,
      "step": 4962
    },
    {
      "epoch": 0.03372034214630861,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009983435720004024,
      "loss": 2.3885,
      "step": 4963
    },
    {
      "epoch": 0.03372713649290267,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998342696878269,
      "loss": 2.3439,
      "step": 4964
    },
    {
      "epoch": 0.033733930839496724,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983418215254085,
      "loss": 2.3945,
      "step": 4965
    },
    {
      "epoch": 0.03374072518609078,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009983409459418215,
      "loss": 2.3623,
      "step": 4966
    },
    {
      "epoch": 0.033747519532684844,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983400701275085,
      "loss": 2.2219,
      "step": 4967
    },
    {
      "epoch": 0.0337543138792789,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009983391940824697,
      "loss": 2.434,
      "step": 4968
    },
    {
      "epoch": 0.03376110822587296,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009983383178067055,
      "loss": 2.2562,
      "step": 4969
    },
    {
      "epoch": 0.033767902572467014,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009983374413002166,
      "loss": 2.2915,
      "step": 4970
    },
    {
      "epoch": 0.03377469691906107,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000998336564563003,
      "loss": 2.3617,
      "step": 4971
    },
    {
      "epoch": 0.03378149126565513,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009983356875950652,
      "loss": 2.4536,
      "step": 4972
    },
    {
      "epoch": 0.03378828561224919,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983348103964039,
      "loss": 2.5098,
      "step": 4973
    },
    {
      "epoch": 0.03379507995884325,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998333932967019,
      "loss": 2.2533,
      "step": 4974
    },
    {
      "epoch": 0.0338018743054373,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009983330553069115,
      "loss": 2.421,
      "step": 4975
    },
    {
      "epoch": 0.03380866865203136,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009983321774160814,
      "loss": 2.3331,
      "step": 4976
    },
    {
      "epoch": 0.033815462998625416,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009983312992945291,
      "loss": 2.3417,
      "step": 4977
    },
    {
      "epoch": 0.03382225734521948,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983304209422552,
      "loss": 2.5928,
      "step": 4978
    },
    {
      "epoch": 0.033829051691813536,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009983295423592602,
      "loss": 2.4372,
      "step": 4979
    },
    {
      "epoch": 0.03383584603840759,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009983286635455442,
      "loss": 2.2708,
      "step": 4980
    },
    {
      "epoch": 0.03384264038500165,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009983277845011075,
      "loss": 2.4217,
      "step": 4981
    },
    {
      "epoch": 0.033849434731595705,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983269052259512,
      "loss": 2.2763,
      "step": 4982
    },
    {
      "epoch": 0.03385622907818976,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000998326025720075,
      "loss": 2.2539,
      "step": 4983
    },
    {
      "epoch": 0.033863023424783825,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009983251459834796,
      "loss": 2.0699,
      "step": 4984
    },
    {
      "epoch": 0.03386981777137788,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009983242660161653,
      "loss": 2.3941,
      "step": 4985
    },
    {
      "epoch": 0.03387661211797194,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009983233858181326,
      "loss": 2.5996,
      "step": 4986
    },
    {
      "epoch": 0.033883406464565995,
      "grad_norm": 0.15625,
      "learning_rate": 0.000998322505389382,
      "loss": 2.2922,
      "step": 4987
    },
    {
      "epoch": 0.03389020081116005,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009983216247299138,
      "loss": 2.4595,
      "step": 4988
    },
    {
      "epoch": 0.033896995157754115,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009983207438397283,
      "loss": 2.5124,
      "step": 4989
    },
    {
      "epoch": 0.03390378950434817,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998319862718826,
      "loss": 2.3847,
      "step": 4990
    },
    {
      "epoch": 0.03391058385094223,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009983189813672073,
      "loss": 2.4262,
      "step": 4991
    },
    {
      "epoch": 0.033917378197536284,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009983180997848726,
      "loss": 2.3729,
      "step": 4992
    },
    {
      "epoch": 0.03392417254413034,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009983172179718225,
      "loss": 2.2955,
      "step": 4993
    },
    {
      "epoch": 0.033930966890724404,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009983163359280572,
      "loss": 2.4023,
      "step": 4994
    },
    {
      "epoch": 0.03393776123731846,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998315453653577,
      "loss": 2.5087,
      "step": 4995
    },
    {
      "epoch": 0.03394455558391252,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009983145711483826,
      "loss": 2.3541,
      "step": 4996
    },
    {
      "epoch": 0.03395134993050657,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009983136884124742,
      "loss": 2.2653,
      "step": 4997
    },
    {
      "epoch": 0.03395814427710063,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009983128054458523,
      "loss": 2.351,
      "step": 4998
    },
    {
      "epoch": 0.033964938623694686,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009983119222485173,
      "loss": 2.3814,
      "step": 4999
    },
    {
      "epoch": 0.03397173297028875,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009983110388204696,
      "loss": 2.2381,
      "step": 5000
    },
    {
      "epoch": 0.033978527316882806,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009983101551617095,
      "loss": 2.2653,
      "step": 5001
    },
    {
      "epoch": 0.03398532166347686,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009983092712722377,
      "loss": 2.3544,
      "step": 5002
    },
    {
      "epoch": 0.03399211601007092,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983083871520544,
      "loss": 2.2849,
      "step": 5003
    },
    {
      "epoch": 0.033998910356664976,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009983075028011598,
      "loss": 2.3257,
      "step": 5004
    },
    {
      "epoch": 0.03400570470325904,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009983066182195548,
      "loss": 2.1651,
      "step": 5005
    },
    {
      "epoch": 0.034012499049853095,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009983057334072393,
      "loss": 2.2155,
      "step": 5006
    },
    {
      "epoch": 0.03401929339644715,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983048483642142,
      "loss": 2.382,
      "step": 5007
    },
    {
      "epoch": 0.03402608774304121,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009983039630904794,
      "loss": 2.2366,
      "step": 5008
    },
    {
      "epoch": 0.034032882089635265,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009983030775860358,
      "loss": 2.3524,
      "step": 5009
    },
    {
      "epoch": 0.03403967643622932,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009983021918508834,
      "loss": 2.2839,
      "step": 5010
    },
    {
      "epoch": 0.034046470782823385,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998301305885023,
      "loss": 2.3222,
      "step": 5011
    },
    {
      "epoch": 0.03405326512941744,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009983004196884547,
      "loss": 2.466,
      "step": 5012
    },
    {
      "epoch": 0.0340600594760115,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009982995332611792,
      "loss": 2.4004,
      "step": 5013
    },
    {
      "epoch": 0.034066853822605554,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009982986466031964,
      "loss": 2.3458,
      "step": 5014
    },
    {
      "epoch": 0.03407364816919961,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009982977597145072,
      "loss": 2.4432,
      "step": 5015
    },
    {
      "epoch": 0.034080442515793674,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998296872595112,
      "loss": 2.515,
      "step": 5016
    },
    {
      "epoch": 0.03408723686238773,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998295985245011,
      "loss": 2.3735,
      "step": 5017
    },
    {
      "epoch": 0.03409403120898179,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009982950976642044,
      "loss": 2.1902,
      "step": 5018
    },
    {
      "epoch": 0.034100825555575844,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000998294209852693,
      "loss": 2.3364,
      "step": 5019
    },
    {
      "epoch": 0.0341076199021699,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009982933218104773,
      "loss": 2.3604,
      "step": 5020
    },
    {
      "epoch": 0.034114414248763957,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009982924335375574,
      "loss": 2.232,
      "step": 5021
    },
    {
      "epoch": 0.03412120859535802,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000998291545033934,
      "loss": 2.2514,
      "step": 5022
    },
    {
      "epoch": 0.034128002941952076,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009982906562996069,
      "loss": 2.5318,
      "step": 5023
    },
    {
      "epoch": 0.03413479728854613,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000998289767334577,
      "loss": 2.4813,
      "step": 5024
    },
    {
      "epoch": 0.03414159163514019,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009982888781388448,
      "loss": 2.2037,
      "step": 5025
    },
    {
      "epoch": 0.034148385981734246,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009982879887124106,
      "loss": 2.4621,
      "step": 5026
    },
    {
      "epoch": 0.03415518032832831,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009982870990552749,
      "loss": 2.3912,
      "step": 5027
    },
    {
      "epoch": 0.034161974674922366,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009982862091674376,
      "loss": 2.3545,
      "step": 5028
    },
    {
      "epoch": 0.03416876902151642,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009982853190489,
      "loss": 2.2851,
      "step": 5029
    },
    {
      "epoch": 0.03417556336811048,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009982844286996616,
      "loss": 2.2517,
      "step": 5030
    },
    {
      "epoch": 0.034182357714704535,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009982835381197235,
      "loss": 2.3912,
      "step": 5031
    },
    {
      "epoch": 0.03418915206129859,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009982826473090855,
      "loss": 2.3781,
      "step": 5032
    },
    {
      "epoch": 0.034195946407892655,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009982817562677487,
      "loss": 2.3365,
      "step": 5033
    },
    {
      "epoch": 0.03420274075448671,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000998280864995713,
      "loss": 2.2713,
      "step": 5034
    },
    {
      "epoch": 0.03420953510108077,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009982799734929789,
      "loss": 2.3901,
      "step": 5035
    },
    {
      "epoch": 0.034216329447674824,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009982790817595468,
      "loss": 2.3794,
      "step": 5036
    },
    {
      "epoch": 0.03422312379426888,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009982781897954175,
      "loss": 2.2572,
      "step": 5037
    },
    {
      "epoch": 0.034229918140862944,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000998277297600591,
      "loss": 2.3051,
      "step": 5038
    },
    {
      "epoch": 0.034236712487457,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009982764051750677,
      "loss": 2.3156,
      "step": 5039
    },
    {
      "epoch": 0.03424350683405106,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009982755125188484,
      "loss": 2.4245,
      "step": 5040
    },
    {
      "epoch": 0.034250301180645114,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000998274619631933,
      "loss": 2.3646,
      "step": 5041
    },
    {
      "epoch": 0.03425709552723917,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000998273726514322,
      "loss": 2.376,
      "step": 5042
    },
    {
      "epoch": 0.03426388987383323,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009982728331660163,
      "loss": 2.4407,
      "step": 5043
    },
    {
      "epoch": 0.03427068422042729,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009982719395870158,
      "loss": 2.5349,
      "step": 5044
    },
    {
      "epoch": 0.03427747856702135,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009982710457773214,
      "loss": 2.3548,
      "step": 5045
    },
    {
      "epoch": 0.0342842729136154,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000998270151736933,
      "loss": 2.3322,
      "step": 5046
    },
    {
      "epoch": 0.03429106726020946,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009982692574658514,
      "loss": 2.3969,
      "step": 5047
    },
    {
      "epoch": 0.034297861606803516,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009982683629640766,
      "loss": 2.385,
      "step": 5048
    },
    {
      "epoch": 0.03430465595339758,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009982674682316092,
      "loss": 2.3156,
      "step": 5049
    },
    {
      "epoch": 0.034311450299991636,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00099826657326845,
      "loss": 2.3752,
      "step": 5050
    },
    {
      "epoch": 0.03431824464658569,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998265678074599,
      "loss": 2.2981,
      "step": 5051
    },
    {
      "epoch": 0.03432503899317975,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009982647826500565,
      "loss": 2.3465,
      "step": 5052
    },
    {
      "epoch": 0.034331833339773805,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009982638869948234,
      "loss": 2.3362,
      "step": 5053
    },
    {
      "epoch": 0.03433862768636787,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009982629911088996,
      "loss": 2.4623,
      "step": 5054
    },
    {
      "epoch": 0.034345422032961925,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009982620949922857,
      "loss": 2.43,
      "step": 5055
    },
    {
      "epoch": 0.03435221637955598,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009982611986449823,
      "loss": 2.3767,
      "step": 5056
    },
    {
      "epoch": 0.03435901072615004,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009982603020669899,
      "loss": 2.4747,
      "step": 5057
    },
    {
      "epoch": 0.034365805072744095,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009982594052583083,
      "loss": 2.3801,
      "step": 5058
    },
    {
      "epoch": 0.03437259941933815,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009982585082189384,
      "loss": 2.328,
      "step": 5059
    },
    {
      "epoch": 0.034379393765932215,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009982576109488807,
      "loss": 2.4316,
      "step": 5060
    },
    {
      "epoch": 0.03438618811252627,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009982567134481353,
      "loss": 2.465,
      "step": 5061
    },
    {
      "epoch": 0.03439298245912033,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009982558157167027,
      "loss": 2.2251,
      "step": 5062
    },
    {
      "epoch": 0.034399776805714384,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009982549177545836,
      "loss": 2.3289,
      "step": 5063
    },
    {
      "epoch": 0.03440657115230844,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998254019561778,
      "loss": 2.3387,
      "step": 5064
    },
    {
      "epoch": 0.034413365498902504,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009982531211382866,
      "loss": 2.1846,
      "step": 5065
    },
    {
      "epoch": 0.03442015984549656,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009982522224841097,
      "loss": 2.2478,
      "step": 5066
    },
    {
      "epoch": 0.03442695419209062,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009982513235992476,
      "loss": 2.4267,
      "step": 5067
    },
    {
      "epoch": 0.03443374853868467,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000998250424483701,
      "loss": 2.3071,
      "step": 5068
    },
    {
      "epoch": 0.03444054288527873,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00099824952513747,
      "loss": 2.2996,
      "step": 5069
    },
    {
      "epoch": 0.034447337231872786,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009982486255605554,
      "loss": 2.389,
      "step": 5070
    },
    {
      "epoch": 0.03445413157846685,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009982477257529571,
      "loss": 2.3239,
      "step": 5071
    },
    {
      "epoch": 0.034460925925060906,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009982468257146763,
      "loss": 2.3368,
      "step": 5072
    },
    {
      "epoch": 0.03446772027165496,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009982459254457125,
      "loss": 2.3233,
      "step": 5073
    },
    {
      "epoch": 0.03447451461824902,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009982450249460665,
      "loss": 2.3327,
      "step": 5074
    },
    {
      "epoch": 0.034481308964843076,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009982441242157392,
      "loss": 2.3958,
      "step": 5075
    },
    {
      "epoch": 0.03448810331143714,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009982432232547302,
      "loss": 2.2887,
      "step": 5076
    },
    {
      "epoch": 0.034494897658031196,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009982423220630404,
      "loss": 2.3113,
      "step": 5077
    },
    {
      "epoch": 0.03450169200462525,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009982414206406703,
      "loss": 2.2077,
      "step": 5078
    },
    {
      "epoch": 0.03450848635121931,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009982405189876199,
      "loss": 2.1772,
      "step": 5079
    },
    {
      "epoch": 0.034515280697813365,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009982396171038898,
      "loss": 2.3143,
      "step": 5080
    },
    {
      "epoch": 0.03452207504440742,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009982387149894806,
      "loss": 2.217,
      "step": 5081
    },
    {
      "epoch": 0.034528869391001485,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009982378126443927,
      "loss": 2.3701,
      "step": 5082
    },
    {
      "epoch": 0.03453566373759554,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009982369100686264,
      "loss": 2.1587,
      "step": 5083
    },
    {
      "epoch": 0.0345424580841896,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000998236007262182,
      "loss": 2.4037,
      "step": 5084
    },
    {
      "epoch": 0.034549252430783654,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009982351042250599,
      "loss": 2.4628,
      "step": 5085
    },
    {
      "epoch": 0.03455604677737771,
      "grad_norm": 0.171875,
      "learning_rate": 0.000998234200957261,
      "loss": 2.4105,
      "step": 5086
    },
    {
      "epoch": 0.034562841123971774,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000998233297458785,
      "loss": 2.4303,
      "step": 5087
    },
    {
      "epoch": 0.03456963547056583,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009982323937296328,
      "loss": 2.3324,
      "step": 5088
    },
    {
      "epoch": 0.03457642981715989,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000998231489769805,
      "loss": 2.4265,
      "step": 5089
    },
    {
      "epoch": 0.034583224163753944,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009982305855793015,
      "loss": 2.5208,
      "step": 5090
    },
    {
      "epoch": 0.034590018510348,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009982296811581232,
      "loss": 2.3804,
      "step": 5091
    },
    {
      "epoch": 0.03459681285694206,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099822877650627,
      "loss": 2.2184,
      "step": 5092
    },
    {
      "epoch": 0.03460360720353612,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009982278716237426,
      "loss": 2.4176,
      "step": 5093
    },
    {
      "epoch": 0.034610401550130176,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009982269665105414,
      "loss": 2.3924,
      "step": 5094
    },
    {
      "epoch": 0.03461719589672423,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009982260611666668,
      "loss": 2.3281,
      "step": 5095
    },
    {
      "epoch": 0.03462399024331829,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009982251555921194,
      "loss": 2.4445,
      "step": 5096
    },
    {
      "epoch": 0.034630784589912346,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009982242497868993,
      "loss": 2.2769,
      "step": 5097
    },
    {
      "epoch": 0.03463757893650641,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009982233437510074,
      "loss": 2.31,
      "step": 5098
    },
    {
      "epoch": 0.034644373283100466,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009982224374844434,
      "loss": 2.2706,
      "step": 5099
    },
    {
      "epoch": 0.03465116762969452,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009982215309872083,
      "loss": 2.2764,
      "step": 5100
    },
    {
      "epoch": 0.03465796197628858,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009982206242593024,
      "loss": 2.4159,
      "step": 5101
    },
    {
      "epoch": 0.034664756322882635,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009982197173007258,
      "loss": 2.3468,
      "step": 5102
    },
    {
      "epoch": 0.03467155066947669,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009982188101114796,
      "loss": 2.381,
      "step": 5103
    },
    {
      "epoch": 0.034678345016070755,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009982179026915635,
      "loss": 2.2528,
      "step": 5104
    },
    {
      "epoch": 0.03468513936266481,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009982169950409783,
      "loss": 2.4129,
      "step": 5105
    },
    {
      "epoch": 0.03469193370925887,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009982160871597244,
      "loss": 2.4747,
      "step": 5106
    },
    {
      "epoch": 0.034698728055852925,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009982151790478022,
      "loss": 2.3415,
      "step": 5107
    },
    {
      "epoch": 0.03470552240244698,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998214270705212,
      "loss": 2.4158,
      "step": 5108
    },
    {
      "epoch": 0.034712316749041044,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009982133621319542,
      "loss": 2.2925,
      "step": 5109
    },
    {
      "epoch": 0.0347191110956351,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009982124533280295,
      "loss": 2.3662,
      "step": 5110
    },
    {
      "epoch": 0.03472590544222916,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009982115442934382,
      "loss": 2.3614,
      "step": 5111
    },
    {
      "epoch": 0.034732699788823214,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009982106350281805,
      "loss": 2.333,
      "step": 5112
    },
    {
      "epoch": 0.03473949413541727,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000998209725532257,
      "loss": 2.2547,
      "step": 5113
    },
    {
      "epoch": 0.034746288482011334,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009982088158056682,
      "loss": 2.3313,
      "step": 5114
    },
    {
      "epoch": 0.03475308282860539,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009982079058484144,
      "loss": 2.2374,
      "step": 5115
    },
    {
      "epoch": 0.03475987717519945,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009982069956604959,
      "loss": 2.321,
      "step": 5116
    },
    {
      "epoch": 0.0347666715217935,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009982060852419135,
      "loss": 2.4005,
      "step": 5117
    },
    {
      "epoch": 0.03477346586838756,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009982051745926672,
      "loss": 2.3886,
      "step": 5118
    },
    {
      "epoch": 0.034780260214981616,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009982042637127577,
      "loss": 2.3972,
      "step": 5119
    },
    {
      "epoch": 0.03478705456157568,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009982033526021853,
      "loss": 2.357,
      "step": 5120
    },
    {
      "epoch": 0.034793848908169736,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009982024412609504,
      "loss": 2.4395,
      "step": 5121
    },
    {
      "epoch": 0.03480064325476379,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009982015296890537,
      "loss": 2.3807,
      "step": 5122
    },
    {
      "epoch": 0.03480743760135785,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009982006178864953,
      "loss": 2.2923,
      "step": 5123
    },
    {
      "epoch": 0.034814231947951906,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009981997058532757,
      "loss": 2.239,
      "step": 5124
    },
    {
      "epoch": 0.03482102629454597,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009981987935893954,
      "loss": 2.1815,
      "step": 5125
    },
    {
      "epoch": 0.034827820641140025,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009981978810948548,
      "loss": 2.3406,
      "step": 5126
    },
    {
      "epoch": 0.03483461498773408,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000998196968369654,
      "loss": 2.4995,
      "step": 5127
    },
    {
      "epoch": 0.03484140933432814,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000998196055413794,
      "loss": 2.2594,
      "step": 5128
    },
    {
      "epoch": 0.034848203680922195,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000998195142227275,
      "loss": 2.3273,
      "step": 5129
    },
    {
      "epoch": 0.03485499802751625,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009981942288100972,
      "loss": 2.4239,
      "step": 5130
    },
    {
      "epoch": 0.034861792374110315,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009981933151622614,
      "loss": 2.4401,
      "step": 5131
    },
    {
      "epoch": 0.03486858672070437,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009981924012837675,
      "loss": 2.4565,
      "step": 5132
    },
    {
      "epoch": 0.03487538106729843,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009981914871746162,
      "loss": 2.5776,
      "step": 5133
    },
    {
      "epoch": 0.034882175413892484,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009981905728348082,
      "loss": 2.4941,
      "step": 5134
    },
    {
      "epoch": 0.03488896976048654,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009981896582643437,
      "loss": 2.3493,
      "step": 5135
    },
    {
      "epoch": 0.034895764107080604,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000998188743463223,
      "loss": 2.4273,
      "step": 5136
    },
    {
      "epoch": 0.03490255845367466,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009981878284314465,
      "loss": 2.4091,
      "step": 5137
    },
    {
      "epoch": 0.03490935280026872,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009981869131690149,
      "loss": 2.4337,
      "step": 5138
    },
    {
      "epoch": 0.034916147146862773,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009981859976759284,
      "loss": 2.3306,
      "step": 5139
    },
    {
      "epoch": 0.03492294149345683,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009981850819521876,
      "loss": 2.477,
      "step": 5140
    },
    {
      "epoch": 0.034929735840050886,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009981841659977928,
      "loss": 2.2225,
      "step": 5141
    },
    {
      "epoch": 0.03493653018664495,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009981832498127444,
      "loss": 2.2776,
      "step": 5142
    },
    {
      "epoch": 0.034943324533239006,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009981823333970428,
      "loss": 2.4622,
      "step": 5143
    },
    {
      "epoch": 0.03495011887983306,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009981814167506883,
      "loss": 2.3441,
      "step": 5144
    },
    {
      "epoch": 0.03495691322642712,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000998180499873682,
      "loss": 2.3228,
      "step": 5145
    },
    {
      "epoch": 0.034963707573021176,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009981795827660237,
      "loss": 2.416,
      "step": 5146
    },
    {
      "epoch": 0.03497050191961524,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009981786654277135,
      "loss": 2.4179,
      "step": 5147
    },
    {
      "epoch": 0.034977296266209296,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009981777478587528,
      "loss": 2.2407,
      "step": 5148
    },
    {
      "epoch": 0.03498409061280335,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009981768300591414,
      "loss": 2.491,
      "step": 5149
    },
    {
      "epoch": 0.03499088495939741,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009981759120288797,
      "loss": 2.3437,
      "step": 5150
    },
    {
      "epoch": 0.034997679305991465,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009981749937679684,
      "loss": 2.2242,
      "step": 5151
    },
    {
      "epoch": 0.03500447365258552,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009981740752764078,
      "loss": 2.5224,
      "step": 5152
    },
    {
      "epoch": 0.035011267999179585,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009981731565541982,
      "loss": 2.3249,
      "step": 5153
    },
    {
      "epoch": 0.03501806234577364,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009981722376013404,
      "loss": 2.2352,
      "step": 5154
    },
    {
      "epoch": 0.0350248566923677,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009981713184178342,
      "loss": 2.3464,
      "step": 5155
    },
    {
      "epoch": 0.035031651038961754,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009981703990036806,
      "loss": 2.4335,
      "step": 5156
    },
    {
      "epoch": 0.03503844538555581,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009981694793588798,
      "loss": 2.3366,
      "step": 5157
    },
    {
      "epoch": 0.035045239732149874,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009981685594834323,
      "loss": 2.3235,
      "step": 5158
    },
    {
      "epoch": 0.03505203407874393,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009981676393773385,
      "loss": 2.4417,
      "step": 5159
    },
    {
      "epoch": 0.03505882842533799,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009981667190405985,
      "loss": 2.3919,
      "step": 5160
    },
    {
      "epoch": 0.035065622771932044,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009981657984732134,
      "loss": 2.3457,
      "step": 5161
    },
    {
      "epoch": 0.0350724171185261,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998164877675183,
      "loss": 2.4496,
      "step": 5162
    },
    {
      "epoch": 0.03507921146512016,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000998163956646508,
      "loss": 2.4141,
      "step": 5163
    },
    {
      "epoch": 0.03508600581171422,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009981630353871889,
      "loss": 2.373,
      "step": 5164
    },
    {
      "epoch": 0.03509280015830828,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000998162113897226,
      "loss": 2.3166,
      "step": 5165
    },
    {
      "epoch": 0.03509959450490233,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009981611921766197,
      "loss": 2.3947,
      "step": 5166
    },
    {
      "epoch": 0.03510638885149639,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009981602702253704,
      "loss": 2.3831,
      "step": 5167
    },
    {
      "epoch": 0.035113183198090446,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009981593480434786,
      "loss": 2.4214,
      "step": 5168
    },
    {
      "epoch": 0.03511997754468451,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009981584256309448,
      "loss": 2.3592,
      "step": 5169
    },
    {
      "epoch": 0.035126771891278566,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009981575029877694,
      "loss": 2.551,
      "step": 5170
    },
    {
      "epoch": 0.03513356623787262,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009981565801139528,
      "loss": 2.3992,
      "step": 5171
    },
    {
      "epoch": 0.03514036058446668,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009981556570094953,
      "loss": 2.3761,
      "step": 5172
    },
    {
      "epoch": 0.035147154931060735,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009981547336743977,
      "loss": 2.4331,
      "step": 5173
    },
    {
      "epoch": 0.0351539492776548,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099815381010866,
      "loss": 2.319,
      "step": 5174
    },
    {
      "epoch": 0.035160743624248855,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009981528863122827,
      "loss": 2.2341,
      "step": 5175
    },
    {
      "epoch": 0.03516753797084291,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009981519622852663,
      "loss": 2.3504,
      "step": 5176
    },
    {
      "epoch": 0.03517433231743697,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009981510380276115,
      "loss": 2.4333,
      "step": 5177
    },
    {
      "epoch": 0.035181126664031025,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009981501135393183,
      "loss": 2.4779,
      "step": 5178
    },
    {
      "epoch": 0.03518792101062508,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009981491888203875,
      "loss": 2.3726,
      "step": 5179
    },
    {
      "epoch": 0.035194715357219145,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009981482638708192,
      "loss": 2.144,
      "step": 5180
    },
    {
      "epoch": 0.0352015097038132,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000998147338690614,
      "loss": 2.4809,
      "step": 5181
    },
    {
      "epoch": 0.03520830405040726,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009981464132797723,
      "loss": 2.3501,
      "step": 5182
    },
    {
      "epoch": 0.035215098397001314,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009981454876382943,
      "loss": 2.3311,
      "step": 5183
    },
    {
      "epoch": 0.03522189274359537,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000998144561766181,
      "loss": 2.4889,
      "step": 5184
    },
    {
      "epoch": 0.035228687090189434,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009981436356634322,
      "loss": 2.2093,
      "step": 5185
    },
    {
      "epoch": 0.03523548143678349,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009981427093300487,
      "loss": 2.446,
      "step": 5186
    },
    {
      "epoch": 0.03524227578337755,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998141782766031,
      "loss": 2.2915,
      "step": 5187
    },
    {
      "epoch": 0.0352490701299716,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000998140855971379,
      "loss": 2.3895,
      "step": 5188
    },
    {
      "epoch": 0.03525586447656566,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009981399289460938,
      "loss": 2.4292,
      "step": 5189
    },
    {
      "epoch": 0.035262658823159716,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009981390016901754,
      "loss": 2.4268,
      "step": 5190
    },
    {
      "epoch": 0.03526945316975378,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009981380742036243,
      "loss": 2.3303,
      "step": 5191
    },
    {
      "epoch": 0.035276247516347836,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998137146486441,
      "loss": 2.2463,
      "step": 5192
    },
    {
      "epoch": 0.03528304186294189,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009981362185386262,
      "loss": 2.3143,
      "step": 5193
    },
    {
      "epoch": 0.03528983620953595,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009981352903601797,
      "loss": 2.4264,
      "step": 5194
    },
    {
      "epoch": 0.035296630556130006,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009981343619511025,
      "loss": 2.2879,
      "step": 5195
    },
    {
      "epoch": 0.03530342490272407,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009981334333113948,
      "loss": 2.4443,
      "step": 5196
    },
    {
      "epoch": 0.035310219249318126,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009981325044410567,
      "loss": 2.1606,
      "step": 5197
    },
    {
      "epoch": 0.03531701359591218,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009981315753400893,
      "loss": 2.4064,
      "step": 5198
    },
    {
      "epoch": 0.03532380794250624,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009981306460084926,
      "loss": 2.3707,
      "step": 5199
    },
    {
      "epoch": 0.035330602289100295,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998129716446267,
      "loss": 2.3556,
      "step": 5200
    },
    {
      "epoch": 0.03533739663569435,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009981287866534132,
      "loss": 2.3748,
      "step": 5201
    },
    {
      "epoch": 0.035344190982288415,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009981278566299314,
      "loss": 2.2861,
      "step": 5202
    },
    {
      "epoch": 0.03535098532888247,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009981269263758222,
      "loss": 2.3558,
      "step": 5203
    },
    {
      "epoch": 0.03535777967547653,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998125995891086,
      "loss": 2.387,
      "step": 5204
    },
    {
      "epoch": 0.035364574022070584,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000998125065175723,
      "loss": 2.3083,
      "step": 5205
    },
    {
      "epoch": 0.03537136836866464,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009981241342297339,
      "loss": 2.4037,
      "step": 5206
    },
    {
      "epoch": 0.035378162715258704,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000998123203053119,
      "loss": 2.2594,
      "step": 5207
    },
    {
      "epoch": 0.03538495706185276,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009981222716458787,
      "loss": 2.3339,
      "step": 5208
    },
    {
      "epoch": 0.03539175140844682,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009981213400080135,
      "loss": 2.3441,
      "step": 5209
    },
    {
      "epoch": 0.035398545755040874,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998120408139524,
      "loss": 2.3657,
      "step": 5210
    },
    {
      "epoch": 0.03540534010163493,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009981194760404104,
      "loss": 2.3472,
      "step": 5211
    },
    {
      "epoch": 0.03541213444822899,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009981185437106732,
      "loss": 2.3429,
      "step": 5212
    },
    {
      "epoch": 0.03541892879482305,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009981176111503128,
      "loss": 2.3025,
      "step": 5213
    },
    {
      "epoch": 0.035425723141417106,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009981166783593297,
      "loss": 2.3425,
      "step": 5214
    },
    {
      "epoch": 0.03543251748801116,
      "grad_norm": 0.140625,
      "learning_rate": 0.000998115745337724,
      "loss": 2.4081,
      "step": 5215
    },
    {
      "epoch": 0.03543931183460522,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009981148120854968,
      "loss": 2.368,
      "step": 5216
    },
    {
      "epoch": 0.035446106181199276,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000998113878602648,
      "loss": 2.3513,
      "step": 5217
    },
    {
      "epoch": 0.03545290052779334,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009981129448891782,
      "loss": 2.2445,
      "step": 5218
    },
    {
      "epoch": 0.035459694874387396,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998112010945088,
      "loss": 2.226,
      "step": 5219
    },
    {
      "epoch": 0.03546648922098145,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009981110767703773,
      "loss": 2.3761,
      "step": 5220
    },
    {
      "epoch": 0.03547328356757551,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000998110142365047,
      "loss": 2.2707,
      "step": 5221
    },
    {
      "epoch": 0.035480077914169565,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009981092077290974,
      "loss": 2.4014,
      "step": 5222
    },
    {
      "epoch": 0.03548687226076362,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000998108272862529,
      "loss": 2.3954,
      "step": 5223
    },
    {
      "epoch": 0.035493666607357685,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009981073377653422,
      "loss": 2.2651,
      "step": 5224
    },
    {
      "epoch": 0.03550046095395174,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009981064024375374,
      "loss": 2.3943,
      "step": 5225
    },
    {
      "epoch": 0.0355072553005458,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000998105466879115,
      "loss": 2.2399,
      "step": 5226
    },
    {
      "epoch": 0.035514049647139855,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009981045310900757,
      "loss": 2.4387,
      "step": 5227
    },
    {
      "epoch": 0.03552084399373391,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009981035950704193,
      "loss": 2.2928,
      "step": 5228
    },
    {
      "epoch": 0.035527638340327974,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009981026588201472,
      "loss": 2.2797,
      "step": 5229
    },
    {
      "epoch": 0.03553443268692203,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009981017223392589,
      "loss": 2.4835,
      "step": 5230
    },
    {
      "epoch": 0.03554122703351609,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009981007856277552,
      "loss": 2.2396,
      "step": 5231
    },
    {
      "epoch": 0.035548021380110144,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009980998486856364,
      "loss": 2.3095,
      "step": 5232
    },
    {
      "epoch": 0.0355548157267042,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009980989115129036,
      "loss": 2.2888,
      "step": 5233
    },
    {
      "epoch": 0.035561610073298264,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009980979741095563,
      "loss": 2.4303,
      "step": 5234
    },
    {
      "epoch": 0.03556840441989232,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980970364755954,
      "loss": 2.3837,
      "step": 5235
    },
    {
      "epoch": 0.03557519876648638,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009980960986110214,
      "loss": 2.2259,
      "step": 5236
    },
    {
      "epoch": 0.03558199311308043,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009980951605158346,
      "loss": 2.3249,
      "step": 5237
    },
    {
      "epoch": 0.03558878745967449,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009980942221900354,
      "loss": 2.2403,
      "step": 5238
    },
    {
      "epoch": 0.035595581806268546,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009980932836336242,
      "loss": 2.2258,
      "step": 5239
    },
    {
      "epoch": 0.03560237615286261,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009980923448466017,
      "loss": 2.3657,
      "step": 5240
    },
    {
      "epoch": 0.035609170499456666,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000998091405828968,
      "loss": 2.418,
      "step": 5241
    },
    {
      "epoch": 0.03561596484605072,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009980904665807238,
      "loss": 2.2511,
      "step": 5242
    },
    {
      "epoch": 0.03562275919264478,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009980895271018695,
      "loss": 2.3643,
      "step": 5243
    },
    {
      "epoch": 0.035629553539238835,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009980885873924052,
      "loss": 2.1919,
      "step": 5244
    },
    {
      "epoch": 0.0356363478858329,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000998087647452332,
      "loss": 2.3912,
      "step": 5245
    },
    {
      "epoch": 0.035643142232426955,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009980867072816495,
      "loss": 2.3647,
      "step": 5246
    },
    {
      "epoch": 0.03564993657902101,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009980857668803586,
      "loss": 2.3652,
      "step": 5247
    },
    {
      "epoch": 0.03565673092561507,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099808482624846,
      "loss": 2.4886,
      "step": 5248
    },
    {
      "epoch": 0.035663525272209125,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009980838853859537,
      "loss": 2.4964,
      "step": 5249
    },
    {
      "epoch": 0.03567031961880318,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009980829442928403,
      "loss": 2.3441,
      "step": 5250
    },
    {
      "epoch": 0.035677113965397245,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00099808200296912,
      "loss": 2.235,
      "step": 5251
    },
    {
      "epoch": 0.0356839083119913,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009980810614147937,
      "loss": 2.2475,
      "step": 5252
    },
    {
      "epoch": 0.03569070265858536,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009980801196298615,
      "loss": 2.3646,
      "step": 5253
    },
    {
      "epoch": 0.035697497005179414,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009980791776143237,
      "loss": 2.3394,
      "step": 5254
    },
    {
      "epoch": 0.03570429135177347,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009980782353681812,
      "loss": 2.3852,
      "step": 5255
    },
    {
      "epoch": 0.035711085698367534,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009980772928914342,
      "loss": 2.4431,
      "step": 5256
    },
    {
      "epoch": 0.03571788004496159,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000998076350184083,
      "loss": 2.4264,
      "step": 5257
    },
    {
      "epoch": 0.03572467439155565,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009980754072461283,
      "loss": 2.2831,
      "step": 5258
    },
    {
      "epoch": 0.0357314687381497,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009980744640775703,
      "loss": 2.3159,
      "step": 5259
    },
    {
      "epoch": 0.03573826308474376,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009980735206784096,
      "loss": 2.3293,
      "step": 5260
    },
    {
      "epoch": 0.035745057431337816,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009980725770486463,
      "loss": 2.299,
      "step": 5261
    },
    {
      "epoch": 0.03575185177793188,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009980716331882812,
      "loss": 2.2205,
      "step": 5262
    },
    {
      "epoch": 0.035758646124525936,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000998070689097315,
      "loss": 2.5783,
      "step": 5263
    },
    {
      "epoch": 0.03576544047111999,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009980697447757475,
      "loss": 2.315,
      "step": 5264
    },
    {
      "epoch": 0.03577223481771405,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009980688002235794,
      "loss": 2.1682,
      "step": 5265
    },
    {
      "epoch": 0.035779029164308106,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009980678554408113,
      "loss": 2.2157,
      "step": 5266
    },
    {
      "epoch": 0.03578582351090217,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980669104274434,
      "loss": 2.3209,
      "step": 5267
    },
    {
      "epoch": 0.035792617857496226,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009980659651834763,
      "loss": 2.3251,
      "step": 5268
    },
    {
      "epoch": 0.03579941220409028,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099806501970891,
      "loss": 2.2473,
      "step": 5269
    },
    {
      "epoch": 0.03580620655068434,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000998064074003746,
      "loss": 2.4571,
      "step": 5270
    },
    {
      "epoch": 0.035813000897278395,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009980631280679835,
      "loss": 2.2241,
      "step": 5271
    },
    {
      "epoch": 0.03581979524387245,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980621819016236,
      "loss": 2.3225,
      "step": 5272
    },
    {
      "epoch": 0.035826589590466515,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980612355046667,
      "loss": 2.4258,
      "step": 5273
    },
    {
      "epoch": 0.03583338393706057,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000998060288877113,
      "loss": 2.4212,
      "step": 5274
    },
    {
      "epoch": 0.03584017828365463,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980593420189632,
      "loss": 2.2588,
      "step": 5275
    },
    {
      "epoch": 0.035846972630248684,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980583949302177,
      "loss": 2.3958,
      "step": 5276
    },
    {
      "epoch": 0.03585376697684274,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009980574476108769,
      "loss": 2.2697,
      "step": 5277
    },
    {
      "epoch": 0.035860561323436804,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000998056500060941,
      "loss": 2.419,
      "step": 5278
    },
    {
      "epoch": 0.03586735567003086,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998055552280411,
      "loss": 2.5335,
      "step": 5279
    },
    {
      "epoch": 0.03587415001662492,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009980546042692867,
      "loss": 2.3538,
      "step": 5280
    },
    {
      "epoch": 0.035880944363218974,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000998053656027569,
      "loss": 2.3927,
      "step": 5281
    },
    {
      "epoch": 0.03588773870981303,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009980527075552578,
      "loss": 2.4873,
      "step": 5282
    },
    {
      "epoch": 0.03589453305640709,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009980517588523542,
      "loss": 2.2353,
      "step": 5283
    },
    {
      "epoch": 0.03590132740300115,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009980508099188584,
      "loss": 2.2635,
      "step": 5284
    },
    {
      "epoch": 0.035908121749595207,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009980498607547706,
      "loss": 2.2991,
      "step": 5285
    },
    {
      "epoch": 0.03591491609618926,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009980489113600916,
      "loss": 2.4422,
      "step": 5286
    },
    {
      "epoch": 0.03592171044278332,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009980479617348215,
      "loss": 2.4388,
      "step": 5287
    },
    {
      "epoch": 0.035928504789377376,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.000998047011878961,
      "loss": 2.2647,
      "step": 5288
    },
    {
      "epoch": 0.03593529913597144,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009980460617925106,
      "loss": 2.2917,
      "step": 5289
    },
    {
      "epoch": 0.035942093482565496,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009980451114754705,
      "loss": 2.4139,
      "step": 5290
    },
    {
      "epoch": 0.03594888782915955,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000998044160927841,
      "loss": 2.3509,
      "step": 5291
    },
    {
      "epoch": 0.03595568217575361,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009980432101496229,
      "loss": 2.3987,
      "step": 5292
    },
    {
      "epoch": 0.035962476522347665,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009980422591408164,
      "loss": 2.3769,
      "step": 5293
    },
    {
      "epoch": 0.03596927086894173,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009980413079014223,
      "loss": 2.3743,
      "step": 5294
    },
    {
      "epoch": 0.035976065215535785,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009980403564314407,
      "loss": 2.3951,
      "step": 5295
    },
    {
      "epoch": 0.03598285956212984,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009980394047308721,
      "loss": 2.217,
      "step": 5296
    },
    {
      "epoch": 0.0359896539087239,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000998038452799717,
      "loss": 2.0813,
      "step": 5297
    },
    {
      "epoch": 0.035996448255317955,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009980375006379755,
      "loss": 2.3453,
      "step": 5298
    },
    {
      "epoch": 0.03600324260191201,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009980365482456486,
      "loss": 2.2719,
      "step": 5299
    },
    {
      "epoch": 0.036010036948506075,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009980355956227364,
      "loss": 2.4629,
      "step": 5300
    },
    {
      "epoch": 0.03601683129510013,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009980346427692396,
      "loss": 2.2968,
      "step": 5301
    },
    {
      "epoch": 0.03602362564169419,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980336896851583,
      "loss": 2.318,
      "step": 5302
    },
    {
      "epoch": 0.036030419988288244,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009980327363704933,
      "loss": 2.305,
      "step": 5303
    },
    {
      "epoch": 0.0360372143348823,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009980317828252449,
      "loss": 2.5475,
      "step": 5304
    },
    {
      "epoch": 0.036044008681476364,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980308290494131,
      "loss": 2.3908,
      "step": 5305
    },
    {
      "epoch": 0.03605080302807042,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000998029875042999,
      "loss": 2.3982,
      "step": 5306
    },
    {
      "epoch": 0.03605759737466448,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980289208060028,
      "loss": 2.3219,
      "step": 5307
    },
    {
      "epoch": 0.03606439172125853,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000998027966338425,
      "loss": 2.3924,
      "step": 5308
    },
    {
      "epoch": 0.03607118606785259,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009980270116402656,
      "loss": 2.186,
      "step": 5309
    },
    {
      "epoch": 0.036077980414446646,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009980260567115256,
      "loss": 2.2903,
      "step": 5310
    },
    {
      "epoch": 0.03608477476104071,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009980251015522053,
      "loss": 2.3392,
      "step": 5311
    },
    {
      "epoch": 0.036091569107634766,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000998024146162305,
      "loss": 2.148,
      "step": 5312
    },
    {
      "epoch": 0.03609836345422882,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009980231905418254,
      "loss": 2.3189,
      "step": 5313
    },
    {
      "epoch": 0.03610515780082288,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009980222346907665,
      "loss": 2.5392,
      "step": 5314
    },
    {
      "epoch": 0.036111952147416936,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009980212786091292,
      "loss": 2.2905,
      "step": 5315
    },
    {
      "epoch": 0.036118746494011,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009980203222969138,
      "loss": 2.3468,
      "step": 5316
    },
    {
      "epoch": 0.036125540840605055,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009980193657541207,
      "loss": 2.3495,
      "step": 5317
    },
    {
      "epoch": 0.03613233518719911,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009980184089807503,
      "loss": 2.362,
      "step": 5318
    },
    {
      "epoch": 0.03613912953379317,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980174519768031,
      "loss": 2.356,
      "step": 5319
    },
    {
      "epoch": 0.036145923880387225,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009980164947422795,
      "loss": 2.5694,
      "step": 5320
    },
    {
      "epoch": 0.03615271822698128,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009980155372771799,
      "loss": 2.2036,
      "step": 5321
    },
    {
      "epoch": 0.036159512573575345,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009980145795815047,
      "loss": 2.3451,
      "step": 5322
    },
    {
      "epoch": 0.0361663069201694,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009980136216552547,
      "loss": 2.5149,
      "step": 5323
    },
    {
      "epoch": 0.03617310126676346,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009980126634984302,
      "loss": 2.3335,
      "step": 5324
    },
    {
      "epoch": 0.036179895613357514,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009980117051110314,
      "loss": 2.4145,
      "step": 5325
    },
    {
      "epoch": 0.03618668995995157,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009980107464930587,
      "loss": 2.3502,
      "step": 5326
    },
    {
      "epoch": 0.036193484306545634,
      "grad_norm": 0.15625,
      "learning_rate": 0.000998009787644513,
      "loss": 2.4162,
      "step": 5327
    },
    {
      "epoch": 0.03620027865313969,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009980088285653943,
      "loss": 2.3831,
      "step": 5328
    },
    {
      "epoch": 0.03620707299973375,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009980078692557034,
      "loss": 2.4164,
      "step": 5329
    },
    {
      "epoch": 0.036213867346327804,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009980069097154404,
      "loss": 2.3038,
      "step": 5330
    },
    {
      "epoch": 0.03622066169292186,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980059499446062,
      "loss": 2.3035,
      "step": 5331
    },
    {
      "epoch": 0.036227456039515916,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009980049899432006,
      "loss": 2.3966,
      "step": 5332
    },
    {
      "epoch": 0.03623425038610998,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009980040297112244,
      "loss": 2.5497,
      "step": 5333
    },
    {
      "epoch": 0.036241044732704036,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009980030692486783,
      "loss": 2.298,
      "step": 5334
    },
    {
      "epoch": 0.03624783907929809,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009980021085555623,
      "loss": 2.4396,
      "step": 5335
    },
    {
      "epoch": 0.03625463342589215,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009980011476318773,
      "loss": 2.3301,
      "step": 5336
    },
    {
      "epoch": 0.036261427772486206,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009980001864776232,
      "loss": 2.2537,
      "step": 5337
    },
    {
      "epoch": 0.03626822211908027,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009979992250928008,
      "loss": 2.405,
      "step": 5338
    },
    {
      "epoch": 0.036275016465674326,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009979982634774106,
      "loss": 2.3873,
      "step": 5339
    },
    {
      "epoch": 0.03628181081226838,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009979973016314527,
      "loss": 2.3216,
      "step": 5340
    },
    {
      "epoch": 0.03628860515886244,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000997996339554928,
      "loss": 2.3273,
      "step": 5341
    },
    {
      "epoch": 0.036295399505456495,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009979953772478363,
      "loss": 2.453,
      "step": 5342
    },
    {
      "epoch": 0.03630219385205055,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009979944147101786,
      "loss": 2.4346,
      "step": 5343
    },
    {
      "epoch": 0.036308988198644615,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009979934519419554,
      "loss": 2.2252,
      "step": 5344
    },
    {
      "epoch": 0.03631578254523867,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009979924889431668,
      "loss": 2.3203,
      "step": 5345
    },
    {
      "epoch": 0.03632257689183273,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009979915257138136,
      "loss": 2.3823,
      "step": 5346
    },
    {
      "epoch": 0.036329371238426784,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009979905622538958,
      "loss": 2.3845,
      "step": 5347
    },
    {
      "epoch": 0.03633616558502084,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009979895985634142,
      "loss": 2.2454,
      "step": 5348
    },
    {
      "epoch": 0.036342959931614904,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997988634642369,
      "loss": 2.085,
      "step": 5349
    },
    {
      "epoch": 0.03634975427820896,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009979876704907609,
      "loss": 2.4898,
      "step": 5350
    },
    {
      "epoch": 0.03635654862480302,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009979867061085902,
      "loss": 2.49,
      "step": 5351
    },
    {
      "epoch": 0.036363342971397074,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009979857414958572,
      "loss": 2.4222,
      "step": 5352
    },
    {
      "epoch": 0.03637013731799113,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009979847766525625,
      "loss": 2.3585,
      "step": 5353
    },
    {
      "epoch": 0.036376931664585194,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009979838115787068,
      "loss": 2.348,
      "step": 5354
    },
    {
      "epoch": 0.03638372601117925,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979828462742901,
      "loss": 2.3421,
      "step": 5355
    },
    {
      "epoch": 0.03639052035777331,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009979818807393133,
      "loss": 2.4701,
      "step": 5356
    },
    {
      "epoch": 0.03639731470436736,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979809149737763,
      "loss": 2.281,
      "step": 5357
    },
    {
      "epoch": 0.03640410905096142,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00099797994897768,
      "loss": 2.3489,
      "step": 5358
    },
    {
      "epoch": 0.036410903397555476,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009979789827510247,
      "loss": 2.2483,
      "step": 5359
    },
    {
      "epoch": 0.03641769774414954,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009979780162938108,
      "loss": 2.4422,
      "step": 5360
    },
    {
      "epoch": 0.036424492090743596,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009979770496060388,
      "loss": 2.383,
      "step": 5361
    },
    {
      "epoch": 0.03643128643733765,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009979760826877091,
      "loss": 2.391,
      "step": 5362
    },
    {
      "epoch": 0.03643808078393171,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009979751155388223,
      "loss": 2.4105,
      "step": 5363
    },
    {
      "epoch": 0.036444875130525765,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009979741481593786,
      "loss": 2.5162,
      "step": 5364
    },
    {
      "epoch": 0.03645166947711983,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009979731805493787,
      "loss": 2.4193,
      "step": 5365
    },
    {
      "epoch": 0.036458463823713885,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000997972212708823,
      "loss": 2.2154,
      "step": 5366
    },
    {
      "epoch": 0.03646525817030794,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009979712446377116,
      "loss": 2.3831,
      "step": 5367
    },
    {
      "epoch": 0.036472052516902,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009979702763360453,
      "loss": 2.2908,
      "step": 5368
    },
    {
      "epoch": 0.036478846863496055,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979693078038246,
      "loss": 2.3928,
      "step": 5369
    },
    {
      "epoch": 0.03648564121009011,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009979683390410497,
      "loss": 2.1917,
      "step": 5370
    },
    {
      "epoch": 0.036492435556684175,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009979673700477212,
      "loss": 2.205,
      "step": 5371
    },
    {
      "epoch": 0.03649922990327823,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009979664008238395,
      "loss": 2.4214,
      "step": 5372
    },
    {
      "epoch": 0.03650602424987229,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997965431369405,
      "loss": 2.3199,
      "step": 5373
    },
    {
      "epoch": 0.036512818596466344,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009979644616844184,
      "loss": 2.3997,
      "step": 5374
    },
    {
      "epoch": 0.0365196129430604,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009979634917688799,
      "loss": 2.3372,
      "step": 5375
    },
    {
      "epoch": 0.036526407289654464,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099796252162279,
      "loss": 2.4303,
      "step": 5376
    },
    {
      "epoch": 0.03653320163624852,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997961551246149,
      "loss": 2.258,
      "step": 5377
    },
    {
      "epoch": 0.03653999598284258,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009979605806389577,
      "loss": 2.3917,
      "step": 5378
    },
    {
      "epoch": 0.03654679032943663,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009979596098012163,
      "loss": 2.2607,
      "step": 5379
    },
    {
      "epoch": 0.03655358467603069,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009979586387329255,
      "loss": 2.2812,
      "step": 5380
    },
    {
      "epoch": 0.036560379022624746,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009979576674340853,
      "loss": 2.4608,
      "step": 5381
    },
    {
      "epoch": 0.03656717336921881,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009979566959046964,
      "loss": 2.3535,
      "step": 5382
    },
    {
      "epoch": 0.036573967715812866,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009979557241447594,
      "loss": 2.3833,
      "step": 5383
    },
    {
      "epoch": 0.03658076206240692,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979547521542747,
      "loss": 2.356,
      "step": 5384
    },
    {
      "epoch": 0.03658755640900098,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009979537799332423,
      "loss": 2.2964,
      "step": 5385
    },
    {
      "epoch": 0.036594350755595036,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009979528074816632,
      "loss": 2.2809,
      "step": 5386
    },
    {
      "epoch": 0.0366011451021891,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009979518347995378,
      "loss": 2.2591,
      "step": 5387
    },
    {
      "epoch": 0.036607939448783156,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009979508618868664,
      "loss": 2.1329,
      "step": 5388
    },
    {
      "epoch": 0.03661473379537721,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979498887436493,
      "loss": 2.3479,
      "step": 5389
    },
    {
      "epoch": 0.03662152814197127,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009979489153698871,
      "loss": 2.2667,
      "step": 5390
    },
    {
      "epoch": 0.036628322488565325,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009979479417655803,
      "loss": 2.3395,
      "step": 5391
    },
    {
      "epoch": 0.03663511683515938,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009979469679307293,
      "loss": 2.3704,
      "step": 5392
    },
    {
      "epoch": 0.036641911181753445,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979459938653347,
      "loss": 2.2885,
      "step": 5393
    },
    {
      "epoch": 0.0366487055283475,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009979450195693967,
      "loss": 2.2519,
      "step": 5394
    },
    {
      "epoch": 0.03665549987494156,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000997944045042916,
      "loss": 2.3776,
      "step": 5395
    },
    {
      "epoch": 0.036662294221535614,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009979430702858926,
      "loss": 2.5733,
      "step": 5396
    },
    {
      "epoch": 0.03666908856812967,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979420952983275,
      "loss": 2.2213,
      "step": 5397
    },
    {
      "epoch": 0.036675882914723734,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997941120080221,
      "loss": 2.29,
      "step": 5398
    },
    {
      "epoch": 0.03668267726131779,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009979401446315733,
      "loss": 2.3347,
      "step": 5399
    },
    {
      "epoch": 0.03668947160791185,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000997939168952385,
      "loss": 2.3223,
      "step": 5400
    },
    {
      "epoch": 0.036696265954505904,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009979381930426568,
      "loss": 2.5384,
      "step": 5401
    },
    {
      "epoch": 0.03670306030109996,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009979372169023888,
      "loss": 2.2625,
      "step": 5402
    },
    {
      "epoch": 0.03670985464769402,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009979362405315815,
      "loss": 2.1673,
      "step": 5403
    },
    {
      "epoch": 0.03671664899428808,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009979352639302354,
      "loss": 2.4317,
      "step": 5404
    },
    {
      "epoch": 0.036723443340882136,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997934287098351,
      "loss": 2.4725,
      "step": 5405
    },
    {
      "epoch": 0.03673023768747619,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000997933310035929,
      "loss": 2.3872,
      "step": 5406
    },
    {
      "epoch": 0.03673703203407025,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009979323327429692,
      "loss": 2.372,
      "step": 5407
    },
    {
      "epoch": 0.036743826380664306,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009979313552194726,
      "loss": 2.3651,
      "step": 5408
    },
    {
      "epoch": 0.03675062072725837,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009979303774654395,
      "loss": 2.2099,
      "step": 5409
    },
    {
      "epoch": 0.036757415073852426,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009979293994808705,
      "loss": 2.1563,
      "step": 5410
    },
    {
      "epoch": 0.03676420942044648,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009979284212657657,
      "loss": 2.2606,
      "step": 5411
    },
    {
      "epoch": 0.03677100376704054,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009979274428201256,
      "loss": 2.2384,
      "step": 5412
    },
    {
      "epoch": 0.036777798113634595,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009979264641439512,
      "loss": 2.3448,
      "step": 5413
    },
    {
      "epoch": 0.03678459246022866,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009979254852372423,
      "loss": 2.2948,
      "step": 5414
    },
    {
      "epoch": 0.036791386806822715,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009979245060999997,
      "loss": 2.2396,
      "step": 5415
    },
    {
      "epoch": 0.03679818115341677,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009979235267322238,
      "loss": 2.3946,
      "step": 5416
    },
    {
      "epoch": 0.03680497550001083,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009979225471339148,
      "loss": 2.2841,
      "step": 5417
    },
    {
      "epoch": 0.036811769846604885,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009979215673050735,
      "loss": 2.4192,
      "step": 5418
    },
    {
      "epoch": 0.03681856419319894,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009979205872457002,
      "loss": 2.4138,
      "step": 5419
    },
    {
      "epoch": 0.036825358539793004,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009979196069557954,
      "loss": 2.3405,
      "step": 5420
    },
    {
      "epoch": 0.03683215288638706,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009979186264353597,
      "loss": 2.3267,
      "step": 5421
    },
    {
      "epoch": 0.03683894723298112,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009979176456843932,
      "loss": 2.3027,
      "step": 5422
    },
    {
      "epoch": 0.036845741579575174,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009979166647028966,
      "loss": 2.2685,
      "step": 5423
    },
    {
      "epoch": 0.03685253592616923,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009979156834908703,
      "loss": 2.2783,
      "step": 5424
    },
    {
      "epoch": 0.036859330272763294,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009979147020483146,
      "loss": 2.3203,
      "step": 5425
    },
    {
      "epoch": 0.03686612461935735,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009979137203752302,
      "loss": 2.4036,
      "step": 5426
    },
    {
      "epoch": 0.03687291896595141,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009979127384716175,
      "loss": 2.4289,
      "step": 5427
    },
    {
      "epoch": 0.03687971331254546,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009979117563374768,
      "loss": 2.4247,
      "step": 5428
    },
    {
      "epoch": 0.03688650765913952,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009979107739728089,
      "loss": 2.4523,
      "step": 5429
    },
    {
      "epoch": 0.036893302005733576,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009979097913776136,
      "loss": 2.311,
      "step": 5430
    },
    {
      "epoch": 0.03690009635232764,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009979088085518921,
      "loss": 2.4003,
      "step": 5431
    },
    {
      "epoch": 0.036906890698921696,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009979078254956444,
      "loss": 2.3149,
      "step": 5432
    },
    {
      "epoch": 0.03691368504551575,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009979068422088712,
      "loss": 2.2181,
      "step": 5433
    },
    {
      "epoch": 0.03692047939210981,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009979058586915728,
      "loss": 2.2878,
      "step": 5434
    },
    {
      "epoch": 0.036927273738703865,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009979048749437498,
      "loss": 2.3222,
      "step": 5435
    },
    {
      "epoch": 0.03693406808529793,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009979038909654024,
      "loss": 2.3968,
      "step": 5436
    },
    {
      "epoch": 0.036940862431891985,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009979029067565311,
      "loss": 2.3777,
      "step": 5437
    },
    {
      "epoch": 0.03694765677848604,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009979019223171365,
      "loss": 2.2282,
      "step": 5438
    },
    {
      "epoch": 0.0369544511250801,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009979009376472193,
      "loss": 2.4302,
      "step": 5439
    },
    {
      "epoch": 0.036961245471674155,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009978999527467795,
      "loss": 2.3043,
      "step": 5440
    },
    {
      "epoch": 0.03696803981826821,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009978989676158176,
      "loss": 2.4063,
      "step": 5441
    },
    {
      "epoch": 0.036974834164862275,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009978979822543343,
      "loss": 2.1867,
      "step": 5442
    },
    {
      "epoch": 0.03698162851145633,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009978969966623302,
      "loss": 2.6281,
      "step": 5443
    },
    {
      "epoch": 0.03698842285805039,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009978960108398053,
      "loss": 2.2677,
      "step": 5444
    },
    {
      "epoch": 0.036995217204644444,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00099789502478676,
      "loss": 2.2709,
      "step": 5445
    },
    {
      "epoch": 0.0370020115512385,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009978940385031952,
      "loss": 2.4215,
      "step": 5446
    },
    {
      "epoch": 0.037008805897832564,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009978930519891113,
      "loss": 2.5203,
      "step": 5447
    },
    {
      "epoch": 0.03701560024442662,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009978920652445086,
      "loss": 2.2146,
      "step": 5448
    },
    {
      "epoch": 0.03702239459102068,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009978910782693874,
      "loss": 2.3745,
      "step": 5449
    },
    {
      "epoch": 0.03702918893761473,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009978900910637483,
      "loss": 2.2976,
      "step": 5450
    },
    {
      "epoch": 0.03703598328420879,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000997889103627592,
      "loss": 2.4229,
      "step": 5451
    },
    {
      "epoch": 0.037042777630802846,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009978881159609188,
      "loss": 2.4677,
      "step": 5452
    },
    {
      "epoch": 0.03704957197739691,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000997887128063729,
      "loss": 2.2651,
      "step": 5453
    },
    {
      "epoch": 0.037056366323990966,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000997886139936023,
      "loss": 2.3949,
      "step": 5454
    },
    {
      "epoch": 0.03706316067058502,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009978851515778015,
      "loss": 2.366,
      "step": 5455
    },
    {
      "epoch": 0.03706995501717908,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997884162989065,
      "loss": 2.266,
      "step": 5456
    },
    {
      "epoch": 0.037076749363773136,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997883174169814,
      "loss": 2.4834,
      "step": 5457
    },
    {
      "epoch": 0.0370835437103672,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009978821851200485,
      "loss": 2.2569,
      "step": 5458
    },
    {
      "epoch": 0.037090338056961256,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009978811958397693,
      "loss": 2.2198,
      "step": 5459
    },
    {
      "epoch": 0.03709713240355531,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000997880206328977,
      "loss": 2.393,
      "step": 5460
    },
    {
      "epoch": 0.03710392675014937,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009978792165876718,
      "loss": 2.3751,
      "step": 5461
    },
    {
      "epoch": 0.037110721096743425,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009978782266158541,
      "loss": 2.2845,
      "step": 5462
    },
    {
      "epoch": 0.03711751544333748,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009978772364135245,
      "loss": 2.2855,
      "step": 5463
    },
    {
      "epoch": 0.037124309789931545,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009978762459806837,
      "loss": 2.2888,
      "step": 5464
    },
    {
      "epoch": 0.0371311041365256,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009978752553173316,
      "loss": 2.3433,
      "step": 5465
    },
    {
      "epoch": 0.03713789848311966,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009978742644234692,
      "loss": 2.3865,
      "step": 5466
    },
    {
      "epoch": 0.037144692829713714,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009978732732990965,
      "loss": 2.2409,
      "step": 5467
    },
    {
      "epoch": 0.03715148717630777,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009978722819442144,
      "loss": 2.3321,
      "step": 5468
    },
    {
      "epoch": 0.037158281522901834,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997871290358823,
      "loss": 2.2157,
      "step": 5469
    },
    {
      "epoch": 0.03716507586949589,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000997870298542923,
      "loss": 2.4484,
      "step": 5470
    },
    {
      "epoch": 0.03717187021608995,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009978693064965146,
      "loss": 2.3986,
      "step": 5471
    },
    {
      "epoch": 0.037178664562684004,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009978683142195985,
      "loss": 2.2392,
      "step": 5472
    },
    {
      "epoch": 0.03718545890927806,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000997867321712175,
      "loss": 2.3183,
      "step": 5473
    },
    {
      "epoch": 0.037192253255872124,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009978663289742446,
      "loss": 2.2969,
      "step": 5474
    },
    {
      "epoch": 0.03719904760246618,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009978653360058082,
      "loss": 2.3273,
      "step": 5475
    },
    {
      "epoch": 0.03720584194906024,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009978643428068654,
      "loss": 2.4106,
      "step": 5476
    },
    {
      "epoch": 0.03721263629565429,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009978633493774175,
      "loss": 2.2879,
      "step": 5477
    },
    {
      "epoch": 0.03721943064224835,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009978623557174641,
      "loss": 2.3391,
      "step": 5478
    },
    {
      "epoch": 0.037226224988842406,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009978613618270066,
      "loss": 2.3845,
      "step": 5479
    },
    {
      "epoch": 0.03723301933543647,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009978603677060448,
      "loss": 2.2967,
      "step": 5480
    },
    {
      "epoch": 0.037239813682030526,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009978593733545793,
      "loss": 2.3256,
      "step": 5481
    },
    {
      "epoch": 0.03724660802862458,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009978583787726108,
      "loss": 2.3425,
      "step": 5482
    },
    {
      "epoch": 0.03725340237521864,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009978573839601393,
      "loss": 2.4036,
      "step": 5483
    },
    {
      "epoch": 0.037260196721812695,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009978563889171657,
      "loss": 2.3058,
      "step": 5484
    },
    {
      "epoch": 0.03726699106840676,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009978553936436903,
      "loss": 2.2847,
      "step": 5485
    },
    {
      "epoch": 0.037273785415000815,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009978543981397137,
      "loss": 2.3159,
      "step": 5486
    },
    {
      "epoch": 0.03728057976159487,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.000997853402405236,
      "loss": 2.4663,
      "step": 5487
    },
    {
      "epoch": 0.03728737410818893,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997852406440258,
      "loss": 2.2823,
      "step": 5488
    },
    {
      "epoch": 0.037294168454782985,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009978514102447799,
      "loss": 2.2136,
      "step": 5489
    },
    {
      "epoch": 0.03730096280137704,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009978504138188026,
      "loss": 2.2734,
      "step": 5490
    },
    {
      "epoch": 0.037307757147971105,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997849417162326,
      "loss": 2.3902,
      "step": 5491
    },
    {
      "epoch": 0.03731455149456516,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997848420275351,
      "loss": 2.4071,
      "step": 5492
    },
    {
      "epoch": 0.03732134584115922,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009978474231578777,
      "loss": 2.3957,
      "step": 5493
    },
    {
      "epoch": 0.037328140187753274,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997846425809907,
      "loss": 2.2858,
      "step": 5494
    },
    {
      "epoch": 0.03733493453434733,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009978454282314388,
      "loss": 2.2757,
      "step": 5495
    },
    {
      "epoch": 0.037341728880941394,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997844430422474,
      "loss": 2.3558,
      "step": 5496
    },
    {
      "epoch": 0.03734852322753545,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997843432383013,
      "loss": 2.2791,
      "step": 5497
    },
    {
      "epoch": 0.03735531757412951,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009978424341130561,
      "loss": 2.4316,
      "step": 5498
    },
    {
      "epoch": 0.03736211192072356,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000997841435612604,
      "loss": 2.2857,
      "step": 5499
    },
    {
      "epoch": 0.03736890626731762,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000997840436881657,
      "loss": 2.3569,
      "step": 5500
    },
    {
      "epoch": 0.037375700613911676,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009978394379202156,
      "loss": 2.4058,
      "step": 5501
    },
    {
      "epoch": 0.03738249496050574,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00099783843872828,
      "loss": 2.2922,
      "step": 5502
    },
    {
      "epoch": 0.037389289307099796,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009978374393058511,
      "loss": 2.3534,
      "step": 5503
    },
    {
      "epoch": 0.03739608365369385,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009978364396529293,
      "loss": 2.3902,
      "step": 5504
    },
    {
      "epoch": 0.03740287800028791,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997835439769515,
      "loss": 2.4258,
      "step": 5505
    },
    {
      "epoch": 0.037409672346881966,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009978344396556083,
      "loss": 2.3544,
      "step": 5506
    },
    {
      "epoch": 0.03741646669347603,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009978334393112099,
      "loss": 2.154,
      "step": 5507
    },
    {
      "epoch": 0.037423261040070085,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009978324387363206,
      "loss": 2.4233,
      "step": 5508
    },
    {
      "epoch": 0.03743005538666414,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009978314379309404,
      "loss": 2.4169,
      "step": 5509
    },
    {
      "epoch": 0.0374368497332582,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009978304368950701,
      "loss": 2.2193,
      "step": 5510
    },
    {
      "epoch": 0.037443644079852255,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00099782943562871,
      "loss": 2.4864,
      "step": 5511
    },
    {
      "epoch": 0.03745043842644631,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009978284341318603,
      "loss": 2.3352,
      "step": 5512
    },
    {
      "epoch": 0.037457232773040375,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000997827432404522,
      "loss": 2.3653,
      "step": 5513
    },
    {
      "epoch": 0.03746402711963443,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009978264304466953,
      "loss": 2.3139,
      "step": 5514
    },
    {
      "epoch": 0.03747082146622849,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009978254282583807,
      "loss": 2.3155,
      "step": 5515
    },
    {
      "epoch": 0.037477615812822544,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009978244258395785,
      "loss": 2.4755,
      "step": 5516
    },
    {
      "epoch": 0.0374844101594166,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009978234231902893,
      "loss": 2.234,
      "step": 5517
    },
    {
      "epoch": 0.037491204506010664,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009978224203105138,
      "loss": 2.3331,
      "step": 5518
    },
    {
      "epoch": 0.03749799885260472,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000997821417200252,
      "loss": 2.2957,
      "step": 5519
    },
    {
      "epoch": 0.03750479319919878,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009978204138595048,
      "loss": 2.4255,
      "step": 5520
    },
    {
      "epoch": 0.037511587545792834,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009978194102882724,
      "loss": 2.5491,
      "step": 5521
    },
    {
      "epoch": 0.03751838189238689,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000997818406486555,
      "loss": 2.2964,
      "step": 5522
    },
    {
      "epoch": 0.037525176238980947,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009978174024543537,
      "loss": 2.3363,
      "step": 5523
    },
    {
      "epoch": 0.03753197058557501,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009978163981916686,
      "loss": 2.4007,
      "step": 5524
    },
    {
      "epoch": 0.037538764932169066,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009978153936985002,
      "loss": 2.375,
      "step": 5525
    },
    {
      "epoch": 0.03754555927876312,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000997814388974849,
      "loss": 2.4824,
      "step": 5526
    },
    {
      "epoch": 0.03755235362535718,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009978133840207154,
      "loss": 2.3272,
      "step": 5527
    },
    {
      "epoch": 0.037559147971951236,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009978123788360997,
      "loss": 2.3403,
      "step": 5528
    },
    {
      "epoch": 0.0375659423185453,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000997811373421003,
      "loss": 2.2102,
      "step": 5529
    },
    {
      "epoch": 0.037572736665139356,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009978103677754251,
      "loss": 2.3899,
      "step": 5530
    },
    {
      "epoch": 0.03757953101173341,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009978093618993667,
      "loss": 2.3995,
      "step": 5531
    },
    {
      "epoch": 0.03758632535832747,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009978083557928283,
      "loss": 2.3482,
      "step": 5532
    },
    {
      "epoch": 0.037593119704921525,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009978073494558105,
      "loss": 2.4007,
      "step": 5533
    },
    {
      "epoch": 0.03759991405151559,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009978063428883132,
      "loss": 2.3244,
      "step": 5534
    },
    {
      "epoch": 0.037606708398109645,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009978053360903377,
      "loss": 2.1646,
      "step": 5535
    },
    {
      "epoch": 0.0376135027447037,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009978043290618838,
      "loss": 2.4926,
      "step": 5536
    },
    {
      "epoch": 0.03762029709129776,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009978033218029524,
      "loss": 2.3828,
      "step": 5537
    },
    {
      "epoch": 0.037627091437891814,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009978023143135436,
      "loss": 2.3292,
      "step": 5538
    },
    {
      "epoch": 0.03763388578448587,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000997801306593658,
      "loss": 2.312,
      "step": 5539
    },
    {
      "epoch": 0.037640680131079934,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009978002986432961,
      "loss": 2.297,
      "step": 5540
    },
    {
      "epoch": 0.03764747447767399,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009977992904624585,
      "loss": 2.4138,
      "step": 5541
    },
    {
      "epoch": 0.03765426882426805,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009977982820511455,
      "loss": 2.3171,
      "step": 5542
    },
    {
      "epoch": 0.037661063170862104,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009977972734093576,
      "loss": 2.2672,
      "step": 5543
    },
    {
      "epoch": 0.03766785751745616,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009977962645370954,
      "loss": 2.2652,
      "step": 5544
    },
    {
      "epoch": 0.037674651864050224,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009977952554343589,
      "loss": 2.4015,
      "step": 5545
    },
    {
      "epoch": 0.03768144621064428,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009977942461011492,
      "loss": 2.3683,
      "step": 5546
    },
    {
      "epoch": 0.03768824055723834,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009977932365374663,
      "loss": 2.463,
      "step": 5547
    },
    {
      "epoch": 0.03769503490383239,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000997792226743311,
      "loss": 2.3124,
      "step": 5548
    },
    {
      "epoch": 0.03770182925042645,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977912167186834,
      "loss": 2.3674,
      "step": 5549
    },
    {
      "epoch": 0.037708623597020506,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009977902064635842,
      "loss": 2.2749,
      "step": 5550
    },
    {
      "epoch": 0.03771541794361457,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009977891959780138,
      "loss": 2.2148,
      "step": 5551
    },
    {
      "epoch": 0.037722212290208626,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997788185261973,
      "loss": 2.3972,
      "step": 5552
    },
    {
      "epoch": 0.03772900663680268,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009977871743154618,
      "loss": 2.3596,
      "step": 5553
    },
    {
      "epoch": 0.03773580098339674,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009977861631384809,
      "loss": 2.2032,
      "step": 5554
    },
    {
      "epoch": 0.037742595329990795,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009977851517310305,
      "loss": 2.4701,
      "step": 5555
    },
    {
      "epoch": 0.03774938967658486,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009977841400931116,
      "loss": 2.2207,
      "step": 5556
    },
    {
      "epoch": 0.037756184023178915,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009977831282247242,
      "loss": 2.2456,
      "step": 5557
    },
    {
      "epoch": 0.03776297836977297,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997782116125869,
      "loss": 2.1858,
      "step": 5558
    },
    {
      "epoch": 0.03776977271636703,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009977811037965461,
      "loss": 2.2865,
      "step": 5559
    },
    {
      "epoch": 0.037776567062961085,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009977800912367564,
      "loss": 2.3395,
      "step": 5560
    },
    {
      "epoch": 0.03778336140955514,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977790784465005,
      "loss": 2.3877,
      "step": 5561
    },
    {
      "epoch": 0.037790155756149205,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009977780654257782,
      "loss": 2.2587,
      "step": 5562
    },
    {
      "epoch": 0.03779695010274326,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009977770521745906,
      "loss": 2.4051,
      "step": 5563
    },
    {
      "epoch": 0.03780374444933732,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009977760386929379,
      "loss": 2.3639,
      "step": 5564
    },
    {
      "epoch": 0.037810538795931374,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009977750249808206,
      "loss": 2.6034,
      "step": 5565
    },
    {
      "epoch": 0.03781733314252543,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000997774011038239,
      "loss": 2.3833,
      "step": 5566
    },
    {
      "epoch": 0.037824127489119494,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000997772996865194,
      "loss": 2.2445,
      "step": 5567
    },
    {
      "epoch": 0.03783092183571355,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009977719824616857,
      "loss": 2.4117,
      "step": 5568
    },
    {
      "epoch": 0.03783771618230761,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009977709678277145,
      "loss": 2.2763,
      "step": 5569
    },
    {
      "epoch": 0.03784451052890166,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009977699529632813,
      "loss": 2.4443,
      "step": 5570
    },
    {
      "epoch": 0.03785130487549572,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009977689378683863,
      "loss": 2.3889,
      "step": 5571
    },
    {
      "epoch": 0.037858099222089776,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099776792254303,
      "loss": 2.4076,
      "step": 5572
    },
    {
      "epoch": 0.03786489356868384,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009977669069872125,
      "loss": 2.2713,
      "step": 5573
    },
    {
      "epoch": 0.037871687915277896,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009977658912009352,
      "loss": 2.4228,
      "step": 5574
    },
    {
      "epoch": 0.03787848226187195,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009977648751841976,
      "loss": 2.3611,
      "step": 5575
    },
    {
      "epoch": 0.03788527660846601,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009977638589370006,
      "loss": 2.2947,
      "step": 5576
    },
    {
      "epoch": 0.037892070955060066,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009977628424593448,
      "loss": 2.5332,
      "step": 5577
    },
    {
      "epoch": 0.03789886530165413,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009977618257512304,
      "loss": 2.3576,
      "step": 5578
    },
    {
      "epoch": 0.037905659648248186,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000997760808812658,
      "loss": 2.3258,
      "step": 5579
    },
    {
      "epoch": 0.03791245399484224,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000997759791643628,
      "loss": 2.3145,
      "step": 5580
    },
    {
      "epoch": 0.0379192483414363,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997758774244141,
      "loss": 2.3651,
      "step": 5581
    },
    {
      "epoch": 0.037926042688030355,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977577566141975,
      "loss": 2.4946,
      "step": 5582
    },
    {
      "epoch": 0.03793283703462441,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977567387537978,
      "loss": 2.3625,
      "step": 5583
    },
    {
      "epoch": 0.037939631381218475,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009977557206629422,
      "loss": 2.3877,
      "step": 5584
    },
    {
      "epoch": 0.03794642572781253,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009977547023416317,
      "loss": 2.4187,
      "step": 5585
    },
    {
      "epoch": 0.03795322007440659,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009977536837898664,
      "loss": 2.3018,
      "step": 5586
    },
    {
      "epoch": 0.037960014421000644,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977526650076468,
      "loss": 2.1747,
      "step": 5587
    },
    {
      "epoch": 0.0379668087675947,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009977516459949733,
      "loss": 2.4136,
      "step": 5588
    },
    {
      "epoch": 0.037973603114188764,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009977506267518468,
      "loss": 2.4742,
      "step": 5589
    },
    {
      "epoch": 0.03798039746078282,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009977496072782672,
      "loss": 2.3637,
      "step": 5590
    },
    {
      "epoch": 0.03798719180737688,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009977485875742353,
      "loss": 2.3627,
      "step": 5591
    },
    {
      "epoch": 0.037993986153970934,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009977475676397516,
      "loss": 2.2074,
      "step": 5592
    },
    {
      "epoch": 0.03800078050056499,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009977465474748164,
      "loss": 2.2077,
      "step": 5593
    },
    {
      "epoch": 0.038007574847159054,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009977455270794304,
      "loss": 2.4532,
      "step": 5594
    },
    {
      "epoch": 0.03801436919375311,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009977445064535938,
      "loss": 2.3311,
      "step": 5595
    },
    {
      "epoch": 0.038021163540347166,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997743485597307,
      "loss": 2.3708,
      "step": 5596
    },
    {
      "epoch": 0.03802795788694122,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997742464510571,
      "loss": 2.2693,
      "step": 5597
    },
    {
      "epoch": 0.03803475223353528,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000997741443193386,
      "loss": 2.3548,
      "step": 5598
    },
    {
      "epoch": 0.038041546580129336,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009977404216457522,
      "loss": 2.2685,
      "step": 5599
    },
    {
      "epoch": 0.0380483409267234,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009977393998676703,
      "loss": 2.3371,
      "step": 5600
    },
    {
      "epoch": 0.038055135273317456,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977383778591409,
      "loss": 2.2729,
      "step": 5601
    },
    {
      "epoch": 0.03806192961991151,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000997737355620164,
      "loss": 2.3413,
      "step": 5602
    },
    {
      "epoch": 0.03806872396650557,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009977363331507407,
      "loss": 2.1018,
      "step": 5603
    },
    {
      "epoch": 0.038075518313099625,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977353104508712,
      "loss": 2.2587,
      "step": 5604
    },
    {
      "epoch": 0.03808231265969369,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977342875205559,
      "loss": 2.4132,
      "step": 5605
    },
    {
      "epoch": 0.038089107006287745,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009977332643597953,
      "loss": 2.4373,
      "step": 5606
    },
    {
      "epoch": 0.0380959013528818,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099773224096859,
      "loss": 2.1798,
      "step": 5607
    },
    {
      "epoch": 0.03810269569947586,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009977312173469403,
      "loss": 2.1278,
      "step": 5608
    },
    {
      "epoch": 0.038109490046069915,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000997730193494847,
      "loss": 2.2963,
      "step": 5609
    },
    {
      "epoch": 0.03811628439266397,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009977291694123099,
      "loss": 2.2462,
      "step": 5610
    },
    {
      "epoch": 0.038123078739258034,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009977281450993301,
      "loss": 2.2447,
      "step": 5611
    },
    {
      "epoch": 0.03812987308585209,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009977271205559078,
      "loss": 2.3599,
      "step": 5612
    },
    {
      "epoch": 0.03813666743244615,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009977260957820437,
      "loss": 2.3338,
      "step": 5613
    },
    {
      "epoch": 0.038143461779040204,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000997725070777738,
      "loss": 2.2532,
      "step": 5614
    },
    {
      "epoch": 0.03815025612563426,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977240455429914,
      "loss": 2.4161,
      "step": 5615
    },
    {
      "epoch": 0.038157050472228324,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000997723020077804,
      "loss": 2.3069,
      "step": 5616
    },
    {
      "epoch": 0.03816384481882238,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009977219943821769,
      "loss": 2.3865,
      "step": 5617
    },
    {
      "epoch": 0.03817063916541644,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009977209684561101,
      "loss": 2.3689,
      "step": 5618
    },
    {
      "epoch": 0.03817743351201049,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009977199422996042,
      "loss": 2.2295,
      "step": 5619
    },
    {
      "epoch": 0.03818422785860455,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009977189159126596,
      "loss": 2.3528,
      "step": 5620
    },
    {
      "epoch": 0.038191022205198606,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997717889295277,
      "loss": 2.5254,
      "step": 5621
    },
    {
      "epoch": 0.03819781655179267,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009977168624474566,
      "loss": 2.4612,
      "step": 5622
    },
    {
      "epoch": 0.038204610898386726,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009977158353691989,
      "loss": 2.4098,
      "step": 5623
    },
    {
      "epoch": 0.03821140524498078,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009977148080605048,
      "loss": 2.4203,
      "step": 5624
    },
    {
      "epoch": 0.03821819959157484,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997713780521374,
      "loss": 2.2722,
      "step": 5625
    },
    {
      "epoch": 0.038224993938168896,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009977127527518077,
      "loss": 2.3732,
      "step": 5626
    },
    {
      "epoch": 0.03823178828476296,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009977117247518062,
      "loss": 2.3327,
      "step": 5627
    },
    {
      "epoch": 0.038238582631357015,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009977106965213696,
      "loss": 2.1269,
      "step": 5628
    },
    {
      "epoch": 0.03824537697795107,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000997709668060499,
      "loss": 2.2068,
      "step": 5629
    },
    {
      "epoch": 0.03825217132454513,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009977086393691942,
      "loss": 2.452,
      "step": 5630
    },
    {
      "epoch": 0.038258965671139185,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009977076104474563,
      "loss": 2.2233,
      "step": 5631
    },
    {
      "epoch": 0.03826576001773324,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009977065812952852,
      "loss": 2.3295,
      "step": 5632
    },
    {
      "epoch": 0.038272554364327305,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009977055519126818,
      "loss": 2.356,
      "step": 5633
    },
    {
      "epoch": 0.03827934871092136,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009977045222996463,
      "loss": 2.3949,
      "step": 5634
    },
    {
      "epoch": 0.03828614305751542,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009977034924561795,
      "loss": 2.4041,
      "step": 5635
    },
    {
      "epoch": 0.038292937404109474,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009977024623822816,
      "loss": 2.3792,
      "step": 5636
    },
    {
      "epoch": 0.03829973175070353,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009977014320779532,
      "loss": 2.2981,
      "step": 5637
    },
    {
      "epoch": 0.038306526097297594,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009977004015431948,
      "loss": 2.2862,
      "step": 5638
    },
    {
      "epoch": 0.03831332044389165,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009976993707780067,
      "loss": 2.3121,
      "step": 5639
    },
    {
      "epoch": 0.03832011479048571,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009976983397823895,
      "loss": 2.4945,
      "step": 5640
    },
    {
      "epoch": 0.038326909137079763,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009976973085563438,
      "loss": 2.3327,
      "step": 5641
    },
    {
      "epoch": 0.03833370348367382,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099769627709987,
      "loss": 2.2922,
      "step": 5642
    },
    {
      "epoch": 0.038340497830267876,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009976952454129682,
      "loss": 2.3402,
      "step": 5643
    },
    {
      "epoch": 0.03834729217686194,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009976942134956395,
      "loss": 2.3455,
      "step": 5644
    },
    {
      "epoch": 0.038354086523455996,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997693181347884,
      "loss": 2.3026,
      "step": 5645
    },
    {
      "epoch": 0.03836088087005005,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009976921489697021,
      "loss": 2.3921,
      "step": 5646
    },
    {
      "epoch": 0.03836767521664411,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009976911163610946,
      "loss": 2.3818,
      "step": 5647
    },
    {
      "epoch": 0.038374469563238166,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000997690083522062,
      "loss": 2.4897,
      "step": 5648
    },
    {
      "epoch": 0.03838126390983223,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009976890504526043,
      "loss": 2.2946,
      "step": 5649
    },
    {
      "epoch": 0.038388058256426286,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009976880171527224,
      "loss": 2.3979,
      "step": 5650
    },
    {
      "epoch": 0.03839485260302034,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009976869836224168,
      "loss": 2.3611,
      "step": 5651
    },
    {
      "epoch": 0.0384016469496144,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009976859498616875,
      "loss": 2.1495,
      "step": 5652
    },
    {
      "epoch": 0.038408441296208455,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009976849158705354,
      "loss": 2.3733,
      "step": 5653
    },
    {
      "epoch": 0.03841523564280252,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009976838816489612,
      "loss": 2.3771,
      "step": 5654
    },
    {
      "epoch": 0.038422029989396575,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009976828471969649,
      "loss": 2.199,
      "step": 5655
    },
    {
      "epoch": 0.03842882433599063,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000997681812514547,
      "loss": 2.5202,
      "step": 5656
    },
    {
      "epoch": 0.03843561868258469,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009976807776017082,
      "loss": 2.4113,
      "step": 5657
    },
    {
      "epoch": 0.038442413029178744,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997679742458449,
      "loss": 2.2723,
      "step": 5658
    },
    {
      "epoch": 0.0384492073757728,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009976787070847695,
      "loss": 2.4092,
      "step": 5659
    },
    {
      "epoch": 0.038456001722366864,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009976776714806707,
      "loss": 2.3559,
      "step": 5660
    },
    {
      "epoch": 0.03846279606896092,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009976766356461528,
      "loss": 2.3968,
      "step": 5661
    },
    {
      "epoch": 0.03846959041555498,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009976755995812164,
      "loss": 2.1816,
      "step": 5662
    },
    {
      "epoch": 0.038476384762149034,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000997674563285862,
      "loss": 2.4184,
      "step": 5663
    },
    {
      "epoch": 0.03848317910874309,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009976735267600897,
      "loss": 2.3826,
      "step": 5664
    },
    {
      "epoch": 0.038489973455337154,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009976724900039003,
      "loss": 2.5118,
      "step": 5665
    },
    {
      "epoch": 0.03849676780193121,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009976714530172944,
      "loss": 2.4638,
      "step": 5666
    },
    {
      "epoch": 0.03850356214852527,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009976704158002722,
      "loss": 2.2966,
      "step": 5667
    },
    {
      "epoch": 0.03851035649511932,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009976693783528344,
      "loss": 2.326,
      "step": 5668
    },
    {
      "epoch": 0.03851715084171338,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009976683406749814,
      "loss": 2.4143,
      "step": 5669
    },
    {
      "epoch": 0.038523945188307436,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009976673027667137,
      "loss": 2.3062,
      "step": 5670
    },
    {
      "epoch": 0.0385307395349015,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009976662646280314,
      "loss": 2.3359,
      "step": 5671
    },
    {
      "epoch": 0.038537533881495556,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009976652262589356,
      "loss": 2.3072,
      "step": 5672
    },
    {
      "epoch": 0.03854432822808961,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009976641876594265,
      "loss": 2.3742,
      "step": 5673
    },
    {
      "epoch": 0.03855112257468367,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009976631488295047,
      "loss": 2.4157,
      "step": 5674
    },
    {
      "epoch": 0.038557916921277725,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009976621097691702,
      "loss": 2.4755,
      "step": 5675
    },
    {
      "epoch": 0.03856471126787179,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997661070478424,
      "loss": 2.2818,
      "step": 5676
    },
    {
      "epoch": 0.038571505614465845,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009976600309572667,
      "loss": 2.4591,
      "step": 5677
    },
    {
      "epoch": 0.0385782999610599,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009976589912056983,
      "loss": 2.3186,
      "step": 5678
    },
    {
      "epoch": 0.03858509430765396,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009976579512237197,
      "loss": 2.3396,
      "step": 5679
    },
    {
      "epoch": 0.038591888654248015,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009976569110113308,
      "loss": 2.4274,
      "step": 5680
    },
    {
      "epoch": 0.03859868300084207,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009976558705685327,
      "loss": 2.2734,
      "step": 5681
    },
    {
      "epoch": 0.038605477347436135,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009976548298953256,
      "loss": 2.3583,
      "step": 5682
    },
    {
      "epoch": 0.03861227169403019,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00099765378899171,
      "loss": 2.2275,
      "step": 5683
    },
    {
      "epoch": 0.03861906604062425,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009976527478576865,
      "loss": 2.2001,
      "step": 5684
    },
    {
      "epoch": 0.038625860387218304,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009976517064932554,
      "loss": 2.4774,
      "step": 5685
    },
    {
      "epoch": 0.03863265473381236,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009976506648984172,
      "loss": 2.3461,
      "step": 5686
    },
    {
      "epoch": 0.038639449080406424,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009976496230731725,
      "loss": 2.3156,
      "step": 5687
    },
    {
      "epoch": 0.03864624342700048,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009976485810175217,
      "loss": 2.4347,
      "step": 5688
    },
    {
      "epoch": 0.03865303777359454,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009976475387314652,
      "loss": 2.3705,
      "step": 5689
    },
    {
      "epoch": 0.03865983212018859,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009976464962150038,
      "loss": 2.3456,
      "step": 5690
    },
    {
      "epoch": 0.03866662646678265,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009976454534681377,
      "loss": 2.323,
      "step": 5691
    },
    {
      "epoch": 0.038673420813376706,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009976444104908674,
      "loss": 2.359,
      "step": 5692
    },
    {
      "epoch": 0.03868021515997077,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009976433672831935,
      "loss": 2.5242,
      "step": 5693
    },
    {
      "epoch": 0.038687009506564826,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009976423238451165,
      "loss": 2.3455,
      "step": 5694
    },
    {
      "epoch": 0.03869380385315888,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009976412801766367,
      "loss": 2.3327,
      "step": 5695
    },
    {
      "epoch": 0.03870059819975294,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997640236277755,
      "loss": 2.2557,
      "step": 5696
    },
    {
      "epoch": 0.038707392546346996,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997639192148471,
      "loss": 2.4213,
      "step": 5697
    },
    {
      "epoch": 0.03871418689294106,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009976381477887861,
      "loss": 2.1421,
      "step": 5698
    },
    {
      "epoch": 0.038720981239535116,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009976371031987007,
      "loss": 2.4105,
      "step": 5699
    },
    {
      "epoch": 0.03872777558612917,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009976360583782147,
      "loss": 2.4435,
      "step": 5700
    },
    {
      "epoch": 0.03873456993272323,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009976350133273288,
      "loss": 2.3237,
      "step": 5701
    },
    {
      "epoch": 0.038741364279317285,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009976339680460439,
      "loss": 2.312,
      "step": 5702
    },
    {
      "epoch": 0.03874815862591134,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099763292253436,
      "loss": 2.0754,
      "step": 5703
    },
    {
      "epoch": 0.038754952972505405,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009976318767922778,
      "loss": 2.2554,
      "step": 5704
    },
    {
      "epoch": 0.03876174731909946,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009976308308197977,
      "loss": 2.3227,
      "step": 5705
    },
    {
      "epoch": 0.03876854166569352,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009976297846169201,
      "loss": 2.3868,
      "step": 5706
    },
    {
      "epoch": 0.038775336012287574,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000997628738183646,
      "loss": 2.1961,
      "step": 5707
    },
    {
      "epoch": 0.03878213035888163,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997627691519975,
      "loss": 2.3239,
      "step": 5708
    },
    {
      "epoch": 0.038788924705475694,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009976266446259086,
      "loss": 2.3163,
      "step": 5709
    },
    {
      "epoch": 0.03879571905206975,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009976255975014466,
      "loss": 2.3855,
      "step": 5710
    },
    {
      "epoch": 0.03880251339866381,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009976245501465894,
      "loss": 2.3821,
      "step": 5711
    },
    {
      "epoch": 0.038809307745257864,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997623502561338,
      "loss": 2.2095,
      "step": 5712
    },
    {
      "epoch": 0.03881610209185192,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009976224547456926,
      "loss": 2.3698,
      "step": 5713
    },
    {
      "epoch": 0.038822896438445983,
      "grad_norm": 0.2021484375,
      "learning_rate": 0.0009976214066996535,
      "loss": 2.3387,
      "step": 5714
    },
    {
      "epoch": 0.03882969078504004,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009976203584232217,
      "loss": 2.4317,
      "step": 5715
    },
    {
      "epoch": 0.038836485131634096,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009976193099163972,
      "loss": 2.4977,
      "step": 5716
    },
    {
      "epoch": 0.03884327947822815,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009976182611791807,
      "loss": 2.4085,
      "step": 5717
    },
    {
      "epoch": 0.03885007382482221,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009976172122115727,
      "loss": 2.3003,
      "step": 5718
    },
    {
      "epoch": 0.038856868171416266,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009976161630135736,
      "loss": 2.0871,
      "step": 5719
    },
    {
      "epoch": 0.03886366251801033,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0009976151135851839,
      "loss": 2.2912,
      "step": 5720
    },
    {
      "epoch": 0.038870456864604386,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000997614063926404,
      "loss": 2.3779,
      "step": 5721
    },
    {
      "epoch": 0.03887725121119844,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009976130140372345,
      "loss": 2.279,
      "step": 5722
    },
    {
      "epoch": 0.0388840455577925,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009976119639176762,
      "loss": 2.2673,
      "step": 5723
    },
    {
      "epoch": 0.038890839904386555,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000997610913567729,
      "loss": 2.3545,
      "step": 5724
    },
    {
      "epoch": 0.03889763425098062,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009976098629873936,
      "loss": 2.363,
      "step": 5725
    },
    {
      "epoch": 0.038904428597574675,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009976088121766706,
      "loss": 2.3209,
      "step": 5726
    },
    {
      "epoch": 0.03891122294416873,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009976077611355606,
      "loss": 2.3077,
      "step": 5727
    },
    {
      "epoch": 0.03891801729076279,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009976067098640639,
      "loss": 2.1905,
      "step": 5728
    },
    {
      "epoch": 0.038924811637356845,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009976056583621807,
      "loss": 2.4098,
      "step": 5729
    },
    {
      "epoch": 0.0389316059839509,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000997604606629912,
      "loss": 2.3079,
      "step": 5730
    },
    {
      "epoch": 0.038938400330544964,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000997603554667258,
      "loss": 2.4077,
      "step": 5731
    },
    {
      "epoch": 0.03894519467713902,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009976025024742193,
      "loss": 2.5378,
      "step": 5732
    },
    {
      "epoch": 0.03895198902373308,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009976014500507962,
      "loss": 2.3618,
      "step": 5733
    },
    {
      "epoch": 0.038958783370327134,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009976003973969896,
      "loss": 2.3799,
      "step": 5734
    },
    {
      "epoch": 0.03896557771692119,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009975993445127996,
      "loss": 2.2604,
      "step": 5735
    },
    {
      "epoch": 0.038972372063515254,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009975982913982269,
      "loss": 2.3776,
      "step": 5736
    },
    {
      "epoch": 0.03897916641010931,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009975972380532716,
      "loss": 2.3018,
      "step": 5737
    },
    {
      "epoch": 0.03898596075670337,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009975961844779346,
      "loss": 2.2815,
      "step": 5738
    },
    {
      "epoch": 0.03899275510329742,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009975951306722163,
      "loss": 2.4121,
      "step": 5739
    },
    {
      "epoch": 0.03899954944989148,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009975940766361172,
      "loss": 2.2889,
      "step": 5740
    },
    {
      "epoch": 0.039006343796485536,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009975930223696378,
      "loss": 2.2102,
      "step": 5741
    },
    {
      "epoch": 0.0390131381430796,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009975919678727786,
      "loss": 2.4119,
      "step": 5742
    },
    {
      "epoch": 0.039019932489673656,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009975909131455399,
      "loss": 2.3951,
      "step": 5743
    },
    {
      "epoch": 0.03902672683626771,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009975898581879223,
      "loss": 2.375,
      "step": 5744
    },
    {
      "epoch": 0.03903352118286177,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009975888029999263,
      "loss": 2.3355,
      "step": 5745
    },
    {
      "epoch": 0.039040315529455825,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009975877475815525,
      "loss": 2.3648,
      "step": 5746
    },
    {
      "epoch": 0.03904710987604989,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009975866919328012,
      "loss": 2.4491,
      "step": 5747
    },
    {
      "epoch": 0.039053904222643945,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009975856360536729,
      "loss": 2.2376,
      "step": 5748
    },
    {
      "epoch": 0.039060698569238,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009975845799441684,
      "loss": 2.1903,
      "step": 5749
    },
    {
      "epoch": 0.03906749291583206,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009975835236042876,
      "loss": 2.337,
      "step": 5750
    },
    {
      "epoch": 0.039074287262426115,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009975824670340314,
      "loss": 2.3044,
      "step": 5751
    },
    {
      "epoch": 0.03908108160902017,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009975814102334004,
      "loss": 2.3985,
      "step": 5752
    },
    {
      "epoch": 0.039087875955614235,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009975803532023947,
      "loss": 2.4704,
      "step": 5753
    },
    {
      "epoch": 0.03909467030220829,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009975792959410154,
      "loss": 2.3544,
      "step": 5754
    },
    {
      "epoch": 0.03910146464880235,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009975782384492621,
      "loss": 2.3512,
      "step": 5755
    },
    {
      "epoch": 0.039108258995396404,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000997577180727136,
      "loss": 2.3364,
      "step": 5756
    },
    {
      "epoch": 0.03911505334199046,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009975761227746375,
      "loss": 2.3278,
      "step": 5757
    },
    {
      "epoch": 0.039121847688584524,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009975750645917667,
      "loss": 2.2922,
      "step": 5758
    },
    {
      "epoch": 0.03912864203517858,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009975740061785245,
      "loss": 2.4806,
      "step": 5759
    },
    {
      "epoch": 0.03913543638177264,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009975729475349112,
      "loss": 2.3253,
      "step": 5760
    },
    {
      "epoch": 0.03914223072836669,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009975718886609272,
      "loss": 2.3769,
      "step": 5761
    },
    {
      "epoch": 0.03914902507496075,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0009975708295565733,
      "loss": 2.4104,
      "step": 5762
    },
    {
      "epoch": 0.039155819421554806,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009975697702218499,
      "loss": 2.2826,
      "step": 5763
    },
    {
      "epoch": 0.03916261376814887,
      "grad_norm": 0.1875,
      "learning_rate": 0.0009975687106567573,
      "loss": 2.2131,
      "step": 5764
    },
    {
      "epoch": 0.039169408114742926,
      "grad_norm": 0.171875,
      "learning_rate": 0.000997567650861296,
      "loss": 2.2937,
      "step": 5765
    },
    {
      "epoch": 0.03917620246133698,
      "grad_norm": 0.1962890625,
      "learning_rate": 0.0009975665908354668,
      "loss": 2.3798,
      "step": 5766
    },
    {
      "epoch": 0.03918299680793104,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009975655305792697,
      "loss": 2.3209,
      "step": 5767
    },
    {
      "epoch": 0.039189791154525096,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009975644700927055,
      "loss": 2.2599,
      "step": 5768
    },
    {
      "epoch": 0.03919658550111916,
      "grad_norm": 0.2060546875,
      "learning_rate": 0.0009975634093757747,
      "loss": 2.4504,
      "step": 5769
    },
    {
      "epoch": 0.039203379847713216,
      "grad_norm": 0.2109375,
      "learning_rate": 0.000997562348428478,
      "loss": 2.4821,
      "step": 5770
    },
    {
      "epoch": 0.03921017419430727,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009975612872508153,
      "loss": 2.303,
      "step": 5771
    },
    {
      "epoch": 0.03921696854090133,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009975602258427876,
      "loss": 2.3722,
      "step": 5772
    },
    {
      "epoch": 0.039223762887495385,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009975591642043952,
      "loss": 2.2094,
      "step": 5773
    },
    {
      "epoch": 0.03923055723408945,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009975581023356386,
      "loss": 2.2618,
      "step": 5774
    },
    {
      "epoch": 0.039237351580683505,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009975570402365182,
      "loss": 2.5094,
      "step": 5775
    },
    {
      "epoch": 0.03924414592727756,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009975559779070347,
      "loss": 2.4334,
      "step": 5776
    },
    {
      "epoch": 0.03925094027387162,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009975549153471886,
      "loss": 2.2,
      "step": 5777
    },
    {
      "epoch": 0.039257734620465674,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00099755385255698,
      "loss": 2.4022,
      "step": 5778
    },
    {
      "epoch": 0.03926452896705973,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009975527895364098,
      "loss": 2.2665,
      "step": 5779
    },
    {
      "epoch": 0.039271323313653794,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009975517262854783,
      "loss": 2.3244,
      "step": 5780
    },
    {
      "epoch": 0.03927811766024785,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009975506628041861,
      "loss": 2.3157,
      "step": 5781
    },
    {
      "epoch": 0.03928491200684191,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009975495990925337,
      "loss": 2.3334,
      "step": 5782
    },
    {
      "epoch": 0.039291706353435964,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009975485351505216,
      "loss": 2.2346,
      "step": 5783
    },
    {
      "epoch": 0.03929850070003002,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009975474709781502,
      "loss": 2.4048,
      "step": 5784
    },
    {
      "epoch": 0.039305295046624084,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00099754640657542,
      "loss": 2.2212,
      "step": 5785
    },
    {
      "epoch": 0.03931208939321814,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009975453419423315,
      "loss": 2.5231,
      "step": 5786
    },
    {
      "epoch": 0.0393188837398122,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009975442770788852,
      "loss": 2.2502,
      "step": 5787
    },
    {
      "epoch": 0.03932567808640625,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009975432119850817,
      "loss": 2.2589,
      "step": 5788
    },
    {
      "epoch": 0.03933247243300031,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009975421466609212,
      "loss": 2.2719,
      "step": 5789
    },
    {
      "epoch": 0.039339266779594366,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009975410811064046,
      "loss": 2.3036,
      "step": 5790
    },
    {
      "epoch": 0.03934606112618843,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000997540015321532,
      "loss": 2.3378,
      "step": 5791
    },
    {
      "epoch": 0.039352855472782486,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000997538949306304,
      "loss": 2.2115,
      "step": 5792
    },
    {
      "epoch": 0.03935964981937654,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009975378830607213,
      "loss": 2.3865,
      "step": 5793
    },
    {
      "epoch": 0.0393664441659706,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009975368165847844,
      "loss": 2.2027,
      "step": 5794
    },
    {
      "epoch": 0.039373238512564655,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009975357498784936,
      "loss": 2.3536,
      "step": 5795
    },
    {
      "epoch": 0.03938003285915872,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009975346829418493,
      "loss": 2.3357,
      "step": 5796
    },
    {
      "epoch": 0.039386827205752775,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009975336157748522,
      "loss": 2.3627,
      "step": 5797
    },
    {
      "epoch": 0.03939362155234683,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009975325483775029,
      "loss": 2.4028,
      "step": 5798
    },
    {
      "epoch": 0.03940041589894089,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009975314807498016,
      "loss": 2.3443,
      "step": 5799
    },
    {
      "epoch": 0.039407210245534945,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997530412891749,
      "loss": 2.4787,
      "step": 5800
    },
    {
      "epoch": 0.039414004592129,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009975293448033454,
      "loss": 2.466,
      "step": 5801
    },
    {
      "epoch": 0.039420798938723065,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009975282764845914,
      "loss": 2.3064,
      "step": 5802
    },
    {
      "epoch": 0.03942759328531712,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009975272079354876,
      "loss": 2.2052,
      "step": 5803
    },
    {
      "epoch": 0.03943438763191118,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009975261391560345,
      "loss": 2.3366,
      "step": 5804
    },
    {
      "epoch": 0.039441181978505234,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009975250701462323,
      "loss": 2.3218,
      "step": 5805
    },
    {
      "epoch": 0.03944797632509929,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997524000906082,
      "loss": 2.3881,
      "step": 5806
    },
    {
      "epoch": 0.039454770671693354,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009975229314355834,
      "loss": 2.3418,
      "step": 5807
    },
    {
      "epoch": 0.03946156501828741,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009975218617347377,
      "loss": 2.3136,
      "step": 5808
    },
    {
      "epoch": 0.03946835936488147,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000997520791803545,
      "loss": 2.3583,
      "step": 5809
    },
    {
      "epoch": 0.03947515371147552,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009975197216420058,
      "loss": 2.4159,
      "step": 5810
    },
    {
      "epoch": 0.03948194805806958,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009975186512501206,
      "loss": 2.2704,
      "step": 5811
    },
    {
      "epoch": 0.039488742404663636,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009975175806278902,
      "loss": 2.3503,
      "step": 5812
    },
    {
      "epoch": 0.0394955367512577,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009975165097753148,
      "loss": 2.2685,
      "step": 5813
    },
    {
      "epoch": 0.039502331097851756,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009975154386923947,
      "loss": 2.1552,
      "step": 5814
    },
    {
      "epoch": 0.03950912544444581,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000997514367379131,
      "loss": 2.4738,
      "step": 5815
    },
    {
      "epoch": 0.03951591979103987,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009975132958355238,
      "loss": 2.1915,
      "step": 5816
    },
    {
      "epoch": 0.039522714137633926,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009975122240615734,
      "loss": 2.2916,
      "step": 5817
    },
    {
      "epoch": 0.03952950848422799,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009975111520572808,
      "loss": 2.4997,
      "step": 5818
    },
    {
      "epoch": 0.039536302830822045,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009975100798226462,
      "loss": 2.4111,
      "step": 5819
    },
    {
      "epoch": 0.0395430971774161,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00099750900735767,
      "loss": 2.3301,
      "step": 5820
    },
    {
      "epoch": 0.03954989152401016,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009975079346623528,
      "loss": 2.2188,
      "step": 5821
    },
    {
      "epoch": 0.039556685870604215,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009975068617366955,
      "loss": 2.2156,
      "step": 5822
    },
    {
      "epoch": 0.03956348021719827,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009975057885806978,
      "loss": 2.465,
      "step": 5823
    },
    {
      "epoch": 0.039570274563792335,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009975047151943607,
      "loss": 2.2286,
      "step": 5824
    },
    {
      "epoch": 0.03957706891038639,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009975036415776847,
      "loss": 2.3112,
      "step": 5825
    },
    {
      "epoch": 0.03958386325698045,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009975025677306704,
      "loss": 2.395,
      "step": 5826
    },
    {
      "epoch": 0.039590657603574504,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997501493653318,
      "loss": 2.2989,
      "step": 5827
    },
    {
      "epoch": 0.03959745195016856,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997500419345628,
      "loss": 2.3653,
      "step": 5828
    },
    {
      "epoch": 0.039604246296762624,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997499344807601,
      "loss": 2.1312,
      "step": 5829
    },
    {
      "epoch": 0.03961104064335668,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009974982700392376,
      "loss": 2.304,
      "step": 5830
    },
    {
      "epoch": 0.03961783498995074,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009974971950405382,
      "loss": 2.3238,
      "step": 5831
    },
    {
      "epoch": 0.039624629336544794,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009974961198115034,
      "loss": 2.3387,
      "step": 5832
    },
    {
      "epoch": 0.03963142368313885,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009974950443521334,
      "loss": 2.3212,
      "step": 5833
    },
    {
      "epoch": 0.03963821802973291,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009974939686624288,
      "loss": 2.3052,
      "step": 5834
    },
    {
      "epoch": 0.03964501237632697,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009974928927423904,
      "loss": 2.3995,
      "step": 5835
    },
    {
      "epoch": 0.039651806722921026,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009974918165920187,
      "loss": 2.4239,
      "step": 5836
    },
    {
      "epoch": 0.03965860106951508,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009974907402113136,
      "loss": 2.4232,
      "step": 5837
    },
    {
      "epoch": 0.03966539541610914,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009974896636002761,
      "loss": 2.4066,
      "step": 5838
    },
    {
      "epoch": 0.039672189762703196,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009974885867589068,
      "loss": 2.2849,
      "step": 5839
    },
    {
      "epoch": 0.03967898410929726,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009974875096872059,
      "loss": 2.3483,
      "step": 5840
    },
    {
      "epoch": 0.039685778455891316,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009974864323851738,
      "loss": 2.4303,
      "step": 5841
    },
    {
      "epoch": 0.03969257280248537,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009974853548528114,
      "loss": 2.4014,
      "step": 5842
    },
    {
      "epoch": 0.03969936714907943,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009974842770901187,
      "loss": 2.3001,
      "step": 5843
    },
    {
      "epoch": 0.039706161495673485,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009974831990970966,
      "loss": 2.4219,
      "step": 5844
    },
    {
      "epoch": 0.03971295584226755,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009974821208737457,
      "loss": 2.4062,
      "step": 5845
    },
    {
      "epoch": 0.039719750188861605,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009974810424200663,
      "loss": 2.1261,
      "step": 5846
    },
    {
      "epoch": 0.03972654453545566,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009974799637360585,
      "loss": 2.2559,
      "step": 5847
    },
    {
      "epoch": 0.03973333888204972,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009974788848217235,
      "loss": 2.2301,
      "step": 5848
    },
    {
      "epoch": 0.039740133228643774,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009974778056770614,
      "loss": 2.3403,
      "step": 5849
    },
    {
      "epoch": 0.03974692757523783,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009974767263020727,
      "loss": 2.3523,
      "step": 5850
    },
    {
      "epoch": 0.039753721921831894,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000997475646696758,
      "loss": 2.3475,
      "step": 5851
    },
    {
      "epoch": 0.03976051626842595,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009974745668611179,
      "loss": 2.3745,
      "step": 5852
    },
    {
      "epoch": 0.03976731061502001,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009974734867951526,
      "loss": 2.4845,
      "step": 5853
    },
    {
      "epoch": 0.039774104961614064,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997472406498863,
      "loss": 2.3551,
      "step": 5854
    },
    {
      "epoch": 0.03978089930820812,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009974713259722492,
      "loss": 2.3996,
      "step": 5855
    },
    {
      "epoch": 0.039787693654802184,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000997470245215312,
      "loss": 2.5159,
      "step": 5856
    },
    {
      "epoch": 0.03979448800139624,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009974691642280515,
      "loss": 2.3225,
      "step": 5857
    },
    {
      "epoch": 0.0398012823479903,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009974680830104688,
      "loss": 2.3288,
      "step": 5858
    },
    {
      "epoch": 0.03980807669458435,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997467001562564,
      "loss": 2.2088,
      "step": 5859
    },
    {
      "epoch": 0.03981487104117841,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009974659198843377,
      "loss": 2.3978,
      "step": 5860
    },
    {
      "epoch": 0.039821665387772466,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009974648379757903,
      "loss": 2.4042,
      "step": 5861
    },
    {
      "epoch": 0.03982845973436653,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009974637558369224,
      "loss": 2.4129,
      "step": 5862
    },
    {
      "epoch": 0.039835254080960586,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009974626734677344,
      "loss": 2.3669,
      "step": 5863
    },
    {
      "epoch": 0.03984204842755464,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997461590868227,
      "loss": 2.1893,
      "step": 5864
    },
    {
      "epoch": 0.0398488427741487,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009974605080384006,
      "loss": 2.2898,
      "step": 5865
    },
    {
      "epoch": 0.039855637120742755,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009974594249782557,
      "loss": 2.3061,
      "step": 5866
    },
    {
      "epoch": 0.03986243146733682,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009974583416877928,
      "loss": 2.374,
      "step": 5867
    },
    {
      "epoch": 0.039869225813930875,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009974572581670123,
      "loss": 2.4514,
      "step": 5868
    },
    {
      "epoch": 0.03987602016052493,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009974561744159147,
      "loss": 2.2004,
      "step": 5869
    },
    {
      "epoch": 0.03988281450711899,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009974550904345008,
      "loss": 2.3734,
      "step": 5870
    },
    {
      "epoch": 0.039889608853713045,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009974540062227708,
      "loss": 2.3556,
      "step": 5871
    },
    {
      "epoch": 0.0398964032003071,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009974529217807253,
      "loss": 2.2386,
      "step": 5872
    },
    {
      "epoch": 0.039903197546901165,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997451837108365,
      "loss": 2.4089,
      "step": 5873
    },
    {
      "epoch": 0.03990999189349522,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009974507522056899,
      "loss": 2.2197,
      "step": 5874
    },
    {
      "epoch": 0.03991678624008928,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000997449667072701,
      "loss": 2.3175,
      "step": 5875
    },
    {
      "epoch": 0.039923580586683334,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009974485817093985,
      "loss": 2.2872,
      "step": 5876
    },
    {
      "epoch": 0.03993037493327739,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000997447496115783,
      "loss": 2.3168,
      "step": 5877
    },
    {
      "epoch": 0.039937169279871454,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997446410291855,
      "loss": 2.4667,
      "step": 5878
    },
    {
      "epoch": 0.03994396362646551,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997445324237615,
      "loss": 2.3512,
      "step": 5879
    },
    {
      "epoch": 0.03995075797305957,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009974442379530636,
      "loss": 2.1873,
      "step": 5880
    },
    {
      "epoch": 0.03995755231965362,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009974431514382014,
      "loss": 2.393,
      "step": 5881
    },
    {
      "epoch": 0.03996434666624768,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009974420646930284,
      "loss": 2.3834,
      "step": 5882
    },
    {
      "epoch": 0.039971141012841736,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009974409777175456,
      "loss": 2.3067,
      "step": 5883
    },
    {
      "epoch": 0.0399779353594358,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009974398905117535,
      "loss": 2.2851,
      "step": 5884
    },
    {
      "epoch": 0.039984729706029856,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000997438803075652,
      "loss": 2.4343,
      "step": 5885
    },
    {
      "epoch": 0.03999152405262391,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009974377154092424,
      "loss": 2.3311,
      "step": 5886
    },
    {
      "epoch": 0.03999831839921797,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009974366275125246,
      "loss": 2.3733,
      "step": 5887
    },
    {
      "epoch": 0.040005112745812026,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009974355393854994,
      "loss": 2.2525,
      "step": 5888
    },
    {
      "epoch": 0.040005112745812026,
      "eval_loss": 2.392270088195801,
      "eval_runtime": 2373.7926,
      "eval_samples_per_second": 15.905,
      "eval_steps_per_second": 3.976,
      "step": 5888
    },
    {
      "epoch": 0.04001190709240609,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009974344510281675,
      "loss": 2.3116,
      "step": 5889
    },
    {
      "epoch": 0.040018701439000146,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009974333624405288,
      "loss": 2.3833,
      "step": 5890
    },
    {
      "epoch": 0.0400254957855942,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009974322736225844,
      "loss": 2.2989,
      "step": 5891
    },
    {
      "epoch": 0.04003229013218826,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009974311845743346,
      "loss": 2.3548,
      "step": 5892
    },
    {
      "epoch": 0.040039084478782315,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009974300952957796,
      "loss": 2.3607,
      "step": 5893
    },
    {
      "epoch": 0.04004587882537638,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009974290057869202,
      "loss": 2.3075,
      "step": 5894
    },
    {
      "epoch": 0.040052673171970435,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009974279160477571,
      "loss": 2.2182,
      "step": 5895
    },
    {
      "epoch": 0.04005946751856449,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009974268260782904,
      "loss": 2.3294,
      "step": 5896
    },
    {
      "epoch": 0.04006626186515855,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009974257358785208,
      "loss": 2.2299,
      "step": 5897
    },
    {
      "epoch": 0.040073056211752604,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009974246454484488,
      "loss": 2.3974,
      "step": 5898
    },
    {
      "epoch": 0.04007985055834666,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000997423554788075,
      "loss": 2.5017,
      "step": 5899
    },
    {
      "epoch": 0.040086644904940724,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009974224638973996,
      "loss": 2.4087,
      "step": 5900
    },
    {
      "epoch": 0.04009343925153478,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009974213727764235,
      "loss": 2.4299,
      "step": 5901
    },
    {
      "epoch": 0.04010023359812884,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009974202814251468,
      "loss": 2.3325,
      "step": 5902
    },
    {
      "epoch": 0.040107027944722894,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009974191898435704,
      "loss": 2.3046,
      "step": 5903
    },
    {
      "epoch": 0.04011382229131695,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009974180980316944,
      "loss": 2.3434,
      "step": 5904
    },
    {
      "epoch": 0.040120616637911014,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009974170059895199,
      "loss": 2.3469,
      "step": 5905
    },
    {
      "epoch": 0.04012741098450507,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009974159137170467,
      "loss": 2.3262,
      "step": 5906
    },
    {
      "epoch": 0.040134205331099126,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009974148212142757,
      "loss": 2.2589,
      "step": 5907
    },
    {
      "epoch": 0.04014099967769318,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009974137284812074,
      "loss": 2.3292,
      "step": 5908
    },
    {
      "epoch": 0.04014779402428724,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009974126355178422,
      "loss": 2.3144,
      "step": 5909
    },
    {
      "epoch": 0.040154588370881296,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009974115423241807,
      "loss": 2.3544,
      "step": 5910
    },
    {
      "epoch": 0.04016138271747536,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009974104489002234,
      "loss": 2.3675,
      "step": 5911
    },
    {
      "epoch": 0.040168177064069416,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009974093552459705,
      "loss": 2.3092,
      "step": 5912
    },
    {
      "epoch": 0.04017497141066347,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009974082613614231,
      "loss": 2.3341,
      "step": 5913
    },
    {
      "epoch": 0.04018176575725753,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009974071672465812,
      "loss": 2.3858,
      "step": 5914
    },
    {
      "epoch": 0.040188560103851585,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009974060729014457,
      "loss": 2.3068,
      "step": 5915
    },
    {
      "epoch": 0.04019535445044565,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009974049783260167,
      "loss": 2.3381,
      "step": 5916
    },
    {
      "epoch": 0.040202148797039705,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009974038835202947,
      "loss": 2.3233,
      "step": 5917
    },
    {
      "epoch": 0.04020894314363376,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009974027884842806,
      "loss": 2.3862,
      "step": 5918
    },
    {
      "epoch": 0.04021573749022782,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009974016932179748,
      "loss": 2.3297,
      "step": 5919
    },
    {
      "epoch": 0.040222531836821875,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009974005977213777,
      "loss": 2.292,
      "step": 5920
    },
    {
      "epoch": 0.04022932618341593,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009973995019944898,
      "loss": 2.4145,
      "step": 5921
    },
    {
      "epoch": 0.040236120530009994,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009973984060373117,
      "loss": 2.399,
      "step": 5922
    },
    {
      "epoch": 0.04024291487660405,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009973973098498436,
      "loss": 2.3137,
      "step": 5923
    },
    {
      "epoch": 0.04024970922319811,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009973962134320865,
      "loss": 2.3764,
      "step": 5924
    },
    {
      "epoch": 0.040256503569792164,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009973951167840404,
      "loss": 2.4477,
      "step": 5925
    },
    {
      "epoch": 0.04026329791638622,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009973940199057064,
      "loss": 2.301,
      "step": 5926
    },
    {
      "epoch": 0.040270092262980284,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009973929227970843,
      "loss": 2.4998,
      "step": 5927
    },
    {
      "epoch": 0.04027688660957434,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009973918254581754,
      "loss": 2.4583,
      "step": 5928
    },
    {
      "epoch": 0.0402836809561684,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009973907278889793,
      "loss": 2.3232,
      "step": 5929
    },
    {
      "epoch": 0.04029047530276245,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973896300894973,
      "loss": 2.5452,
      "step": 5930
    },
    {
      "epoch": 0.04029726964935651,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009973885320597295,
      "loss": 2.3074,
      "step": 5931
    },
    {
      "epoch": 0.040304063995950566,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009973874337996767,
      "loss": 2.347,
      "step": 5932
    },
    {
      "epoch": 0.04031085834254463,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009973863353093392,
      "loss": 2.2819,
      "step": 5933
    },
    {
      "epoch": 0.040317652689138686,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009973852365887173,
      "loss": 2.2685,
      "step": 5934
    },
    {
      "epoch": 0.04032444703573274,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973841376378118,
      "loss": 2.411,
      "step": 5935
    },
    {
      "epoch": 0.0403312413823268,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009973830384566232,
      "loss": 2.2541,
      "step": 5936
    },
    {
      "epoch": 0.040338035728920855,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000997381939045152,
      "loss": 2.5083,
      "step": 5937
    },
    {
      "epoch": 0.04034483007551492,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009973808394033985,
      "loss": 2.4298,
      "step": 5938
    },
    {
      "epoch": 0.040351624422108975,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009973797395313635,
      "loss": 2.1617,
      "step": 5939
    },
    {
      "epoch": 0.04035841876870303,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009973786394290474,
      "loss": 2.38,
      "step": 5940
    },
    {
      "epoch": 0.04036521311529709,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009973775390964507,
      "loss": 2.357,
      "step": 5941
    },
    {
      "epoch": 0.040372007461891145,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997376438533574,
      "loss": 2.2183,
      "step": 5942
    },
    {
      "epoch": 0.0403788018084852,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009973753377404174,
      "loss": 2.3904,
      "step": 5943
    },
    {
      "epoch": 0.040385596155079265,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000997374236716982,
      "loss": 2.4011,
      "step": 5944
    },
    {
      "epoch": 0.04039239050167332,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973731354632677,
      "loss": 2.2817,
      "step": 5945
    },
    {
      "epoch": 0.04039918484826738,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009973720339792758,
      "loss": 2.2994,
      "step": 5946
    },
    {
      "epoch": 0.040405979194861434,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000997370932265006,
      "loss": 2.2679,
      "step": 5947
    },
    {
      "epoch": 0.04041277354145549,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009973698303204594,
      "loss": 2.4166,
      "step": 5948
    },
    {
      "epoch": 0.040419567888049554,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997368728145636,
      "loss": 2.2373,
      "step": 5949
    },
    {
      "epoch": 0.04042636223464361,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000997367625740537,
      "loss": 2.3385,
      "step": 5950
    },
    {
      "epoch": 0.04043315658123767,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997366523105162,
      "loss": 2.4672,
      "step": 5951
    },
    {
      "epoch": 0.04043995092783172,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009973654202395123,
      "loss": 2.3809,
      "step": 5952
    },
    {
      "epoch": 0.04044674527442578,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997364317143588,
      "loss": 2.3579,
      "step": 5953
    },
    {
      "epoch": 0.04045353962101984,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009973632138173898,
      "loss": 2.3319,
      "step": 5954
    },
    {
      "epoch": 0.0404603339676139,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973621102609181,
      "loss": 2.4691,
      "step": 5955
    },
    {
      "epoch": 0.040467128314207956,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009973610064741734,
      "loss": 2.3595,
      "step": 5956
    },
    {
      "epoch": 0.04047392266080201,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009973599024571563,
      "loss": 2.3218,
      "step": 5957
    },
    {
      "epoch": 0.04048071700739607,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973587982098674,
      "loss": 2.3114,
      "step": 5958
    },
    {
      "epoch": 0.040487511353990126,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009973576937323068,
      "loss": 2.3568,
      "step": 5959
    },
    {
      "epoch": 0.04049430570058419,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009973565890244755,
      "loss": 2.3466,
      "step": 5960
    },
    {
      "epoch": 0.040501100047178246,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009973554840863738,
      "loss": 2.4681,
      "step": 5961
    },
    {
      "epoch": 0.0405078943937723,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000997354378918002,
      "loss": 2.2803,
      "step": 5962
    },
    {
      "epoch": 0.04051468874036636,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000997353273519361,
      "loss": 2.2098,
      "step": 5963
    },
    {
      "epoch": 0.040521483086960415,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997352167890451,
      "loss": 2.4085,
      "step": 5964
    },
    {
      "epoch": 0.04052827743355448,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997351062031273,
      "loss": 2.4314,
      "step": 5965
    },
    {
      "epoch": 0.040535071780148535,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009973499559418268,
      "loss": 2.381,
      "step": 5966
    },
    {
      "epoch": 0.04054186612674259,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009973488496221134,
      "loss": 2.3704,
      "step": 5967
    },
    {
      "epoch": 0.04054866047333665,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009973477430721331,
      "loss": 2.4103,
      "step": 5968
    },
    {
      "epoch": 0.040555454819930704,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009973466362918867,
      "loss": 2.5159,
      "step": 5969
    },
    {
      "epoch": 0.04056224916652476,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009973455292813743,
      "loss": 2.2991,
      "step": 5970
    },
    {
      "epoch": 0.040569043513118824,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009973444220405967,
      "loss": 2.2601,
      "step": 5971
    },
    {
      "epoch": 0.04057583785971288,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009973433145695544,
      "loss": 2.3405,
      "step": 5972
    },
    {
      "epoch": 0.04058263220630694,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009973422068682478,
      "loss": 2.4175,
      "step": 5973
    },
    {
      "epoch": 0.040589426552900994,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009973410989366773,
      "loss": 2.3553,
      "step": 5974
    },
    {
      "epoch": 0.04059622089949505,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009973399907748438,
      "loss": 2.1951,
      "step": 5975
    },
    {
      "epoch": 0.040603015246089114,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009973388823827476,
      "loss": 2.2234,
      "step": 5976
    },
    {
      "epoch": 0.04060980959268317,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000997337773760389,
      "loss": 2.2936,
      "step": 5977
    },
    {
      "epoch": 0.04061660393927723,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.000997336664907769,
      "loss": 2.5433,
      "step": 5978
    },
    {
      "epoch": 0.04062339828587128,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009973355558248874,
      "loss": 2.2983,
      "step": 5979
    },
    {
      "epoch": 0.04063019263246534,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009973344465117454,
      "loss": 2.4153,
      "step": 5980
    },
    {
      "epoch": 0.040636986979059396,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009973333369683432,
      "loss": 2.2958,
      "step": 5981
    },
    {
      "epoch": 0.04064378132565346,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009973322271946813,
      "loss": 2.276,
      "step": 5982
    },
    {
      "epoch": 0.040650575672247516,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009973311171907605,
      "loss": 2.2844,
      "step": 5983
    },
    {
      "epoch": 0.04065737001884157,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997330006956581,
      "loss": 2.1346,
      "step": 5984
    },
    {
      "epoch": 0.04066416436543563,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009973288964921433,
      "loss": 2.1936,
      "step": 5985
    },
    {
      "epoch": 0.040670958712029685,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997327785797448,
      "loss": 2.3563,
      "step": 5986
    },
    {
      "epoch": 0.04067775305862375,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009973266748724958,
      "loss": 2.3381,
      "step": 5987
    },
    {
      "epoch": 0.040684547405217805,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973255637172871,
      "loss": 2.374,
      "step": 5988
    },
    {
      "epoch": 0.04069134175181186,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973244523318221,
      "loss": 2.4098,
      "step": 5989
    },
    {
      "epoch": 0.04069813609840592,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009973233407161019,
      "loss": 2.2834,
      "step": 5990
    },
    {
      "epoch": 0.040704930444999975,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009973222288701265,
      "loss": 2.3192,
      "step": 5991
    },
    {
      "epoch": 0.04071172479159403,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009973211167938966,
      "loss": 2.2944,
      "step": 5992
    },
    {
      "epoch": 0.040718519138188095,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000997320004487413,
      "loss": 2.3678,
      "step": 5993
    },
    {
      "epoch": 0.04072531348478215,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009973188919506755,
      "loss": 2.4209,
      "step": 5994
    },
    {
      "epoch": 0.04073210783137621,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973177791836852,
      "loss": 2.2384,
      "step": 5995
    },
    {
      "epoch": 0.040738902177970264,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009973166661864426,
      "loss": 2.4369,
      "step": 5996
    },
    {
      "epoch": 0.04074569652456432,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997315552958948,
      "loss": 2.2385,
      "step": 5997
    },
    {
      "epoch": 0.040752490871158384,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000997314439501202,
      "loss": 2.0906,
      "step": 5998
    },
    {
      "epoch": 0.04075928521775244,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009973133258132053,
      "loss": 2.1516,
      "step": 5999
    },
    {
      "epoch": 0.0407660795643465,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000997312211894958,
      "loss": 2.3012,
      "step": 6000
    },
    {
      "epoch": 0.04077287391094055,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009973110977464607,
      "loss": 2.418,
      "step": 6001
    },
    {
      "epoch": 0.04077966825753461,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009973099833677143,
      "loss": 2.4419,
      "step": 6002
    },
    {
      "epoch": 0.040786462604128666,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009973088687587192,
      "loss": 2.196,
      "step": 6003
    },
    {
      "epoch": 0.04079325695072273,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009973077539194757,
      "loss": 2.2895,
      "step": 6004
    },
    {
      "epoch": 0.040800051297316786,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009973066388499843,
      "loss": 2.3468,
      "step": 6005
    },
    {
      "epoch": 0.04080684564391084,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009973055235502457,
      "loss": 2.4288,
      "step": 6006
    },
    {
      "epoch": 0.0408136399905049,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009973044080202602,
      "loss": 2.3433,
      "step": 6007
    },
    {
      "epoch": 0.040820434337098956,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009973032922600286,
      "loss": 2.2434,
      "step": 6008
    },
    {
      "epoch": 0.04082722868369302,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009973021762695513,
      "loss": 2.2598,
      "step": 6009
    },
    {
      "epoch": 0.040834023030287075,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009973010600488289,
      "loss": 2.4158,
      "step": 6010
    },
    {
      "epoch": 0.04084081737688113,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009972999435978616,
      "loss": 2.3916,
      "step": 6011
    },
    {
      "epoch": 0.04084761172347519,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009972988269166501,
      "loss": 2.3688,
      "step": 6012
    },
    {
      "epoch": 0.040854406070069245,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009972977100051951,
      "loss": 2.5031,
      "step": 6013
    },
    {
      "epoch": 0.04086120041666331,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997296592863497,
      "loss": 2.4674,
      "step": 6014
    },
    {
      "epoch": 0.040867994763257365,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997295475491556,
      "loss": 2.2852,
      "step": 6015
    },
    {
      "epoch": 0.04087478910985142,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009972943578893732,
      "loss": 2.4094,
      "step": 6016
    },
    {
      "epoch": 0.04088158345644548,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009972932400569487,
      "loss": 2.3217,
      "step": 6017
    },
    {
      "epoch": 0.040888377803039534,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997292121994283,
      "loss": 2.3229,
      "step": 6018
    },
    {
      "epoch": 0.04089517214963359,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009972910037013769,
      "loss": 2.31,
      "step": 6019
    },
    {
      "epoch": 0.040901966496227654,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009972898851782308,
      "loss": 2.3822,
      "step": 6020
    },
    {
      "epoch": 0.04090876084282171,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997288766424845,
      "loss": 2.299,
      "step": 6021
    },
    {
      "epoch": 0.04091555518941577,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009972876474412206,
      "loss": 2.3118,
      "step": 6022
    },
    {
      "epoch": 0.040922349536009824,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009972865282273573,
      "loss": 2.1661,
      "step": 6023
    },
    {
      "epoch": 0.04092914388260388,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009972854087832563,
      "loss": 2.2928,
      "step": 6024
    },
    {
      "epoch": 0.04093593822919794,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009972842891089178,
      "loss": 2.2339,
      "step": 6025
    },
    {
      "epoch": 0.040942732575792,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009972831692043422,
      "loss": 2.4738,
      "step": 6026
    },
    {
      "epoch": 0.040949526922386056,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009972820490695303,
      "loss": 2.2917,
      "step": 6027
    },
    {
      "epoch": 0.04095632126898011,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009972809287044826,
      "loss": 2.2875,
      "step": 6028
    },
    {
      "epoch": 0.04096311561557417,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009972798081091994,
      "loss": 2.2896,
      "step": 6029
    },
    {
      "epoch": 0.040969909962168226,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009972786872836815,
      "loss": 2.3791,
      "step": 6030
    },
    {
      "epoch": 0.04097670430876229,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009972775662279289,
      "loss": 2.1383,
      "step": 6031
    },
    {
      "epoch": 0.040983498655356346,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997276444941943,
      "loss": 2.4492,
      "step": 6032
    },
    {
      "epoch": 0.0409902930019504,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009972753234257234,
      "loss": 2.4617,
      "step": 6033
    },
    {
      "epoch": 0.04099708734854446,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009972742016792712,
      "loss": 2.4811,
      "step": 6034
    },
    {
      "epoch": 0.041003881695138515,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009972730797025869,
      "loss": 2.4429,
      "step": 6035
    },
    {
      "epoch": 0.04101067604173258,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009972719574956705,
      "loss": 2.2367,
      "step": 6036
    },
    {
      "epoch": 0.041017470388326635,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009972708350585232,
      "loss": 2.2743,
      "step": 6037
    },
    {
      "epoch": 0.04102426473492069,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009972697123911453,
      "loss": 2.2932,
      "step": 6038
    },
    {
      "epoch": 0.04103105908151475,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009972685894935368,
      "loss": 2.2148,
      "step": 6039
    },
    {
      "epoch": 0.041037853428108804,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997267466365699,
      "loss": 2.313,
      "step": 6040
    },
    {
      "epoch": 0.04104464777470286,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009972663430076318,
      "loss": 2.4585,
      "step": 6041
    },
    {
      "epoch": 0.041051442121296924,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009972652194193362,
      "loss": 2.2781,
      "step": 6042
    },
    {
      "epoch": 0.04105823646789098,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009972640956008123,
      "loss": 2.2676,
      "step": 6043
    },
    {
      "epoch": 0.04106503081448504,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009972629715520609,
      "loss": 2.3718,
      "step": 6044
    },
    {
      "epoch": 0.041071825161079094,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009972618472730826,
      "loss": 2.333,
      "step": 6045
    },
    {
      "epoch": 0.04107861950767315,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009972607227638776,
      "loss": 2.2998,
      "step": 6046
    },
    {
      "epoch": 0.041085413854267214,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009972595980244467,
      "loss": 2.4295,
      "step": 6047
    },
    {
      "epoch": 0.04109220820086127,
      "grad_norm": 0.3125,
      "learning_rate": 0.0009972584730547902,
      "loss": 2.2585,
      "step": 6048
    },
    {
      "epoch": 0.04109900254745533,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009972573478549088,
      "loss": 2.2703,
      "step": 6049
    },
    {
      "epoch": 0.04110579689404938,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997256222424803,
      "loss": 2.3501,
      "step": 6050
    },
    {
      "epoch": 0.04111259124064344,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009972550967644732,
      "loss": 2.2998,
      "step": 6051
    },
    {
      "epoch": 0.041119385587237496,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00099725397087392,
      "loss": 2.2563,
      "step": 6052
    },
    {
      "epoch": 0.04112617993383156,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009972528447531436,
      "loss": 2.3394,
      "step": 6053
    },
    {
      "epoch": 0.041132974280425616,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009972517184021453,
      "loss": 2.4677,
      "step": 6054
    },
    {
      "epoch": 0.04113976862701967,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009972505918209248,
      "loss": 2.3901,
      "step": 6055
    },
    {
      "epoch": 0.04114656297361373,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009972494650094834,
      "loss": 2.3233,
      "step": 6056
    },
    {
      "epoch": 0.041153357320207785,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009972483379678207,
      "loss": 2.1427,
      "step": 6057
    },
    {
      "epoch": 0.04116015166680185,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000997247210695938,
      "loss": 2.2669,
      "step": 6058
    },
    {
      "epoch": 0.041166946013395905,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009972460831938357,
      "loss": 2.2444,
      "step": 6059
    },
    {
      "epoch": 0.04117374035998996,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997244955461514,
      "loss": 2.3023,
      "step": 6060
    },
    {
      "epoch": 0.04118053470658402,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009972438274989735,
      "loss": 2.2798,
      "step": 6061
    },
    {
      "epoch": 0.041187329053178075,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000997242699306215,
      "loss": 2.394,
      "step": 6062
    },
    {
      "epoch": 0.04119412339977213,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009972415708832385,
      "loss": 2.3107,
      "step": 6063
    },
    {
      "epoch": 0.041200917746366195,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009972404422300453,
      "loss": 2.3733,
      "step": 6064
    },
    {
      "epoch": 0.04120771209296025,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009972393133466351,
      "loss": 2.3573,
      "step": 6065
    },
    {
      "epoch": 0.04121450643955431,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997238184233009,
      "loss": 2.3629,
      "step": 6066
    },
    {
      "epoch": 0.041221300786148364,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009972370548891672,
      "loss": 2.3289,
      "step": 6067
    },
    {
      "epoch": 0.04122809513274242,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009972359253151105,
      "loss": 2.3948,
      "step": 6068
    },
    {
      "epoch": 0.041234889479336484,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009972347955108391,
      "loss": 2.3945,
      "step": 6069
    },
    {
      "epoch": 0.04124168382593054,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009972336654763536,
      "loss": 2.4322,
      "step": 6070
    },
    {
      "epoch": 0.0412484781725246,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000997232535211655,
      "loss": 2.4723,
      "step": 6071
    },
    {
      "epoch": 0.04125527251911865,
      "grad_norm": 0.15625,
      "learning_rate": 0.000997231404716743,
      "loss": 2.3227,
      "step": 6072
    },
    {
      "epoch": 0.04126206686571271,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009972302739916189,
      "loss": 2.2573,
      "step": 6073
    },
    {
      "epoch": 0.04126886121230677,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009972291430362828,
      "loss": 2.327,
      "step": 6074
    },
    {
      "epoch": 0.04127565555890083,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009972280118507352,
      "loss": 2.3539,
      "step": 6075
    },
    {
      "epoch": 0.041282449905494886,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009972268804349768,
      "loss": 2.3677,
      "step": 6076
    },
    {
      "epoch": 0.04128924425208894,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000997225748789008,
      "loss": 2.3138,
      "step": 6077
    },
    {
      "epoch": 0.041296038598683,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009972246169128293,
      "loss": 2.1813,
      "step": 6078
    },
    {
      "epoch": 0.041302832945277056,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009972234848064417,
      "loss": 2.2798,
      "step": 6079
    },
    {
      "epoch": 0.04130962729187112,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009972223524698448,
      "loss": 2.3353,
      "step": 6080
    },
    {
      "epoch": 0.041316421638465176,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00099722121990304,
      "loss": 2.276,
      "step": 6081
    },
    {
      "epoch": 0.04132321598505923,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009972200871060273,
      "loss": 2.3097,
      "step": 6082
    },
    {
      "epoch": 0.04133001033165329,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009972189540788075,
      "loss": 2.3136,
      "step": 6083
    },
    {
      "epoch": 0.041336804678247345,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997217820821381,
      "loss": 2.4237,
      "step": 6084
    },
    {
      "epoch": 0.04134359902484141,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009972166873337485,
      "loss": 2.3204,
      "step": 6085
    },
    {
      "epoch": 0.041350393371435465,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009972155536159101,
      "loss": 2.2783,
      "step": 6086
    },
    {
      "epoch": 0.04135718771802952,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009972144196678668,
      "loss": 2.5178,
      "step": 6087
    },
    {
      "epoch": 0.04136398206462358,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997213285489619,
      "loss": 2.2472,
      "step": 6088
    },
    {
      "epoch": 0.041370776411217634,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009972121510811668,
      "loss": 2.2761,
      "step": 6089
    },
    {
      "epoch": 0.04137757075781169,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009972110164425114,
      "loss": 2.2866,
      "step": 6090
    },
    {
      "epoch": 0.041384365104405754,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009972098815736528,
      "loss": 2.3838,
      "step": 6091
    },
    {
      "epoch": 0.04139115945099981,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009972087464745919,
      "loss": 2.3998,
      "step": 6092
    },
    {
      "epoch": 0.04139795379759387,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000997207611145329,
      "loss": 2.3562,
      "step": 6093
    },
    {
      "epoch": 0.041404748144187924,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009972064755858646,
      "loss": 2.3409,
      "step": 6094
    },
    {
      "epoch": 0.04141154249078198,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009972053397961993,
      "loss": 2.2155,
      "step": 6095
    },
    {
      "epoch": 0.041418336837376044,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009972042037763335,
      "loss": 2.4023,
      "step": 6096
    },
    {
      "epoch": 0.0414251311839701,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997203067526268,
      "loss": 2.4894,
      "step": 6097
    },
    {
      "epoch": 0.041431925530564157,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009972019310460034,
      "loss": 2.4052,
      "step": 6098
    },
    {
      "epoch": 0.04143871987715821,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009972007943355396,
      "loss": 2.4378,
      "step": 6099
    },
    {
      "epoch": 0.04144551422375227,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009971996573948777,
      "loss": 2.2258,
      "step": 6100
    },
    {
      "epoch": 0.041452308570346326,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997198520224018,
      "loss": 2.1696,
      "step": 6101
    },
    {
      "epoch": 0.04145910291694039,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009971973828229615,
      "loss": 2.3563,
      "step": 6102
    },
    {
      "epoch": 0.041465897263534446,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009971962451917078,
      "loss": 2.2256,
      "step": 6103
    },
    {
      "epoch": 0.0414726916101285,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009971951073302582,
      "loss": 2.4227,
      "step": 6104
    },
    {
      "epoch": 0.04147948595672256,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997193969238613,
      "loss": 2.3645,
      "step": 6105
    },
    {
      "epoch": 0.041486280303316615,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971928309167725,
      "loss": 2.3602,
      "step": 6106
    },
    {
      "epoch": 0.04149307464991068,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009971916923647375,
      "loss": 2.3614,
      "step": 6107
    },
    {
      "epoch": 0.041499868996504735,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009971905535825084,
      "loss": 2.3349,
      "step": 6108
    },
    {
      "epoch": 0.04150666334309879,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009971894145700858,
      "loss": 2.3796,
      "step": 6109
    },
    {
      "epoch": 0.04151345768969285,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009971882753274702,
      "loss": 2.2803,
      "step": 6110
    },
    {
      "epoch": 0.041520252036286905,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009971871358546622,
      "loss": 2.3121,
      "step": 6111
    },
    {
      "epoch": 0.04152704638288096,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009971859961516623,
      "loss": 2.3916,
      "step": 6112
    },
    {
      "epoch": 0.041533840729475024,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971848562184708,
      "loss": 2.389,
      "step": 6113
    },
    {
      "epoch": 0.04154063507606908,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009971837160550885,
      "loss": 2.3515,
      "step": 6114
    },
    {
      "epoch": 0.04154742942266314,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009971825756615157,
      "loss": 2.3618,
      "step": 6115
    },
    {
      "epoch": 0.041554223769257194,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009971814350377534,
      "loss": 2.2997,
      "step": 6116
    },
    {
      "epoch": 0.04156101811585125,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009971802941838015,
      "loss": 2.2499,
      "step": 6117
    },
    {
      "epoch": 0.041567812462445314,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000997179153099661,
      "loss": 2.2745,
      "step": 6118
    },
    {
      "epoch": 0.04157460680903937,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997178011785332,
      "loss": 2.3561,
      "step": 6119
    },
    {
      "epoch": 0.04158140115563343,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009971768702408156,
      "loss": 2.2058,
      "step": 6120
    },
    {
      "epoch": 0.04158819550222748,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997175728466112,
      "loss": 2.3158,
      "step": 6121
    },
    {
      "epoch": 0.04159498984882154,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009971745864612216,
      "loss": 2.2998,
      "step": 6122
    },
    {
      "epoch": 0.041601784195415596,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009971734442261449,
      "loss": 2.2415,
      "step": 6123
    },
    {
      "epoch": 0.04160857854200966,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000997172301760883,
      "loss": 2.3768,
      "step": 6124
    },
    {
      "epoch": 0.041615372888603716,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971711590654356,
      "loss": 2.4512,
      "step": 6125
    },
    {
      "epoch": 0.04162216723519777,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009971700161398038,
      "loss": 2.4186,
      "step": 6126
    },
    {
      "epoch": 0.04162896158179183,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009971688729839881,
      "loss": 2.4498,
      "step": 6127
    },
    {
      "epoch": 0.041635755928385886,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009971677295979888,
      "loss": 2.3252,
      "step": 6128
    },
    {
      "epoch": 0.04164255027497995,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009971665859818065,
      "loss": 2.2021,
      "step": 6129
    },
    {
      "epoch": 0.041649344621574005,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000997165442135442,
      "loss": 2.3028,
      "step": 6130
    },
    {
      "epoch": 0.04165613896816806,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009971642980588955,
      "loss": 2.3387,
      "step": 6131
    },
    {
      "epoch": 0.04166293331476212,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009971631537521674,
      "loss": 2.4413,
      "step": 6132
    },
    {
      "epoch": 0.041669727661356175,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009971620092152588,
      "loss": 2.3308,
      "step": 6133
    },
    {
      "epoch": 0.04167652200795024,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009971608644481697,
      "loss": 2.3762,
      "step": 6134
    },
    {
      "epoch": 0.041683316354544295,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009971597194509008,
      "loss": 2.1294,
      "step": 6135
    },
    {
      "epoch": 0.04169011070113835,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009971585742234528,
      "loss": 2.3487,
      "step": 6136
    },
    {
      "epoch": 0.04169690504773241,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009971574287658258,
      "loss": 2.3561,
      "step": 6137
    },
    {
      "epoch": 0.041703699394326464,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997156283078021,
      "loss": 2.2651,
      "step": 6138
    },
    {
      "epoch": 0.04171049374092052,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009971551371600383,
      "loss": 2.1398,
      "step": 6139
    },
    {
      "epoch": 0.041717288087514584,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009971539910118784,
      "loss": 2.319,
      "step": 6140
    },
    {
      "epoch": 0.04172408243410864,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000997152844633542,
      "loss": 2.3583,
      "step": 6141
    },
    {
      "epoch": 0.0417308767807027,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009971516980250296,
      "loss": 2.4193,
      "step": 6142
    },
    {
      "epoch": 0.041737671127296753,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009971505511863416,
      "loss": 2.1732,
      "step": 6143
    },
    {
      "epoch": 0.04174446547389081,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971494041174788,
      "loss": 2.3599,
      "step": 6144
    },
    {
      "epoch": 0.04175125982048487,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997148256818441,
      "loss": 2.2765,
      "step": 6145
    },
    {
      "epoch": 0.04175805416707893,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009971471092892297,
      "loss": 2.4268,
      "step": 6146
    },
    {
      "epoch": 0.041764848513672986,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009971459615298447,
      "loss": 2.4235,
      "step": 6147
    },
    {
      "epoch": 0.04177164286026704,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009971448135402872,
      "loss": 2.4239,
      "step": 6148
    },
    {
      "epoch": 0.0417784372068611,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997143665320557,
      "loss": 2.2912,
      "step": 6149
    },
    {
      "epoch": 0.041785231553455156,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009971425168706551,
      "loss": 2.1987,
      "step": 6150
    },
    {
      "epoch": 0.04179202590004922,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009971413681905822,
      "loss": 2.2106,
      "step": 6151
    },
    {
      "epoch": 0.041798820246643276,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009971402192803381,
      "loss": 2.2749,
      "step": 6152
    },
    {
      "epoch": 0.04180561459323733,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009971390701399238,
      "loss": 2.2578,
      "step": 6153
    },
    {
      "epoch": 0.04181240893983139,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00099713792076934,
      "loss": 2.2304,
      "step": 6154
    },
    {
      "epoch": 0.041819203286425445,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009971367711685871,
      "loss": 2.3082,
      "step": 6155
    },
    {
      "epoch": 0.04182599763301951,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009971356213376655,
      "loss": 2.3006,
      "step": 6156
    },
    {
      "epoch": 0.041832791979613565,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009971344712765757,
      "loss": 2.3588,
      "step": 6157
    },
    {
      "epoch": 0.04183958632620762,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009971333209853185,
      "loss": 2.3661,
      "step": 6158
    },
    {
      "epoch": 0.04184638067280168,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009971321704638943,
      "loss": 2.4255,
      "step": 6159
    },
    {
      "epoch": 0.041853175019395734,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009971310197123035,
      "loss": 2.2442,
      "step": 6160
    },
    {
      "epoch": 0.04185996936598979,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009971298687305467,
      "loss": 2.2642,
      "step": 6161
    },
    {
      "epoch": 0.041866763712583854,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009971287175186245,
      "loss": 2.3858,
      "step": 6162
    },
    {
      "epoch": 0.04187355805917791,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009971275660765374,
      "loss": 2.4732,
      "step": 6163
    },
    {
      "epoch": 0.04188035240577197,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997126414404286,
      "loss": 2.1827,
      "step": 6164
    },
    {
      "epoch": 0.041887146752366024,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971252625018707,
      "loss": 2.2575,
      "step": 6165
    },
    {
      "epoch": 0.04189394109896008,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000997124110369292,
      "loss": 2.4036,
      "step": 6166
    },
    {
      "epoch": 0.041900735445554144,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971229580065506,
      "loss": 2.372,
      "step": 6167
    },
    {
      "epoch": 0.0419075297921482,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009971218054136471,
      "loss": 2.287,
      "step": 6168
    },
    {
      "epoch": 0.04191432413874226,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009971206525905818,
      "loss": 2.2431,
      "step": 6169
    },
    {
      "epoch": 0.04192111848533631,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009971194995373556,
      "loss": 2.2553,
      "step": 6170
    },
    {
      "epoch": 0.04192791283193037,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009971183462539684,
      "loss": 2.2472,
      "step": 6171
    },
    {
      "epoch": 0.041934707178524426,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009971171927404213,
      "loss": 2.3513,
      "step": 6172
    },
    {
      "epoch": 0.04194150152511849,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009971160389967146,
      "loss": 2.3505,
      "step": 6173
    },
    {
      "epoch": 0.041948295871712546,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000997114885022849,
      "loss": 2.3445,
      "step": 6174
    },
    {
      "epoch": 0.0419550902183066,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971137308188245,
      "loss": 2.4073,
      "step": 6175
    },
    {
      "epoch": 0.04196188456490066,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009971125763846424,
      "loss": 2.3853,
      "step": 6176
    },
    {
      "epoch": 0.041968678911494715,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009971114217203027,
      "loss": 2.3416,
      "step": 6177
    },
    {
      "epoch": 0.04197547325808878,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009971102668258063,
      "loss": 2.3598,
      "step": 6178
    },
    {
      "epoch": 0.041982267604682835,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009971091117011535,
      "loss": 2.374,
      "step": 6179
    },
    {
      "epoch": 0.04198906195127689,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009971079563463447,
      "loss": 2.2319,
      "step": 6180
    },
    {
      "epoch": 0.04199585629787095,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009971068007613808,
      "loss": 2.3535,
      "step": 6181
    },
    {
      "epoch": 0.042002650644465005,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997105644946262,
      "loss": 2.3201,
      "step": 6182
    },
    {
      "epoch": 0.04200944499105906,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000997104488900989,
      "loss": 2.3956,
      "step": 6183
    },
    {
      "epoch": 0.042016239337653125,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009971033326255626,
      "loss": 2.3749,
      "step": 6184
    },
    {
      "epoch": 0.04202303368424718,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009971021761199827,
      "loss": 2.3227,
      "step": 6185
    },
    {
      "epoch": 0.04202982803084124,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009971010193842503,
      "loss": 2.4074,
      "step": 6186
    },
    {
      "epoch": 0.042036622377435294,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009970998624183659,
      "loss": 2.4104,
      "step": 6187
    },
    {
      "epoch": 0.04204341672402935,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009970987052223298,
      "loss": 2.3221,
      "step": 6188
    },
    {
      "epoch": 0.042050211070623414,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000997097547796143,
      "loss": 2.2216,
      "step": 6189
    },
    {
      "epoch": 0.04205700541721747,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009970963901398055,
      "loss": 2.4339,
      "step": 6190
    },
    {
      "epoch": 0.04206379976381153,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009970952322533183,
      "loss": 2.291,
      "step": 6191
    },
    {
      "epoch": 0.04207059411040558,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009970940741366814,
      "loss": 2.2702,
      "step": 6192
    },
    {
      "epoch": 0.04207738845699964,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009970929157898957,
      "loss": 2.4581,
      "step": 6193
    },
    {
      "epoch": 0.0420841828035937,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000997091757212962,
      "loss": 2.4196,
      "step": 6194
    },
    {
      "epoch": 0.04209097715018776,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009970905984058801,
      "loss": 2.327,
      "step": 6195
    },
    {
      "epoch": 0.042097771496781816,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997089439368651,
      "loss": 2.2606,
      "step": 6196
    },
    {
      "epoch": 0.04210456584337587,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009970882801012754,
      "loss": 2.2487,
      "step": 6197
    },
    {
      "epoch": 0.04211136018996993,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009970871206037536,
      "loss": 2.4173,
      "step": 6198
    },
    {
      "epoch": 0.042118154536563986,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009970859608760861,
      "loss": 2.3443,
      "step": 6199
    },
    {
      "epoch": 0.04212494888315805,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009970848009182734,
      "loss": 2.3206,
      "step": 6200
    },
    {
      "epoch": 0.042131743229752106,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009970836407303164,
      "loss": 2.3162,
      "step": 6201
    },
    {
      "epoch": 0.04213853757634616,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009970824803122152,
      "loss": 2.4439,
      "step": 6202
    },
    {
      "epoch": 0.04214533192294022,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009970813196639706,
      "loss": 2.2688,
      "step": 6203
    },
    {
      "epoch": 0.042152126269534275,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009970801587855828,
      "loss": 2.2743,
      "step": 6204
    },
    {
      "epoch": 0.04215892061612834,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009970789976770527,
      "loss": 2.4612,
      "step": 6205
    },
    {
      "epoch": 0.042165714962722395,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009970778363383808,
      "loss": 2.3797,
      "step": 6206
    },
    {
      "epoch": 0.04217250930931645,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009970766747695676,
      "loss": 2.4064,
      "step": 6207
    },
    {
      "epoch": 0.04217930365591051,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009970755129706135,
      "loss": 2.242,
      "step": 6208
    },
    {
      "epoch": 0.042186098002504564,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009970743509415193,
      "loss": 2.4028,
      "step": 6209
    },
    {
      "epoch": 0.04219289234909862,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000997073188682285,
      "loss": 2.3295,
      "step": 6210
    },
    {
      "epoch": 0.042199686695692684,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009970720261929119,
      "loss": 2.1435,
      "step": 6211
    },
    {
      "epoch": 0.04220648104228674,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009970708634734,
      "loss": 2.3048,
      "step": 6212
    },
    {
      "epoch": 0.0422132753888808,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00099706970052375,
      "loss": 2.3596,
      "step": 6213
    },
    {
      "epoch": 0.042220069735474854,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009970685373439623,
      "loss": 2.4501,
      "step": 6214
    },
    {
      "epoch": 0.04222686408206891,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000997067373934038,
      "loss": 2.3912,
      "step": 6215
    },
    {
      "epoch": 0.042233658428662973,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009970662102939767,
      "loss": 2.4084,
      "step": 6216
    },
    {
      "epoch": 0.04224045277525703,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009970650464237796,
      "loss": 2.2719,
      "step": 6217
    },
    {
      "epoch": 0.042247247121851086,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970638823234472,
      "loss": 2.3825,
      "step": 6218
    },
    {
      "epoch": 0.04225404146844514,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099706271799298,
      "loss": 2.2899,
      "step": 6219
    },
    {
      "epoch": 0.0422608358150392,
      "grad_norm": 0.140625,
      "learning_rate": 0.000997061553432378,
      "loss": 2.2932,
      "step": 6220
    },
    {
      "epoch": 0.042267630161633256,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970603886416427,
      "loss": 2.2781,
      "step": 6221
    },
    {
      "epoch": 0.04227442450822732,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009970592236207738,
      "loss": 2.4939,
      "step": 6222
    },
    {
      "epoch": 0.042281218854821376,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009970580583697724,
      "loss": 2.5378,
      "step": 6223
    },
    {
      "epoch": 0.04228801320141543,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970568928886387,
      "loss": 2.2712,
      "step": 6224
    },
    {
      "epoch": 0.04229480754800949,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009970557271773734,
      "loss": 2.2924,
      "step": 6225
    },
    {
      "epoch": 0.042301601894603545,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000997054561235977,
      "loss": 2.4069,
      "step": 6226
    },
    {
      "epoch": 0.04230839624119761,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099705339506445,
      "loss": 2.2554,
      "step": 6227
    },
    {
      "epoch": 0.042315190587791665,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000997052228662793,
      "loss": 2.4162,
      "step": 6228
    },
    {
      "epoch": 0.04232198493438572,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009970510620310065,
      "loss": 2.441,
      "step": 6229
    },
    {
      "epoch": 0.04232877928097978,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009970498951690913,
      "loss": 2.5254,
      "step": 6230
    },
    {
      "epoch": 0.042335573627573835,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009970487280770474,
      "loss": 2.4006,
      "step": 6231
    },
    {
      "epoch": 0.04234236797416789,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009970475607548755,
      "loss": 2.4252,
      "step": 6232
    },
    {
      "epoch": 0.042349162320761954,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009970463932025765,
      "loss": 2.235,
      "step": 6233
    },
    {
      "epoch": 0.04235595666735601,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009970452254201507,
      "loss": 2.2479,
      "step": 6234
    },
    {
      "epoch": 0.04236275101395007,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009970440574075987,
      "loss": 2.2352,
      "step": 6235
    },
    {
      "epoch": 0.042369545360544124,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000997042889164921,
      "loss": 2.5828,
      "step": 6236
    },
    {
      "epoch": 0.04237633970713818,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009970417206921181,
      "loss": 2.3237,
      "step": 6237
    },
    {
      "epoch": 0.042383134053732244,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009970405519891905,
      "loss": 2.2462,
      "step": 6238
    },
    {
      "epoch": 0.0423899284003263,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000997039383056139,
      "loss": 2.4258,
      "step": 6239
    },
    {
      "epoch": 0.04239672274692036,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000997038213892964,
      "loss": 2.1913,
      "step": 6240
    },
    {
      "epoch": 0.04240351709351441,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009970370444996658,
      "loss": 2.1968,
      "step": 6241
    },
    {
      "epoch": 0.04241031144010847,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997035874876245,
      "loss": 2.3262,
      "step": 6242
    },
    {
      "epoch": 0.042417105786702526,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970347050227027,
      "loss": 2.3199,
      "step": 6243
    },
    {
      "epoch": 0.04242390013329659,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009970335349390388,
      "loss": 2.3115,
      "step": 6244
    },
    {
      "epoch": 0.042430694479890646,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009970323646252542,
      "loss": 2.3939,
      "step": 6245
    },
    {
      "epoch": 0.0424374888264847,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009970311940813493,
      "loss": 2.2935,
      "step": 6246
    },
    {
      "epoch": 0.04244428317307876,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009970300233073245,
      "loss": 2.2679,
      "step": 6247
    },
    {
      "epoch": 0.042451077519672815,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009970288523031806,
      "loss": 2.2949,
      "step": 6248
    },
    {
      "epoch": 0.04245787186626688,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000997027681068918,
      "loss": 2.1901,
      "step": 6249
    },
    {
      "epoch": 0.042464666212860935,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009970265096045372,
      "loss": 2.2817,
      "step": 6250
    },
    {
      "epoch": 0.04247146055945499,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000997025337910039,
      "loss": 2.1095,
      "step": 6251
    },
    {
      "epoch": 0.04247825490604905,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009970241659854238,
      "loss": 2.3826,
      "step": 6252
    },
    {
      "epoch": 0.042485049252643105,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009970229938306918,
      "loss": 2.3664,
      "step": 6253
    },
    {
      "epoch": 0.04249184359923717,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009970218214458441,
      "loss": 2.2576,
      "step": 6254
    },
    {
      "epoch": 0.042498637945831225,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000997020648830881,
      "loss": 2.2182,
      "step": 6255
    },
    {
      "epoch": 0.04250543229242528,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000997019475985803,
      "loss": 2.3845,
      "step": 6256
    },
    {
      "epoch": 0.04251222663901934,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970183029106108,
      "loss": 2.2034,
      "step": 6257
    },
    {
      "epoch": 0.042519020985613394,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009970171296053046,
      "loss": 2.2738,
      "step": 6258
    },
    {
      "epoch": 0.04252581533220745,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009970159560698853,
      "loss": 2.4049,
      "step": 6259
    },
    {
      "epoch": 0.042532609678801514,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000997014782304353,
      "loss": 2.2243,
      "step": 6260
    },
    {
      "epoch": 0.04253940402539557,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009970136083087091,
      "loss": 2.2423,
      "step": 6261
    },
    {
      "epoch": 0.04254619837198963,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009970124340829532,
      "loss": 2.3617,
      "step": 6262
    },
    {
      "epoch": 0.04255299271858368,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970112596270864,
      "loss": 2.4541,
      "step": 6263
    },
    {
      "epoch": 0.04255978706517774,
      "grad_norm": 0.1875,
      "learning_rate": 0.000997010084941109,
      "loss": 2.2772,
      "step": 6264
    },
    {
      "epoch": 0.0425665814117718,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009970089100250217,
      "loss": 2.4377,
      "step": 6265
    },
    {
      "epoch": 0.04257337575836586,
      "grad_norm": 0.15625,
      "learning_rate": 0.000997007734878825,
      "loss": 2.5164,
      "step": 6266
    },
    {
      "epoch": 0.042580170104959916,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970065595025193,
      "loss": 2.2222,
      "step": 6267
    },
    {
      "epoch": 0.04258696445155397,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009970053838961053,
      "loss": 2.3535,
      "step": 6268
    },
    {
      "epoch": 0.04259375879814803,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009970042080595836,
      "loss": 2.3561,
      "step": 6269
    },
    {
      "epoch": 0.042600553144742086,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009970030319929545,
      "loss": 2.1844,
      "step": 6270
    },
    {
      "epoch": 0.04260734749133615,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000997001855696219,
      "loss": 2.2118,
      "step": 6271
    },
    {
      "epoch": 0.042614141837930206,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009970006791693768,
      "loss": 2.4545,
      "step": 6272
    },
    {
      "epoch": 0.04262093618452426,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009969995024124295,
      "loss": 2.3158,
      "step": 6273
    },
    {
      "epoch": 0.04262773053111832,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000996998325425377,
      "loss": 2.4847,
      "step": 6274
    },
    {
      "epoch": 0.042634524877712375,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009969971482082197,
      "loss": 2.2105,
      "step": 6275
    },
    {
      "epoch": 0.04264131922430644,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009969959707609587,
      "loss": 2.335,
      "step": 6276
    },
    {
      "epoch": 0.042648113570900495,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000996994793083594,
      "loss": 2.2846,
      "step": 6277
    },
    {
      "epoch": 0.04265490791749455,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009969936151761267,
      "loss": 2.3978,
      "step": 6278
    },
    {
      "epoch": 0.04266170226408861,
      "grad_norm": 0.140625,
      "learning_rate": 0.000996992437038557,
      "loss": 2.3325,
      "step": 6279
    },
    {
      "epoch": 0.042668496610682664,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009969912586708853,
      "loss": 2.3348,
      "step": 6280
    },
    {
      "epoch": 0.04267529095727672,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009969900800731128,
      "loss": 2.2224,
      "step": 6281
    },
    {
      "epoch": 0.042682085303870784,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009969889012452391,
      "loss": 2.2594,
      "step": 6282
    },
    {
      "epoch": 0.04268887965046484,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009969877221872655,
      "loss": 2.2686,
      "step": 6283
    },
    {
      "epoch": 0.0426956739970589,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000996986542899192,
      "loss": 2.2634,
      "step": 6284
    },
    {
      "epoch": 0.042702468343652954,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009969853633810196,
      "loss": 2.3108,
      "step": 6285
    },
    {
      "epoch": 0.04270926269024701,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009969841836327488,
      "loss": 2.2559,
      "step": 6286
    },
    {
      "epoch": 0.042716057036841074,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009969830036543799,
      "loss": 2.2164,
      "step": 6287
    },
    {
      "epoch": 0.04272285138343513,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009969818234459135,
      "loss": 2.3015,
      "step": 6288
    },
    {
      "epoch": 0.04272964573002919,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009969806430073504,
      "loss": 2.2674,
      "step": 6289
    },
    {
      "epoch": 0.04273644007662324,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009969794623386908,
      "loss": 2.2789,
      "step": 6290
    },
    {
      "epoch": 0.0427432344232173,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009969782814399355,
      "loss": 2.3447,
      "step": 6291
    },
    {
      "epoch": 0.042750028769811356,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000996977100311085,
      "loss": 2.3304,
      "step": 6292
    },
    {
      "epoch": 0.04275682311640542,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009969759189521396,
      "loss": 2.4733,
      "step": 6293
    },
    {
      "epoch": 0.042763617462999476,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009969747373631001,
      "loss": 2.1448,
      "step": 6294
    },
    {
      "epoch": 0.04277041180959353,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009969735555439671,
      "loss": 2.3404,
      "step": 6295
    },
    {
      "epoch": 0.04277720615618759,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996972373494741,
      "loss": 2.3709,
      "step": 6296
    },
    {
      "epoch": 0.042784000502781645,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009969711912154225,
      "loss": 2.2267,
      "step": 6297
    },
    {
      "epoch": 0.04279079484937571,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009969700087060118,
      "loss": 2.2929,
      "step": 6298
    },
    {
      "epoch": 0.042797589195969765,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099696882596651,
      "loss": 2.392,
      "step": 6299
    },
    {
      "epoch": 0.04280438354256382,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996967642996917,
      "loss": 2.2907,
      "step": 6300
    },
    {
      "epoch": 0.04281117788915788,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009969664597972339,
      "loss": 2.2627,
      "step": 6301
    },
    {
      "epoch": 0.042817972235751935,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009969652763674609,
      "loss": 2.307,
      "step": 6302
    },
    {
      "epoch": 0.04282476658234599,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009969640927075987,
      "loss": 2.3534,
      "step": 6303
    },
    {
      "epoch": 0.042831560928940055,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996962908817648,
      "loss": 2.2654,
      "step": 6304
    },
    {
      "epoch": 0.04283835527553411,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996961724697609,
      "loss": 2.2105,
      "step": 6305
    },
    {
      "epoch": 0.04284514962212817,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009969605403474823,
      "loss": 2.2657,
      "step": 6306
    },
    {
      "epoch": 0.042851943968722224,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009969593557672687,
      "loss": 2.368,
      "step": 6307
    },
    {
      "epoch": 0.04285873831531628,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009969581709569687,
      "loss": 2.2814,
      "step": 6308
    },
    {
      "epoch": 0.042865532661910344,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009969569859165825,
      "loss": 2.2626,
      "step": 6309
    },
    {
      "epoch": 0.0428723270085044,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009969558006461111,
      "loss": 2.284,
      "step": 6310
    },
    {
      "epoch": 0.04287912135509846,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996954615145555,
      "loss": 2.4533,
      "step": 6311
    },
    {
      "epoch": 0.04288591570169251,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009969534294149145,
      "loss": 2.4002,
      "step": 6312
    },
    {
      "epoch": 0.04289271004828657,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009969522434541901,
      "loss": 2.2897,
      "step": 6313
    },
    {
      "epoch": 0.04289950439488063,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009969510572633827,
      "loss": 2.4,
      "step": 6314
    },
    {
      "epoch": 0.04290629874147469,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009969498708424927,
      "loss": 2.2012,
      "step": 6315
    },
    {
      "epoch": 0.042913093088068746,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009969486841915204,
      "loss": 2.3717,
      "step": 6316
    },
    {
      "epoch": 0.0429198874346628,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009969474973104667,
      "loss": 2.4534,
      "step": 6317
    },
    {
      "epoch": 0.04292668178125686,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000996946310199332,
      "loss": 2.3745,
      "step": 6318
    },
    {
      "epoch": 0.042933476127850916,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009969451228581169,
      "loss": 2.2268,
      "step": 6319
    },
    {
      "epoch": 0.04294027047444498,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009969439352868216,
      "loss": 2.4077,
      "step": 6320
    },
    {
      "epoch": 0.042947064821039035,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009969427474854472,
      "loss": 2.1959,
      "step": 6321
    },
    {
      "epoch": 0.04295385916763309,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996941559453994,
      "loss": 2.2921,
      "step": 6322
    },
    {
      "epoch": 0.04296065351422715,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009969403711924626,
      "loss": 2.3626,
      "step": 6323
    },
    {
      "epoch": 0.042967447860821205,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009969391827008534,
      "loss": 2.3713,
      "step": 6324
    },
    {
      "epoch": 0.04297424220741527,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000996937993979167,
      "loss": 2.3385,
      "step": 6325
    },
    {
      "epoch": 0.042981036554009325,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996936805027404,
      "loss": 2.2654,
      "step": 6326
    },
    {
      "epoch": 0.04298783090060338,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009969356158455653,
      "loss": 2.3255,
      "step": 6327
    },
    {
      "epoch": 0.04299462524719744,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009969344264336509,
      "loss": 2.1836,
      "step": 6328
    },
    {
      "epoch": 0.043001419593791494,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009969332367916616,
      "loss": 2.3969,
      "step": 6329
    },
    {
      "epoch": 0.04300821394038555,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009969320469195976,
      "loss": 2.2984,
      "step": 6330
    },
    {
      "epoch": 0.043015008286979614,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099693085681746,
      "loss": 2.3338,
      "step": 6331
    },
    {
      "epoch": 0.04302180263357367,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009969296664852493,
      "loss": 2.255,
      "step": 6332
    },
    {
      "epoch": 0.04302859698016773,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009969284759229655,
      "loss": 2.2628,
      "step": 6333
    },
    {
      "epoch": 0.043035391326761784,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009969272851306098,
      "loss": 2.3964,
      "step": 6334
    },
    {
      "epoch": 0.04304218567335584,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009969260941081822,
      "loss": 2.4297,
      "step": 6335
    },
    {
      "epoch": 0.0430489800199499,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009969249028556836,
      "loss": 2.3823,
      "step": 6336
    },
    {
      "epoch": 0.04305577436654396,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009969237113731147,
      "loss": 2.3121,
      "step": 6337
    },
    {
      "epoch": 0.043062568713138016,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009969225196604756,
      "loss": 2.2918,
      "step": 6338
    },
    {
      "epoch": 0.04306936305973207,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009969213277177672,
      "loss": 2.393,
      "step": 6339
    },
    {
      "epoch": 0.04307615740632613,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009969201355449898,
      "loss": 2.4648,
      "step": 6340
    },
    {
      "epoch": 0.043082951752920186,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000996918943142144,
      "loss": 2.302,
      "step": 6341
    },
    {
      "epoch": 0.04308974609951425,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009969177505092307,
      "loss": 2.2036,
      "step": 6342
    },
    {
      "epoch": 0.043096540446108306,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00099691655764625,
      "loss": 2.3783,
      "step": 6343
    },
    {
      "epoch": 0.04310333479270236,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009969153645532024,
      "loss": 2.232,
      "step": 6344
    },
    {
      "epoch": 0.04311012913929642,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009969141712300891,
      "loss": 2.33,
      "step": 6345
    },
    {
      "epoch": 0.043116923485890475,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099691297767691,
      "loss": 2.262,
      "step": 6346
    },
    {
      "epoch": 0.04312371783248454,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000996911783893666,
      "loss": 2.2729,
      "step": 6347
    },
    {
      "epoch": 0.043130512179078595,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009969105898803574,
      "loss": 2.4588,
      "step": 6348
    },
    {
      "epoch": 0.04313730652567265,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000996909395636985,
      "loss": 2.3567,
      "step": 6349
    },
    {
      "epoch": 0.04314410087226671,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009969082011635493,
      "loss": 2.2116,
      "step": 6350
    },
    {
      "epoch": 0.043150895218860764,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009969070064600508,
      "loss": 2.305,
      "step": 6351
    },
    {
      "epoch": 0.04315768956545482,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009969058115264899,
      "loss": 2.3454,
      "step": 6352
    },
    {
      "epoch": 0.043164483912048884,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009969046163628675,
      "loss": 2.3192,
      "step": 6353
    },
    {
      "epoch": 0.04317127825864294,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009969034209691837,
      "loss": 2.205,
      "step": 6354
    },
    {
      "epoch": 0.043178072605237,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009969022253454396,
      "loss": 2.3661,
      "step": 6355
    },
    {
      "epoch": 0.043184866951831054,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009969010294916352,
      "loss": 2.3256,
      "step": 6356
    },
    {
      "epoch": 0.04319166129842511,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009968998334077715,
      "loss": 2.2814,
      "step": 6357
    },
    {
      "epoch": 0.043198455645019174,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009968986370938488,
      "loss": 2.2631,
      "step": 6358
    },
    {
      "epoch": 0.04320524999161323,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009968974405498677,
      "loss": 2.2219,
      "step": 6359
    },
    {
      "epoch": 0.04321204433820729,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009968962437758289,
      "loss": 2.2774,
      "step": 6360
    },
    {
      "epoch": 0.04321883868480134,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009968950467717328,
      "loss": 2.5316,
      "step": 6361
    },
    {
      "epoch": 0.0432256330313954,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009968938495375798,
      "loss": 2.3951,
      "step": 6362
    },
    {
      "epoch": 0.043232427377989456,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996892652073371,
      "loss": 2.3754,
      "step": 6363
    },
    {
      "epoch": 0.04323922172458352,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009968914543791064,
      "loss": 2.3977,
      "step": 6364
    },
    {
      "epoch": 0.043246016071177576,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009968902564547865,
      "loss": 2.3163,
      "step": 6365
    },
    {
      "epoch": 0.04325281041777163,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009968890583004126,
      "loss": 2.3172,
      "step": 6366
    },
    {
      "epoch": 0.04325960476436569,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009968878599159845,
      "loss": 2.378,
      "step": 6367
    },
    {
      "epoch": 0.043266399110959745,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009968866613015031,
      "loss": 2.1702,
      "step": 6368
    },
    {
      "epoch": 0.04327319345755381,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009968854624569689,
      "loss": 2.3518,
      "step": 6369
    },
    {
      "epoch": 0.043279987804147865,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996884263382382,
      "loss": 2.3845,
      "step": 6370
    },
    {
      "epoch": 0.04328678215074192,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009968830640777438,
      "loss": 2.3133,
      "step": 6371
    },
    {
      "epoch": 0.04329357649733598,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009968818645430544,
      "loss": 2.5232,
      "step": 6372
    },
    {
      "epoch": 0.043300370843930035,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009968806647783143,
      "loss": 2.3242,
      "step": 6373
    },
    {
      "epoch": 0.0433071651905241,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009968794647835244,
      "loss": 2.3786,
      "step": 6374
    },
    {
      "epoch": 0.043313959537118155,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009968782645586847,
      "loss": 2.2896,
      "step": 6375
    },
    {
      "epoch": 0.04332075388371221,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009968770641037962,
      "loss": 2.2789,
      "step": 6376
    },
    {
      "epoch": 0.04332754823030627,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009968758634188593,
      "loss": 2.2834,
      "step": 6377
    },
    {
      "epoch": 0.043334342576900324,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009968746625038745,
      "loss": 2.4515,
      "step": 6378
    },
    {
      "epoch": 0.04334113692349438,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009968734613588425,
      "loss": 2.2773,
      "step": 6379
    },
    {
      "epoch": 0.043347931270088444,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009968722599837637,
      "loss": 2.402,
      "step": 6380
    },
    {
      "epoch": 0.0433547256166825,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009968710583786388,
      "loss": 2.2699,
      "step": 6381
    },
    {
      "epoch": 0.04336151996327656,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009968698565434684,
      "loss": 2.3145,
      "step": 6382
    },
    {
      "epoch": 0.04336831430987061,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009968686544782528,
      "loss": 2.4998,
      "step": 6383
    },
    {
      "epoch": 0.04337510865646467,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000996867452182993,
      "loss": 2.4538,
      "step": 6384
    },
    {
      "epoch": 0.04338190300305873,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009968662496576887,
      "loss": 2.1896,
      "step": 6385
    },
    {
      "epoch": 0.04338869734965279,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009968650469023414,
      "loss": 2.3418,
      "step": 6386
    },
    {
      "epoch": 0.043395491696246846,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009968638439169513,
      "loss": 2.1959,
      "step": 6387
    },
    {
      "epoch": 0.0434022860428409,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009968626407015188,
      "loss": 2.2824,
      "step": 6388
    },
    {
      "epoch": 0.04340908038943496,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009968614372560447,
      "loss": 2.3905,
      "step": 6389
    },
    {
      "epoch": 0.043415874736029016,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009968602335805295,
      "loss": 2.2965,
      "step": 6390
    },
    {
      "epoch": 0.04342266908262308,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009968590296749735,
      "loss": 2.3768,
      "step": 6391
    },
    {
      "epoch": 0.043429463429217136,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009968578255393776,
      "loss": 2.2498,
      "step": 6392
    },
    {
      "epoch": 0.04343625777581119,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009968566211737424,
      "loss": 2.2943,
      "step": 6393
    },
    {
      "epoch": 0.04344305212240525,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996855416578068,
      "loss": 2.4399,
      "step": 6394
    },
    {
      "epoch": 0.043449846468999305,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009968542117523554,
      "loss": 2.3645,
      "step": 6395
    },
    {
      "epoch": 0.04345664081559337,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009968530066966049,
      "loss": 2.271,
      "step": 6396
    },
    {
      "epoch": 0.043463435162187425,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009968518014108173,
      "loss": 2.2987,
      "step": 6397
    },
    {
      "epoch": 0.04347022950878148,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009968505958949927,
      "loss": 2.2861,
      "step": 6398
    },
    {
      "epoch": 0.04347702385537554,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009968493901491321,
      "loss": 2.4063,
      "step": 6399
    },
    {
      "epoch": 0.043483818201969594,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009968481841732362,
      "loss": 2.3953,
      "step": 6400
    },
    {
      "epoch": 0.04349061254856365,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996846977967305,
      "loss": 2.4016,
      "step": 6401
    },
    {
      "epoch": 0.043497406895157714,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009968457715313394,
      "loss": 2.3454,
      "step": 6402
    },
    {
      "epoch": 0.04350420124175177,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009968445648653398,
      "loss": 2.2502,
      "step": 6403
    },
    {
      "epoch": 0.04351099558834583,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996843357969307,
      "loss": 2.4045,
      "step": 6404
    },
    {
      "epoch": 0.043517789934939884,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009968421508432415,
      "loss": 2.3154,
      "step": 6405
    },
    {
      "epoch": 0.04352458428153394,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009968409434871438,
      "loss": 2.4229,
      "step": 6406
    },
    {
      "epoch": 0.043531378628128004,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009968397359010142,
      "loss": 2.1524,
      "step": 6407
    },
    {
      "epoch": 0.04353817297472206,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009968385280848537,
      "loss": 2.3356,
      "step": 6408
    },
    {
      "epoch": 0.043544967321316116,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009968373200386627,
      "loss": 2.2758,
      "step": 6409
    },
    {
      "epoch": 0.04355176166791017,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009968361117624414,
      "loss": 2.4185,
      "step": 6410
    },
    {
      "epoch": 0.04355855601450423,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000996834903256191,
      "loss": 2.4774,
      "step": 6411
    },
    {
      "epoch": 0.043565350361098286,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009968336945199116,
      "loss": 2.3055,
      "step": 6412
    },
    {
      "epoch": 0.04357214470769235,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009968324855536038,
      "loss": 2.3793,
      "step": 6413
    },
    {
      "epoch": 0.043578939054286406,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009968312763572683,
      "loss": 2.3998,
      "step": 6414
    },
    {
      "epoch": 0.04358573340088046,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996830066930906,
      "loss": 2.3079,
      "step": 6415
    },
    {
      "epoch": 0.04359252774747452,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009968288572745165,
      "loss": 2.4101,
      "step": 6416
    },
    {
      "epoch": 0.043599322094068575,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009968276473881013,
      "loss": 2.271,
      "step": 6417
    },
    {
      "epoch": 0.04360611644066264,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009968264372716604,
      "loss": 2.2933,
      "step": 6418
    },
    {
      "epoch": 0.043612910787256695,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009968252269251948,
      "loss": 2.3845,
      "step": 6419
    },
    {
      "epoch": 0.04361970513385075,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009968240163487046,
      "loss": 2.5086,
      "step": 6420
    },
    {
      "epoch": 0.04362649948044481,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009968228055421905,
      "loss": 2.3216,
      "step": 6421
    },
    {
      "epoch": 0.043633293827038865,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009968215945056534,
      "loss": 2.3706,
      "step": 6422
    },
    {
      "epoch": 0.04364008817363292,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009968203832390935,
      "loss": 2.493,
      "step": 6423
    },
    {
      "epoch": 0.043646882520226984,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009968191717425114,
      "loss": 2.4447,
      "step": 6424
    },
    {
      "epoch": 0.04365367686682104,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0009968179600159078,
      "loss": 2.3978,
      "step": 6425
    },
    {
      "epoch": 0.0436604712134151,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009968167480592831,
      "loss": 2.1156,
      "step": 6426
    },
    {
      "epoch": 0.043667265560009154,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009968155358726378,
      "loss": 2.307,
      "step": 6427
    },
    {
      "epoch": 0.04367405990660321,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009968143234559729,
      "loss": 2.3572,
      "step": 6428
    },
    {
      "epoch": 0.043680854253197274,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009968131108092884,
      "loss": 2.2793,
      "step": 6429
    },
    {
      "epoch": 0.04368764859979133,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009968118979325853,
      "loss": 2.5377,
      "step": 6430
    },
    {
      "epoch": 0.04369444294638539,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996810684825864,
      "loss": 2.3395,
      "step": 6431
    },
    {
      "epoch": 0.04370123729297944,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009968094714891248,
      "loss": 2.2919,
      "step": 6432
    },
    {
      "epoch": 0.0437080316395735,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009968082579223689,
      "loss": 2.263,
      "step": 6433
    },
    {
      "epoch": 0.043714825986167556,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009968070441255964,
      "loss": 2.4567,
      "step": 6434
    },
    {
      "epoch": 0.04372162033276162,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009968058300988077,
      "loss": 2.3324,
      "step": 6435
    },
    {
      "epoch": 0.043728414679355676,
      "grad_norm": 0.208984375,
      "learning_rate": 0.0009968046158420037,
      "loss": 2.4111,
      "step": 6436
    },
    {
      "epoch": 0.04373520902594973,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009968034013551849,
      "loss": 2.2424,
      "step": 6437
    },
    {
      "epoch": 0.04374200337254379,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009968021866383517,
      "loss": 2.3179,
      "step": 6438
    },
    {
      "epoch": 0.043748797719137845,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000996800971691505,
      "loss": 2.217,
      "step": 6439
    },
    {
      "epoch": 0.04375559206573191,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996799756514645,
      "loss": 2.3423,
      "step": 6440
    },
    {
      "epoch": 0.043762386412325965,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009967985411077724,
      "loss": 2.3154,
      "step": 6441
    },
    {
      "epoch": 0.04376918075892002,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009967973254708877,
      "loss": 2.2298,
      "step": 6442
    },
    {
      "epoch": 0.04377597510551408,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009967961096039918,
      "loss": 2.3125,
      "step": 6443
    },
    {
      "epoch": 0.043782769452108135,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009967948935070848,
      "loss": 2.321,
      "step": 6444
    },
    {
      "epoch": 0.0437895637987022,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009967936771801674,
      "loss": 2.376,
      "step": 6445
    },
    {
      "epoch": 0.043796358145296255,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009967924606232403,
      "loss": 2.4367,
      "step": 6446
    },
    {
      "epoch": 0.04380315249189031,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009967912438363041,
      "loss": 2.3684,
      "step": 6447
    },
    {
      "epoch": 0.04380994683848437,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996790026819359,
      "loss": 2.4602,
      "step": 6448
    },
    {
      "epoch": 0.043816741185078424,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996788809572406,
      "loss": 2.2905,
      "step": 6449
    },
    {
      "epoch": 0.04382353553167248,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009967875920954454,
      "loss": 2.3728,
      "step": 6450
    },
    {
      "epoch": 0.043830329878266544,
      "grad_norm": 0.140625,
      "learning_rate": 0.000996786374388478,
      "loss": 2.3754,
      "step": 6451
    },
    {
      "epoch": 0.0438371242248606,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009967851564515041,
      "loss": 2.4452,
      "step": 6452
    },
    {
      "epoch": 0.04384391857145466,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009967839382845243,
      "loss": 2.452,
      "step": 6453
    },
    {
      "epoch": 0.04385071291804871,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009967827198875392,
      "loss": 2.3732,
      "step": 6454
    },
    {
      "epoch": 0.04385750726464277,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009967815012605495,
      "loss": 2.299,
      "step": 6455
    },
    {
      "epoch": 0.04386430161123683,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009967802824035557,
      "loss": 2.2721,
      "step": 6456
    },
    {
      "epoch": 0.04387109595783089,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009967790633165583,
      "loss": 2.2693,
      "step": 6457
    },
    {
      "epoch": 0.043877890304424946,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009967778439995579,
      "loss": 2.2997,
      "step": 6458
    },
    {
      "epoch": 0.043884684651019,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996776624452555,
      "loss": 2.2878,
      "step": 6459
    },
    {
      "epoch": 0.04389147899761306,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099677540467555,
      "loss": 2.2995,
      "step": 6460
    },
    {
      "epoch": 0.043898273344207116,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009967741846685439,
      "loss": 2.29,
      "step": 6461
    },
    {
      "epoch": 0.04390506769080118,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009967729644315372,
      "loss": 2.1703,
      "step": 6462
    },
    {
      "epoch": 0.043911862037395236,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00099677174396453,
      "loss": 2.2343,
      "step": 6463
    },
    {
      "epoch": 0.04391865638398929,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009967705232675234,
      "loss": 2.4059,
      "step": 6464
    },
    {
      "epoch": 0.04392545073058335,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009967693023405179,
      "loss": 2.3846,
      "step": 6465
    },
    {
      "epoch": 0.043932245077177405,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009967680811835136,
      "loss": 2.4297,
      "step": 6466
    },
    {
      "epoch": 0.04393903942377147,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009967668597965114,
      "loss": 2.2582,
      "step": 6467
    },
    {
      "epoch": 0.043945833770365525,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009967656381795118,
      "loss": 2.4348,
      "step": 6468
    },
    {
      "epoch": 0.04395262811695958,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009967644163325156,
      "loss": 2.3215,
      "step": 6469
    },
    {
      "epoch": 0.04395942246355364,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996763194255523,
      "loss": 2.3158,
      "step": 6470
    },
    {
      "epoch": 0.043966216810147694,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996761971948535,
      "loss": 2.4197,
      "step": 6471
    },
    {
      "epoch": 0.04397301115674175,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009967607494115517,
      "loss": 2.2748,
      "step": 6472
    },
    {
      "epoch": 0.043979805503335814,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009967595266445737,
      "loss": 2.4559,
      "step": 6473
    },
    {
      "epoch": 0.04398659984992987,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000996758303647602,
      "loss": 2.3458,
      "step": 6474
    },
    {
      "epoch": 0.04399339419652393,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009967570804206368,
      "loss": 2.2849,
      "step": 6475
    },
    {
      "epoch": 0.044000188543117984,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009967558569636788,
      "loss": 2.2903,
      "step": 6476
    },
    {
      "epoch": 0.04400698288971204,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009967546332767285,
      "loss": 2.3224,
      "step": 6477
    },
    {
      "epoch": 0.044013777236306104,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009967534093597865,
      "loss": 2.4478,
      "step": 6478
    },
    {
      "epoch": 0.04402057158290016,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009967521852128536,
      "loss": 2.4091,
      "step": 6479
    },
    {
      "epoch": 0.04402736592949422,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009967509608359297,
      "loss": 2.1473,
      "step": 6480
    },
    {
      "epoch": 0.04403416027608827,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000996749736229016,
      "loss": 2.3925,
      "step": 6481
    },
    {
      "epoch": 0.04404095462268233,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009967485113921131,
      "loss": 2.3103,
      "step": 6482
    },
    {
      "epoch": 0.044047748969276386,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996747286325221,
      "loss": 2.4285,
      "step": 6483
    },
    {
      "epoch": 0.04405454331587045,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009967460610283408,
      "loss": 2.2632,
      "step": 6484
    },
    {
      "epoch": 0.044061337662464506,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009967448355014728,
      "loss": 2.355,
      "step": 6485
    },
    {
      "epoch": 0.04406813200905856,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009967436097446177,
      "loss": 2.2533,
      "step": 6486
    },
    {
      "epoch": 0.04407492635565262,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009967423837577759,
      "loss": 2.5697,
      "step": 6487
    },
    {
      "epoch": 0.044081720702246675,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996741157540948,
      "loss": 2.3835,
      "step": 6488
    },
    {
      "epoch": 0.04408851504884074,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009967399310941347,
      "loss": 2.3457,
      "step": 6489
    },
    {
      "epoch": 0.044095309395434795,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009967387044173366,
      "loss": 2.24,
      "step": 6490
    },
    {
      "epoch": 0.04410210374202885,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009967374775105542,
      "loss": 2.3265,
      "step": 6491
    },
    {
      "epoch": 0.04410889808862291,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000996736250373788,
      "loss": 2.266,
      "step": 6492
    },
    {
      "epoch": 0.044115692435216965,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009967350230070384,
      "loss": 2.2623,
      "step": 6493
    },
    {
      "epoch": 0.04412248678181102,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009967337954103065,
      "loss": 2.1775,
      "step": 6494
    },
    {
      "epoch": 0.044129281128405085,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009967325675835923,
      "loss": 2.1497,
      "step": 6495
    },
    {
      "epoch": 0.04413607547499914,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009967313395268967,
      "loss": 2.2761,
      "step": 6496
    },
    {
      "epoch": 0.0441428698215932,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009967301112402203,
      "loss": 2.3743,
      "step": 6497
    },
    {
      "epoch": 0.044149664168187254,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009967288827235634,
      "loss": 2.1683,
      "step": 6498
    },
    {
      "epoch": 0.04415645851478131,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009967276539769269,
      "loss": 2.3685,
      "step": 6499
    },
    {
      "epoch": 0.044163252861375374,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000996726425000311,
      "loss": 2.2316,
      "step": 6500
    },
    {
      "epoch": 0.04417004720796943,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009967251957937165,
      "loss": 2.3915,
      "step": 6501
    },
    {
      "epoch": 0.04417684155456349,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996723966357144,
      "loss": 2.2207,
      "step": 6502
    },
    {
      "epoch": 0.04418363590115754,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000996722736690594,
      "loss": 2.353,
      "step": 6503
    },
    {
      "epoch": 0.0441904302477516,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009967215067940669,
      "loss": 2.4119,
      "step": 6504
    },
    {
      "epoch": 0.04419722459434566,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009967202766675636,
      "loss": 2.2512,
      "step": 6505
    },
    {
      "epoch": 0.04420401894093972,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009967190463110844,
      "loss": 2.2998,
      "step": 6506
    },
    {
      "epoch": 0.044210813287533776,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009967178157246302,
      "loss": 2.328,
      "step": 6507
    },
    {
      "epoch": 0.04421760763412783,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009967165849082012,
      "loss": 2.3934,
      "step": 6508
    },
    {
      "epoch": 0.04422440198072189,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009967153538617982,
      "loss": 2.2088,
      "step": 6509
    },
    {
      "epoch": 0.044231196327315946,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009967141225854215,
      "loss": 2.2807,
      "step": 6510
    },
    {
      "epoch": 0.04423799067391001,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000996712891079072,
      "loss": 2.2379,
      "step": 6511
    },
    {
      "epoch": 0.044244785020504065,
      "grad_norm": 0.1640625,
      "learning_rate": 0.00099671165934275,
      "loss": 2.3211,
      "step": 6512
    },
    {
      "epoch": 0.04425157936709812,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009967104273764563,
      "loss": 2.2841,
      "step": 6513
    },
    {
      "epoch": 0.04425837371369218,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009967091951801912,
      "loss": 2.4165,
      "step": 6514
    },
    {
      "epoch": 0.044265168060286235,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009967079627539557,
      "loss": 2.3828,
      "step": 6515
    },
    {
      "epoch": 0.0442719624068803,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.00099670673009775,
      "loss": 2.195,
      "step": 6516
    },
    {
      "epoch": 0.044278756753474355,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009967054972115747,
      "loss": 2.3803,
      "step": 6517
    },
    {
      "epoch": 0.04428555110006841,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009967042640954304,
      "loss": 2.3237,
      "step": 6518
    },
    {
      "epoch": 0.04429234544666247,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009967030307493178,
      "loss": 2.2809,
      "step": 6519
    },
    {
      "epoch": 0.044299139793256524,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009967017971732376,
      "loss": 2.2064,
      "step": 6520
    },
    {
      "epoch": 0.04430593413985058,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00099670056336719,
      "loss": 2.201,
      "step": 6521
    },
    {
      "epoch": 0.044312728486444644,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009966993293311756,
      "loss": 2.1988,
      "step": 6522
    },
    {
      "epoch": 0.0443195228330387,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009966980950651954,
      "loss": 2.3313,
      "step": 6523
    },
    {
      "epoch": 0.04432631717963276,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009966968605692494,
      "loss": 2.3086,
      "step": 6524
    },
    {
      "epoch": 0.044333111526226814,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009966956258433387,
      "loss": 2.2943,
      "step": 6525
    },
    {
      "epoch": 0.04433990587282087,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009966943908874632,
      "loss": 2.3033,
      "step": 6526
    },
    {
      "epoch": 0.04434670021941493,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009966931557016244,
      "loss": 2.3011,
      "step": 6527
    },
    {
      "epoch": 0.04435349456600899,
      "grad_norm": 0.140625,
      "learning_rate": 0.000996691920285822,
      "loss": 2.243,
      "step": 6528
    },
    {
      "epoch": 0.044360288912603046,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996690684640057,
      "loss": 2.1073,
      "step": 6529
    },
    {
      "epoch": 0.0443670832591971,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00099668944876433,
      "loss": 2.4004,
      "step": 6530
    },
    {
      "epoch": 0.04437387760579116,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009966882126586415,
      "loss": 2.3709,
      "step": 6531
    },
    {
      "epoch": 0.044380671952385216,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996686976322992,
      "loss": 2.3296,
      "step": 6532
    },
    {
      "epoch": 0.04438746629897928,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996685739757382,
      "loss": 2.2278,
      "step": 6533
    },
    {
      "epoch": 0.044394260645573336,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009966845029618122,
      "loss": 2.4117,
      "step": 6534
    },
    {
      "epoch": 0.04440105499216739,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009966832659362834,
      "loss": 2.2925,
      "step": 6535
    },
    {
      "epoch": 0.04440784933876145,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009966820286807958,
      "loss": 2.3245,
      "step": 6536
    },
    {
      "epoch": 0.044414643685355505,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009966807911953503,
      "loss": 2.3466,
      "step": 6537
    },
    {
      "epoch": 0.04442143803194957,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000996679553479947,
      "loss": 2.2348,
      "step": 6538
    },
    {
      "epoch": 0.044428232378543625,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996678315534587,
      "loss": 2.2868,
      "step": 6539
    },
    {
      "epoch": 0.04443502672513768,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009966770773592705,
      "loss": 2.2074,
      "step": 6540
    },
    {
      "epoch": 0.04444182107173174,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009966758389539983,
      "loss": 2.395,
      "step": 6541
    },
    {
      "epoch": 0.044448615418325794,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009966746003187707,
      "loss": 2.3151,
      "step": 6542
    },
    {
      "epoch": 0.04445540976491985,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009966733614535886,
      "loss": 2.4666,
      "step": 6543
    },
    {
      "epoch": 0.044462204111513914,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009966721223584525,
      "loss": 2.2185,
      "step": 6544
    },
    {
      "epoch": 0.04446899845810797,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009966708830333626,
      "loss": 2.2358,
      "step": 6545
    },
    {
      "epoch": 0.04447579280470203,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099666964347832,
      "loss": 2.4572,
      "step": 6546
    },
    {
      "epoch": 0.044482587151296084,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009966684036933251,
      "loss": 2.3072,
      "step": 6547
    },
    {
      "epoch": 0.04448938149789014,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009966671636783785,
      "loss": 2.3919,
      "step": 6548
    },
    {
      "epoch": 0.044496175844484204,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009966659234334804,
      "loss": 2.3879,
      "step": 6549
    },
    {
      "epoch": 0.04450297019107826,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996664682958632,
      "loss": 2.2987,
      "step": 6550
    },
    {
      "epoch": 0.04450976453767232,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009966634422538332,
      "loss": 2.4417,
      "step": 6551
    },
    {
      "epoch": 0.04451655888426637,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000996662201319085,
      "loss": 2.3636,
      "step": 6552
    },
    {
      "epoch": 0.04452335323086043,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996660960154388,
      "loss": 2.3166,
      "step": 6553
    },
    {
      "epoch": 0.044530147577454486,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009966597187597427,
      "loss": 2.2603,
      "step": 6554
    },
    {
      "epoch": 0.04453694192404855,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009966584771351496,
      "loss": 2.313,
      "step": 6555
    },
    {
      "epoch": 0.044543736270642606,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009966572352806092,
      "loss": 2.4532,
      "step": 6556
    },
    {
      "epoch": 0.04455053061723666,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009966559931961223,
      "loss": 2.1295,
      "step": 6557
    },
    {
      "epoch": 0.04455732496383072,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009966547508816895,
      "loss": 2.2705,
      "step": 6558
    },
    {
      "epoch": 0.044564119310424775,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996653508337311,
      "loss": 2.4093,
      "step": 6559
    },
    {
      "epoch": 0.04457091365701884,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009966522655629876,
      "loss": 2.2158,
      "step": 6560
    },
    {
      "epoch": 0.044577708003612895,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009966510225587202,
      "loss": 2.2115,
      "step": 6561
    },
    {
      "epoch": 0.04458450235020695,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009966497793245088,
      "loss": 2.3457,
      "step": 6562
    },
    {
      "epoch": 0.04459129669680101,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009966485358603545,
      "loss": 2.3258,
      "step": 6563
    },
    {
      "epoch": 0.044598091043395065,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009966472921662576,
      "loss": 2.361,
      "step": 6564
    },
    {
      "epoch": 0.04460488538998913,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009966460482422183,
      "loss": 2.2495,
      "step": 6565
    },
    {
      "epoch": 0.044611679736583185,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000996644804088238,
      "loss": 2.3636,
      "step": 6566
    },
    {
      "epoch": 0.04461847408317724,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009966435597043167,
      "loss": 2.314,
      "step": 6567
    },
    {
      "epoch": 0.0446252684297713,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000996642315090455,
      "loss": 2.2466,
      "step": 6568
    },
    {
      "epoch": 0.044632062776365354,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009966410702466537,
      "loss": 2.1907,
      "step": 6569
    },
    {
      "epoch": 0.04463885712295941,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009966398251729134,
      "loss": 2.1812,
      "step": 6570
    },
    {
      "epoch": 0.044645651469553474,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009966385798692344,
      "loss": 2.2347,
      "step": 6571
    },
    {
      "epoch": 0.04465244581614753,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009966373343356174,
      "loss": 2.2947,
      "step": 6572
    },
    {
      "epoch": 0.04465924016274159,
      "grad_norm": 0.140625,
      "learning_rate": 0.000996636088572063,
      "loss": 2.3056,
      "step": 6573
    },
    {
      "epoch": 0.04466603450933564,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009966348425785722,
      "loss": 2.2055,
      "step": 6574
    },
    {
      "epoch": 0.0446728288559297,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009966335963551448,
      "loss": 2.2526,
      "step": 6575
    },
    {
      "epoch": 0.04467962320252376,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009966323499017817,
      "loss": 2.258,
      "step": 6576
    },
    {
      "epoch": 0.04468641754911782,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009966311032184837,
      "loss": 2.417,
      "step": 6577
    },
    {
      "epoch": 0.044693211895711876,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009966298563052508,
      "loss": 2.207,
      "step": 6578
    },
    {
      "epoch": 0.04470000624230593,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009966286091620843,
      "loss": 2.4238,
      "step": 6579
    },
    {
      "epoch": 0.04470680058889999,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009966273617889846,
      "loss": 2.1836,
      "step": 6580
    },
    {
      "epoch": 0.044713594935494046,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009966261141859517,
      "loss": 2.3172,
      "step": 6581
    },
    {
      "epoch": 0.04472038928208811,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000996624866352987,
      "loss": 2.2144,
      "step": 6582
    },
    {
      "epoch": 0.044727183628682166,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009966236182900904,
      "loss": 2.2901,
      "step": 6583
    },
    {
      "epoch": 0.04473397797527622,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009966223699972628,
      "loss": 2.3228,
      "step": 6584
    },
    {
      "epoch": 0.04474077232187028,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000996621121474505,
      "loss": 2.2203,
      "step": 6585
    },
    {
      "epoch": 0.044747566668464335,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996619872721817,
      "loss": 2.3777,
      "step": 6586
    },
    {
      "epoch": 0.0447543610150584,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009966186237392,
      "loss": 2.3666,
      "step": 6587
    },
    {
      "epoch": 0.044761155361652455,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009966173745266538,
      "loss": 2.2985,
      "step": 6588
    },
    {
      "epoch": 0.04476794970824651,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009966161250841799,
      "loss": 2.3553,
      "step": 6589
    },
    {
      "epoch": 0.04477474405484057,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996614875411778,
      "loss": 2.4017,
      "step": 6590
    },
    {
      "epoch": 0.044781538401434624,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009966136255094495,
      "loss": 2.3183,
      "step": 6591
    },
    {
      "epoch": 0.04478833274802868,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009966123753771944,
      "loss": 2.3274,
      "step": 6592
    },
    {
      "epoch": 0.044795127094622744,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009966111250150136,
      "loss": 2.27,
      "step": 6593
    },
    {
      "epoch": 0.0448019214412168,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009966098744229074,
      "loss": 2.3831,
      "step": 6594
    },
    {
      "epoch": 0.04480871578781086,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009966086236008766,
      "loss": 2.413,
      "step": 6595
    },
    {
      "epoch": 0.044815510134404914,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009966073725489217,
      "loss": 2.3621,
      "step": 6596
    },
    {
      "epoch": 0.04482230448099897,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009966061212670432,
      "loss": 2.4342,
      "step": 6597
    },
    {
      "epoch": 0.044829098827593034,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996604869755242,
      "loss": 2.2716,
      "step": 6598
    },
    {
      "epoch": 0.04483589317418709,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009966036180135182,
      "loss": 2.2799,
      "step": 6599
    },
    {
      "epoch": 0.044842687520781147,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009966023660418727,
      "loss": 2.5252,
      "step": 6600
    },
    {
      "epoch": 0.0448494818673752,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009966011138403058,
      "loss": 2.2553,
      "step": 6601
    },
    {
      "epoch": 0.04485627621396926,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009965998614088185,
      "loss": 2.2826,
      "step": 6602
    },
    {
      "epoch": 0.044863070560563316,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000996598608747411,
      "loss": 2.2077,
      "step": 6603
    },
    {
      "epoch": 0.04486986490715738,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009965973558560843,
      "loss": 2.2423,
      "step": 6604
    },
    {
      "epoch": 0.044876659253751436,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009965961027348384,
      "loss": 2.2796,
      "step": 6605
    },
    {
      "epoch": 0.04488345360034549,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009965948493836745,
      "loss": 2.507,
      "step": 6606
    },
    {
      "epoch": 0.04489024794693955,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009965935958025926,
      "loss": 2.3122,
      "step": 6607
    },
    {
      "epoch": 0.044897042293533605,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009965923419915938,
      "loss": 2.4097,
      "step": 6608
    },
    {
      "epoch": 0.04490383664012767,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009965910879506782,
      "loss": 2.3494,
      "step": 6609
    },
    {
      "epoch": 0.044910630986721725,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996589833679847,
      "loss": 2.2387,
      "step": 6610
    },
    {
      "epoch": 0.04491742533331578,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009965885791791002,
      "loss": 2.3409,
      "step": 6611
    },
    {
      "epoch": 0.04492421967990984,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009965873244484383,
      "loss": 2.226,
      "step": 6612
    },
    {
      "epoch": 0.044931014026503895,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009965860694878623,
      "loss": 2.3749,
      "step": 6613
    },
    {
      "epoch": 0.04493780837309795,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009965848142973728,
      "loss": 2.2913,
      "step": 6614
    },
    {
      "epoch": 0.044944602719692014,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009965835588769702,
      "loss": 2.3913,
      "step": 6615
    },
    {
      "epoch": 0.04495139706628607,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009965823032266551,
      "loss": 2.205,
      "step": 6616
    },
    {
      "epoch": 0.04495819141288013,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996581047346428,
      "loss": 2.3059,
      "step": 6617
    },
    {
      "epoch": 0.044964985759474184,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009965797912362896,
      "loss": 2.2732,
      "step": 6618
    },
    {
      "epoch": 0.04497178010606824,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009965785348962405,
      "loss": 2.1935,
      "step": 6619
    },
    {
      "epoch": 0.044978574452662304,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009965772783262813,
      "loss": 2.3062,
      "step": 6620
    },
    {
      "epoch": 0.04498536879925636,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009965760215264123,
      "loss": 2.1758,
      "step": 6621
    },
    {
      "epoch": 0.04499216314585042,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009965747644966345,
      "loss": 2.2867,
      "step": 6622
    },
    {
      "epoch": 0.04499895749244447,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009965735072369481,
      "loss": 2.3042,
      "step": 6623
    },
    {
      "epoch": 0.04500575183903853,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000996572249747354,
      "loss": 2.2574,
      "step": 6624
    },
    {
      "epoch": 0.04501254618563259,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009965709920278525,
      "loss": 2.3009,
      "step": 6625
    },
    {
      "epoch": 0.04501934053222665,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009965697340784446,
      "loss": 2.3312,
      "step": 6626
    },
    {
      "epoch": 0.045026134878820706,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009965684758991302,
      "loss": 2.3173,
      "step": 6627
    },
    {
      "epoch": 0.04503292922541476,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009965672174899105,
      "loss": 2.2382,
      "step": 6628
    },
    {
      "epoch": 0.04503972357200882,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009965659588507861,
      "loss": 2.2639,
      "step": 6629
    },
    {
      "epoch": 0.045046517918602876,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000996564699981757,
      "loss": 2.4204,
      "step": 6630
    },
    {
      "epoch": 0.04505331226519694,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009965634408828244,
      "loss": 2.3507,
      "step": 6631
    },
    {
      "epoch": 0.045060106611790995,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009965621815539883,
      "loss": 2.3875,
      "step": 6632
    },
    {
      "epoch": 0.04506690095838505,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009965609219952497,
      "loss": 2.3088,
      "step": 6633
    },
    {
      "epoch": 0.04507369530497911,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009965596622066094,
      "loss": 2.524,
      "step": 6634
    },
    {
      "epoch": 0.045080489651573165,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009965584021880674,
      "loss": 2.3543,
      "step": 6635
    },
    {
      "epoch": 0.04508728399816723,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009965571419396245,
      "loss": 2.1679,
      "step": 6636
    },
    {
      "epoch": 0.045094078344761285,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009965558814612814,
      "loss": 2.3086,
      "step": 6637
    },
    {
      "epoch": 0.04510087269135534,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009965546207530388,
      "loss": 2.199,
      "step": 6638
    },
    {
      "epoch": 0.0451076670379494,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996553359814897,
      "loss": 2.2575,
      "step": 6639
    },
    {
      "epoch": 0.045114461384543454,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009965520986468567,
      "loss": 2.3881,
      "step": 6640
    },
    {
      "epoch": 0.04512125573113751,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009965508372489183,
      "loss": 2.348,
      "step": 6641
    },
    {
      "epoch": 0.045128050077731574,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009965495756210827,
      "loss": 2.1664,
      "step": 6642
    },
    {
      "epoch": 0.04513484442432563,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009965483137633502,
      "loss": 2.3503,
      "step": 6643
    },
    {
      "epoch": 0.04514163877091969,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009965470516757216,
      "loss": 2.3922,
      "step": 6644
    },
    {
      "epoch": 0.045148433117513743,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009965457893581975,
      "loss": 2.2931,
      "step": 6645
    },
    {
      "epoch": 0.0451552274641078,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009965445268107784,
      "loss": 2.3484,
      "step": 6646
    },
    {
      "epoch": 0.04516202181070186,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996543264033465,
      "loss": 2.3076,
      "step": 6647
    },
    {
      "epoch": 0.04516881615729592,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009965420010262574,
      "loss": 2.2148,
      "step": 6648
    },
    {
      "epoch": 0.045175610503889976,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009965407377891566,
      "loss": 2.3392,
      "step": 6649
    },
    {
      "epoch": 0.04518240485048403,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009965394743221632,
      "loss": 2.286,
      "step": 6650
    },
    {
      "epoch": 0.04518919919707809,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009965382106252778,
      "loss": 2.438,
      "step": 6651
    },
    {
      "epoch": 0.045195993543672146,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996536946698501,
      "loss": 2.2468,
      "step": 6652
    },
    {
      "epoch": 0.04520278789026621,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000996535682541833,
      "loss": 2.263,
      "step": 6653
    },
    {
      "epoch": 0.045209582236860266,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009965344181552748,
      "loss": 2.3422,
      "step": 6654
    },
    {
      "epoch": 0.04521637658345432,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009965331535388268,
      "loss": 2.2105,
      "step": 6655
    },
    {
      "epoch": 0.04522317093004838,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099653188869249,
      "loss": 2.3579,
      "step": 6656
    },
    {
      "epoch": 0.045229965276642435,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009965306236162641,
      "loss": 2.3072,
      "step": 6657
    },
    {
      "epoch": 0.0452367596232365,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009965293583101508,
      "loss": 2.1719,
      "step": 6658
    },
    {
      "epoch": 0.045243553969830555,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009965280927741495,
      "loss": 2.3327,
      "step": 6659
    },
    {
      "epoch": 0.04525034831642461,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009965268270082618,
      "loss": 2.4862,
      "step": 6660
    },
    {
      "epoch": 0.04525714266301867,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009965255610124876,
      "loss": 2.4348,
      "step": 6661
    },
    {
      "epoch": 0.045263937009612724,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000996524294786828,
      "loss": 2.252,
      "step": 6662
    },
    {
      "epoch": 0.04527073135620678,
      "grad_norm": 0.15625,
      "learning_rate": 0.000996523028331283,
      "loss": 2.1894,
      "step": 6663
    },
    {
      "epoch": 0.045277525702800844,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000996521761645854,
      "loss": 2.2881,
      "step": 6664
    },
    {
      "epoch": 0.0452843200493949,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009965204947305407,
      "loss": 2.3832,
      "step": 6665
    },
    {
      "epoch": 0.04529111439598896,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009965192275853442,
      "loss": 2.2455,
      "step": 6666
    },
    {
      "epoch": 0.045297908742583014,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009965179602102652,
      "loss": 2.4207,
      "step": 6667
    },
    {
      "epoch": 0.04530470308917707,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996516692605304,
      "loss": 2.2346,
      "step": 6668
    },
    {
      "epoch": 0.045311497435771134,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009965154247704611,
      "loss": 2.3237,
      "step": 6669
    },
    {
      "epoch": 0.04531829178236519,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009965141567057374,
      "loss": 2.2653,
      "step": 6670
    },
    {
      "epoch": 0.04532508612895925,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009965128884111333,
      "loss": 2.3701,
      "step": 6671
    },
    {
      "epoch": 0.0453318804755533,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009965116198866494,
      "loss": 2.2798,
      "step": 6672
    },
    {
      "epoch": 0.04533867482214736,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009965103511322864,
      "loss": 2.4273,
      "step": 6673
    },
    {
      "epoch": 0.045345469168741416,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009965090821480447,
      "loss": 2.3509,
      "step": 6674
    },
    {
      "epoch": 0.04535226351533548,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009965078129339248,
      "loss": 2.1116,
      "step": 6675
    },
    {
      "epoch": 0.045359057861929536,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009965065434899278,
      "loss": 2.2927,
      "step": 6676
    },
    {
      "epoch": 0.04536585220852359,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009965052738160538,
      "loss": 2.3898,
      "step": 6677
    },
    {
      "epoch": 0.04537264655511765,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009965040039123037,
      "loss": 2.3475,
      "step": 6678
    },
    {
      "epoch": 0.045379440901711705,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009965027337786778,
      "loss": 2.4202,
      "step": 6679
    },
    {
      "epoch": 0.04538623524830577,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996501463415177,
      "loss": 2.3406,
      "step": 6680
    },
    {
      "epoch": 0.045393029594899825,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009965001928218015,
      "loss": 2.1663,
      "step": 6681
    },
    {
      "epoch": 0.04539982394149388,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009964989219985521,
      "loss": 2.3531,
      "step": 6682
    },
    {
      "epoch": 0.04540661828808794,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009964976509454296,
      "loss": 2.4407,
      "step": 6683
    },
    {
      "epoch": 0.045413412634681995,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009964963796624342,
      "loss": 2.3473,
      "step": 6684
    },
    {
      "epoch": 0.04542020698127606,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009964951081495668,
      "loss": 2.2187,
      "step": 6685
    },
    {
      "epoch": 0.045427001327870115,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009964938364068277,
      "loss": 2.3824,
      "step": 6686
    },
    {
      "epoch": 0.04543379567446417,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009964925644342178,
      "loss": 2.3156,
      "step": 6687
    },
    {
      "epoch": 0.04544059002105823,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009964912922317375,
      "loss": 2.3131,
      "step": 6688
    },
    {
      "epoch": 0.045447384367652284,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009964900197993874,
      "loss": 2.4117,
      "step": 6689
    },
    {
      "epoch": 0.04545417871424634,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000996488747137168,
      "loss": 2.3476,
      "step": 6690
    },
    {
      "epoch": 0.045460973060840404,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00099648747424508,
      "loss": 2.297,
      "step": 6691
    },
    {
      "epoch": 0.04546776740743446,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009964862011231242,
      "loss": 2.1907,
      "step": 6692
    },
    {
      "epoch": 0.04547456175402852,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000996484927771301,
      "loss": 2.3766,
      "step": 6693
    },
    {
      "epoch": 0.04548135610062257,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009964836541896107,
      "loss": 2.4088,
      "step": 6694
    },
    {
      "epoch": 0.04548815044721663,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009964823803780543,
      "loss": 2.3382,
      "step": 6695
    },
    {
      "epoch": 0.04549494479381069,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009964811063366322,
      "loss": 2.3173,
      "step": 6696
    },
    {
      "epoch": 0.04550173914040475,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009964798320653452,
      "loss": 2.2369,
      "step": 6697
    },
    {
      "epoch": 0.045508533486998806,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009964785575641937,
      "loss": 2.2821,
      "step": 6698
    },
    {
      "epoch": 0.04551532783359286,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000996477282833178,
      "loss": 2.3965,
      "step": 6699
    },
    {
      "epoch": 0.04552212218018692,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009964760078722994,
      "loss": 2.3624,
      "step": 6700
    },
    {
      "epoch": 0.045528916526780976,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996474732681558,
      "loss": 2.175,
      "step": 6701
    },
    {
      "epoch": 0.04553571087337504,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009964734572609545,
      "loss": 2.3798,
      "step": 6702
    },
    {
      "epoch": 0.045542505219969096,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009964721816104894,
      "loss": 2.3114,
      "step": 6703
    },
    {
      "epoch": 0.04554929956656315,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009964709057301633,
      "loss": 2.2771,
      "step": 6704
    },
    {
      "epoch": 0.04555609391315721,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009964696296199769,
      "loss": 2.3348,
      "step": 6705
    },
    {
      "epoch": 0.045562888259751265,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009964683532799308,
      "loss": 2.1612,
      "step": 6706
    },
    {
      "epoch": 0.04556968260634533,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009964670767100257,
      "loss": 2.3849,
      "step": 6707
    },
    {
      "epoch": 0.045576476952939385,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009964657999102617,
      "loss": 2.2502,
      "step": 6708
    },
    {
      "epoch": 0.04558327129953344,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099646452288064,
      "loss": 2.3432,
      "step": 6709
    },
    {
      "epoch": 0.0455900656461275,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996463245621161,
      "loss": 2.2589,
      "step": 6710
    },
    {
      "epoch": 0.045596859992721554,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009964619681318248,
      "loss": 2.1063,
      "step": 6711
    },
    {
      "epoch": 0.04560365433931561,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009964606904126327,
      "loss": 2.4326,
      "step": 6712
    },
    {
      "epoch": 0.045610448685909674,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009964594124635849,
      "loss": 2.2789,
      "step": 6713
    },
    {
      "epoch": 0.04561724303250373,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000996458134284682,
      "loss": 2.1731,
      "step": 6714
    },
    {
      "epoch": 0.04562403737909779,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996456855875925,
      "loss": 2.3761,
      "step": 6715
    },
    {
      "epoch": 0.045630831725691844,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996455577237314,
      "loss": 2.421,
      "step": 6716
    },
    {
      "epoch": 0.0456376260722859,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009964542983688495,
      "loss": 2.2459,
      "step": 6717
    },
    {
      "epoch": 0.045644420418879963,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009964530192705329,
      "loss": 2.2631,
      "step": 6718
    },
    {
      "epoch": 0.04565121476547402,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009964517399423639,
      "loss": 2.3407,
      "step": 6719
    },
    {
      "epoch": 0.045658009112068076,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009964504603843433,
      "loss": 2.346,
      "step": 6720
    },
    {
      "epoch": 0.04566480345866213,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996449180596472,
      "loss": 2.1804,
      "step": 6721
    },
    {
      "epoch": 0.04567159780525619,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009964479005787504,
      "loss": 2.4084,
      "step": 6722
    },
    {
      "epoch": 0.045678392151850246,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000996446620331179,
      "loss": 2.1953,
      "step": 6723
    },
    {
      "epoch": 0.04568518649844431,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009964453398537587,
      "loss": 2.3303,
      "step": 6724
    },
    {
      "epoch": 0.045691980845038366,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009964440591464898,
      "loss": 2.0731,
      "step": 6725
    },
    {
      "epoch": 0.04569877519163242,
      "grad_norm": 0.1796875,
      "learning_rate": 0.000996442778209373,
      "loss": 2.2569,
      "step": 6726
    },
    {
      "epoch": 0.04570556953822648,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000996441497042409,
      "loss": 2.2618,
      "step": 6727
    },
    {
      "epoch": 0.045712363884820535,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000996440215645598,
      "loss": 2.3389,
      "step": 6728
    },
    {
      "epoch": 0.0457191582314146,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009964389340189413,
      "loss": 2.2248,
      "step": 6729
    },
    {
      "epoch": 0.045725952578008655,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009964376521624388,
      "loss": 2.2279,
      "step": 6730
    },
    {
      "epoch": 0.04573274692460271,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009964363700760911,
      "loss": 2.4455,
      "step": 6731
    },
    {
      "epoch": 0.04573954127119677,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009964350877598996,
      "loss": 2.3447,
      "step": 6732
    },
    {
      "epoch": 0.045746335617790825,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000996433805213864,
      "loss": 2.2906,
      "step": 6733
    },
    {
      "epoch": 0.04575312996438488,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009964325224379852,
      "loss": 2.2332,
      "step": 6734
    },
    {
      "epoch": 0.045759924310978944,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000996431239432264,
      "loss": 2.4108,
      "step": 6735
    },
    {
      "epoch": 0.045766718657573,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009964299561967007,
      "loss": 2.4107,
      "step": 6736
    },
    {
      "epoch": 0.04577351300416706,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009964286727312962,
      "loss": 2.301,
      "step": 6737
    },
    {
      "epoch": 0.045780307350761114,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009964273890360508,
      "loss": 2.4145,
      "step": 6738
    },
    {
      "epoch": 0.04578710169735517,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009964261051109652,
      "loss": 2.2155,
      "step": 6739
    },
    {
      "epoch": 0.045793896043949234,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00099642482095604,
      "loss": 2.2569,
      "step": 6740
    },
    {
      "epoch": 0.04580069039054329,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009964235365712759,
      "loss": 2.299,
      "step": 6741
    },
    {
      "epoch": 0.04580748473713735,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009964222519566733,
      "loss": 2.2809,
      "step": 6742
    },
    {
      "epoch": 0.0458142790837314,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009964209671122328,
      "loss": 2.3878,
      "step": 6743
    },
    {
      "epoch": 0.04582107343032546,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000996419682037955,
      "loss": 2.2262,
      "step": 6744
    },
    {
      "epoch": 0.04582786777691952,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009964183967338409,
      "loss": 2.2411,
      "step": 6745
    },
    {
      "epoch": 0.04583466212351358,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009964171111998908,
      "loss": 2.2786,
      "step": 6746
    },
    {
      "epoch": 0.045841456470107636,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009964158254361052,
      "loss": 2.3232,
      "step": 6747
    },
    {
      "epoch": 0.04584825081670169,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009964145394424845,
      "loss": 2.2619,
      "step": 6748
    },
    {
      "epoch": 0.04585504516329575,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009964132532190298,
      "loss": 2.2918,
      "step": 6749
    },
    {
      "epoch": 0.045861839509889805,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009964119667657415,
      "loss": 2.2757,
      "step": 6750
    },
    {
      "epoch": 0.04586863385648387,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009964106800826199,
      "loss": 2.385,
      "step": 6751
    },
    {
      "epoch": 0.045875428203077925,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009964093931696658,
      "loss": 2.2927,
      "step": 6752
    },
    {
      "epoch": 0.04588222254967198,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.00099640810602688,
      "loss": 2.5748,
      "step": 6753
    },
    {
      "epoch": 0.04588901689626604,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009964068186542631,
      "loss": 2.4054,
      "step": 6754
    },
    {
      "epoch": 0.045895811242860095,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009964055310518155,
      "loss": 2.5468,
      "step": 6755
    },
    {
      "epoch": 0.04590260558945416,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009964042432195376,
      "loss": 2.3406,
      "step": 6756
    },
    {
      "epoch": 0.045909399936048215,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009964029551574303,
      "loss": 2.2974,
      "step": 6757
    },
    {
      "epoch": 0.04591619428264227,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009964016668654943,
      "loss": 2.468,
      "step": 6758
    },
    {
      "epoch": 0.04592298862923633,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099640037834373,
      "loss": 2.4111,
      "step": 6759
    },
    {
      "epoch": 0.045929782975830384,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009963990895921378,
      "loss": 2.184,
      "step": 6760
    },
    {
      "epoch": 0.04593657732242444,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009963978006107186,
      "loss": 2.3617,
      "step": 6761
    },
    {
      "epoch": 0.045943371669018504,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000996396511399473,
      "loss": 2.3859,
      "step": 6762
    },
    {
      "epoch": 0.04595016601561256,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009963952219584013,
      "loss": 2.345,
      "step": 6763
    },
    {
      "epoch": 0.04595696036220662,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009963939322875045,
      "loss": 2.166,
      "step": 6764
    },
    {
      "epoch": 0.04596375470880067,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009963926423867831,
      "loss": 2.3624,
      "step": 6765
    },
    {
      "epoch": 0.04597054905539473,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009963913522562373,
      "loss": 2.4204,
      "step": 6766
    },
    {
      "epoch": 0.04597734340198879,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009963900618958682,
      "loss": 2.3423,
      "step": 6767
    },
    {
      "epoch": 0.04598413774858285,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000996388771305676,
      "loss": 2.356,
      "step": 6768
    },
    {
      "epoch": 0.045990932095176906,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009963874804856617,
      "loss": 2.0915,
      "step": 6769
    },
    {
      "epoch": 0.04599772644177096,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009963861894358257,
      "loss": 2.3364,
      "step": 6770
    },
    {
      "epoch": 0.04600452078836502,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009963848981561685,
      "loss": 2.2724,
      "step": 6771
    },
    {
      "epoch": 0.046011315134959076,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009963836066466908,
      "loss": 2.2448,
      "step": 6772
    },
    {
      "epoch": 0.04601810948155314,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996382314907393,
      "loss": 2.3918,
      "step": 6773
    },
    {
      "epoch": 0.046024903828147196,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996381022938276,
      "loss": 2.3541,
      "step": 6774
    },
    {
      "epoch": 0.04603169817474125,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009963797307393403,
      "loss": 2.296,
      "step": 6775
    },
    {
      "epoch": 0.04603849252133531,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009963784383105864,
      "loss": 2.4204,
      "step": 6776
    },
    {
      "epoch": 0.046045286867929365,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009963771456520153,
      "loss": 2.2962,
      "step": 6777
    },
    {
      "epoch": 0.04605208121452343,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996375852763627,
      "loss": 2.361,
      "step": 6778
    },
    {
      "epoch": 0.046058875561117485,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009963745596454224,
      "loss": 2.3202,
      "step": 6779
    },
    {
      "epoch": 0.04606566990771154,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000996373266297402,
      "loss": 2.3283,
      "step": 6780
    },
    {
      "epoch": 0.0460724642543056,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009963719727195666,
      "loss": 2.2965,
      "step": 6781
    },
    {
      "epoch": 0.046079258600899654,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009963706789119166,
      "loss": 2.4557,
      "step": 6782
    },
    {
      "epoch": 0.04608605294749371,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009963693848744526,
      "loss": 2.3728,
      "step": 6783
    },
    {
      "epoch": 0.046092847294087774,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009963680906071753,
      "loss": 2.2654,
      "step": 6784
    },
    {
      "epoch": 0.04609964164068183,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009963667961100852,
      "loss": 2.3403,
      "step": 6785
    },
    {
      "epoch": 0.04610643598727589,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009963655013831832,
      "loss": 2.2344,
      "step": 6786
    },
    {
      "epoch": 0.046113230333869944,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009963642064264695,
      "loss": 2.2985,
      "step": 6787
    },
    {
      "epoch": 0.046120024680464,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996362911239945,
      "loss": 2.4059,
      "step": 6788
    },
    {
      "epoch": 0.046126819027058064,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009963616158236098,
      "loss": 2.326,
      "step": 6789
    },
    {
      "epoch": 0.04613361337365212,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009963603201774654,
      "loss": 2.2618,
      "step": 6790
    },
    {
      "epoch": 0.04614040772024618,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009963590243015116,
      "loss": 2.4553,
      "step": 6791
    },
    {
      "epoch": 0.04614720206684023,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009963577281957492,
      "loss": 2.1683,
      "step": 6792
    },
    {
      "epoch": 0.04615399641343429,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009963564318601788,
      "loss": 2.3384,
      "step": 6793
    },
    {
      "epoch": 0.046160790760028346,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009963551352948012,
      "loss": 2.5673,
      "step": 6794
    },
    {
      "epoch": 0.04616758510662241,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000996353838499617,
      "loss": 2.3679,
      "step": 6795
    },
    {
      "epoch": 0.046174379453216466,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009963525414746265,
      "loss": 2.3008,
      "step": 6796
    },
    {
      "epoch": 0.04618117379981052,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009963512442198303,
      "loss": 2.2573,
      "step": 6797
    },
    {
      "epoch": 0.04618796814640458,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009963499467352294,
      "loss": 2.3472,
      "step": 6798
    },
    {
      "epoch": 0.046194762492998635,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996348649020824,
      "loss": 2.2864,
      "step": 6799
    },
    {
      "epoch": 0.0462015568395927,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000996347351076615,
      "loss": 2.3026,
      "step": 6800
    },
    {
      "epoch": 0.046208351186186755,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009963460529026028,
      "loss": 2.2981,
      "step": 6801
    },
    {
      "epoch": 0.04621514553278081,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996344754498788,
      "loss": 2.3325,
      "step": 6802
    },
    {
      "epoch": 0.04622193987937487,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009963434558651714,
      "loss": 2.3265,
      "step": 6803
    },
    {
      "epoch": 0.046228734225968925,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009963421570017535,
      "loss": 2.2577,
      "step": 6804
    },
    {
      "epoch": 0.04623552857256299,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009963408579085349,
      "loss": 2.1733,
      "step": 6805
    },
    {
      "epoch": 0.046242322919157045,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000996339558585516,
      "loss": 2.3676,
      "step": 6806
    },
    {
      "epoch": 0.0462491172657511,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009963382590326977,
      "loss": 2.2877,
      "step": 6807
    },
    {
      "epoch": 0.04625591161234516,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009963369592500804,
      "loss": 2.1942,
      "step": 6808
    },
    {
      "epoch": 0.046262705958939214,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009963356592376648,
      "loss": 2.3634,
      "step": 6809
    },
    {
      "epoch": 0.04626950030553327,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009963343589954514,
      "loss": 2.3091,
      "step": 6810
    },
    {
      "epoch": 0.046276294652127334,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000996333058523441,
      "loss": 2.3595,
      "step": 6811
    },
    {
      "epoch": 0.04628308899872139,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.000996331757821634,
      "loss": 2.1203,
      "step": 6812
    },
    {
      "epoch": 0.04628988334531545,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009963304568900311,
      "loss": 2.351,
      "step": 6813
    },
    {
      "epoch": 0.0462966776919095,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000996329155728633,
      "loss": 2.4133,
      "step": 6814
    },
    {
      "epoch": 0.04630347203850356,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009963278543374402,
      "loss": 2.2933,
      "step": 6815
    },
    {
      "epoch": 0.04631026638509762,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009963265527164532,
      "loss": 2.2798,
      "step": 6816
    },
    {
      "epoch": 0.04631706073169168,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009963252508656726,
      "loss": 2.2591,
      "step": 6817
    },
    {
      "epoch": 0.046323855078285736,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009963239487850992,
      "loss": 2.2488,
      "step": 6818
    },
    {
      "epoch": 0.04633064942487979,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009963226464747336,
      "loss": 2.2609,
      "step": 6819
    },
    {
      "epoch": 0.04633744377147385,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996321343934576,
      "loss": 2.3849,
      "step": 6820
    },
    {
      "epoch": 0.046344238118067906,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009963200411646277,
      "loss": 2.3666,
      "step": 6821
    },
    {
      "epoch": 0.04635103246466197,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009963187381648885,
      "loss": 2.4486,
      "step": 6822
    },
    {
      "epoch": 0.046357826811256025,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009963174349353597,
      "loss": 2.31,
      "step": 6823
    },
    {
      "epoch": 0.04636462115785008,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009963161314760416,
      "loss": 2.2834,
      "step": 6824
    },
    {
      "epoch": 0.04637141550444414,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009963148277869348,
      "loss": 2.3684,
      "step": 6825
    },
    {
      "epoch": 0.046378209851038195,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009963135238680398,
      "loss": 2.4262,
      "step": 6826
    },
    {
      "epoch": 0.04638500419763226,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009963122197193576,
      "loss": 2.3786,
      "step": 6827
    },
    {
      "epoch": 0.046391798544226315,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009963109153408882,
      "loss": 2.3008,
      "step": 6828
    },
    {
      "epoch": 0.04639859289082037,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009963096107326328,
      "loss": 2.153,
      "step": 6829
    },
    {
      "epoch": 0.04640538723741443,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009963083058945916,
      "loss": 2.2361,
      "step": 6830
    },
    {
      "epoch": 0.046412181584008484,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009963070008267654,
      "loss": 2.1025,
      "step": 6831
    },
    {
      "epoch": 0.04641897593060254,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009963056955291547,
      "loss": 2.3032,
      "step": 6832
    },
    {
      "epoch": 0.046425770277196604,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009963043900017604,
      "loss": 2.1428,
      "step": 6833
    },
    {
      "epoch": 0.04643256462379066,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009963030842445827,
      "loss": 2.3694,
      "step": 6834
    },
    {
      "epoch": 0.04643935897038472,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009963017782576222,
      "loss": 2.2607,
      "step": 6835
    },
    {
      "epoch": 0.046446153316978774,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009963004720408798,
      "loss": 2.3752,
      "step": 6836
    },
    {
      "epoch": 0.04645294766357283,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996299165594356,
      "loss": 2.157,
      "step": 6837
    },
    {
      "epoch": 0.04645974201016689,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009962978589180512,
      "loss": 2.3476,
      "step": 6838
    },
    {
      "epoch": 0.04646653635676095,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009962965520119663,
      "loss": 2.2758,
      "step": 6839
    },
    {
      "epoch": 0.046473330703355006,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009962952448761017,
      "loss": 2.3787,
      "step": 6840
    },
    {
      "epoch": 0.04648012504994906,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009962939375104582,
      "loss": 2.2238,
      "step": 6841
    },
    {
      "epoch": 0.04648691939654312,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009962926299150363,
      "loss": 2.3472,
      "step": 6842
    },
    {
      "epoch": 0.046493713743137176,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009962913220898366,
      "loss": 2.4259,
      "step": 6843
    },
    {
      "epoch": 0.04650050808973124,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009962900140348596,
      "loss": 2.4128,
      "step": 6844
    },
    {
      "epoch": 0.046507302436325296,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009962887057501063,
      "loss": 2.4335,
      "step": 6845
    },
    {
      "epoch": 0.04651409678291935,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009962873972355767,
      "loss": 2.3218,
      "step": 6846
    },
    {
      "epoch": 0.04652089112951341,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000996286088491272,
      "loss": 2.3068,
      "step": 6847
    },
    {
      "epoch": 0.046527685476107465,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009962847795171922,
      "loss": 2.2543,
      "step": 6848
    },
    {
      "epoch": 0.04653447982270153,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009962834703133383,
      "loss": 2.2978,
      "step": 6849
    },
    {
      "epoch": 0.046541274169295585,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000996282160879711,
      "loss": 2.2445,
      "step": 6850
    },
    {
      "epoch": 0.04654806851588964,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009962808512163108,
      "loss": 2.2446,
      "step": 6851
    },
    {
      "epoch": 0.0465548628624837,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000996279541323138,
      "loss": 2.1134,
      "step": 6852
    },
    {
      "epoch": 0.046561657209077754,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009962782312001937,
      "loss": 2.1635,
      "step": 6853
    },
    {
      "epoch": 0.04656845155567181,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009962769208474783,
      "loss": 2.351,
      "step": 6854
    },
    {
      "epoch": 0.046575245902265874,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009962756102649922,
      "loss": 2.4022,
      "step": 6855
    },
    {
      "epoch": 0.04658204024885993,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009962742994527364,
      "loss": 2.3145,
      "step": 6856
    },
    {
      "epoch": 0.04658883459545399,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996272988410711,
      "loss": 2.2829,
      "step": 6857
    },
    {
      "epoch": 0.046595628942048044,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009962716771389172,
      "loss": 2.153,
      "step": 6858
    },
    {
      "epoch": 0.0466024232886421,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009962703656373549,
      "loss": 2.2917,
      "step": 6859
    },
    {
      "epoch": 0.046609217635236164,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009962690539060254,
      "loss": 2.1324,
      "step": 6860
    },
    {
      "epoch": 0.04661601198183022,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000996267741944929,
      "loss": 2.1765,
      "step": 6861
    },
    {
      "epoch": 0.04662280632842428,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009962664297540663,
      "loss": 2.4411,
      "step": 6862
    },
    {
      "epoch": 0.04662960067501833,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009962651173334378,
      "loss": 2.2936,
      "step": 6863
    },
    {
      "epoch": 0.04663639502161239,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009962638046830444,
      "loss": 2.2971,
      "step": 6864
    },
    {
      "epoch": 0.04664318936820645,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009962624918028865,
      "loss": 2.2678,
      "step": 6865
    },
    {
      "epoch": 0.04664998371480051,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009962611786929647,
      "loss": 2.2466,
      "step": 6866
    },
    {
      "epoch": 0.046656778061394566,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099625986535328,
      "loss": 2.2646,
      "step": 6867
    },
    {
      "epoch": 0.04666357240798862,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009962585517838322,
      "loss": 2.3806,
      "step": 6868
    },
    {
      "epoch": 0.04667036675458268,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009962572379846226,
      "loss": 2.1936,
      "step": 6869
    },
    {
      "epoch": 0.046677161101176735,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009962559239556515,
      "loss": 2.3177,
      "step": 6870
    },
    {
      "epoch": 0.0466839554477708,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009962546096969197,
      "loss": 2.5406,
      "step": 6871
    },
    {
      "epoch": 0.046690749794364855,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009962532952084277,
      "loss": 2.2412,
      "step": 6872
    },
    {
      "epoch": 0.04669754414095891,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009962519804901761,
      "loss": 2.1986,
      "step": 6873
    },
    {
      "epoch": 0.04670433848755297,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009962506655421654,
      "loss": 2.4101,
      "step": 6874
    },
    {
      "epoch": 0.046711132834147025,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009962493503643966,
      "loss": 2.2359,
      "step": 6875
    },
    {
      "epoch": 0.04671792718074109,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009962480349568698,
      "loss": 2.305,
      "step": 6876
    },
    {
      "epoch": 0.046724721527335145,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009962467193195858,
      "loss": 2.0995,
      "step": 6877
    },
    {
      "epoch": 0.0467315158739292,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009962454034525454,
      "loss": 2.2659,
      "step": 6878
    },
    {
      "epoch": 0.04673831022052326,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996244087355749,
      "loss": 2.1925,
      "step": 6879
    },
    {
      "epoch": 0.046745104567117314,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009962427710291976,
      "loss": 2.4689,
      "step": 6880
    },
    {
      "epoch": 0.04675189891371137,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000996241454472891,
      "loss": 2.4362,
      "step": 6881
    },
    {
      "epoch": 0.046758693260305434,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009962401376868307,
      "loss": 2.2654,
      "step": 6882
    },
    {
      "epoch": 0.04676548760689949,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009962388206710166,
      "loss": 2.191,
      "step": 6883
    },
    {
      "epoch": 0.04677228195349355,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009962375034254496,
      "loss": 2.3713,
      "step": 6884
    },
    {
      "epoch": 0.0467790763000876,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009962361859501305,
      "loss": 2.1918,
      "step": 6885
    },
    {
      "epoch": 0.04678587064668166,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009962348682450598,
      "loss": 2.2358,
      "step": 6886
    },
    {
      "epoch": 0.04679266499327572,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000996233550310238,
      "loss": 2.2533,
      "step": 6887
    },
    {
      "epoch": 0.04679945933986978,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009962322321456656,
      "loss": 2.2731,
      "step": 6888
    },
    {
      "epoch": 0.046806253686463836,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009962309137513435,
      "loss": 2.2506,
      "step": 6889
    },
    {
      "epoch": 0.04681304803305789,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000996229595127272,
      "loss": 2.3147,
      "step": 6890
    },
    {
      "epoch": 0.04681984237965195,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000996228276273452,
      "loss": 2.3229,
      "step": 6891
    },
    {
      "epoch": 0.046826636726246006,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000996226957189884,
      "loss": 2.2878,
      "step": 6892
    },
    {
      "epoch": 0.04683343107284007,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009962256378765687,
      "loss": 2.3964,
      "step": 6893
    },
    {
      "epoch": 0.046840225419434126,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009962243183335067,
      "loss": 2.2836,
      "step": 6894
    },
    {
      "epoch": 0.04684701976602818,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009962229985606983,
      "loss": 2.5232,
      "step": 6895
    },
    {
      "epoch": 0.04685381411262224,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009962216785581444,
      "loss": 2.4058,
      "step": 6896
    },
    {
      "epoch": 0.046860608459216295,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009962203583258455,
      "loss": 2.3055,
      "step": 6897
    },
    {
      "epoch": 0.04686740280581036,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009962190378638024,
      "loss": 2.3726,
      "step": 6898
    },
    {
      "epoch": 0.046874197152404415,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009962177171720155,
      "loss": 2.2737,
      "step": 6899
    },
    {
      "epoch": 0.04688099149899847,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009962163962504854,
      "loss": 2.3323,
      "step": 6900
    },
    {
      "epoch": 0.04688778584559253,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996215075099213,
      "loss": 2.2833,
      "step": 6901
    },
    {
      "epoch": 0.046894580192186584,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009962137537181987,
      "loss": 2.2349,
      "step": 6902
    },
    {
      "epoch": 0.04690137453878064,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996212432107443,
      "loss": 2.2989,
      "step": 6903
    },
    {
      "epoch": 0.046908168885374704,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009962111102669465,
      "loss": 2.3059,
      "step": 6904
    },
    {
      "epoch": 0.04691496323196876,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00099620978819671,
      "loss": 2.3411,
      "step": 6905
    },
    {
      "epoch": 0.04692175757856282,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009962084658967341,
      "loss": 2.3523,
      "step": 6906
    },
    {
      "epoch": 0.046928551925156874,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009962071433670195,
      "loss": 2.219,
      "step": 6907
    },
    {
      "epoch": 0.04693534627175093,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009962058206075665,
      "loss": 2.2714,
      "step": 6908
    },
    {
      "epoch": 0.046942140618344994,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996204497618376,
      "loss": 2.3307,
      "step": 6909
    },
    {
      "epoch": 0.04694893496493905,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009962031743994486,
      "loss": 2.5365,
      "step": 6910
    },
    {
      "epoch": 0.046955729311533106,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009962018509507846,
      "loss": 2.3645,
      "step": 6911
    },
    {
      "epoch": 0.04696252365812716,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000996200527272385,
      "loss": 2.2034,
      "step": 6912
    },
    {
      "epoch": 0.04696931800472122,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961992033642503,
      "loss": 2.3577,
      "step": 6913
    },
    {
      "epoch": 0.046976112351315276,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009961978792263808,
      "loss": 2.3711,
      "step": 6914
    },
    {
      "epoch": 0.04698290669790934,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961965548587776,
      "loss": 2.2581,
      "step": 6915
    },
    {
      "epoch": 0.046989701044503396,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000996195230261441,
      "loss": 2.2694,
      "step": 6916
    },
    {
      "epoch": 0.04699649539109745,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009961939054343716,
      "loss": 2.2129,
      "step": 6917
    },
    {
      "epoch": 0.04700328973769151,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009961925803775703,
      "loss": 2.3656,
      "step": 6918
    },
    {
      "epoch": 0.047010084084285565,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009961912550910375,
      "loss": 2.3462,
      "step": 6919
    },
    {
      "epoch": 0.04701687843087963,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009961899295747737,
      "loss": 2.4714,
      "step": 6920
    },
    {
      "epoch": 0.047023672777473685,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009961886038287798,
      "loss": 2.3607,
      "step": 6921
    },
    {
      "epoch": 0.04703046712406774,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000996187277853056,
      "loss": 2.1763,
      "step": 6922
    },
    {
      "epoch": 0.0470372614706618,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009961859516476035,
      "loss": 2.4473,
      "step": 6923
    },
    {
      "epoch": 0.047044055817255855,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009961846252124224,
      "loss": 2.4071,
      "step": 6924
    },
    {
      "epoch": 0.04705085016384992,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009961832985475136,
      "loss": 2.3115,
      "step": 6925
    },
    {
      "epoch": 0.047057644510443974,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009961819716528778,
      "loss": 2.3752,
      "step": 6926
    },
    {
      "epoch": 0.04706443885703803,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009961806445285151,
      "loss": 2.098,
      "step": 6927
    },
    {
      "epoch": 0.04707123320363209,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009961793171744265,
      "loss": 2.3195,
      "step": 6928
    },
    {
      "epoch": 0.047078027550226144,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961779895906126,
      "loss": 2.2864,
      "step": 6929
    },
    {
      "epoch": 0.0470848218968202,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009961766617770742,
      "loss": 2.3511,
      "step": 6930
    },
    {
      "epoch": 0.047091616243414264,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009961753337338114,
      "loss": 2.6145,
      "step": 6931
    },
    {
      "epoch": 0.04709841059000832,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009961740054608252,
      "loss": 2.426,
      "step": 6932
    },
    {
      "epoch": 0.04710520493660238,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961726769581161,
      "loss": 2.2124,
      "step": 6933
    },
    {
      "epoch": 0.04711199928319643,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000996171348225685,
      "loss": 2.3886,
      "step": 6934
    },
    {
      "epoch": 0.04711879362979049,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996170019263532,
      "loss": 2.3391,
      "step": 6935
    },
    {
      "epoch": 0.04712558797638455,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000996168690071658,
      "loss": 2.2772,
      "step": 6936
    },
    {
      "epoch": 0.04713238232297861,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009961673606500635,
      "loss": 2.3908,
      "step": 6937
    },
    {
      "epoch": 0.047139176669572666,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009961660309987493,
      "loss": 2.0485,
      "step": 6938
    },
    {
      "epoch": 0.04714597101616672,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961647011177159,
      "loss": 2.2715,
      "step": 6939
    },
    {
      "epoch": 0.04715276536276078,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009961633710069639,
      "loss": 2.3665,
      "step": 6940
    },
    {
      "epoch": 0.047159559709354835,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000996162040666494,
      "loss": 2.3406,
      "step": 6941
    },
    {
      "epoch": 0.0471663540559489,
      "grad_norm": 0.140625,
      "learning_rate": 0.000996160710096307,
      "loss": 2.2045,
      "step": 6942
    },
    {
      "epoch": 0.047173148402542955,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961593792964028,
      "loss": 2.3138,
      "step": 6943
    },
    {
      "epoch": 0.04717994274913701,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009961580482667827,
      "loss": 2.2134,
      "step": 6944
    },
    {
      "epoch": 0.04718673709573107,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.0009961567170074473,
      "loss": 2.2158,
      "step": 6945
    },
    {
      "epoch": 0.047193531442325125,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009961553855183966,
      "loss": 2.3478,
      "step": 6946
    },
    {
      "epoch": 0.04720032578891919,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996154053799632,
      "loss": 2.2759,
      "step": 6947
    },
    {
      "epoch": 0.047207120135513245,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009961527218511537,
      "loss": 2.354,
      "step": 6948
    },
    {
      "epoch": 0.0472139144821073,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009961513896729623,
      "loss": 2.2985,
      "step": 6949
    },
    {
      "epoch": 0.04722070882870136,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009961500572650585,
      "loss": 2.2457,
      "step": 6950
    },
    {
      "epoch": 0.047227503175295414,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009961487246274432,
      "loss": 2.4607,
      "step": 6951
    },
    {
      "epoch": 0.04723429752188947,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009961473917601163,
      "loss": 2.2555,
      "step": 6952
    },
    {
      "epoch": 0.047241091868483534,
      "grad_norm": 0.140625,
      "learning_rate": 0.000996146058663079,
      "loss": 2.2309,
      "step": 6953
    },
    {
      "epoch": 0.04724788621507759,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000996144725336332,
      "loss": 2.2984,
      "step": 6954
    },
    {
      "epoch": 0.04725468056167165,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009961433917798754,
      "loss": 2.5382,
      "step": 6955
    },
    {
      "epoch": 0.0472614749082657,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.00099614205799371,
      "loss": 2.3579,
      "step": 6956
    },
    {
      "epoch": 0.04726826925485976,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009961407239778368,
      "loss": 2.2706,
      "step": 6957
    },
    {
      "epoch": 0.04727506360145382,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000996139389732256,
      "loss": 2.4497,
      "step": 6958
    },
    {
      "epoch": 0.04728185794804788,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009961380552569684,
      "loss": 2.2842,
      "step": 6959
    },
    {
      "epoch": 0.047288652294641936,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009961367205519745,
      "loss": 2.3761,
      "step": 6960
    },
    {
      "epoch": 0.04729544664123599,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000996135385617275,
      "loss": 2.3892,
      "step": 6961
    },
    {
      "epoch": 0.04730224098783005,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009961340504528707,
      "loss": 2.2174,
      "step": 6962
    },
    {
      "epoch": 0.047309035334424106,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009961327150587618,
      "loss": 2.3596,
      "step": 6963
    },
    {
      "epoch": 0.04731582968101817,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996131379434949,
      "loss": 2.3729,
      "step": 6964
    },
    {
      "epoch": 0.047322624027612226,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009961300435814334,
      "loss": 2.3243,
      "step": 6965
    },
    {
      "epoch": 0.04732941837420628,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009961287074982152,
      "loss": 2.3459,
      "step": 6966
    },
    {
      "epoch": 0.04733621272080034,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996127371185295,
      "loss": 2.2328,
      "step": 6967
    },
    {
      "epoch": 0.047343007067394395,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009961260346426735,
      "loss": 2.4349,
      "step": 6968
    },
    {
      "epoch": 0.04734980141398846,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009961246978703515,
      "loss": 2.3448,
      "step": 6969
    },
    {
      "epoch": 0.047356595760582515,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009961233608683292,
      "loss": 2.315,
      "step": 6970
    },
    {
      "epoch": 0.04736339010717657,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009961220236366075,
      "loss": 2.3658,
      "step": 6971
    },
    {
      "epoch": 0.04737018445377063,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009961206861751873,
      "loss": 2.3559,
      "step": 6972
    },
    {
      "epoch": 0.047376978800364684,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009961193484840688,
      "loss": 2.3869,
      "step": 6973
    },
    {
      "epoch": 0.04738377314695874,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009961180105632524,
      "loss": 2.4093,
      "step": 6974
    },
    {
      "epoch": 0.047390567493552804,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961166724127392,
      "loss": 2.2326,
      "step": 6975
    },
    {
      "epoch": 0.04739736184014686,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00099611533403253,
      "loss": 2.3638,
      "step": 6976
    },
    {
      "epoch": 0.04740415618674092,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009961139954226246,
      "loss": 2.4332,
      "step": 6977
    },
    {
      "epoch": 0.047410950533334974,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009961126565830244,
      "loss": 2.4875,
      "step": 6978
    },
    {
      "epoch": 0.04741774487992903,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009961113175137298,
      "loss": 2.3813,
      "step": 6979
    },
    {
      "epoch": 0.047424539226523094,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009961099782147412,
      "loss": 2.3356,
      "step": 6980
    },
    {
      "epoch": 0.04743133357311715,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009961086386860594,
      "loss": 2.3227,
      "step": 6981
    },
    {
      "epoch": 0.04743812791971121,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009961072989276849,
      "loss": 2.3735,
      "step": 6982
    },
    {
      "epoch": 0.04744492226630526,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009961059589396184,
      "loss": 2.3484,
      "step": 6983
    },
    {
      "epoch": 0.04745171661289932,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009961046187218607,
      "loss": 2.3159,
      "step": 6984
    },
    {
      "epoch": 0.04745851095949338,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009961032782744121,
      "loss": 2.2881,
      "step": 6985
    },
    {
      "epoch": 0.04746530530608744,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009961019375972736,
      "loss": 2.1609,
      "step": 6986
    },
    {
      "epoch": 0.047472099652681496,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009961005966904453,
      "loss": 2.3451,
      "step": 6987
    },
    {
      "epoch": 0.04747889399927555,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009960992555539284,
      "loss": 2.3472,
      "step": 6988
    },
    {
      "epoch": 0.04748568834586961,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000996097914187723,
      "loss": 2.3802,
      "step": 6989
    },
    {
      "epoch": 0.047492482692463665,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00099609657259183,
      "loss": 2.4005,
      "step": 6990
    },
    {
      "epoch": 0.04749927703905773,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009960952307662499,
      "loss": 2.2903,
      "step": 6991
    },
    {
      "epoch": 0.047506071385651785,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009960938887109835,
      "loss": 2.3437,
      "step": 6992
    },
    {
      "epoch": 0.04751286573224584,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009960925464260312,
      "loss": 2.2816,
      "step": 6993
    },
    {
      "epoch": 0.0475196600788399,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000996091203911394,
      "loss": 2.3407,
      "step": 6994
    },
    {
      "epoch": 0.047526454425433955,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000996089861167072,
      "loss": 2.359,
      "step": 6995
    },
    {
      "epoch": 0.04753324877202802,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009960885181930662,
      "loss": 2.3956,
      "step": 6996
    },
    {
      "epoch": 0.047540043118622075,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009960871749893773,
      "loss": 2.3462,
      "step": 6997
    },
    {
      "epoch": 0.04754683746521613,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009960858315560053,
      "loss": 2.3049,
      "step": 6998
    },
    {
      "epoch": 0.04755363181181019,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009960844878929517,
      "loss": 2.1345,
      "step": 6999
    },
    {
      "epoch": 0.047560426158404244,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009960831440002163,
      "loss": 2.3668,
      "step": 7000
    },
    {
      "epoch": 0.0475672205049983,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009960817998778003,
      "loss": 2.2108,
      "step": 7001
    },
    {
      "epoch": 0.047574014851592364,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009960804555257038,
      "loss": 2.2971,
      "step": 7002
    },
    {
      "epoch": 0.04758080919818642,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996079110943928,
      "loss": 2.3843,
      "step": 7003
    },
    {
      "epoch": 0.04758760354478048,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009960777661324734,
      "loss": 2.3875,
      "step": 7004
    },
    {
      "epoch": 0.04759439789137453,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009960764210913402,
      "loss": 2.3081,
      "step": 7005
    },
    {
      "epoch": 0.04760119223796859,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009960750758205295,
      "loss": 2.3338,
      "step": 7006
    },
    {
      "epoch": 0.04760798658456265,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009960737303200415,
      "loss": 2.138,
      "step": 7007
    },
    {
      "epoch": 0.04761478093115671,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009960723845898773,
      "loss": 2.3947,
      "step": 7008
    },
    {
      "epoch": 0.047621575277750766,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009960710386300372,
      "loss": 2.3004,
      "step": 7009
    },
    {
      "epoch": 0.04762836962434482,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009960696924405218,
      "loss": 2.2523,
      "step": 7010
    },
    {
      "epoch": 0.04763516397093888,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000996068346021332,
      "loss": 2.2973,
      "step": 7011
    },
    {
      "epoch": 0.047641958317532936,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009960669993724681,
      "loss": 2.358,
      "step": 7012
    },
    {
      "epoch": 0.047648752664127,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009960656524939308,
      "loss": 2.1016,
      "step": 7013
    },
    {
      "epoch": 0.047655547010721055,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000996064305385721,
      "loss": 2.3106,
      "step": 7014
    },
    {
      "epoch": 0.04766234135731511,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009960629580478388,
      "loss": 2.3486,
      "step": 7015
    },
    {
      "epoch": 0.04766913570390917,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009960616104802854,
      "loss": 2.2672,
      "step": 7016
    },
    {
      "epoch": 0.047675930050503225,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000996060262683061,
      "loss": 2.4436,
      "step": 7017
    },
    {
      "epoch": 0.04768272439709729,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009960589146561665,
      "loss": 2.4646,
      "step": 7018
    },
    {
      "epoch": 0.047689518743691345,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009960575663996026,
      "loss": 2.1492,
      "step": 7019
    },
    {
      "epoch": 0.0476963130902854,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009960562179133694,
      "loss": 2.3309,
      "step": 7020
    },
    {
      "epoch": 0.04770310743687946,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996054869197468,
      "loss": 2.4292,
      "step": 7021
    },
    {
      "epoch": 0.047709901783473514,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009960535202518989,
      "loss": 2.1338,
      "step": 7022
    },
    {
      "epoch": 0.04771669613006757,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009960521710766625,
      "loss": 2.2961,
      "step": 7023
    },
    {
      "epoch": 0.047723490476661634,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009960508216717598,
      "loss": 2.335,
      "step": 7024
    },
    {
      "epoch": 0.04773028482325569,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009960494720371914,
      "loss": 2.2013,
      "step": 7025
    },
    {
      "epoch": 0.04773707916984975,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009960481221729575,
      "loss": 2.4211,
      "step": 7026
    },
    {
      "epoch": 0.047743873516443804,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009960467720790593,
      "loss": 2.2647,
      "step": 7027
    },
    {
      "epoch": 0.04775066786303786,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000996045421755497,
      "loss": 2.2239,
      "step": 7028
    },
    {
      "epoch": 0.04775746220963192,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009960440712022713,
      "loss": 2.355,
      "step": 7029
    },
    {
      "epoch": 0.04776425655622598,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000996042720419383,
      "loss": 2.1774,
      "step": 7030
    },
    {
      "epoch": 0.047771050902820036,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009960413694068324,
      "loss": 2.2771,
      "step": 7031
    },
    {
      "epoch": 0.04777784524941409,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009960400181646205,
      "loss": 2.2759,
      "step": 7032
    },
    {
      "epoch": 0.04778463959600815,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009960386666927478,
      "loss": 2.3696,
      "step": 7033
    },
    {
      "epoch": 0.047791433942602206,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009960373149912147,
      "loss": 2.2747,
      "step": 7034
    },
    {
      "epoch": 0.04779822828919627,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009960359630600223,
      "loss": 2.27,
      "step": 7035
    },
    {
      "epoch": 0.047805022635790326,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009960346108991707,
      "loss": 2.2981,
      "step": 7036
    },
    {
      "epoch": 0.04781181698238438,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009960332585086606,
      "loss": 2.3735,
      "step": 7037
    },
    {
      "epoch": 0.04781861132897844,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000996031905888493,
      "loss": 2.2406,
      "step": 7038
    },
    {
      "epoch": 0.047825405675572495,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009960305530386684,
      "loss": 2.2665,
      "step": 7039
    },
    {
      "epoch": 0.04783220002216656,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009960291999591873,
      "loss": 2.2814,
      "step": 7040
    },
    {
      "epoch": 0.047838994368760615,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009960278466500502,
      "loss": 2.3001,
      "step": 7041
    },
    {
      "epoch": 0.04784578871535467,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000996026493111258,
      "loss": 2.2714,
      "step": 7042
    },
    {
      "epoch": 0.04785258306194873,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009960251393428112,
      "loss": 2.3436,
      "step": 7043
    },
    {
      "epoch": 0.047859377408542784,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009960237853447107,
      "loss": 2.4155,
      "step": 7044
    },
    {
      "epoch": 0.04786617175513685,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009960224311169566,
      "loss": 2.3147,
      "step": 7045
    },
    {
      "epoch": 0.047872966101730904,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00099602107665955,
      "loss": 2.3776,
      "step": 7046
    },
    {
      "epoch": 0.04787976044832496,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000996019721972491,
      "loss": 2.3516,
      "step": 7047
    },
    {
      "epoch": 0.04788655479491902,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009960183670557807,
      "loss": 2.1757,
      "step": 7048
    },
    {
      "epoch": 0.047893349141513074,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009960170119094197,
      "loss": 2.262,
      "step": 7049
    },
    {
      "epoch": 0.04790014348810713,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009960156565334084,
      "loss": 2.3562,
      "step": 7050
    },
    {
      "epoch": 0.047906937834701194,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009960143009277475,
      "loss": 2.3312,
      "step": 7051
    },
    {
      "epoch": 0.04791373218129525,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009960129450924378,
      "loss": 2.3873,
      "step": 7052
    },
    {
      "epoch": 0.04792052652788931,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009960115890274798,
      "loss": 2.1531,
      "step": 7053
    },
    {
      "epoch": 0.04792732087448336,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009960102327328739,
      "loss": 2.4083,
      "step": 7054
    },
    {
      "epoch": 0.04793411522107742,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009960088762086213,
      "loss": 2.3053,
      "step": 7055
    },
    {
      "epoch": 0.04794090956767148,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000996007519454722,
      "loss": 2.361,
      "step": 7056
    },
    {
      "epoch": 0.04794770391426554,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000996006162471177,
      "loss": 2.402,
      "step": 7057
    },
    {
      "epoch": 0.047954498260859596,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009960048052579868,
      "loss": 2.3071,
      "step": 7058
    },
    {
      "epoch": 0.04796129260745365,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000996003447815152,
      "loss": 2.2102,
      "step": 7059
    },
    {
      "epoch": 0.04796808695404771,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009960020901426734,
      "loss": 2.2867,
      "step": 7060
    },
    {
      "epoch": 0.047974881300641765,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009960007322405515,
      "loss": 2.227,
      "step": 7061
    },
    {
      "epoch": 0.04798167564723583,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000995999374108787,
      "loss": 2.3744,
      "step": 7062
    },
    {
      "epoch": 0.047988469993829885,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009959980157473803,
      "loss": 2.3927,
      "step": 7063
    },
    {
      "epoch": 0.04799526434042394,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009959966571563325,
      "loss": 2.2139,
      "step": 7064
    },
    {
      "epoch": 0.048002058687018,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009959952983356435,
      "loss": 2.3482,
      "step": 7065
    },
    {
      "epoch": 0.048008853033612055,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009959939392853146,
      "loss": 2.2609,
      "step": 7066
    },
    {
      "epoch": 0.04801564738020612,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009959925800053463,
      "loss": 2.2384,
      "step": 7067
    },
    {
      "epoch": 0.048022441726800175,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000995991220495739,
      "loss": 2.3083,
      "step": 7068
    },
    {
      "epoch": 0.04802923607339423,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009959898607564935,
      "loss": 2.1985,
      "step": 7069
    },
    {
      "epoch": 0.04803603041998829,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009959885007876103,
      "loss": 2.3313,
      "step": 7070
    },
    {
      "epoch": 0.048042824766582344,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009959871405890901,
      "loss": 2.5755,
      "step": 7071
    },
    {
      "epoch": 0.0480496191131764,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009959857801609337,
      "loss": 2.3496,
      "step": 7072
    },
    {
      "epoch": 0.048056413459770464,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009959844195031415,
      "loss": 2.2267,
      "step": 7073
    },
    {
      "epoch": 0.04806320780636452,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000995983058615714,
      "loss": 2.2923,
      "step": 7074
    },
    {
      "epoch": 0.04807000215295858,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009959816974986523,
      "loss": 2.3638,
      "step": 7075
    },
    {
      "epoch": 0.04807679649955263,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009959803361519568,
      "loss": 2.2981,
      "step": 7076
    },
    {
      "epoch": 0.04808359084614669,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000995978974575628,
      "loss": 2.3008,
      "step": 7077
    },
    {
      "epoch": 0.04809038519274075,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009959776127696665,
      "loss": 2.4926,
      "step": 7078
    },
    {
      "epoch": 0.04809717953933481,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000995976250734073,
      "loss": 2.3176,
      "step": 7079
    },
    {
      "epoch": 0.048103973885928866,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009959748884688484,
      "loss": 2.1421,
      "step": 7080
    },
    {
      "epoch": 0.04811076823252292,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009959735259739931,
      "loss": 2.298,
      "step": 7081
    },
    {
      "epoch": 0.04811756257911698,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009959721632495076,
      "loss": 2.2832,
      "step": 7082
    },
    {
      "epoch": 0.048124356925711036,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009959708002953926,
      "loss": 2.1787,
      "step": 7083
    },
    {
      "epoch": 0.0481311512723051,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000995969437111649,
      "loss": 2.1095,
      "step": 7084
    },
    {
      "epoch": 0.048137945618899156,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009959680736982772,
      "loss": 2.4445,
      "step": 7085
    },
    {
      "epoch": 0.04814473996549321,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009959667100552779,
      "loss": 2.3716,
      "step": 7086
    },
    {
      "epoch": 0.04815153431208727,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009959653461826517,
      "loss": 2.4046,
      "step": 7087
    },
    {
      "epoch": 0.048158328658681325,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995963982080399,
      "loss": 2.4076,
      "step": 7088
    },
    {
      "epoch": 0.04816512300527539,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000995962617748521,
      "loss": 2.3702,
      "step": 7089
    },
    {
      "epoch": 0.048171917351869445,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009959612531870179,
      "loss": 2.2654,
      "step": 7090
    },
    {
      "epoch": 0.0481787116984635,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009959598883958903,
      "loss": 2.3758,
      "step": 7091
    },
    {
      "epoch": 0.04818550604505756,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000995958523375139,
      "loss": 2.2377,
      "step": 7092
    },
    {
      "epoch": 0.048192300391651614,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009959571581247646,
      "loss": 2.2068,
      "step": 7093
    },
    {
      "epoch": 0.04819909473824567,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009959557926447677,
      "loss": 2.3076,
      "step": 7094
    },
    {
      "epoch": 0.048205889084839734,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000995954426935149,
      "loss": 2.1675,
      "step": 7095
    },
    {
      "epoch": 0.04821268343143379,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000995953060995909,
      "loss": 2.2961,
      "step": 7096
    },
    {
      "epoch": 0.04821947777802785,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009959516948270487,
      "loss": 2.3818,
      "step": 7097
    },
    {
      "epoch": 0.048226272124621904,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009959503284285682,
      "loss": 2.1988,
      "step": 7098
    },
    {
      "epoch": 0.04823306647121596,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009959489618004685,
      "loss": 2.4636,
      "step": 7099
    },
    {
      "epoch": 0.048239860817810024,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009959475949427499,
      "loss": 2.393,
      "step": 7100
    },
    {
      "epoch": 0.04824665516440408,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009959462278554135,
      "loss": 2.4011,
      "step": 7101
    },
    {
      "epoch": 0.048253449510998137,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009959448605384594,
      "loss": 2.3546,
      "step": 7102
    },
    {
      "epoch": 0.04826024385759219,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009959434929918886,
      "loss": 2.2809,
      "step": 7103
    },
    {
      "epoch": 0.04826703820418625,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009959421252157018,
      "loss": 2.2853,
      "step": 7104
    },
    {
      "epoch": 0.04827383255078031,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009959407572098994,
      "loss": 2.233,
      "step": 7105
    },
    {
      "epoch": 0.04828062689737437,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009959393889744821,
      "loss": 2.4638,
      "step": 7106
    },
    {
      "epoch": 0.048287421243968426,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009959380205094507,
      "loss": 2.2264,
      "step": 7107
    },
    {
      "epoch": 0.04829421559056248,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009959366518148057,
      "loss": 2.2363,
      "step": 7108
    },
    {
      "epoch": 0.04830100993715654,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009959352828905475,
      "loss": 2.3371,
      "step": 7109
    },
    {
      "epoch": 0.048307804283750595,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009959339137366769,
      "loss": 2.2495,
      "step": 7110
    },
    {
      "epoch": 0.04831459863034466,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009959325443531948,
      "loss": 2.3545,
      "step": 7111
    },
    {
      "epoch": 0.048321392976938715,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009959311747401015,
      "loss": 2.4214,
      "step": 7112
    },
    {
      "epoch": 0.04832818732353277,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009959298048973979,
      "loss": 2.2224,
      "step": 7113
    },
    {
      "epoch": 0.04833498167012683,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009959284348250842,
      "loss": 2.258,
      "step": 7114
    },
    {
      "epoch": 0.048341776016720885,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009959270645231615,
      "loss": 2.2234,
      "step": 7115
    },
    {
      "epoch": 0.04834857036331495,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009959256939916304,
      "loss": 2.1767,
      "step": 7116
    },
    {
      "epoch": 0.048355364709909004,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000995924323230491,
      "loss": 2.1737,
      "step": 7117
    },
    {
      "epoch": 0.04836215905650306,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009959229522397447,
      "loss": 2.3896,
      "step": 7118
    },
    {
      "epoch": 0.04836895340309712,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009959215810193915,
      "loss": 2.2996,
      "step": 7119
    },
    {
      "epoch": 0.048375747749691174,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009959202095694327,
      "loss": 2.1488,
      "step": 7120
    },
    {
      "epoch": 0.04838254209628523,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000995918837889868,
      "loss": 2.1312,
      "step": 7121
    },
    {
      "epoch": 0.048389336442879294,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000995917465980699,
      "loss": 2.2838,
      "step": 7122
    },
    {
      "epoch": 0.04839613078947335,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009959160938419257,
      "loss": 2.4446,
      "step": 7123
    },
    {
      "epoch": 0.04840292513606741,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000995914721473549,
      "loss": 2.4982,
      "step": 7124
    },
    {
      "epoch": 0.04840971948266146,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009959133488755692,
      "loss": 2.3534,
      "step": 7125
    },
    {
      "epoch": 0.04841651382925552,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009959119760479874,
      "loss": 2.2971,
      "step": 7126
    },
    {
      "epoch": 0.04842330817584958,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000995910602990804,
      "loss": 2.3767,
      "step": 7127
    },
    {
      "epoch": 0.04843010252244364,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009959092297040198,
      "loss": 2.2258,
      "step": 7128
    },
    {
      "epoch": 0.048436896869037696,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009959078561876352,
      "loss": 2.341,
      "step": 7129
    },
    {
      "epoch": 0.04844369121563175,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000995906482441651,
      "loss": 2.4682,
      "step": 7130
    },
    {
      "epoch": 0.04845048556222581,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009959051084660678,
      "loss": 2.3456,
      "step": 7131
    },
    {
      "epoch": 0.048457279908819866,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009959037342608863,
      "loss": 2.3463,
      "step": 7132
    },
    {
      "epoch": 0.04846407425541393,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009959023598261067,
      "loss": 2.4037,
      "step": 7133
    },
    {
      "epoch": 0.048470868602007985,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009959009851617305,
      "loss": 2.3037,
      "step": 7134
    },
    {
      "epoch": 0.04847766294860204,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009958996102677575,
      "loss": 2.3452,
      "step": 7135
    },
    {
      "epoch": 0.0484844572951961,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009958982351441887,
      "loss": 2.4,
      "step": 7136
    },
    {
      "epoch": 0.048491251641790155,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009958968597910246,
      "loss": 2.2328,
      "step": 7137
    },
    {
      "epoch": 0.04849804598838422,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000995895484208266,
      "loss": 2.2611,
      "step": 7138
    },
    {
      "epoch": 0.048504840334978275,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009958941083959138,
      "loss": 2.3224,
      "step": 7139
    },
    {
      "epoch": 0.04851163468157233,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995892732353968,
      "loss": 2.2372,
      "step": 7140
    },
    {
      "epoch": 0.04851842902816639,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009958913560824297,
      "loss": 2.3116,
      "step": 7141
    },
    {
      "epoch": 0.048525223374760444,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009958899795812991,
      "loss": 2.3349,
      "step": 7142
    },
    {
      "epoch": 0.0485320177213545,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009958886028505773,
      "loss": 2.1869,
      "step": 7143
    },
    {
      "epoch": 0.048538812067948564,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000995887225890265,
      "loss": 2.3009,
      "step": 7144
    },
    {
      "epoch": 0.04854560641454262,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009958858487003623,
      "loss": 2.3025,
      "step": 7145
    },
    {
      "epoch": 0.04855240076113668,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009958844712808703,
      "loss": 2.2628,
      "step": 7146
    },
    {
      "epoch": 0.048559195107730733,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009958830936317893,
      "loss": 2.4091,
      "step": 7147
    },
    {
      "epoch": 0.04856598945432479,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009958817157531203,
      "loss": 2.2752,
      "step": 7148
    },
    {
      "epoch": 0.04857278380091885,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009958803376448637,
      "loss": 2.1445,
      "step": 7149
    },
    {
      "epoch": 0.04857957814751291,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00099587895930702,
      "loss": 2.4329,
      "step": 7150
    },
    {
      "epoch": 0.048586372494106966,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009958775807395901,
      "loss": 2.1794,
      "step": 7151
    },
    {
      "epoch": 0.04859316684070102,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009958762019425748,
      "loss": 2.4562,
      "step": 7152
    },
    {
      "epoch": 0.04859996118729508,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009958748229159744,
      "loss": 2.2958,
      "step": 7153
    },
    {
      "epoch": 0.048606755533889136,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009958734436597897,
      "loss": 2.3407,
      "step": 7154
    },
    {
      "epoch": 0.0486135498804832,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000995872064174021,
      "loss": 2.1516,
      "step": 7155
    },
    {
      "epoch": 0.048620344227077256,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009958706844586693,
      "loss": 2.1712,
      "step": 7156
    },
    {
      "epoch": 0.04862713857367131,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009958693045137353,
      "loss": 2.3283,
      "step": 7157
    },
    {
      "epoch": 0.04863393292026537,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009958679243392194,
      "loss": 2.3789,
      "step": 7158
    },
    {
      "epoch": 0.048640727266859425,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009958665439351224,
      "loss": 2.2581,
      "step": 7159
    },
    {
      "epoch": 0.04864752161345349,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995865163301445,
      "loss": 2.3051,
      "step": 7160
    },
    {
      "epoch": 0.048654315960047545,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009958637824381876,
      "loss": 2.2297,
      "step": 7161
    },
    {
      "epoch": 0.0486611103066416,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009958624013453508,
      "loss": 2.1612,
      "step": 7162
    },
    {
      "epoch": 0.04866790465323566,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009958610200229354,
      "loss": 2.5515,
      "step": 7163
    },
    {
      "epoch": 0.048674698999829714,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009958596384709421,
      "loss": 2.3405,
      "step": 7164
    },
    {
      "epoch": 0.04868149334642378,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009958582566893717,
      "loss": 2.3783,
      "step": 7165
    },
    {
      "epoch": 0.048688287693017834,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009958568746782244,
      "loss": 2.2321,
      "step": 7166
    },
    {
      "epoch": 0.04869508203961189,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995855492437501,
      "loss": 2.2582,
      "step": 7167
    },
    {
      "epoch": 0.04870187638620595,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009958541099672025,
      "loss": 2.0175,
      "step": 7168
    },
    {
      "epoch": 0.048708670732800004,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009958527272673288,
      "loss": 2.2769,
      "step": 7169
    },
    {
      "epoch": 0.04871546507939406,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009958513443378812,
      "loss": 2.2843,
      "step": 7170
    },
    {
      "epoch": 0.048722259425988124,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099584996117886,
      "loss": 2.2853,
      "step": 7171
    },
    {
      "epoch": 0.04872905377258218,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000995848577790266,
      "loss": 2.2201,
      "step": 7172
    },
    {
      "epoch": 0.04873584811917624,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009958471941721,
      "loss": 2.4442,
      "step": 7173
    },
    {
      "epoch": 0.04874264246577029,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009958458103243623,
      "loss": 2.1823,
      "step": 7174
    },
    {
      "epoch": 0.04874943681236435,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009958444262470536,
      "loss": 2.3382,
      "step": 7175
    },
    {
      "epoch": 0.04875623115895841,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009958430419401748,
      "loss": 2.3231,
      "step": 7176
    },
    {
      "epoch": 0.04876302550555247,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009958416574037262,
      "loss": 2.382,
      "step": 7177
    },
    {
      "epoch": 0.048769819852146526,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009958402726377084,
      "loss": 2.1861,
      "step": 7178
    },
    {
      "epoch": 0.04877661419874058,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009958388876421225,
      "loss": 2.2238,
      "step": 7179
    },
    {
      "epoch": 0.04878340854533464,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000995837502416969,
      "loss": 2.2404,
      "step": 7180
    },
    {
      "epoch": 0.048790202891928695,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009958361169622482,
      "loss": 2.3683,
      "step": 7181
    },
    {
      "epoch": 0.04879699723852276,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995834731277961,
      "loss": 2.3501,
      "step": 7182
    },
    {
      "epoch": 0.048803791585116815,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995833345364108,
      "loss": 2.3707,
      "step": 7183
    },
    {
      "epoch": 0.04881058593171087,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009958319592206898,
      "loss": 2.2162,
      "step": 7184
    },
    {
      "epoch": 0.04881738027830493,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009958305728477073,
      "loss": 2.3714,
      "step": 7185
    },
    {
      "epoch": 0.048824174624898985,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009958291862451607,
      "loss": 2.2333,
      "step": 7186
    },
    {
      "epoch": 0.04883096897149305,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009958277994130509,
      "loss": 2.3273,
      "step": 7187
    },
    {
      "epoch": 0.048837763318087105,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009958264123513785,
      "loss": 2.2518,
      "step": 7188
    },
    {
      "epoch": 0.04884455766468116,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009958250250601444,
      "loss": 2.256,
      "step": 7189
    },
    {
      "epoch": 0.04885135201127522,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009958236375393489,
      "loss": 2.184,
      "step": 7190
    },
    {
      "epoch": 0.048858146357869274,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009958222497889926,
      "loss": 2.3544,
      "step": 7191
    },
    {
      "epoch": 0.04886494070446333,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009958208618090763,
      "loss": 2.4007,
      "step": 7192
    },
    {
      "epoch": 0.048871735051057394,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009958194735996007,
      "loss": 2.4352,
      "step": 7193
    },
    {
      "epoch": 0.04887852939765145,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009958180851605662,
      "loss": 2.2531,
      "step": 7194
    },
    {
      "epoch": 0.04888532374424551,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009958166964919737,
      "loss": 2.1464,
      "step": 7195
    },
    {
      "epoch": 0.04889211809083956,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009958153075938238,
      "loss": 2.3468,
      "step": 7196
    },
    {
      "epoch": 0.04889891243743362,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009958139184661173,
      "loss": 2.3719,
      "step": 7197
    },
    {
      "epoch": 0.04890570678402768,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009958125291088544,
      "loss": 2.3943,
      "step": 7198
    },
    {
      "epoch": 0.04891250113062174,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000995811139522036,
      "loss": 2.2274,
      "step": 7199
    },
    {
      "epoch": 0.048919295477215796,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009958097497056626,
      "loss": 2.2598,
      "step": 7200
    },
    {
      "epoch": 0.04892608982380985,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000995808359659735,
      "loss": 2.2779,
      "step": 7201
    },
    {
      "epoch": 0.04893288417040391,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009958069693842542,
      "loss": 2.2621,
      "step": 7202
    },
    {
      "epoch": 0.048939678516997966,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009958055788792202,
      "loss": 2.446,
      "step": 7203
    },
    {
      "epoch": 0.04894647286359203,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009958041881446339,
      "loss": 2.2448,
      "step": 7204
    },
    {
      "epoch": 0.048953267210186086,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000995802797180496,
      "loss": 2.3255,
      "step": 7205
    },
    {
      "epoch": 0.04896006155678014,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009958014059868068,
      "loss": 2.2744,
      "step": 7206
    },
    {
      "epoch": 0.0489668559033742,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009958000145635676,
      "loss": 2.3118,
      "step": 7207
    },
    {
      "epoch": 0.048973650249968255,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009957986229107785,
      "loss": 2.3019,
      "step": 7208
    },
    {
      "epoch": 0.04898044459656232,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009957972310284404,
      "loss": 2.3641,
      "step": 7209
    },
    {
      "epoch": 0.048987238943156375,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009957958389165537,
      "loss": 2.2932,
      "step": 7210
    },
    {
      "epoch": 0.04899403328975043,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009957944465751195,
      "loss": 2.3519,
      "step": 7211
    },
    {
      "epoch": 0.04900082763634449,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995793054004138,
      "loss": 2.2764,
      "step": 7212
    },
    {
      "epoch": 0.049007621982938544,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009957916612036099,
      "loss": 2.3058,
      "step": 7213
    },
    {
      "epoch": 0.0490144163295326,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000995790268173536,
      "loss": 2.3249,
      "step": 7214
    },
    {
      "epoch": 0.049021210676126664,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000995788874913917,
      "loss": 2.2084,
      "step": 7215
    },
    {
      "epoch": 0.04902800502272072,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009957874814247533,
      "loss": 2.3796,
      "step": 7216
    },
    {
      "epoch": 0.04903479936931478,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009957860877060456,
      "loss": 2.3629,
      "step": 7217
    },
    {
      "epoch": 0.049041593715908834,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009957846937577947,
      "loss": 2.3098,
      "step": 7218
    },
    {
      "epoch": 0.04904838806250289,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009957832995800013,
      "loss": 2.2714,
      "step": 7219
    },
    {
      "epoch": 0.049055182409096953,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000995781905172666,
      "loss": 2.3333,
      "step": 7220
    },
    {
      "epoch": 0.04906197675569101,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009957805105357891,
      "loss": 2.3305,
      "step": 7221
    },
    {
      "epoch": 0.049068771102285066,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009957791156693717,
      "loss": 2.309,
      "step": 7222
    },
    {
      "epoch": 0.04907556544887912,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009957777205734143,
      "loss": 2.2125,
      "step": 7223
    },
    {
      "epoch": 0.04908235979547318,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009957763252479172,
      "loss": 2.4752,
      "step": 7224
    },
    {
      "epoch": 0.04908915414206724,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009957749296928815,
      "loss": 2.2601,
      "step": 7225
    },
    {
      "epoch": 0.0490959484886613,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009957735339083077,
      "loss": 2.3092,
      "step": 7226
    },
    {
      "epoch": 0.049102742835255356,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009957721378941963,
      "loss": 2.4016,
      "step": 7227
    },
    {
      "epoch": 0.04910953718184941,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009957707416505485,
      "loss": 2.3241,
      "step": 7228
    },
    {
      "epoch": 0.04911633152844347,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009957693451773642,
      "loss": 2.3396,
      "step": 7229
    },
    {
      "epoch": 0.049123125875037525,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009957679484746446,
      "loss": 2.3019,
      "step": 7230
    },
    {
      "epoch": 0.04912992022163159,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009957665515423898,
      "loss": 2.3544,
      "step": 7231
    },
    {
      "epoch": 0.049136714568225645,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000995765154380601,
      "loss": 2.2822,
      "step": 7232
    },
    {
      "epoch": 0.0491435089148197,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009957637569892785,
      "loss": 2.256,
      "step": 7233
    },
    {
      "epoch": 0.04915030326141376,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009957623593684231,
      "loss": 2.3048,
      "step": 7234
    },
    {
      "epoch": 0.049157097608007815,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009957609615180354,
      "loss": 2.1978,
      "step": 7235
    },
    {
      "epoch": 0.04916389195460188,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995759563438116,
      "loss": 2.2556,
      "step": 7236
    },
    {
      "epoch": 0.049170686301195934,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009957581651286657,
      "loss": 2.3371,
      "step": 7237
    },
    {
      "epoch": 0.04917748064778999,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009957567665896852,
      "loss": 2.3657,
      "step": 7238
    },
    {
      "epoch": 0.04918427499438405,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009957553678211746,
      "loss": 2.3421,
      "step": 7239
    },
    {
      "epoch": 0.049191069340978104,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009957539688231353,
      "loss": 2.2319,
      "step": 7240
    },
    {
      "epoch": 0.04919786368757216,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009957525695955675,
      "loss": 2.3466,
      "step": 7241
    },
    {
      "epoch": 0.049204658034166224,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000995751170138472,
      "loss": 2.2101,
      "step": 7242
    },
    {
      "epoch": 0.04921145238076028,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009957497704518492,
      "loss": 2.2755,
      "step": 7243
    },
    {
      "epoch": 0.04921824672735434,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009957483705357,
      "loss": 2.2543,
      "step": 7244
    },
    {
      "epoch": 0.04922504107394839,
      "grad_norm": 0.15625,
      "learning_rate": 0.000995746970390025,
      "loss": 2.2306,
      "step": 7245
    },
    {
      "epoch": 0.04923183542054245,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000995745570014825,
      "loss": 2.3189,
      "step": 7246
    },
    {
      "epoch": 0.04923862976713651,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009957441694101003,
      "loss": 2.3347,
      "step": 7247
    },
    {
      "epoch": 0.04924542411373057,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000995742768575852,
      "loss": 2.4357,
      "step": 7248
    },
    {
      "epoch": 0.049252218460324626,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009957413675120803,
      "loss": 2.382,
      "step": 7249
    },
    {
      "epoch": 0.04925901280691868,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000995739966218786,
      "loss": 2.2615,
      "step": 7250
    },
    {
      "epoch": 0.04926580715351274,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009957385646959696,
      "loss": 2.3572,
      "step": 7251
    },
    {
      "epoch": 0.049272601500106795,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009957371629436323,
      "loss": 2.3656,
      "step": 7252
    },
    {
      "epoch": 0.04927939584670086,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000995735760961774,
      "loss": 2.2324,
      "step": 7253
    },
    {
      "epoch": 0.049286190193294915,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000995734358750396,
      "loss": 2.3361,
      "step": 7254
    },
    {
      "epoch": 0.04929298453988897,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009957329563094986,
      "loss": 2.3597,
      "step": 7255
    },
    {
      "epoch": 0.04929977888648303,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009957315536390826,
      "loss": 2.1636,
      "step": 7256
    },
    {
      "epoch": 0.049306573233077085,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009957301507391486,
      "loss": 2.0897,
      "step": 7257
    },
    {
      "epoch": 0.04931336757967115,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000995728747609697,
      "loss": 2.3761,
      "step": 7258
    },
    {
      "epoch": 0.049320161926265205,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000995727344250729,
      "loss": 2.3513,
      "step": 7259
    },
    {
      "epoch": 0.04932695627285926,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009957259406622445,
      "loss": 2.3065,
      "step": 7260
    },
    {
      "epoch": 0.04933375061945332,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009957245368442448,
      "loss": 2.4035,
      "step": 7261
    },
    {
      "epoch": 0.049340544966047374,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009957231327967302,
      "loss": 2.3175,
      "step": 7262
    },
    {
      "epoch": 0.04934733931264143,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009957217285197018,
      "loss": 2.3238,
      "step": 7263
    },
    {
      "epoch": 0.049354133659235494,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009957203240131597,
      "loss": 2.3717,
      "step": 7264
    },
    {
      "epoch": 0.04936092800582955,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009957189192771048,
      "loss": 2.2872,
      "step": 7265
    },
    {
      "epoch": 0.04936772235242361,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009957175143115375,
      "loss": 2.3322,
      "step": 7266
    },
    {
      "epoch": 0.04937451669901766,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009957161091164591,
      "loss": 2.2885,
      "step": 7267
    },
    {
      "epoch": 0.04938131104561172,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009957147036918695,
      "loss": 2.3489,
      "step": 7268
    },
    {
      "epoch": 0.04938810539220578,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009957132980377699,
      "loss": 2.3289,
      "step": 7269
    },
    {
      "epoch": 0.04939489973879984,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009957118921541605,
      "loss": 2.2399,
      "step": 7270
    },
    {
      "epoch": 0.049401694085393896,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009957104860410422,
      "loss": 2.2677,
      "step": 7271
    },
    {
      "epoch": 0.04940848843198795,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009957090796984158,
      "loss": 2.2063,
      "step": 7272
    },
    {
      "epoch": 0.04941528277858201,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009957076731262817,
      "loss": 2.3208,
      "step": 7273
    },
    {
      "epoch": 0.049422077125176066,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009957062663246406,
      "loss": 2.3977,
      "step": 7274
    },
    {
      "epoch": 0.04942887147177013,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009957048592934932,
      "loss": 2.4101,
      "step": 7275
    },
    {
      "epoch": 0.049435665818364186,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00099570345203284,
      "loss": 2.5327,
      "step": 7276
    },
    {
      "epoch": 0.04944246016495824,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000995702044542682,
      "loss": 2.3056,
      "step": 7277
    },
    {
      "epoch": 0.0494492545115523,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009957006368230196,
      "loss": 2.4485,
      "step": 7278
    },
    {
      "epoch": 0.049456048858146355,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009956992288738533,
      "loss": 2.2417,
      "step": 7279
    },
    {
      "epoch": 0.04946284320474042,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009956978206951842,
      "loss": 2.2757,
      "step": 7280
    },
    {
      "epoch": 0.049469637551334475,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009956964122870126,
      "loss": 2.3555,
      "step": 7281
    },
    {
      "epoch": 0.04947643189792853,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000995695003649339,
      "loss": 2.2306,
      "step": 7282
    },
    {
      "epoch": 0.04948322624452259,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009956935947821647,
      "loss": 2.1903,
      "step": 7283
    },
    {
      "epoch": 0.049490020591116644,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009956921856854896,
      "loss": 2.2775,
      "step": 7284
    },
    {
      "epoch": 0.04949681493771071,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000995690776359315,
      "loss": 2.2015,
      "step": 7285
    },
    {
      "epoch": 0.049503609284304764,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009956893668036412,
      "loss": 2.3435,
      "step": 7286
    },
    {
      "epoch": 0.04951040363089882,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009956879570184687,
      "loss": 2.2021,
      "step": 7287
    },
    {
      "epoch": 0.04951719797749288,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009956865470037986,
      "loss": 2.2028,
      "step": 7288
    },
    {
      "epoch": 0.049523992324086934,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000995685136759631,
      "loss": 2.2165,
      "step": 7289
    },
    {
      "epoch": 0.04953078667068099,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009956837262859672,
      "loss": 2.2139,
      "step": 7290
    },
    {
      "epoch": 0.049537581017275054,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009956823155828074,
      "loss": 2.1454,
      "step": 7291
    },
    {
      "epoch": 0.04954437536386911,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009956809046501522,
      "loss": 2.1991,
      "step": 7292
    },
    {
      "epoch": 0.04955116971046317,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009956794934880026,
      "loss": 2.3615,
      "step": 7293
    },
    {
      "epoch": 0.04955796405705722,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995678082096359,
      "loss": 2.3559,
      "step": 7294
    },
    {
      "epoch": 0.04956475840365128,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009956766704752225,
      "loss": 2.2403,
      "step": 7295
    },
    {
      "epoch": 0.04957155275024534,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000995675258624593,
      "loss": 2.353,
      "step": 7296
    },
    {
      "epoch": 0.0495783470968394,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009956738465444715,
      "loss": 2.2826,
      "step": 7297
    },
    {
      "epoch": 0.049585141443433456,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995672434234859,
      "loss": 2.3916,
      "step": 7298
    },
    {
      "epoch": 0.04959193579002751,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009956710216957555,
      "loss": 2.2844,
      "step": 7299
    },
    {
      "epoch": 0.04959873013662157,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995669608927162,
      "loss": 2.2935,
      "step": 7300
    },
    {
      "epoch": 0.049605524483215625,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009956681959290795,
      "loss": 2.2975,
      "step": 7301
    },
    {
      "epoch": 0.04961231882980969,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000995666782701508,
      "loss": 2.2451,
      "step": 7302
    },
    {
      "epoch": 0.049619113176403745,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009956653692444486,
      "loss": 2.2514,
      "step": 7303
    },
    {
      "epoch": 0.0496259075229978,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009956639555579016,
      "loss": 2.3818,
      "step": 7304
    },
    {
      "epoch": 0.04963270186959186,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000995662541641868,
      "loss": 2.3598,
      "step": 7305
    },
    {
      "epoch": 0.049639496216185915,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009956611274963484,
      "loss": 2.2652,
      "step": 7306
    },
    {
      "epoch": 0.04964629056277998,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009956597131213435,
      "loss": 2.3339,
      "step": 7307
    },
    {
      "epoch": 0.049653084909374035,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009956582985168536,
      "loss": 2.2125,
      "step": 7308
    },
    {
      "epoch": 0.04965987925596809,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0009956568836828798,
      "loss": 2.3366,
      "step": 7309
    },
    {
      "epoch": 0.04966667360256215,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009956554686194224,
      "loss": 2.4338,
      "step": 7310
    },
    {
      "epoch": 0.049673467949156204,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009956540533264822,
      "loss": 2.3774,
      "step": 7311
    },
    {
      "epoch": 0.04968026229575026,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009956526378040599,
      "loss": 2.505,
      "step": 7312
    },
    {
      "epoch": 0.049687056642344324,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009956512220521561,
      "loss": 2.2426,
      "step": 7313
    },
    {
      "epoch": 0.04969385098893838,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009956498060707715,
      "loss": 2.2341,
      "step": 7314
    },
    {
      "epoch": 0.04970064533553244,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009956483898599067,
      "loss": 2.4448,
      "step": 7315
    },
    {
      "epoch": 0.04970743968212649,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009956469734195624,
      "loss": 2.2427,
      "step": 7316
    },
    {
      "epoch": 0.04971423402872055,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000995645556749739,
      "loss": 2.156,
      "step": 7317
    },
    {
      "epoch": 0.04972102837531461,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009956441398504376,
      "loss": 2.2278,
      "step": 7318
    },
    {
      "epoch": 0.04972782272190867,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009956427227216587,
      "loss": 2.288,
      "step": 7319
    },
    {
      "epoch": 0.049734617068502726,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009956413053634028,
      "loss": 2.3717,
      "step": 7320
    },
    {
      "epoch": 0.04974141141509678,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009956398877756707,
      "loss": 2.4345,
      "step": 7321
    },
    {
      "epoch": 0.04974820576169084,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000995638469958463,
      "loss": 2.395,
      "step": 7322
    },
    {
      "epoch": 0.049755000108284896,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009956370519117804,
      "loss": 2.2143,
      "step": 7323
    },
    {
      "epoch": 0.04976179445487896,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009956356336356235,
      "loss": 2.3628,
      "step": 7324
    },
    {
      "epoch": 0.049768588801473015,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995634215129993,
      "loss": 2.3191,
      "step": 7325
    },
    {
      "epoch": 0.04977538314806707,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009956327963948895,
      "loss": 2.3317,
      "step": 7326
    },
    {
      "epoch": 0.04978217749466113,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009956313774303136,
      "loss": 2.352,
      "step": 7327
    },
    {
      "epoch": 0.049788971841255185,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009956299582362664,
      "loss": 2.3825,
      "step": 7328
    },
    {
      "epoch": 0.04979576618784925,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995628538812748,
      "loss": 2.248,
      "step": 7329
    },
    {
      "epoch": 0.049802560534443305,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009956271191597593,
      "loss": 2.2437,
      "step": 7330
    },
    {
      "epoch": 0.04980935488103736,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009956256992773008,
      "loss": 2.3711,
      "step": 7331
    },
    {
      "epoch": 0.04981614922763142,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009956242791653734,
      "loss": 2.255,
      "step": 7332
    },
    {
      "epoch": 0.049822943574225474,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009956228588239775,
      "loss": 2.2685,
      "step": 7333
    },
    {
      "epoch": 0.04982973792081953,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000995621438253114,
      "loss": 2.2626,
      "step": 7334
    },
    {
      "epoch": 0.049836532267413594,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009956200174527836,
      "loss": 2.4057,
      "step": 7335
    },
    {
      "epoch": 0.04984332661400765,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009956185964229865,
      "loss": 2.3135,
      "step": 7336
    },
    {
      "epoch": 0.04985012096060171,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009956171751637239,
      "loss": 2.3614,
      "step": 7337
    },
    {
      "epoch": 0.049856915307195764,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009956157536749962,
      "loss": 2.2896,
      "step": 7338
    },
    {
      "epoch": 0.04986370965378982,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000995614331956804,
      "loss": 2.2118,
      "step": 7339
    },
    {
      "epoch": 0.04987050400038388,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009956129100091482,
      "loss": 2.2101,
      "step": 7340
    },
    {
      "epoch": 0.04987729834697794,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009956114878320293,
      "loss": 2.1602,
      "step": 7341
    },
    {
      "epoch": 0.049884092693571996,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000995610065425448,
      "loss": 2.265,
      "step": 7342
    },
    {
      "epoch": 0.04989088704016605,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0009956086427894046,
      "loss": 2.2649,
      "step": 7343
    },
    {
      "epoch": 0.04989768138676011,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009956072199239002,
      "loss": 2.3918,
      "step": 7344
    },
    {
      "epoch": 0.04990447573335417,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009956057968289354,
      "loss": 2.4781,
      "step": 7345
    },
    {
      "epoch": 0.04991127007994823,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000995604373504511,
      "loss": 2.3296,
      "step": 7346
    },
    {
      "epoch": 0.049918064426542286,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009956029499506272,
      "loss": 2.2886,
      "step": 7347
    },
    {
      "epoch": 0.04992485877313634,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009956015261672851,
      "loss": 2.3169,
      "step": 7348
    },
    {
      "epoch": 0.0499316531197304,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009956001021544849,
      "loss": 2.1907,
      "step": 7349
    },
    {
      "epoch": 0.049938447466324455,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009955986779122278,
      "loss": 2.3711,
      "step": 7350
    },
    {
      "epoch": 0.04994524181291852,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000995597253440514,
      "loss": 2.4165,
      "step": 7351
    },
    {
      "epoch": 0.049952036159512575,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009955958287393445,
      "loss": 2.2341,
      "step": 7352
    },
    {
      "epoch": 0.04995883050610663,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009955944038087198,
      "loss": 2.6274,
      "step": 7353
    },
    {
      "epoch": 0.04996562485270069,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009955929786486404,
      "loss": 2.3242,
      "step": 7354
    },
    {
      "epoch": 0.049972419199294744,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009955915532591072,
      "loss": 2.2045,
      "step": 7355
    },
    {
      "epoch": 0.04997921354588881,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000995590127640121,
      "loss": 2.223,
      "step": 7356
    },
    {
      "epoch": 0.049986007892482864,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009955887017916822,
      "loss": 2.2971,
      "step": 7357
    },
    {
      "epoch": 0.04999280223907692,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009955872757137915,
      "loss": 2.2898,
      "step": 7358
    },
    {
      "epoch": 0.04999959658567098,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009955858494064493,
      "loss": 2.1831,
      "step": 7359
    },
    {
      "epoch": 0.050006390932265034,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009955844228696567,
      "loss": 2.4532,
      "step": 7360
    },
    {
      "epoch": 0.05001318527885909,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009955829961034142,
      "loss": 2.4003,
      "step": 7361
    },
    {
      "epoch": 0.050019979625453154,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009955815691077225,
      "loss": 2.3192,
      "step": 7362
    },
    {
      "epoch": 0.05002677397204721,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009955801418825823,
      "loss": 2.3012,
      "step": 7363
    },
    {
      "epoch": 0.05003356831864127,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000995578714427994,
      "loss": 2.2166,
      "step": 7364
    },
    {
      "epoch": 0.05004036266523532,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009955772867439585,
      "loss": 2.4304,
      "step": 7365
    },
    {
      "epoch": 0.05004715701182938,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009955758588304764,
      "loss": 2.2609,
      "step": 7366
    },
    {
      "epoch": 0.05005395135842344,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009955744306875485,
      "loss": 2.257,
      "step": 7367
    },
    {
      "epoch": 0.0500607457050175,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009955730023151752,
      "loss": 2.3506,
      "step": 7368
    },
    {
      "epoch": 0.050067540051611556,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009955715737133572,
      "loss": 2.2638,
      "step": 7369
    },
    {
      "epoch": 0.05007433439820561,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009955701448820953,
      "loss": 2.3419,
      "step": 7370
    },
    {
      "epoch": 0.05008112874479967,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00099556871582139,
      "loss": 2.1298,
      "step": 7371
    },
    {
      "epoch": 0.050087923091393725,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009955672865312422,
      "loss": 2.1949,
      "step": 7372
    },
    {
      "epoch": 0.05009471743798779,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009955658570116525,
      "loss": 2.3673,
      "step": 7373
    },
    {
      "epoch": 0.050101511784581845,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009955644272626213,
      "loss": 2.3732,
      "step": 7374
    },
    {
      "epoch": 0.0501083061311759,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009955629972841495,
      "loss": 2.3786,
      "step": 7375
    },
    {
      "epoch": 0.05011510047776996,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009955615670762378,
      "loss": 2.3059,
      "step": 7376
    },
    {
      "epoch": 0.050121894824364015,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000995560136638887,
      "loss": 2.2556,
      "step": 7377
    },
    {
      "epoch": 0.05012868917095808,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995558705972097,
      "loss": 2.1957,
      "step": 7378
    },
    {
      "epoch": 0.050135483517552135,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009955572750758694,
      "loss": 2.2563,
      "step": 7379
    },
    {
      "epoch": 0.05014227786414619,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009955558439502044,
      "loss": 2.3206,
      "step": 7380
    },
    {
      "epoch": 0.05014907221074025,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009955544125951027,
      "loss": 2.3632,
      "step": 7381
    },
    {
      "epoch": 0.050155866557334304,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009955529810105647,
      "loss": 2.4442,
      "step": 7382
    },
    {
      "epoch": 0.05016266090392836,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009955515491965918,
      "loss": 2.2102,
      "step": 7383
    },
    {
      "epoch": 0.050169455250522424,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995550117153184,
      "loss": 2.2227,
      "step": 7384
    },
    {
      "epoch": 0.05017624959711648,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995548684880342,
      "loss": 2.3419,
      "step": 7385
    },
    {
      "epoch": 0.05018304394371054,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995547252378067,
      "loss": 2.3638,
      "step": 7386
    },
    {
      "epoch": 0.05018983829030459,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000995545819646359,
      "loss": 2.3584,
      "step": 7387
    },
    {
      "epoch": 0.05019663263689865,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009955443866852191,
      "loss": 2.2541,
      "step": 7388
    },
    {
      "epoch": 0.05020342698349271,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000995542953494648,
      "loss": 2.3383,
      "step": 7389
    },
    {
      "epoch": 0.05021022133008677,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000995541520074646,
      "loss": 2.3528,
      "step": 7390
    },
    {
      "epoch": 0.050217015676680826,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000995540086425214,
      "loss": 2.4387,
      "step": 7391
    },
    {
      "epoch": 0.05022381002327488,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009955386525463525,
      "loss": 2.2993,
      "step": 7392
    },
    {
      "epoch": 0.05023060436986894,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009955372184380625,
      "loss": 2.3349,
      "step": 7393
    },
    {
      "epoch": 0.050237398716462996,
      "grad_norm": 0.220703125,
      "learning_rate": 0.0009955357841003442,
      "loss": 2.2832,
      "step": 7394
    },
    {
      "epoch": 0.05024419306305706,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009955343495331988,
      "loss": 2.2313,
      "step": 7395
    },
    {
      "epoch": 0.050250987409651116,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009955329147366266,
      "loss": 2.4011,
      "step": 7396
    },
    {
      "epoch": 0.05025778175624517,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009955314797106282,
      "loss": 2.4418,
      "step": 7397
    },
    {
      "epoch": 0.05026457610283923,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009955300444552044,
      "loss": 2.1684,
      "step": 7398
    },
    {
      "epoch": 0.050271370449433285,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000995528608970356,
      "loss": 2.3806,
      "step": 7399
    },
    {
      "epoch": 0.05027816479602735,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009955271732560835,
      "loss": 2.2502,
      "step": 7400
    },
    {
      "epoch": 0.050284959142621405,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009955257373123875,
      "loss": 2.3463,
      "step": 7401
    },
    {
      "epoch": 0.05029175348921546,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000995524301139269,
      "loss": 2.2307,
      "step": 7402
    },
    {
      "epoch": 0.05029854783580952,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009955228647367282,
      "loss": 2.3747,
      "step": 7403
    },
    {
      "epoch": 0.050305342182403574,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009955214281047661,
      "loss": 2.24,
      "step": 7404
    },
    {
      "epoch": 0.05031213652899764,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009955199912433832,
      "loss": 2.2438,
      "step": 7405
    },
    {
      "epoch": 0.050318930875591694,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009955185541525803,
      "loss": 2.2833,
      "step": 7406
    },
    {
      "epoch": 0.05032572522218575,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009955171168323578,
      "loss": 2.2585,
      "step": 7407
    },
    {
      "epoch": 0.05033251956877981,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009955156792827169,
      "loss": 2.2249,
      "step": 7408
    },
    {
      "epoch": 0.050339313915373864,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009955142415036576,
      "loss": 2.3374,
      "step": 7409
    },
    {
      "epoch": 0.05034610826196792,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995512803495181,
      "loss": 2.2303,
      "step": 7410
    },
    {
      "epoch": 0.050352902608561984,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009955113652572879,
      "loss": 2.1956,
      "step": 7411
    },
    {
      "epoch": 0.05035969695515604,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009955099267899784,
      "loss": 2.3754,
      "step": 7412
    },
    {
      "epoch": 0.050366491301750096,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009955084880932535,
      "loss": 2.379,
      "step": 7413
    },
    {
      "epoch": 0.05037328564834415,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009955070491671138,
      "loss": 2.3408,
      "step": 7414
    },
    {
      "epoch": 0.05038007999493821,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009955056100115602,
      "loss": 2.2724,
      "step": 7415
    },
    {
      "epoch": 0.05038687434153227,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995504170626593,
      "loss": 2.2923,
      "step": 7416
    },
    {
      "epoch": 0.05039366868812633,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009955027310122133,
      "loss": 2.2683,
      "step": 7417
    },
    {
      "epoch": 0.050400463034720386,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009955012911684214,
      "loss": 2.37,
      "step": 7418
    },
    {
      "epoch": 0.05040725738131444,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000995499851095218,
      "loss": 2.3745,
      "step": 7419
    },
    {
      "epoch": 0.0504140517279085,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000995498410792604,
      "loss": 2.3109,
      "step": 7420
    },
    {
      "epoch": 0.050420846074502555,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009954969702605798,
      "loss": 2.322,
      "step": 7421
    },
    {
      "epoch": 0.05042764042109662,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009954955294991462,
      "loss": 2.2647,
      "step": 7422
    },
    {
      "epoch": 0.050434434767690675,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000995494088508304,
      "loss": 2.3216,
      "step": 7423
    },
    {
      "epoch": 0.05044122911428473,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009954926472880534,
      "loss": 2.1088,
      "step": 7424
    },
    {
      "epoch": 0.05044802346087879,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009954912058383956,
      "loss": 2.2544,
      "step": 7425
    },
    {
      "epoch": 0.050454817807472845,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000995489764159331,
      "loss": 2.3206,
      "step": 7426
    },
    {
      "epoch": 0.05046161215406691,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009954883222508604,
      "loss": 2.3915,
      "step": 7427
    },
    {
      "epoch": 0.050468406500660964,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009954868801129843,
      "loss": 2.234,
      "step": 7428
    },
    {
      "epoch": 0.05047520084725502,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009954854377457036,
      "loss": 2.2653,
      "step": 7429
    },
    {
      "epoch": 0.05048199519384908,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009954839951490187,
      "loss": 2.3419,
      "step": 7430
    },
    {
      "epoch": 0.050488789540443134,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009954825523229305,
      "loss": 2.312,
      "step": 7431
    },
    {
      "epoch": 0.05049558388703719,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009954811092674394,
      "loss": 2.3843,
      "step": 7432
    },
    {
      "epoch": 0.050502378233631254,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009954796659825463,
      "loss": 2.3088,
      "step": 7433
    },
    {
      "epoch": 0.05050917258022531,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009954782224682517,
      "loss": 2.0987,
      "step": 7434
    },
    {
      "epoch": 0.05051596692681937,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009954767787245565,
      "loss": 2.1981,
      "step": 7435
    },
    {
      "epoch": 0.05052276127341342,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009954753347514611,
      "loss": 2.5242,
      "step": 7436
    },
    {
      "epoch": 0.05052955562000748,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009954738905489664,
      "loss": 2.1761,
      "step": 7437
    },
    {
      "epoch": 0.05053634996660154,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000995472446117073,
      "loss": 2.2248,
      "step": 7438
    },
    {
      "epoch": 0.0505431443131956,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009954710014557814,
      "loss": 2.3198,
      "step": 7439
    },
    {
      "epoch": 0.050549938659789656,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009954695565650923,
      "loss": 2.3781,
      "step": 7440
    },
    {
      "epoch": 0.05055673300638371,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009954681114450069,
      "loss": 2.2282,
      "step": 7441
    },
    {
      "epoch": 0.05056352735297777,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995466666095525,
      "loss": 2.349,
      "step": 7442
    },
    {
      "epoch": 0.050570321699571825,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009954652205166482,
      "loss": 2.2832,
      "step": 7443
    },
    {
      "epoch": 0.05057711604616589,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009954637747083763,
      "loss": 2.3041,
      "step": 7444
    },
    {
      "epoch": 0.050583910392759945,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009954623286707106,
      "loss": 2.1559,
      "step": 7445
    },
    {
      "epoch": 0.050590704739354,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009954608824036513,
      "loss": 2.1918,
      "step": 7446
    },
    {
      "epoch": 0.05059749908594806,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009954594359071994,
      "loss": 2.2456,
      "step": 7447
    },
    {
      "epoch": 0.050604293432542115,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009954579891813553,
      "loss": 2.316,
      "step": 7448
    },
    {
      "epoch": 0.05061108777913618,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099545654222612,
      "loss": 2.256,
      "step": 7449
    },
    {
      "epoch": 0.050617882125730235,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009954550950414941,
      "loss": 2.2294,
      "step": 7450
    },
    {
      "epoch": 0.05062467647232429,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000995453647627478,
      "loss": 2.2984,
      "step": 7451
    },
    {
      "epoch": 0.05063147081891835,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009954521999840725,
      "loss": 2.2691,
      "step": 7452
    },
    {
      "epoch": 0.050638265165512404,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009954507521112784,
      "loss": 2.2672,
      "step": 7453
    },
    {
      "epoch": 0.05064505951210646,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009954493040090961,
      "loss": 2.2226,
      "step": 7454
    },
    {
      "epoch": 0.050651853858700524,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009954478556775269,
      "loss": 2.2985,
      "step": 7455
    },
    {
      "epoch": 0.05065864820529458,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009954464071165706,
      "loss": 2.326,
      "step": 7456
    },
    {
      "epoch": 0.05066544255188864,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009954449583262283,
      "loss": 2.2433,
      "step": 7457
    },
    {
      "epoch": 0.05067223689848269,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000995443509306501,
      "loss": 2.455,
      "step": 7458
    },
    {
      "epoch": 0.05067903124507675,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009954420600573888,
      "loss": 2.3676,
      "step": 7459
    },
    {
      "epoch": 0.05068582559167081,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009954406105788928,
      "loss": 2.3323,
      "step": 7460
    },
    {
      "epoch": 0.05069261993826487,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009954391608710132,
      "loss": 2.3844,
      "step": 7461
    },
    {
      "epoch": 0.050699414284858926,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995437710933751,
      "loss": 2.3549,
      "step": 7462
    },
    {
      "epoch": 0.05070620863145298,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009954362607671068,
      "loss": 2.3027,
      "step": 7463
    },
    {
      "epoch": 0.05071300297804704,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009954348103710816,
      "loss": 2.268,
      "step": 7464
    },
    {
      "epoch": 0.0507197973246411,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009954333597456755,
      "loss": 2.2673,
      "step": 7465
    },
    {
      "epoch": 0.05072659167123516,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009954319088908894,
      "loss": 2.2915,
      "step": 7466
    },
    {
      "epoch": 0.050733386017829216,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009954304578067242,
      "loss": 2.2098,
      "step": 7467
    },
    {
      "epoch": 0.05074018036442327,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099542900649318,
      "loss": 2.3166,
      "step": 7468
    },
    {
      "epoch": 0.05074697471101733,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009954275549502582,
      "loss": 2.4091,
      "step": 7469
    },
    {
      "epoch": 0.050753769057611385,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009954261031779592,
      "loss": 2.3214,
      "step": 7470
    },
    {
      "epoch": 0.05076056340420545,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009954246511762834,
      "loss": 2.5262,
      "step": 7471
    },
    {
      "epoch": 0.050767357750799505,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009954231989452317,
      "loss": 2.3197,
      "step": 7472
    },
    {
      "epoch": 0.05077415209739356,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009954217464848048,
      "loss": 2.3128,
      "step": 7473
    },
    {
      "epoch": 0.05078094644398762,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009954202937950032,
      "loss": 2.2837,
      "step": 7474
    },
    {
      "epoch": 0.050787740790581674,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009954188408758276,
      "loss": 2.3482,
      "step": 7475
    },
    {
      "epoch": 0.05079453513717574,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000995417387727279,
      "loss": 2.2956,
      "step": 7476
    },
    {
      "epoch": 0.050801329483769794,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009954159343493576,
      "loss": 2.2368,
      "step": 7477
    },
    {
      "epoch": 0.05080812383036385,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009954144807420645,
      "loss": 2.2344,
      "step": 7478
    },
    {
      "epoch": 0.05081491817695791,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009954130269054001,
      "loss": 2.2141,
      "step": 7479
    },
    {
      "epoch": 0.050821712523551964,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009954115728393652,
      "loss": 2.3216,
      "step": 7480
    },
    {
      "epoch": 0.05082850687014602,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009954101185439604,
      "loss": 2.2487,
      "step": 7481
    },
    {
      "epoch": 0.050835301216740084,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009954086640191861,
      "loss": 2.3812,
      "step": 7482
    },
    {
      "epoch": 0.05084209556333414,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009954072092650437,
      "loss": 2.2647,
      "step": 7483
    },
    {
      "epoch": 0.0508488899099282,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009954057542815333,
      "loss": 2.4583,
      "step": 7484
    },
    {
      "epoch": 0.05085568425652225,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009954042990686556,
      "loss": 2.3222,
      "step": 7485
    },
    {
      "epoch": 0.05086247860311631,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009954028436264116,
      "loss": 2.3069,
      "step": 7486
    },
    {
      "epoch": 0.05086927294971037,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009954013879548015,
      "loss": 2.1333,
      "step": 7487
    },
    {
      "epoch": 0.05087606729630443,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009953999320538264,
      "loss": 2.2246,
      "step": 7488
    },
    {
      "epoch": 0.050882861642898486,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009953984759234868,
      "loss": 2.4616,
      "step": 7489
    },
    {
      "epoch": 0.05088965598949254,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009953970195637834,
      "loss": 2.1706,
      "step": 7490
    },
    {
      "epoch": 0.0508964503360866,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009953955629747166,
      "loss": 2.4131,
      "step": 7491
    },
    {
      "epoch": 0.050903244682680655,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009953941061562877,
      "loss": 2.1912,
      "step": 7492
    },
    {
      "epoch": 0.05091003902927472,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009953926491084969,
      "loss": 2.1671,
      "step": 7493
    },
    {
      "epoch": 0.050916833375868775,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000995391191831345,
      "loss": 2.2079,
      "step": 7494
    },
    {
      "epoch": 0.05092362772246283,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009953897343248327,
      "loss": 2.2686,
      "step": 7495
    },
    {
      "epoch": 0.05093042206905689,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009953882765889603,
      "loss": 2.276,
      "step": 7496
    },
    {
      "epoch": 0.050937216415650945,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009953868186237291,
      "loss": 2.4214,
      "step": 7497
    },
    {
      "epoch": 0.05094401076224501,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009953853604291393,
      "loss": 2.3349,
      "step": 7498
    },
    {
      "epoch": 0.050950805108839065,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000995383902005192,
      "loss": 2.2406,
      "step": 7499
    },
    {
      "epoch": 0.05095759945543312,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009953824433518875,
      "loss": 2.2741,
      "step": 7500
    },
    {
      "epoch": 0.05096439380202718,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009953809844692268,
      "loss": 2.323,
      "step": 7501
    },
    {
      "epoch": 0.050971188148621234,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099537952535721,
      "loss": 2.2762,
      "step": 7502
    },
    {
      "epoch": 0.05097798249521529,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009953780660158385,
      "loss": 2.3639,
      "step": 7503
    },
    {
      "epoch": 0.050984776841809354,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009953766064451125,
      "loss": 2.3188,
      "step": 7504
    },
    {
      "epoch": 0.05099157118840341,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009953751466450329,
      "loss": 2.3874,
      "step": 7505
    },
    {
      "epoch": 0.05099836553499747,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009953736866156002,
      "loss": 2.3292,
      "step": 7506
    },
    {
      "epoch": 0.05100515988159152,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995372226356815,
      "loss": 2.2688,
      "step": 7507
    },
    {
      "epoch": 0.05101195422818558,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009953707658686784,
      "loss": 2.2933,
      "step": 7508
    },
    {
      "epoch": 0.05101874857477964,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009953693051511906,
      "loss": 2.1701,
      "step": 7509
    },
    {
      "epoch": 0.0510255429213737,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009953678442043527,
      "loss": 2.1771,
      "step": 7510
    },
    {
      "epoch": 0.051032337267967756,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000995366383028165,
      "loss": 2.2167,
      "step": 7511
    },
    {
      "epoch": 0.05103913161456181,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009953649216226286,
      "loss": 2.3907,
      "step": 7512
    },
    {
      "epoch": 0.05104592596115587,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009953634599877436,
      "loss": 2.3371,
      "step": 7513
    },
    {
      "epoch": 0.051052720307749926,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000995361998123511,
      "loss": 2.3731,
      "step": 7514
    },
    {
      "epoch": 0.05105951465434399,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009953605360299319,
      "loss": 2.317,
      "step": 7515
    },
    {
      "epoch": 0.051066309000938045,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009953590737070061,
      "loss": 1.9872,
      "step": 7516
    },
    {
      "epoch": 0.0510731033475321,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995357611154735,
      "loss": 2.2926,
      "step": 7517
    },
    {
      "epoch": 0.05107989769412616,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009953561483731189,
      "loss": 2.1261,
      "step": 7518
    },
    {
      "epoch": 0.051086692040720215,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009953546853621585,
      "loss": 2.3075,
      "step": 7519
    },
    {
      "epoch": 0.05109348638731428,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009953532221218546,
      "loss": 2.303,
      "step": 7520
    },
    {
      "epoch": 0.051100280733908335,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995351758652208,
      "loss": 2.4467,
      "step": 7521
    },
    {
      "epoch": 0.05110707508050239,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009953502949532188,
      "loss": 2.33,
      "step": 7522
    },
    {
      "epoch": 0.05111386942709645,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009953488310248885,
      "loss": 2.4136,
      "step": 7523
    },
    {
      "epoch": 0.051120663773690504,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009953473668672173,
      "loss": 2.281,
      "step": 7524
    },
    {
      "epoch": 0.05112745812028457,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000995345902480206,
      "loss": 2.2065,
      "step": 7525
    },
    {
      "epoch": 0.051134252466878624,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000995344437863855,
      "loss": 2.3465,
      "step": 7526
    },
    {
      "epoch": 0.05114104681347268,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009953429730181654,
      "loss": 2.4006,
      "step": 7527
    },
    {
      "epoch": 0.05114784116006674,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009953415079431377,
      "loss": 2.2626,
      "step": 7528
    },
    {
      "epoch": 0.051154635506660794,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009953400426387724,
      "loss": 2.1871,
      "step": 7529
    },
    {
      "epoch": 0.05116142985325485,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009953385771050705,
      "loss": 2.3596,
      "step": 7530
    },
    {
      "epoch": 0.05116822419984891,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009953371113420324,
      "loss": 2.238,
      "step": 7531
    },
    {
      "epoch": 0.05117501854644297,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009953356453496589,
      "loss": 2.2967,
      "step": 7532
    },
    {
      "epoch": 0.051181812893037026,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009953341791279506,
      "loss": 2.4098,
      "step": 7533
    },
    {
      "epoch": 0.05118860723963108,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009953327126769084,
      "loss": 2.2967,
      "step": 7534
    },
    {
      "epoch": 0.05119540158622514,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009953312459965327,
      "loss": 2.2647,
      "step": 7535
    },
    {
      "epoch": 0.0512021959328192,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009953297790868244,
      "loss": 2.2485,
      "step": 7536
    },
    {
      "epoch": 0.05120899027941326,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009953283119477842,
      "loss": 2.3145,
      "step": 7537
    },
    {
      "epoch": 0.051215784626007316,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009953268445794126,
      "loss": 2.4198,
      "step": 7538
    },
    {
      "epoch": 0.05122257897260137,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009953253769817104,
      "loss": 2.451,
      "step": 7539
    },
    {
      "epoch": 0.05122937331919543,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000995323909154678,
      "loss": 2.2185,
      "step": 7540
    },
    {
      "epoch": 0.051236167665789485,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009953224410983164,
      "loss": 2.1323,
      "step": 7541
    },
    {
      "epoch": 0.05124296201238355,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009953209728126263,
      "loss": 2.3435,
      "step": 7542
    },
    {
      "epoch": 0.051249756358977605,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009953195042976082,
      "loss": 2.2723,
      "step": 7543
    },
    {
      "epoch": 0.05125655070557166,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009953180355532628,
      "loss": 2.2897,
      "step": 7544
    },
    {
      "epoch": 0.05126334505216572,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995316566579591,
      "loss": 2.1332,
      "step": 7545
    },
    {
      "epoch": 0.051270139398759774,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000995315097376593,
      "loss": 2.308,
      "step": 7546
    },
    {
      "epoch": 0.05127693374535384,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099531362794427,
      "loss": 2.4713,
      "step": 7547
    },
    {
      "epoch": 0.051283728091947894,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009953121582826226,
      "loss": 2.3432,
      "step": 7548
    },
    {
      "epoch": 0.05129052243854195,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000995310688391651,
      "loss": 2.3252,
      "step": 7549
    },
    {
      "epoch": 0.05129731678513601,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009953092182713565,
      "loss": 2.498,
      "step": 7550
    },
    {
      "epoch": 0.051304111131730064,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009953077479217395,
      "loss": 2.4705,
      "step": 7551
    },
    {
      "epoch": 0.05131090547832412,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009953062773428005,
      "loss": 2.3988,
      "step": 7552
    },
    {
      "epoch": 0.051317699824918184,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009953048065345406,
      "loss": 2.3579,
      "step": 7553
    },
    {
      "epoch": 0.05132449417151224,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009953033354969602,
      "loss": 2.2975,
      "step": 7554
    },
    {
      "epoch": 0.0513312885181063,
      "grad_norm": 0.169921875,
      "learning_rate": 0.00099530186423006,
      "loss": 2.3472,
      "step": 7555
    },
    {
      "epoch": 0.05133808286470035,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009953003927338406,
      "loss": 2.2256,
      "step": 7556
    },
    {
      "epoch": 0.05134487721129441,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000995298921008303,
      "loss": 2.3007,
      "step": 7557
    },
    {
      "epoch": 0.05135167155788847,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009952974490534474,
      "loss": 2.322,
      "step": 7558
    },
    {
      "epoch": 0.05135846590448253,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009952959768692751,
      "loss": 2.2457,
      "step": 7559
    },
    {
      "epoch": 0.051365260251076586,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009952945044557863,
      "loss": 2.2281,
      "step": 7560
    },
    {
      "epoch": 0.05137205459767064,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000995293031812982,
      "loss": 2.2468,
      "step": 7561
    },
    {
      "epoch": 0.0513788489442647,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009952915589408623,
      "loss": 2.3512,
      "step": 7562
    },
    {
      "epoch": 0.051385643290858755,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009952900858394285,
      "loss": 2.2952,
      "step": 7563
    },
    {
      "epoch": 0.05139243763745282,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995288612508681,
      "loss": 2.1417,
      "step": 7564
    },
    {
      "epoch": 0.051399231984046875,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009952871389486207,
      "loss": 2.3744,
      "step": 7565
    },
    {
      "epoch": 0.05140602633064093,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009952856651592478,
      "loss": 2.1381,
      "step": 7566
    },
    {
      "epoch": 0.05141282067723499,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009952841911405636,
      "loss": 2.298,
      "step": 7567
    },
    {
      "epoch": 0.051419615023829045,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009952827168925686,
      "loss": 2.347,
      "step": 7568
    },
    {
      "epoch": 0.05142640937042311,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009952812424152632,
      "loss": 2.3023,
      "step": 7569
    },
    {
      "epoch": 0.051433203717017165,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009952797677086483,
      "loss": 2.2618,
      "step": 7570
    },
    {
      "epoch": 0.05143999806361122,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009952782927727245,
      "loss": 2.1871,
      "step": 7571
    },
    {
      "epoch": 0.05144679241020528,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009952768176074926,
      "loss": 2.3531,
      "step": 7572
    },
    {
      "epoch": 0.051453586756799334,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009952753422129532,
      "loss": 2.2314,
      "step": 7573
    },
    {
      "epoch": 0.05146038110339339,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995273866589107,
      "loss": 2.3552,
      "step": 7574
    },
    {
      "epoch": 0.051467175449987454,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009952723907359546,
      "loss": 2.2131,
      "step": 7575
    },
    {
      "epoch": 0.05147396979658151,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009952709146534967,
      "loss": 2.3081,
      "step": 7576
    },
    {
      "epoch": 0.05148076414317557,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009952694383417342,
      "loss": 2.2679,
      "step": 7577
    },
    {
      "epoch": 0.05148755848976962,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009952679618006677,
      "loss": 2.4324,
      "step": 7578
    },
    {
      "epoch": 0.05149435283636368,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009952664850302977,
      "loss": 2.2496,
      "step": 7579
    },
    {
      "epoch": 0.05150114718295774,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995265008030625,
      "loss": 2.2162,
      "step": 7580
    },
    {
      "epoch": 0.0515079415295518,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009952635308016502,
      "loss": 2.3022,
      "step": 7581
    },
    {
      "epoch": 0.051514735876145856,
      "grad_norm": 0.2158203125,
      "learning_rate": 0.0009952620533433742,
      "loss": 2.1857,
      "step": 7582
    },
    {
      "epoch": 0.05152153022273991,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009952605756557975,
      "loss": 2.3006,
      "step": 7583
    },
    {
      "epoch": 0.05152832456933397,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009952590977389208,
      "loss": 2.3246,
      "step": 7584
    },
    {
      "epoch": 0.05153511891592803,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009952576195927448,
      "loss": 2.1496,
      "step": 7585
    },
    {
      "epoch": 0.05154191326252209,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009952561412172702,
      "loss": 2.3861,
      "step": 7586
    },
    {
      "epoch": 0.051548707609116146,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.000995254662612498,
      "loss": 2.5233,
      "step": 7587
    },
    {
      "epoch": 0.0515555019557102,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995253183778428,
      "loss": 2.239,
      "step": 7588
    },
    {
      "epoch": 0.05156229630230426,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009952517047150619,
      "loss": 2.2948,
      "step": 7589
    },
    {
      "epoch": 0.051569090648898315,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009952502254223995,
      "loss": 2.2808,
      "step": 7590
    },
    {
      "epoch": 0.05157588499549238,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009952487459004425,
      "loss": 2.2853,
      "step": 7591
    },
    {
      "epoch": 0.051582679342086435,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009952472661491907,
      "loss": 2.3069,
      "step": 7592
    },
    {
      "epoch": 0.05158947368868049,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000995245786168645,
      "loss": 2.429,
      "step": 7593
    },
    {
      "epoch": 0.05159626803527455,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009952443059588064,
      "loss": 2.2321,
      "step": 7594
    },
    {
      "epoch": 0.051603062381868604,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000995242825519675,
      "loss": 2.5416,
      "step": 7595
    },
    {
      "epoch": 0.05160985672846267,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009952413448512524,
      "loss": 2.4046,
      "step": 7596
    },
    {
      "epoch": 0.051616651075056724,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009952398639535382,
      "loss": 2.3701,
      "step": 7597
    },
    {
      "epoch": 0.05162344542165078,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000995238382826534,
      "loss": 2.4217,
      "step": 7598
    },
    {
      "epoch": 0.05163023976824484,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009952369014702399,
      "loss": 2.4781,
      "step": 7599
    },
    {
      "epoch": 0.051637034114838894,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000995235419884657,
      "loss": 2.3161,
      "step": 7600
    },
    {
      "epoch": 0.05164382846143295,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009952339380697854,
      "loss": 2.3462,
      "step": 7601
    },
    {
      "epoch": 0.051650622808027014,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009952324560256265,
      "loss": 2.2068,
      "step": 7602
    },
    {
      "epoch": 0.05165741715462107,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009952309737521806,
      "loss": 2.3325,
      "step": 7603
    },
    {
      "epoch": 0.051664211501215127,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009952294912494483,
      "loss": 2.442,
      "step": 7604
    },
    {
      "epoch": 0.05167100584780918,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009952280085174306,
      "loss": 2.4038,
      "step": 7605
    },
    {
      "epoch": 0.05167780019440324,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009952265255561278,
      "loss": 2.2771,
      "step": 7606
    },
    {
      "epoch": 0.0516845945409973,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000995225042365541,
      "loss": 2.3818,
      "step": 7607
    },
    {
      "epoch": 0.05169138888759136,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009952235589456705,
      "loss": 2.4071,
      "step": 7608
    },
    {
      "epoch": 0.051698183234185416,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009952220752965173,
      "loss": 2.4852,
      "step": 7609
    },
    {
      "epoch": 0.05170497758077947,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009952205914180818,
      "loss": 2.2066,
      "step": 7610
    },
    {
      "epoch": 0.05171177192737353,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009952191073103651,
      "loss": 2.1617,
      "step": 7611
    },
    {
      "epoch": 0.051718566273967585,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009952176229733674,
      "loss": 2.3302,
      "step": 7612
    },
    {
      "epoch": 0.05172536062056165,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009952161384070898,
      "loss": 2.2778,
      "step": 7613
    },
    {
      "epoch": 0.051732154967155705,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009952146536115326,
      "loss": 2.4101,
      "step": 7614
    },
    {
      "epoch": 0.05173894931374976,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000995213168586697,
      "loss": 2.3999,
      "step": 7615
    },
    {
      "epoch": 0.05174574366034382,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000995211683332583,
      "loss": 2.3613,
      "step": 7616
    },
    {
      "epoch": 0.051752538006937875,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000995210197849192,
      "loss": 2.2724,
      "step": 7617
    },
    {
      "epoch": 0.05175933235353194,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000995208712136524,
      "loss": 2.3865,
      "step": 7618
    },
    {
      "epoch": 0.051766126700125994,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009952072261945803,
      "loss": 2.4592,
      "step": 7619
    },
    {
      "epoch": 0.05177292104672005,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009952057400233615,
      "loss": 2.1934,
      "step": 7620
    },
    {
      "epoch": 0.05177971539331411,
      "grad_norm": 0.15625,
      "learning_rate": 0.000995204253622868,
      "loss": 2.4255,
      "step": 7621
    },
    {
      "epoch": 0.051786509739908164,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009952027669931003,
      "loss": 2.239,
      "step": 7622
    },
    {
      "epoch": 0.05179330408650222,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009952012801340596,
      "loss": 2.2996,
      "step": 7623
    },
    {
      "epoch": 0.051800098433096284,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009951997930457465,
      "loss": 2.2408,
      "step": 7624
    },
    {
      "epoch": 0.05180689277969034,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009951983057281614,
      "loss": 2.2254,
      "step": 7625
    },
    {
      "epoch": 0.0518136871262844,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009951968181813053,
      "loss": 2.191,
      "step": 7626
    },
    {
      "epoch": 0.05182048147287845,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009951953304051786,
      "loss": 2.325,
      "step": 7627
    },
    {
      "epoch": 0.05182727581947251,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009951938423997822,
      "loss": 2.2149,
      "step": 7628
    },
    {
      "epoch": 0.05183407016606657,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009951923541651167,
      "loss": 2.3392,
      "step": 7629
    },
    {
      "epoch": 0.05184086451266063,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000995190865701183,
      "loss": 2.4779,
      "step": 7630
    },
    {
      "epoch": 0.051847658859254686,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009951893770079813,
      "loss": 2.2296,
      "step": 7631
    },
    {
      "epoch": 0.05185445320584874,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995187888085513,
      "loss": 2.3904,
      "step": 7632
    },
    {
      "epoch": 0.0518612475524428,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995186398933778,
      "loss": 2.2678,
      "step": 7633
    },
    {
      "epoch": 0.051868041899036856,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009951849095527776,
      "loss": 2.236,
      "step": 7634
    },
    {
      "epoch": 0.05187483624563092,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009951834199425121,
      "loss": 2.2462,
      "step": 7635
    },
    {
      "epoch": 0.051881630592224975,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009951819301029826,
      "loss": 2.1983,
      "step": 7636
    },
    {
      "epoch": 0.05188842493881903,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009951804400341892,
      "loss": 2.1994,
      "step": 7637
    },
    {
      "epoch": 0.05189521928541309,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009951789497361333,
      "loss": 2.218,
      "step": 7638
    },
    {
      "epoch": 0.051902013632007145,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000995177459208815,
      "loss": 2.2885,
      "step": 7639
    },
    {
      "epoch": 0.05190880797860121,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009951759684522353,
      "loss": 2.3065,
      "step": 7640
    },
    {
      "epoch": 0.051915602325195265,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009951744774663947,
      "loss": 2.1842,
      "step": 7641
    },
    {
      "epoch": 0.05192239667178932,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995172986251294,
      "loss": 2.2315,
      "step": 7642
    },
    {
      "epoch": 0.05192919101838338,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000995171494806934,
      "loss": 2.2004,
      "step": 7643
    },
    {
      "epoch": 0.051935985364977434,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009951700031333153,
      "loss": 2.2501,
      "step": 7644
    },
    {
      "epoch": 0.0519427797115715,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009951685112304385,
      "loss": 2.3594,
      "step": 7645
    },
    {
      "epoch": 0.051949574058165554,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009951670190983042,
      "loss": 2.2333,
      "step": 7646
    },
    {
      "epoch": 0.05195636840475961,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009951655267369133,
      "loss": 2.1648,
      "step": 7647
    },
    {
      "epoch": 0.05196316275135367,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009951640341462667,
      "loss": 2.2758,
      "step": 7648
    },
    {
      "epoch": 0.051969957097947724,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009951625413263645,
      "loss": 2.3551,
      "step": 7649
    },
    {
      "epoch": 0.05197675144454178,
      "grad_norm": 0.140625,
      "learning_rate": 0.000995161048277208,
      "loss": 2.2481,
      "step": 7650
    },
    {
      "epoch": 0.05198354579113584,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009951595549987976,
      "loss": 2.2555,
      "step": 7651
    },
    {
      "epoch": 0.0519903401377299,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009951580614911337,
      "loss": 2.3423,
      "step": 7652
    },
    {
      "epoch": 0.051997134484323956,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009951565677542176,
      "loss": 2.3072,
      "step": 7653
    },
    {
      "epoch": 0.05200392883091801,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009951550737880494,
      "loss": 2.5174,
      "step": 7654
    },
    {
      "epoch": 0.05201072317751207,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009951535795926303,
      "loss": 2.436,
      "step": 7655
    },
    {
      "epoch": 0.05201751752410613,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009951520851679608,
      "loss": 2.2336,
      "step": 7656
    },
    {
      "epoch": 0.05202431187070019,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009951505905140415,
      "loss": 2.357,
      "step": 7657
    },
    {
      "epoch": 0.052031106217294246,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000995149095630873,
      "loss": 2.2529,
      "step": 7658
    },
    {
      "epoch": 0.0520379005638883,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009951476005184564,
      "loss": 2.2295,
      "step": 7659
    },
    {
      "epoch": 0.05204469491048236,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995146105176792,
      "loss": 2.3583,
      "step": 7660
    },
    {
      "epoch": 0.052051489257076415,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009951446096058804,
      "loss": 2.1852,
      "step": 7661
    },
    {
      "epoch": 0.05205828360367048,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000995143113805723,
      "loss": 2.325,
      "step": 7662
    },
    {
      "epoch": 0.052065077950264535,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009951416177763196,
      "loss": 2.3129,
      "step": 7663
    },
    {
      "epoch": 0.05207187229685859,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009951401215176716,
      "loss": 2.2346,
      "step": 7664
    },
    {
      "epoch": 0.05207866664345265,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009951386250297792,
      "loss": 2.3277,
      "step": 7665
    },
    {
      "epoch": 0.052085460990046704,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009951371283126434,
      "loss": 2.1737,
      "step": 7666
    },
    {
      "epoch": 0.05209225533664077,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009951356313662647,
      "loss": 2.2875,
      "step": 7667
    },
    {
      "epoch": 0.052099049683234824,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000995134134190644,
      "loss": 2.2456,
      "step": 7668
    },
    {
      "epoch": 0.05210584402982888,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009951326367857818,
      "loss": 2.2111,
      "step": 7669
    },
    {
      "epoch": 0.05211263837642294,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995131139151679,
      "loss": 2.2536,
      "step": 7670
    },
    {
      "epoch": 0.052119432723016994,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000995129641288336,
      "loss": 2.304,
      "step": 7671
    },
    {
      "epoch": 0.05212622706961105,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009951281431957537,
      "loss": 2.3116,
      "step": 7672
    },
    {
      "epoch": 0.052133021416205114,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009951266448739327,
      "loss": 2.2123,
      "step": 7673
    },
    {
      "epoch": 0.05213981576279917,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009951251463228739,
      "loss": 2.3028,
      "step": 7674
    },
    {
      "epoch": 0.05214661010939323,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009951236475425776,
      "loss": 2.1443,
      "step": 7675
    },
    {
      "epoch": 0.05215340445598728,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009951221485330448,
      "loss": 2.1931,
      "step": 7676
    },
    {
      "epoch": 0.05216019880258134,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995120649294276,
      "loss": 2.3473,
      "step": 7677
    },
    {
      "epoch": 0.0521669931491754,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009951191498262723,
      "loss": 2.3375,
      "step": 7678
    },
    {
      "epoch": 0.05217378749576946,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995117650129034,
      "loss": 2.2529,
      "step": 7679
    },
    {
      "epoch": 0.052180581842363516,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009951161502025619,
      "loss": 2.416,
      "step": 7680
    },
    {
      "epoch": 0.05218737618895757,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009951146500468567,
      "loss": 2.1095,
      "step": 7681
    },
    {
      "epoch": 0.05219417053555163,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000995113149661919,
      "loss": 2.4269,
      "step": 7682
    },
    {
      "epoch": 0.052200964882145685,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00099511164904775,
      "loss": 2.2664,
      "step": 7683
    },
    {
      "epoch": 0.05220775922873975,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009951101482043494,
      "loss": 2.4413,
      "step": 7684
    },
    {
      "epoch": 0.052214553575333805,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009951086471317188,
      "loss": 2.3055,
      "step": 7685
    },
    {
      "epoch": 0.05222134792192786,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009951071458298585,
      "loss": 2.1138,
      "step": 7686
    },
    {
      "epoch": 0.05222814226852192,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009951056442987693,
      "loss": 2.2477,
      "step": 7687
    },
    {
      "epoch": 0.052234936615115975,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009951041425384517,
      "loss": 2.2406,
      "step": 7688
    },
    {
      "epoch": 0.05224173096171004,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009951026405489068,
      "loss": 2.3537,
      "step": 7689
    },
    {
      "epoch": 0.052248525308304095,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009951011383301349,
      "loss": 2.2607,
      "step": 7690
    },
    {
      "epoch": 0.05225531965489815,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009950996358821368,
      "loss": 2.464,
      "step": 7691
    },
    {
      "epoch": 0.05226211400149221,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009950981332049133,
      "loss": 2.3856,
      "step": 7692
    },
    {
      "epoch": 0.052268908348086264,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009950966302984651,
      "loss": 2.2883,
      "step": 7693
    },
    {
      "epoch": 0.05227570269468032,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000995095127162793,
      "loss": 2.3168,
      "step": 7694
    },
    {
      "epoch": 0.052282497041274384,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009950936237978973,
      "loss": 2.3291,
      "step": 7695
    },
    {
      "epoch": 0.05228929138786844,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995092120203779,
      "loss": 2.3279,
      "step": 7696
    },
    {
      "epoch": 0.0522960857344625,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009950906163804388,
      "loss": 2.2731,
      "step": 7697
    },
    {
      "epoch": 0.05230288008105655,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995089112327877,
      "loss": 2.3465,
      "step": 7698
    },
    {
      "epoch": 0.05230967442765061,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009950876080460948,
      "loss": 2.2162,
      "step": 7699
    },
    {
      "epoch": 0.05231646877424467,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009950861035350929,
      "loss": 2.2989,
      "step": 7700
    },
    {
      "epoch": 0.05232326312083873,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009950845987948716,
      "loss": 2.3171,
      "step": 7701
    },
    {
      "epoch": 0.052330057467432786,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000995083093825432,
      "loss": 2.3288,
      "step": 7702
    },
    {
      "epoch": 0.05233685181402684,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009950815886267744,
      "loss": 2.3113,
      "step": 7703
    },
    {
      "epoch": 0.0523436461606209,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009950800831988998,
      "loss": 2.133,
      "step": 7704
    },
    {
      "epoch": 0.05235044050721496,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009950785775418088,
      "loss": 2.2576,
      "step": 7705
    },
    {
      "epoch": 0.05235723485380902,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000995077071655502,
      "loss": 2.4327,
      "step": 7706
    },
    {
      "epoch": 0.052364029200403076,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009950755655399803,
      "loss": 2.2559,
      "step": 7707
    },
    {
      "epoch": 0.05237082354699713,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009950740591952443,
      "loss": 2.3694,
      "step": 7708
    },
    {
      "epoch": 0.05237761789359119,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009950725526212945,
      "loss": 2.0904,
      "step": 7709
    },
    {
      "epoch": 0.052384412240185245,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000995071045818132,
      "loss": 2.3087,
      "step": 7710
    },
    {
      "epoch": 0.05239120658677931,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009950695387857572,
      "loss": 2.3499,
      "step": 7711
    },
    {
      "epoch": 0.052398000933373365,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009950680315241709,
      "loss": 2.3223,
      "step": 7712
    },
    {
      "epoch": 0.05240479527996742,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009950665240333735,
      "loss": 2.332,
      "step": 7713
    },
    {
      "epoch": 0.05241158962656148,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009950650163133663,
      "loss": 2.2525,
      "step": 7714
    },
    {
      "epoch": 0.052418383973155534,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009950635083641496,
      "loss": 2.2136,
      "step": 7715
    },
    {
      "epoch": 0.0524251783197496,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009950620001857242,
      "loss": 2.2819,
      "step": 7716
    },
    {
      "epoch": 0.052431972666343654,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009950604917780908,
      "loss": 2.3742,
      "step": 7717
    },
    {
      "epoch": 0.05243876701293771,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009950589831412499,
      "loss": 2.1767,
      "step": 7718
    },
    {
      "epoch": 0.05244556135953177,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009950574742752026,
      "loss": 2.3861,
      "step": 7719
    },
    {
      "epoch": 0.052452355706125824,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009950559651799493,
      "loss": 2.3066,
      "step": 7720
    },
    {
      "epoch": 0.05245915005271988,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009950544558554905,
      "loss": 2.1668,
      "step": 7721
    },
    {
      "epoch": 0.052465944399313943,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009950529463018274,
      "loss": 2.3488,
      "step": 7722
    },
    {
      "epoch": 0.052472738745908,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009950514365189606,
      "loss": 2.4061,
      "step": 7723
    },
    {
      "epoch": 0.052479533092502056,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009950499265068903,
      "loss": 2.2367,
      "step": 7724
    },
    {
      "epoch": 0.05248632743909611,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000995048416265618,
      "loss": 2.2133,
      "step": 7725
    },
    {
      "epoch": 0.05249312178569017,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009950469057951436,
      "loss": 2.2653,
      "step": 7726
    },
    {
      "epoch": 0.05249991613228423,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009950453950954682,
      "loss": 2.2927,
      "step": 7727
    },
    {
      "epoch": 0.05250671047887829,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009950438841665925,
      "loss": 2.2392,
      "step": 7728
    },
    {
      "epoch": 0.052513504825472346,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000995042373008517,
      "loss": 2.2355,
      "step": 7729
    },
    {
      "epoch": 0.0525202991720664,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009950408616212429,
      "loss": 2.2565,
      "step": 7730
    },
    {
      "epoch": 0.05252709351866046,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009950393500047705,
      "loss": 2.3126,
      "step": 7731
    },
    {
      "epoch": 0.052533887865254515,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009950378381591003,
      "loss": 2.3318,
      "step": 7732
    },
    {
      "epoch": 0.05254068221184858,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009950363260842335,
      "loss": 2.3699,
      "step": 7733
    },
    {
      "epoch": 0.052547476558442635,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009950348137801705,
      "loss": 2.3076,
      "step": 7734
    },
    {
      "epoch": 0.05255427090503669,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000995033301246912,
      "loss": 2.4966,
      "step": 7735
    },
    {
      "epoch": 0.05256106525163075,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009950317884844589,
      "loss": 2.375,
      "step": 7736
    },
    {
      "epoch": 0.052567859598224805,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009950302754928116,
      "loss": 2.2768,
      "step": 7737
    },
    {
      "epoch": 0.05257465394481887,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009950287622719708,
      "loss": 2.3727,
      "step": 7738
    },
    {
      "epoch": 0.052581448291412924,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009950272488219376,
      "loss": 2.4447,
      "step": 7739
    },
    {
      "epoch": 0.05258824263800698,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009950257351427126,
      "loss": 2.4082,
      "step": 7740
    },
    {
      "epoch": 0.05259503698460104,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000995024221234296,
      "loss": 2.364,
      "step": 7741
    },
    {
      "epoch": 0.052601831331195094,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000995022707096689,
      "loss": 2.3403,
      "step": 7742
    },
    {
      "epoch": 0.05260862567778915,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009950211927298922,
      "loss": 2.3985,
      "step": 7743
    },
    {
      "epoch": 0.052615420024383214,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009950196781339062,
      "loss": 2.2206,
      "step": 7744
    },
    {
      "epoch": 0.05262221437097727,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009950181633087318,
      "loss": 2.2475,
      "step": 7745
    },
    {
      "epoch": 0.05262900871757133,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00099501664825437,
      "loss": 2.1802,
      "step": 7746
    },
    {
      "epoch": 0.05263580306416538,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009950151329708207,
      "loss": 2.2743,
      "step": 7747
    },
    {
      "epoch": 0.05264259741075944,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009950136174580852,
      "loss": 2.2515,
      "step": 7748
    },
    {
      "epoch": 0.0526493917573535,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009950121017161641,
      "loss": 2.2151,
      "step": 7749
    },
    {
      "epoch": 0.05265618610394756,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000995010585745058,
      "loss": 2.2381,
      "step": 7750
    },
    {
      "epoch": 0.052662980450541616,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009950090695447678,
      "loss": 2.2513,
      "step": 7751
    },
    {
      "epoch": 0.05266977479713567,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000995007553115294,
      "loss": 2.2274,
      "step": 7752
    },
    {
      "epoch": 0.05267656914372973,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009950060364566373,
      "loss": 2.2545,
      "step": 7753
    },
    {
      "epoch": 0.052683363490323785,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009950045195687986,
      "loss": 2.4023,
      "step": 7754
    },
    {
      "epoch": 0.05269015783691785,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009950030024517785,
      "loss": 2.243,
      "step": 7755
    },
    {
      "epoch": 0.052696952183511905,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009950014851055776,
      "loss": 2.3213,
      "step": 7756
    },
    {
      "epoch": 0.05270374653010596,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009949999675301967,
      "loss": 2.365,
      "step": 7757
    },
    {
      "epoch": 0.05271054087670002,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009949984497256365,
      "loss": 2.2949,
      "step": 7758
    },
    {
      "epoch": 0.052717335223294075,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009949969316918978,
      "loss": 2.2805,
      "step": 7759
    },
    {
      "epoch": 0.05272412956988814,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994995413428981,
      "loss": 2.1171,
      "step": 7760
    },
    {
      "epoch": 0.052730923916482195,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994993894936887,
      "loss": 2.2659,
      "step": 7761
    },
    {
      "epoch": 0.05273771826307625,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009949923762156167,
      "loss": 2.3362,
      "step": 7762
    },
    {
      "epoch": 0.05274451260967031,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009949908572651706,
      "loss": 2.3167,
      "step": 7763
    },
    {
      "epoch": 0.052751306956264364,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009949893380855493,
      "loss": 2.2868,
      "step": 7764
    },
    {
      "epoch": 0.05275810130285843,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009949878186767537,
      "loss": 2.1984,
      "step": 7765
    },
    {
      "epoch": 0.052764895649452484,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009949862990387842,
      "loss": 2.15,
      "step": 7766
    },
    {
      "epoch": 0.05277168999604654,
      "grad_norm": 0.171875,
      "learning_rate": 0.000994984779171642,
      "loss": 2.3072,
      "step": 7767
    },
    {
      "epoch": 0.0527784843426406,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009949832590753272,
      "loss": 2.0526,
      "step": 7768
    },
    {
      "epoch": 0.05278527868923465,
      "grad_norm": 0.15625,
      "learning_rate": 0.000994981738749841,
      "loss": 2.3222,
      "step": 7769
    },
    {
      "epoch": 0.05279207303582871,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000994980218195184,
      "loss": 2.2593,
      "step": 7770
    },
    {
      "epoch": 0.05279886738242277,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009949786974113568,
      "loss": 2.2514,
      "step": 7771
    },
    {
      "epoch": 0.05280566172901683,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00099497717639836,
      "loss": 2.4215,
      "step": 7772
    },
    {
      "epoch": 0.052812456075610886,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009949756551561946,
      "loss": 2.3662,
      "step": 7773
    },
    {
      "epoch": 0.05281925042220494,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009949741336848612,
      "loss": 2.4789,
      "step": 7774
    },
    {
      "epoch": 0.052826044768799,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009949726119843603,
      "loss": 2.286,
      "step": 7775
    },
    {
      "epoch": 0.05283283911539306,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009949710900546927,
      "loss": 2.3076,
      "step": 7776
    },
    {
      "epoch": 0.05283963346198712,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009949695678958593,
      "loss": 2.4089,
      "step": 7777
    },
    {
      "epoch": 0.052846427808581176,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009949680455078607,
      "loss": 2.2017,
      "step": 7778
    },
    {
      "epoch": 0.05285322215517523,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009949665228906974,
      "loss": 2.4025,
      "step": 7779
    },
    {
      "epoch": 0.05286001650176929,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009949650000443703,
      "loss": 2.155,
      "step": 7780
    },
    {
      "epoch": 0.052866810848363345,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009949634769688802,
      "loss": 2.3063,
      "step": 7781
    },
    {
      "epoch": 0.05287360519495741,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009949619536642276,
      "loss": 2.2799,
      "step": 7782
    },
    {
      "epoch": 0.052880399541551465,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009949604301304133,
      "loss": 2.2106,
      "step": 7783
    },
    {
      "epoch": 0.05288719388814552,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009949589063674381,
      "loss": 2.2189,
      "step": 7784
    },
    {
      "epoch": 0.05289398823473958,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009949573823753027,
      "loss": 2.3043,
      "step": 7785
    },
    {
      "epoch": 0.052900782581333634,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009949558581540075,
      "loss": 2.1783,
      "step": 7786
    },
    {
      "epoch": 0.0529075769279277,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009949543337035534,
      "loss": 2.3431,
      "step": 7787
    },
    {
      "epoch": 0.052914371274521754,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994952809023941,
      "loss": 2.0894,
      "step": 7788
    },
    {
      "epoch": 0.05292116562111581,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009949512841151713,
      "loss": 2.1842,
      "step": 7789
    },
    {
      "epoch": 0.05292795996770987,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994949758977245,
      "loss": 2.3226,
      "step": 7790
    },
    {
      "epoch": 0.052934754314303924,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009949482336101623,
      "loss": 2.1799,
      "step": 7791
    },
    {
      "epoch": 0.05294154866089798,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009949467080139244,
      "loss": 2.3101,
      "step": 7792
    },
    {
      "epoch": 0.052948343007492044,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009949451821885317,
      "loss": 2.1292,
      "step": 7793
    },
    {
      "epoch": 0.0529551373540861,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009949436561339853,
      "loss": 2.2754,
      "step": 7794
    },
    {
      "epoch": 0.05296193170068016,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009949421298502857,
      "loss": 2.3002,
      "step": 7795
    },
    {
      "epoch": 0.05296872604727421,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009949406033374333,
      "loss": 2.3099,
      "step": 7796
    },
    {
      "epoch": 0.05297552039386827,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009949390765954292,
      "loss": 2.2679,
      "step": 7797
    },
    {
      "epoch": 0.05298231474046233,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994937549624274,
      "loss": 2.3221,
      "step": 7798
    },
    {
      "epoch": 0.05298910908705639,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009949360224239684,
      "loss": 2.4895,
      "step": 7799
    },
    {
      "epoch": 0.052995903433650446,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009949344949945129,
      "loss": 2.2082,
      "step": 7800
    },
    {
      "epoch": 0.0530026977802445,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009949329673359087,
      "loss": 2.2504,
      "step": 7801
    },
    {
      "epoch": 0.05300949212683856,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009949314394481561,
      "loss": 2.3163,
      "step": 7802
    },
    {
      "epoch": 0.053016286473432615,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994929911331256,
      "loss": 2.2567,
      "step": 7803
    },
    {
      "epoch": 0.05302308082002668,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009949283829852086,
      "loss": 2.3427,
      "step": 7804
    },
    {
      "epoch": 0.053029875166620735,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009949268544100155,
      "loss": 2.3049,
      "step": 7805
    },
    {
      "epoch": 0.05303666951321479,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009949253256056768,
      "loss": 2.1817,
      "step": 7806
    },
    {
      "epoch": 0.05304346385980885,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009949237965721934,
      "loss": 2.4037,
      "step": 7807
    },
    {
      "epoch": 0.053050258206402905,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009949222673095658,
      "loss": 2.3421,
      "step": 7808
    },
    {
      "epoch": 0.05305705255299697,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000994920737817795,
      "loss": 2.5197,
      "step": 7809
    },
    {
      "epoch": 0.053063846899591025,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009949192080968814,
      "loss": 2.3185,
      "step": 7810
    },
    {
      "epoch": 0.05307064124618508,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009949176781468261,
      "loss": 2.3495,
      "step": 7811
    },
    {
      "epoch": 0.05307743559277914,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009949161479676297,
      "loss": 2.3269,
      "step": 7812
    },
    {
      "epoch": 0.053084229939373194,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009949146175592925,
      "loss": 2.2922,
      "step": 7813
    },
    {
      "epoch": 0.05309102428596725,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009949130869218155,
      "loss": 2.3438,
      "step": 7814
    },
    {
      "epoch": 0.053097818632561314,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009949115560551995,
      "loss": 2.3817,
      "step": 7815
    },
    {
      "epoch": 0.05310461297915537,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009949100249594452,
      "loss": 2.2661,
      "step": 7816
    },
    {
      "epoch": 0.05311140732574943,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009949084936345532,
      "loss": 2.3321,
      "step": 7817
    },
    {
      "epoch": 0.05311820167234348,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009949069620805242,
      "loss": 2.32,
      "step": 7818
    },
    {
      "epoch": 0.05312499601893754,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994905430297359,
      "loss": 2.2014,
      "step": 7819
    },
    {
      "epoch": 0.0531317903655316,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009949038982850582,
      "loss": 2.302,
      "step": 7820
    },
    {
      "epoch": 0.05313858471212566,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009949023660436227,
      "loss": 2.1989,
      "step": 7821
    },
    {
      "epoch": 0.053145379058719716,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994900833573053,
      "loss": 2.4016,
      "step": 7822
    },
    {
      "epoch": 0.05315217340531377,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009948993008733497,
      "loss": 2.1355,
      "step": 7823
    },
    {
      "epoch": 0.05315896775190783,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994897767944514,
      "loss": 2.2175,
      "step": 7824
    },
    {
      "epoch": 0.05316576209850189,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000994896234786546,
      "loss": 2.24,
      "step": 7825
    },
    {
      "epoch": 0.05317255644509595,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009948947013994468,
      "loss": 2.4228,
      "step": 7826
    },
    {
      "epoch": 0.053179350791690005,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009948931677832172,
      "loss": 2.3647,
      "step": 7827
    },
    {
      "epoch": 0.05318614513828406,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009948916339378579,
      "loss": 2.2742,
      "step": 7828
    },
    {
      "epoch": 0.05319293948487812,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009948900998633689,
      "loss": 2.2054,
      "step": 7829
    },
    {
      "epoch": 0.053199733831472175,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000994888565559752,
      "loss": 2.3629,
      "step": 7830
    },
    {
      "epoch": 0.05320652817806624,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994887031027007,
      "loss": 2.3373,
      "step": 7831
    },
    {
      "epoch": 0.053213322524660295,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994885496265135,
      "loss": 2.3626,
      "step": 7832
    },
    {
      "epoch": 0.05322011687125435,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000994883961274137,
      "loss": 2.3186,
      "step": 7833
    },
    {
      "epoch": 0.05322691121784841,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000994882426054013,
      "loss": 2.3095,
      "step": 7834
    },
    {
      "epoch": 0.053233705564442464,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009948808906047645,
      "loss": 2.3145,
      "step": 7835
    },
    {
      "epoch": 0.05324049991103653,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009948793549263914,
      "loss": 2.2381,
      "step": 7836
    },
    {
      "epoch": 0.053247294257630584,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009948778190188952,
      "loss": 2.431,
      "step": 7837
    },
    {
      "epoch": 0.05325408860422464,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994876282882276,
      "loss": 2.1498,
      "step": 7838
    },
    {
      "epoch": 0.0532608829508187,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009948747465165348,
      "loss": 2.2322,
      "step": 7839
    },
    {
      "epoch": 0.053267677297412754,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009948732099216723,
      "loss": 2.3358,
      "step": 7840
    },
    {
      "epoch": 0.05327447164400681,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009948716730976892,
      "loss": 2.1155,
      "step": 7841
    },
    {
      "epoch": 0.05328126599060087,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000994870136044586,
      "loss": 2.2123,
      "step": 7842
    },
    {
      "epoch": 0.05328806033719493,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994868598762364,
      "loss": 2.2329,
      "step": 7843
    },
    {
      "epoch": 0.053294854683788986,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009948670612510233,
      "loss": 2.2658,
      "step": 7844
    },
    {
      "epoch": 0.05330164903038304,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009948655235105647,
      "loss": 2.2186,
      "step": 7845
    },
    {
      "epoch": 0.0533084433769771,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009948639855409894,
      "loss": 2.3909,
      "step": 7846
    },
    {
      "epoch": 0.05331523772357116,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009948624473422973,
      "loss": 2.1751,
      "step": 7847
    },
    {
      "epoch": 0.05332203207016522,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009948609089144899,
      "loss": 2.3042,
      "step": 7848
    },
    {
      "epoch": 0.053328826416759276,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009948593702575675,
      "loss": 2.3587,
      "step": 7849
    },
    {
      "epoch": 0.05333562076335333,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009948578313715307,
      "loss": 2.427,
      "step": 7850
    },
    {
      "epoch": 0.05334241510994739,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009948562922563807,
      "loss": 2.4226,
      "step": 7851
    },
    {
      "epoch": 0.053349209456541445,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009948547529121175,
      "loss": 2.3042,
      "step": 7852
    },
    {
      "epoch": 0.05335600380313551,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009948532133387425,
      "loss": 2.5056,
      "step": 7853
    },
    {
      "epoch": 0.053362798149729565,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000994851673536256,
      "loss": 2.2341,
      "step": 7854
    },
    {
      "epoch": 0.05336959249632362,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994850133504659,
      "loss": 2.3153,
      "step": 7855
    },
    {
      "epoch": 0.05337638684291768,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000994848593243952,
      "loss": 2.306,
      "step": 7856
    },
    {
      "epoch": 0.053383181189511734,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009948470527541358,
      "loss": 2.1944,
      "step": 7857
    },
    {
      "epoch": 0.0533899755361058,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009948455120352112,
      "loss": 2.4583,
      "step": 7858
    },
    {
      "epoch": 0.053396769882699854,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009948439710871786,
      "loss": 2.2957,
      "step": 7859
    },
    {
      "epoch": 0.05340356422929391,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000994842429910039,
      "loss": 2.2697,
      "step": 7860
    },
    {
      "epoch": 0.05341035857588797,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000994840888503793,
      "loss": 2.1921,
      "step": 7861
    },
    {
      "epoch": 0.053417152922482024,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009948393468684413,
      "loss": 2.1722,
      "step": 7862
    },
    {
      "epoch": 0.05342394726907608,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009948378050039846,
      "loss": 2.448,
      "step": 7863
    },
    {
      "epoch": 0.053430741615670144,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994836262910424,
      "loss": 2.1457,
      "step": 7864
    },
    {
      "epoch": 0.0534375359622642,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009948347205877594,
      "loss": 2.1942,
      "step": 7865
    },
    {
      "epoch": 0.05344433030885826,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009948331780359923,
      "loss": 2.4457,
      "step": 7866
    },
    {
      "epoch": 0.05345112465545231,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000994831635255123,
      "loss": 2.3072,
      "step": 7867
    },
    {
      "epoch": 0.05345791900204637,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009948300922451524,
      "loss": 2.2905,
      "step": 7868
    },
    {
      "epoch": 0.05346471334864043,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009948285490060812,
      "loss": 2.3489,
      "step": 7869
    },
    {
      "epoch": 0.05347150769523449,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00099482700553791,
      "loss": 2.1481,
      "step": 7870
    },
    {
      "epoch": 0.053478302041828546,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009948254618406395,
      "loss": 2.1864,
      "step": 7871
    },
    {
      "epoch": 0.0534850963884226,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009948239179142706,
      "loss": 2.2824,
      "step": 7872
    },
    {
      "epoch": 0.05349189073501666,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009948223737588039,
      "loss": 2.2385,
      "step": 7873
    },
    {
      "epoch": 0.053498685081610715,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009948208293742399,
      "loss": 2.1838,
      "step": 7874
    },
    {
      "epoch": 0.05350547942820478,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009948192847605798,
      "loss": 2.1753,
      "step": 7875
    },
    {
      "epoch": 0.053512273774798835,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009948177399178237,
      "loss": 2.202,
      "step": 7876
    },
    {
      "epoch": 0.05351906812139289,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994816194845973,
      "loss": 2.3983,
      "step": 7877
    },
    {
      "epoch": 0.05352586246798695,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000994814649545028,
      "loss": 2.2665,
      "step": 7878
    },
    {
      "epoch": 0.053532656814581005,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009948131040149895,
      "loss": 2.3849,
      "step": 7879
    },
    {
      "epoch": 0.05353945116117507,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009948115582558582,
      "loss": 2.2528,
      "step": 7880
    },
    {
      "epoch": 0.053546245507769125,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009948100122676347,
      "loss": 2.3599,
      "step": 7881
    },
    {
      "epoch": 0.05355303985436318,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009948084660503199,
      "loss": 2.3158,
      "step": 7882
    },
    {
      "epoch": 0.05355983420095724,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009948069196039145,
      "loss": 2.2324,
      "step": 7883
    },
    {
      "epoch": 0.053566628547551294,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994805372928419,
      "loss": 2.3062,
      "step": 7884
    },
    {
      "epoch": 0.05357342289414536,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009948038260238345,
      "loss": 2.4876,
      "step": 7885
    },
    {
      "epoch": 0.053580217240739414,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009948022788901613,
      "loss": 2.4568,
      "step": 7886
    },
    {
      "epoch": 0.05358701158733347,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009948007315274004,
      "loss": 2.3556,
      "step": 7887
    },
    {
      "epoch": 0.05359380593392753,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009947991839355525,
      "loss": 2.0933,
      "step": 7888
    },
    {
      "epoch": 0.05360060028052158,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009947976361146182,
      "loss": 2.3441,
      "step": 7889
    },
    {
      "epoch": 0.05360739462711564,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009947960880645984,
      "loss": 2.2831,
      "step": 7890
    },
    {
      "epoch": 0.0536141889737097,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009947945397854934,
      "loss": 2.2553,
      "step": 7891
    },
    {
      "epoch": 0.05362098332030376,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009947929912773046,
      "loss": 2.271,
      "step": 7892
    },
    {
      "epoch": 0.053627777666897816,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994791442540032,
      "loss": 2.1525,
      "step": 7893
    },
    {
      "epoch": 0.05363457201349187,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009947898935736767,
      "loss": 2.2773,
      "step": 7894
    },
    {
      "epoch": 0.05364136636008593,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009947883443782394,
      "loss": 2.225,
      "step": 7895
    },
    {
      "epoch": 0.05364816070667999,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009947867949537208,
      "loss": 2.2449,
      "step": 7896
    },
    {
      "epoch": 0.05365495505327405,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009947852453001213,
      "loss": 2.0981,
      "step": 7897
    },
    {
      "epoch": 0.053661749399868106,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009947836954174423,
      "loss": 2.2799,
      "step": 7898
    },
    {
      "epoch": 0.05366854374646216,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000994782145305684,
      "loss": 2.3203,
      "step": 7899
    },
    {
      "epoch": 0.05367533809305622,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009947805949648472,
      "loss": 2.1532,
      "step": 7900
    },
    {
      "epoch": 0.053682132439650275,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009947790443949325,
      "loss": 2.2046,
      "step": 7901
    },
    {
      "epoch": 0.05368892678624434,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994777493595941,
      "loss": 2.3193,
      "step": 7902
    },
    {
      "epoch": 0.053695721132838395,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994775942567873,
      "loss": 2.1762,
      "step": 7903
    },
    {
      "epoch": 0.05370251547943245,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009947743913107296,
      "loss": 2.3182,
      "step": 7904
    },
    {
      "epoch": 0.05370930982602651,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009947728398245113,
      "loss": 2.2263,
      "step": 7905
    },
    {
      "epoch": 0.053716104172620564,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009947712881092188,
      "loss": 2.1416,
      "step": 7906
    },
    {
      "epoch": 0.05372289851921463,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000994769736164853,
      "loss": 2.4549,
      "step": 7907
    },
    {
      "epoch": 0.053729692865808684,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009947681839914142,
      "loss": 2.2764,
      "step": 7908
    },
    {
      "epoch": 0.05373648721240274,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009947666315889036,
      "loss": 2.4501,
      "step": 7909
    },
    {
      "epoch": 0.0537432815589968,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009947650789573217,
      "loss": 2.2724,
      "step": 7910
    },
    {
      "epoch": 0.053750075905590854,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009947635260966691,
      "loss": 2.3336,
      "step": 7911
    },
    {
      "epoch": 0.05375687025218491,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009947619730069468,
      "loss": 2.465,
      "step": 7912
    },
    {
      "epoch": 0.053763664598778974,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009947604196881555,
      "loss": 2.4255,
      "step": 7913
    },
    {
      "epoch": 0.05377045894537303,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009947588661402954,
      "loss": 2.2344,
      "step": 7914
    },
    {
      "epoch": 0.053777253291967086,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994757312363368,
      "loss": 2.3863,
      "step": 7915
    },
    {
      "epoch": 0.05378404763856114,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009947557583573736,
      "loss": 2.2762,
      "step": 7916
    },
    {
      "epoch": 0.0537908419851552,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009947542041223128,
      "loss": 2.1303,
      "step": 7917
    },
    {
      "epoch": 0.05379763633174926,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009947526496581866,
      "loss": 2.2919,
      "step": 7918
    },
    {
      "epoch": 0.05380443067834332,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009947510949649956,
      "loss": 2.1132,
      "step": 7919
    },
    {
      "epoch": 0.053811225024937376,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009947495400427404,
      "loss": 2.2094,
      "step": 7920
    },
    {
      "epoch": 0.05381801937153143,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994747984891422,
      "loss": 2.2107,
      "step": 7921
    },
    {
      "epoch": 0.05382481371812549,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009947464295110406,
      "loss": 2.3383,
      "step": 7922
    },
    {
      "epoch": 0.053831608064719545,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009947448739015977,
      "loss": 2.134,
      "step": 7923
    },
    {
      "epoch": 0.05383840241131361,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009947433180630934,
      "loss": 2.2415,
      "step": 7924
    },
    {
      "epoch": 0.053845196757907665,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009947417619955285,
      "loss": 2.2726,
      "step": 7925
    },
    {
      "epoch": 0.05385199110450172,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009947402056989041,
      "loss": 2.1254,
      "step": 7926
    },
    {
      "epoch": 0.05385878545109578,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009947386491732204,
      "loss": 2.3924,
      "step": 7927
    },
    {
      "epoch": 0.053865579797689835,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009947370924184786,
      "loss": 2.1563,
      "step": 7928
    },
    {
      "epoch": 0.0538723741442839,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994735535434679,
      "loss": 2.3099,
      "step": 7929
    },
    {
      "epoch": 0.053879168490877954,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009947339782218226,
      "loss": 2.321,
      "step": 7930
    },
    {
      "epoch": 0.05388596283747201,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009947324207799102,
      "loss": 2.2829,
      "step": 7931
    },
    {
      "epoch": 0.05389275718406607,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009947308631089423,
      "loss": 2.3055,
      "step": 7932
    },
    {
      "epoch": 0.053899551530660124,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009947293052089195,
      "loss": 2.1939,
      "step": 7933
    },
    {
      "epoch": 0.05390634587725418,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009947277470798428,
      "loss": 2.3751,
      "step": 7934
    },
    {
      "epoch": 0.053913140223848244,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009947261887217128,
      "loss": 2.4578,
      "step": 7935
    },
    {
      "epoch": 0.0539199345704423,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009947246301345303,
      "loss": 2.3729,
      "step": 7936
    },
    {
      "epoch": 0.05392672891703636,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000994723071318296,
      "loss": 2.4035,
      "step": 7937
    },
    {
      "epoch": 0.05393352326363041,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009947215122730105,
      "loss": 2.0689,
      "step": 7938
    },
    {
      "epoch": 0.05394031761022447,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009947199529986746,
      "loss": 2.1797,
      "step": 7939
    },
    {
      "epoch": 0.05394711195681853,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994718393495289,
      "loss": 2.2783,
      "step": 7940
    },
    {
      "epoch": 0.05395390630341259,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009947168337628546,
      "loss": 2.416,
      "step": 7941
    },
    {
      "epoch": 0.053960700650006646,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009947152738013717,
      "loss": 2.2074,
      "step": 7942
    },
    {
      "epoch": 0.0539674949966007,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009947137136108418,
      "loss": 2.3665,
      "step": 7943
    },
    {
      "epoch": 0.05397428934319476,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009947121531912648,
      "loss": 2.1204,
      "step": 7944
    },
    {
      "epoch": 0.05398108368978882,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009947105925426417,
      "loss": 2.248,
      "step": 7945
    },
    {
      "epoch": 0.05398787803638288,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009947090316649735,
      "loss": 2.2385,
      "step": 7946
    },
    {
      "epoch": 0.053994672382976935,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009947074705582604,
      "loss": 2.3484,
      "step": 7947
    },
    {
      "epoch": 0.05400146672957099,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009947059092225035,
      "loss": 2.3887,
      "step": 7948
    },
    {
      "epoch": 0.05400826107616505,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009947043476577035,
      "loss": 2.2612,
      "step": 7949
    },
    {
      "epoch": 0.054015055422759105,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994702785863861,
      "loss": 2.2688,
      "step": 7950
    },
    {
      "epoch": 0.05402184976935317,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000994701223840977,
      "loss": 2.2677,
      "step": 7951
    },
    {
      "epoch": 0.054028644115947225,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009946996615890517,
      "loss": 2.3784,
      "step": 7952
    },
    {
      "epoch": 0.05403543846254128,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009946980991080863,
      "loss": 2.3245,
      "step": 7953
    },
    {
      "epoch": 0.05404223280913534,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009946965363980813,
      "loss": 2.2644,
      "step": 7954
    },
    {
      "epoch": 0.054049027155729394,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009946949734590375,
      "loss": 2.2759,
      "step": 7955
    },
    {
      "epoch": 0.05405582150232346,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009946934102909556,
      "loss": 2.3536,
      "step": 7956
    },
    {
      "epoch": 0.054062615848917514,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009946918468938361,
      "loss": 2.261,
      "step": 7957
    },
    {
      "epoch": 0.05406941019551157,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009946902832676803,
      "loss": 2.394,
      "step": 7958
    },
    {
      "epoch": 0.05407620454210563,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009946887194124884,
      "loss": 2.2273,
      "step": 7959
    },
    {
      "epoch": 0.054082998888699683,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009946871553282613,
      "loss": 2.3206,
      "step": 7960
    },
    {
      "epoch": 0.05408979323529374,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009946855910149999,
      "loss": 2.4569,
      "step": 7961
    },
    {
      "epoch": 0.0540965875818878,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009946840264727045,
      "loss": 2.2116,
      "step": 7962
    },
    {
      "epoch": 0.05410338192848186,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009946824617013761,
      "loss": 2.4616,
      "step": 7963
    },
    {
      "epoch": 0.054110176275075916,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009946808967010155,
      "loss": 2.1487,
      "step": 7964
    },
    {
      "epoch": 0.05411697062166997,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009946793314716232,
      "loss": 2.2863,
      "step": 7965
    },
    {
      "epoch": 0.05412376496826403,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009946777660132,
      "loss": 2.2487,
      "step": 7966
    },
    {
      "epoch": 0.05413055931485809,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009946762003257468,
      "loss": 2.1666,
      "step": 7967
    },
    {
      "epoch": 0.05413735366145215,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009946746344092641,
      "loss": 2.3576,
      "step": 7968
    },
    {
      "epoch": 0.054144148008046206,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000994673068263753,
      "loss": 2.3792,
      "step": 7969
    },
    {
      "epoch": 0.05415094235464026,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009946715018892137,
      "loss": 2.2388,
      "step": 7970
    },
    {
      "epoch": 0.05415773670123432,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994669935285647,
      "loss": 2.5012,
      "step": 7971
    },
    {
      "epoch": 0.054164531047828375,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000994668368453054,
      "loss": 2.3282,
      "step": 7972
    },
    {
      "epoch": 0.05417132539442244,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009946668013914352,
      "loss": 2.3035,
      "step": 7973
    },
    {
      "epoch": 0.054178119741016495,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009946652341007913,
      "loss": 2.1948,
      "step": 7974
    },
    {
      "epoch": 0.05418491408761055,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994663666581123,
      "loss": 2.3148,
      "step": 7975
    },
    {
      "epoch": 0.05419170843420461,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009946620988324313,
      "loss": 2.3504,
      "step": 7976
    },
    {
      "epoch": 0.054198502780798664,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009946605308547166,
      "loss": 2.2529,
      "step": 7977
    },
    {
      "epoch": 0.05420529712739273,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009946589626479796,
      "loss": 2.1928,
      "step": 7978
    },
    {
      "epoch": 0.054212091473986784,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009946573942122214,
      "loss": 2.3476,
      "step": 7979
    },
    {
      "epoch": 0.05421888582058084,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009946558255474426,
      "loss": 2.2274,
      "step": 7980
    },
    {
      "epoch": 0.0542256801671749,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009946542566536434,
      "loss": 2.3885,
      "step": 7981
    },
    {
      "epoch": 0.054232474513768954,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009946526875308252,
      "loss": 2.1809,
      "step": 7982
    },
    {
      "epoch": 0.05423926886036301,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009946511181789886,
      "loss": 2.3206,
      "step": 7983
    },
    {
      "epoch": 0.054246063206957074,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000994649548598134,
      "loss": 2.3084,
      "step": 7984
    },
    {
      "epoch": 0.05425285755355113,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009946479787882622,
      "loss": 2.4327,
      "step": 7985
    },
    {
      "epoch": 0.05425965190014519,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009946464087493745,
      "loss": 2.2995,
      "step": 7986
    },
    {
      "epoch": 0.05426644624673924,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009946448384814707,
      "loss": 2.2968,
      "step": 7987
    },
    {
      "epoch": 0.0542732405933333,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009946432679845524,
      "loss": 2.397,
      "step": 7988
    },
    {
      "epoch": 0.05428003493992736,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009946416972586199,
      "loss": 2.2783,
      "step": 7989
    },
    {
      "epoch": 0.05428682928652142,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009946401263036737,
      "loss": 2.0974,
      "step": 7990
    },
    {
      "epoch": 0.054293623633115476,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009946385551197151,
      "loss": 2.2533,
      "step": 7991
    },
    {
      "epoch": 0.05430041797970953,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009946369837067442,
      "loss": 2.2407,
      "step": 7992
    },
    {
      "epoch": 0.05430721232630359,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009946354120647623,
      "loss": 2.1928,
      "step": 7993
    },
    {
      "epoch": 0.054314006672897645,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009946338401937698,
      "loss": 2.3982,
      "step": 7994
    },
    {
      "epoch": 0.05432080101949171,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009946322680937673,
      "loss": 2.1744,
      "step": 7995
    },
    {
      "epoch": 0.054327595366085765,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000994630695764756,
      "loss": 2.2575,
      "step": 7996
    },
    {
      "epoch": 0.05433438971267982,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009946291232067364,
      "loss": 2.3875,
      "step": 7997
    },
    {
      "epoch": 0.05434118405927388,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000994627550419709,
      "loss": 2.2213,
      "step": 7998
    },
    {
      "epoch": 0.054347978405867935,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009946259774036748,
      "loss": 2.2461,
      "step": 7999
    },
    {
      "epoch": 0.054354772752462,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009946244041586345,
      "loss": 2.2176,
      "step": 8000
    },
    {
      "epoch": 0.054361567099056055,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009946228306845885,
      "loss": 2.2004,
      "step": 8001
    },
    {
      "epoch": 0.05436836144565011,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000994621256981538,
      "loss": 2.1454,
      "step": 8002
    },
    {
      "epoch": 0.05437515579224417,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009946196830494835,
      "loss": 2.3622,
      "step": 8003
    },
    {
      "epoch": 0.054381950138838224,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000994618108888426,
      "loss": 2.3387,
      "step": 8004
    },
    {
      "epoch": 0.05438874448543229,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009946165344983657,
      "loss": 2.2077,
      "step": 8005
    },
    {
      "epoch": 0.054395538832026344,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009946149598793036,
      "loss": 2.1932,
      "step": 8006
    },
    {
      "epoch": 0.0544023331786204,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009946133850312405,
      "loss": 2.4078,
      "step": 8007
    },
    {
      "epoch": 0.05440912752521446,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009946118099541771,
      "loss": 2.1179,
      "step": 8008
    },
    {
      "epoch": 0.05441592187180851,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009946102346481142,
      "loss": 2.2053,
      "step": 8009
    },
    {
      "epoch": 0.05442271621840257,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009946086591130522,
      "loss": 2.2409,
      "step": 8010
    },
    {
      "epoch": 0.05442951056499663,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009946070833489922,
      "loss": 2.276,
      "step": 8011
    },
    {
      "epoch": 0.05443630491159069,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009946055073559348,
      "loss": 2.2322,
      "step": 8012
    },
    {
      "epoch": 0.054443099258184746,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009946039311338809,
      "loss": 2.1407,
      "step": 8013
    },
    {
      "epoch": 0.0544498936047788,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009946023546828307,
      "loss": 2.3493,
      "step": 8014
    },
    {
      "epoch": 0.05445668795137286,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009946007780027855,
      "loss": 2.4839,
      "step": 8015
    },
    {
      "epoch": 0.05446348229796692,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009945992010937457,
      "loss": 2.0453,
      "step": 8016
    },
    {
      "epoch": 0.05447027664456098,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009945976239557122,
      "loss": 2.285,
      "step": 8017
    },
    {
      "epoch": 0.054477070991155035,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009945960465886855,
      "loss": 2.255,
      "step": 8018
    },
    {
      "epoch": 0.05448386533774909,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009945944689926669,
      "loss": 2.2188,
      "step": 8019
    },
    {
      "epoch": 0.05449065968434315,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009945928911676565,
      "loss": 2.2531,
      "step": 8020
    },
    {
      "epoch": 0.054497454030937205,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009945913131136552,
      "loss": 2.2714,
      "step": 8021
    },
    {
      "epoch": 0.05450424837753127,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994589734830664,
      "loss": 2.2487,
      "step": 8022
    },
    {
      "epoch": 0.054511042724125325,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000994588156318683,
      "loss": 2.1639,
      "step": 8023
    },
    {
      "epoch": 0.05451783707071938,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009945865775777137,
      "loss": 2.3708,
      "step": 8024
    },
    {
      "epoch": 0.05452463141731344,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009945849986077563,
      "loss": 2.3628,
      "step": 8025
    },
    {
      "epoch": 0.054531425763907494,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000994583419408812,
      "loss": 2.1847,
      "step": 8026
    },
    {
      "epoch": 0.05453822011050156,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994581839980881,
      "loss": 2.2196,
      "step": 8027
    },
    {
      "epoch": 0.054545014457095614,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009945802603239643,
      "loss": 2.2711,
      "step": 8028
    },
    {
      "epoch": 0.05455180880368967,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009945786804380628,
      "loss": 2.3786,
      "step": 8029
    },
    {
      "epoch": 0.05455860315028373,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009945771003231768,
      "loss": 2.3488,
      "step": 8030
    },
    {
      "epoch": 0.054565397496877784,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009945755199793074,
      "loss": 2.1153,
      "step": 8031
    },
    {
      "epoch": 0.05457219184347184,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000994573939406455,
      "loss": 2.1262,
      "step": 8032
    },
    {
      "epoch": 0.0545789861900659,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009945723586046208,
      "loss": 2.1907,
      "step": 8033
    },
    {
      "epoch": 0.05458578053665996,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994570777573805,
      "loss": 2.2338,
      "step": 8034
    },
    {
      "epoch": 0.054592574883254016,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994569196314009,
      "loss": 2.3549,
      "step": 8035
    },
    {
      "epoch": 0.05459936922984807,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009945676148252328,
      "loss": 2.3304,
      "step": 8036
    },
    {
      "epoch": 0.05460616357644213,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009945660331074776,
      "loss": 2.3568,
      "step": 8037
    },
    {
      "epoch": 0.05461295792303619,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994564451160744,
      "loss": 2.309,
      "step": 8038
    },
    {
      "epoch": 0.05461975226963025,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009945628689850326,
      "loss": 2.3289,
      "step": 8039
    },
    {
      "epoch": 0.054626546616224306,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009945612865803445,
      "loss": 2.357,
      "step": 8040
    },
    {
      "epoch": 0.05463334096281836,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009945597039466801,
      "loss": 2.1349,
      "step": 8041
    },
    {
      "epoch": 0.05464013530941242,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009945581210840401,
      "loss": 2.3713,
      "step": 8042
    },
    {
      "epoch": 0.054646929656006475,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009945565379924255,
      "loss": 2.2559,
      "step": 8043
    },
    {
      "epoch": 0.05465372400260054,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009945549546718367,
      "loss": 2.2388,
      "step": 8044
    },
    {
      "epoch": 0.054660518349194595,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009945533711222749,
      "loss": 2.2969,
      "step": 8045
    },
    {
      "epoch": 0.05466731269578865,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009945517873437403,
      "loss": 2.2415,
      "step": 8046
    },
    {
      "epoch": 0.05467410704238271,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009945502033362342,
      "loss": 2.2908,
      "step": 8047
    },
    {
      "epoch": 0.054680901388976764,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009945486190997567,
      "loss": 2.3796,
      "step": 8048
    },
    {
      "epoch": 0.05468769573557083,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994547034634309,
      "loss": 2.2724,
      "step": 8049
    },
    {
      "epoch": 0.054694490082164884,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009945454499398917,
      "loss": 2.1195,
      "step": 8050
    },
    {
      "epoch": 0.05470128442875894,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009945438650165055,
      "loss": 2.3099,
      "step": 8051
    },
    {
      "epoch": 0.054708078775353,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009945422798641513,
      "loss": 2.3124,
      "step": 8052
    },
    {
      "epoch": 0.054714873121947054,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009945406944828295,
      "loss": 2.3187,
      "step": 8053
    },
    {
      "epoch": 0.05472166746854111,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994539108872541,
      "loss": 2.2752,
      "step": 8054
    },
    {
      "epoch": 0.054728461815135174,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009945375230332868,
      "loss": 2.3181,
      "step": 8055
    },
    {
      "epoch": 0.05473525616172923,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009945359369650671,
      "loss": 2.1182,
      "step": 8056
    },
    {
      "epoch": 0.05474205050832329,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994534350667883,
      "loss": 2.2027,
      "step": 8057
    },
    {
      "epoch": 0.05474884485491734,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009945327641417354,
      "loss": 2.1751,
      "step": 8058
    },
    {
      "epoch": 0.0547556392015114,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009945311773866246,
      "loss": 2.3102,
      "step": 8059
    },
    {
      "epoch": 0.05476243354810546,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009945295904025516,
      "loss": 2.2643,
      "step": 8060
    },
    {
      "epoch": 0.05476922789469952,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009945280031895172,
      "loss": 2.1705,
      "step": 8061
    },
    {
      "epoch": 0.054776022241293576,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009945264157475218,
      "loss": 2.3775,
      "step": 8062
    },
    {
      "epoch": 0.05478281658788763,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009945248280765664,
      "loss": 2.3255,
      "step": 8063
    },
    {
      "epoch": 0.05478961093448169,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009945232401766515,
      "loss": 2.2123,
      "step": 8064
    },
    {
      "epoch": 0.054796405281075745,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009945216520477781,
      "loss": 2.4007,
      "step": 8065
    },
    {
      "epoch": 0.05480319962766981,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009945200636899472,
      "loss": 2.354,
      "step": 8066
    },
    {
      "epoch": 0.054809993974263865,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009945184751031589,
      "loss": 2.2949,
      "step": 8067
    },
    {
      "epoch": 0.05481678832085792,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000994516886287414,
      "loss": 2.2562,
      "step": 8068
    },
    {
      "epoch": 0.05482358266745198,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009945152972427138,
      "loss": 2.2561,
      "step": 8069
    },
    {
      "epoch": 0.054830377014046035,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009945137079690584,
      "loss": 2.3623,
      "step": 8070
    },
    {
      "epoch": 0.0548371713606401,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009945121184664492,
      "loss": 2.3736,
      "step": 8071
    },
    {
      "epoch": 0.054843965707234155,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009945105287348863,
      "loss": 2.2818,
      "step": 8072
    },
    {
      "epoch": 0.05485076005382821,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009945089387743706,
      "loss": 2.2483,
      "step": 8073
    },
    {
      "epoch": 0.05485755440042227,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009945073485849033,
      "loss": 2.171,
      "step": 8074
    },
    {
      "epoch": 0.054864348747016324,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009945057581664846,
      "loss": 2.2404,
      "step": 8075
    },
    {
      "epoch": 0.05487114309361039,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009945041675191151,
      "loss": 2.3254,
      "step": 8076
    },
    {
      "epoch": 0.054877937440204444,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009945025766427962,
      "loss": 2.2334,
      "step": 8077
    },
    {
      "epoch": 0.0548847317867985,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994500985537528,
      "loss": 2.3949,
      "step": 8078
    },
    {
      "epoch": 0.05489152613339256,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009944993942033117,
      "loss": 2.402,
      "step": 8079
    },
    {
      "epoch": 0.05489832047998661,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000994497802640148,
      "loss": 2.3506,
      "step": 8080
    },
    {
      "epoch": 0.05490511482658067,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994496210848037,
      "loss": 2.3474,
      "step": 8081
    },
    {
      "epoch": 0.05491190917317473,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009944946188269804,
      "loss": 2.2704,
      "step": 8082
    },
    {
      "epoch": 0.05491870351976879,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009944930265769782,
      "loss": 2.2132,
      "step": 8083
    },
    {
      "epoch": 0.054925497866362846,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009944914340980315,
      "loss": 2.204,
      "step": 8084
    },
    {
      "epoch": 0.0549322922129569,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000994489841390141,
      "loss": 2.2867,
      "step": 8085
    },
    {
      "epoch": 0.05493908655955096,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009944882484533074,
      "loss": 2.2011,
      "step": 8086
    },
    {
      "epoch": 0.05494588090614502,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009944866552875312,
      "loss": 2.2608,
      "step": 8087
    },
    {
      "epoch": 0.05495267525273908,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009944850618928134,
      "loss": 2.4007,
      "step": 8088
    },
    {
      "epoch": 0.054959469599333136,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009944834682691548,
      "loss": 2.2889,
      "step": 8089
    },
    {
      "epoch": 0.05496626394592719,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009944818744165558,
      "loss": 2.1038,
      "step": 8090
    },
    {
      "epoch": 0.05497305829252125,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009944802803350176,
      "loss": 2.4675,
      "step": 8091
    },
    {
      "epoch": 0.054979852639115305,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009944786860245406,
      "loss": 2.308,
      "step": 8092
    },
    {
      "epoch": 0.05498664698570937,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009944770914851256,
      "loss": 2.3346,
      "step": 8093
    },
    {
      "epoch": 0.054993441332303425,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009944754967167735,
      "loss": 2.312,
      "step": 8094
    },
    {
      "epoch": 0.05500023567889748,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009944739017194847,
      "loss": 2.2871,
      "step": 8095
    },
    {
      "epoch": 0.05500703002549154,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009944723064932604,
      "loss": 2.2017,
      "step": 8096
    },
    {
      "epoch": 0.055013824372085594,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009944707110381011,
      "loss": 2.2552,
      "step": 8097
    },
    {
      "epoch": 0.05502061871867966,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009944691153540074,
      "loss": 2.355,
      "step": 8098
    },
    {
      "epoch": 0.055027413065273714,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009944675194409802,
      "loss": 2.3159,
      "step": 8099
    },
    {
      "epoch": 0.05503420741186777,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00099446592329902,
      "loss": 2.2487,
      "step": 8100
    },
    {
      "epoch": 0.05504100175846183,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000994464326928128,
      "loss": 2.2475,
      "step": 8101
    },
    {
      "epoch": 0.055047796105055884,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009944627303283048,
      "loss": 2.2875,
      "step": 8102
    },
    {
      "epoch": 0.05505459045164994,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000994461133499551,
      "loss": 2.2,
      "step": 8103
    },
    {
      "epoch": 0.055061384798244004,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994459536441867,
      "loss": 2.2866,
      "step": 8104
    },
    {
      "epoch": 0.05506817914483806,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000994457939155254,
      "loss": 2.1513,
      "step": 8105
    },
    {
      "epoch": 0.055074973491432117,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009944563416397127,
      "loss": 2.3502,
      "step": 8106
    },
    {
      "epoch": 0.05508176783802617,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994454743895244,
      "loss": 2.1872,
      "step": 8107
    },
    {
      "epoch": 0.05508856218462023,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009944531459218481,
      "loss": 2.3015,
      "step": 8108
    },
    {
      "epoch": 0.05509535653121429,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009944515477195263,
      "loss": 2.4059,
      "step": 8109
    },
    {
      "epoch": 0.05510215087780835,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994449949288279,
      "loss": 2.3267,
      "step": 8110
    },
    {
      "epoch": 0.055108945224402406,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994448350628107,
      "loss": 2.2133,
      "step": 8111
    },
    {
      "epoch": 0.05511573957099646,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009944467517390113,
      "loss": 2.49,
      "step": 8112
    },
    {
      "epoch": 0.05512253391759052,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009944451526209923,
      "loss": 2.3133,
      "step": 8113
    },
    {
      "epoch": 0.055129328264184575,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009944435532740507,
      "loss": 2.2717,
      "step": 8114
    },
    {
      "epoch": 0.05513612261077864,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009944419536981877,
      "loss": 2.3819,
      "step": 8115
    },
    {
      "epoch": 0.055142916957372695,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009944403538934034,
      "loss": 2.2051,
      "step": 8116
    },
    {
      "epoch": 0.05514971130396675,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009944387538596993,
      "loss": 2.3969,
      "step": 8117
    },
    {
      "epoch": 0.05515650565056081,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009944371535970753,
      "loss": 2.191,
      "step": 8118
    },
    {
      "epoch": 0.055163299997154865,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009944355531055328,
      "loss": 2.3314,
      "step": 8119
    },
    {
      "epoch": 0.05517009434374893,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009944339523850724,
      "loss": 2.1788,
      "step": 8120
    },
    {
      "epoch": 0.055176888690342984,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009944323514356945,
      "loss": 2.2851,
      "step": 8121
    },
    {
      "epoch": 0.05518368303693704,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009944307502574002,
      "loss": 2.2918,
      "step": 8122
    },
    {
      "epoch": 0.0551904773835311,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009944291488501902,
      "loss": 2.3025,
      "step": 8123
    },
    {
      "epoch": 0.055197271730125154,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994427547214065,
      "loss": 2.3469,
      "step": 8124
    },
    {
      "epoch": 0.05520406607671921,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009944259453490258,
      "loss": 2.2238,
      "step": 8125
    },
    {
      "epoch": 0.055210860423313274,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009944243432550728,
      "loss": 2.3472,
      "step": 8126
    },
    {
      "epoch": 0.05521765476990733,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000994422740932207,
      "loss": 2.2943,
      "step": 8127
    },
    {
      "epoch": 0.05522444911650139,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009944211383804294,
      "loss": 2.1229,
      "step": 8128
    },
    {
      "epoch": 0.05523124346309544,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009944195355997402,
      "loss": 2.2537,
      "step": 8129
    },
    {
      "epoch": 0.0552380378096895,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009944179325901406,
      "loss": 2.3028,
      "step": 8130
    },
    {
      "epoch": 0.05524483215628356,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009944163293516312,
      "loss": 2.1758,
      "step": 8131
    },
    {
      "epoch": 0.05525162650287762,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009944147258842124,
      "loss": 2.368,
      "step": 8132
    },
    {
      "epoch": 0.055258420849471676,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009944131221878854,
      "loss": 2.1904,
      "step": 8133
    },
    {
      "epoch": 0.05526521519606573,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009944115182626508,
      "loss": 2.1707,
      "step": 8134
    },
    {
      "epoch": 0.05527200954265979,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009944099141085095,
      "loss": 2.3925,
      "step": 8135
    },
    {
      "epoch": 0.05527880388925385,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009944083097254619,
      "loss": 2.4129,
      "step": 8136
    },
    {
      "epoch": 0.05528559823584791,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994406705113509,
      "loss": 2.3884,
      "step": 8137
    },
    {
      "epoch": 0.055292392582441965,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009944051002726514,
      "loss": 2.2577,
      "step": 8138
    },
    {
      "epoch": 0.05529918692903602,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00099440349520289,
      "loss": 2.3111,
      "step": 8139
    },
    {
      "epoch": 0.05530598127563008,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009944018899042254,
      "loss": 2.31,
      "step": 8140
    },
    {
      "epoch": 0.055312775622224135,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009944002843766584,
      "loss": 2.3709,
      "step": 8141
    },
    {
      "epoch": 0.0553195699688182,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009943986786201896,
      "loss": 2.2759,
      "step": 8142
    },
    {
      "epoch": 0.055326364315412255,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099439707263482,
      "loss": 2.2407,
      "step": 8143
    },
    {
      "epoch": 0.05533315866200631,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009943954664205503,
      "loss": 2.3206,
      "step": 8144
    },
    {
      "epoch": 0.05533995300860037,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994393859977381,
      "loss": 2.3691,
      "step": 8145
    },
    {
      "epoch": 0.055346747355194424,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009943922533053131,
      "loss": 2.1764,
      "step": 8146
    },
    {
      "epoch": 0.05535354170178849,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009943906464043473,
      "loss": 2.3687,
      "step": 8147
    },
    {
      "epoch": 0.055360336048382544,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009943890392744843,
      "loss": 2.4497,
      "step": 8148
    },
    {
      "epoch": 0.0553671303949766,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009943874319157248,
      "loss": 2.4224,
      "step": 8149
    },
    {
      "epoch": 0.05537392474157066,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009943858243280695,
      "loss": 2.3279,
      "step": 8150
    },
    {
      "epoch": 0.055380719088164714,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009943842165115193,
      "loss": 2.148,
      "step": 8151
    },
    {
      "epoch": 0.05538751343475877,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994382608466075,
      "loss": 2.2049,
      "step": 8152
    },
    {
      "epoch": 0.05539430778135283,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994381000191737,
      "loss": 2.2992,
      "step": 8153
    },
    {
      "epoch": 0.05540110212794689,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009943793916885064,
      "loss": 2.27,
      "step": 8154
    },
    {
      "epoch": 0.055407896474540946,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009943777829563838,
      "loss": 2.2938,
      "step": 8155
    },
    {
      "epoch": 0.055414690821135,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00099437617399537,
      "loss": 2.3447,
      "step": 8156
    },
    {
      "epoch": 0.05542148516772906,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009943745648054655,
      "loss": 2.2694,
      "step": 8157
    },
    {
      "epoch": 0.05542827951432312,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009943729553866716,
      "loss": 2.2409,
      "step": 8158
    },
    {
      "epoch": 0.05543507386091718,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009943713457389884,
      "loss": 2.4166,
      "step": 8159
    },
    {
      "epoch": 0.055441868207511236,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994369735862417,
      "loss": 2.3532,
      "step": 8160
    },
    {
      "epoch": 0.05544866255410529,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994368125756958,
      "loss": 2.3206,
      "step": 8161
    },
    {
      "epoch": 0.05545545690069935,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009943665154226124,
      "loss": 2.3192,
      "step": 8162
    },
    {
      "epoch": 0.055462251247293405,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009943649048593805,
      "loss": 2.2451,
      "step": 8163
    },
    {
      "epoch": 0.05546904559388747,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009943632940672637,
      "loss": 2.328,
      "step": 8164
    },
    {
      "epoch": 0.055475839940481525,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009943616830462622,
      "loss": 2.1462,
      "step": 8165
    },
    {
      "epoch": 0.05548263428707558,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009943600717963767,
      "loss": 2.0517,
      "step": 8166
    },
    {
      "epoch": 0.05548942863366964,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009943584603176083,
      "loss": 2.3542,
      "step": 8167
    },
    {
      "epoch": 0.055496222980263694,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009943568486099578,
      "loss": 2.3189,
      "step": 8168
    },
    {
      "epoch": 0.05550301732685776,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009943552366734256,
      "loss": 2.1824,
      "step": 8169
    },
    {
      "epoch": 0.055509811673451814,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009943536245080125,
      "loss": 2.1584,
      "step": 8170
    },
    {
      "epoch": 0.05551660602004587,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009943520121137193,
      "loss": 2.213,
      "step": 8171
    },
    {
      "epoch": 0.05552340036663993,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000994350399490547,
      "loss": 2.2591,
      "step": 8172
    },
    {
      "epoch": 0.055530194713233984,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994348786638496,
      "loss": 2.1623,
      "step": 8173
    },
    {
      "epoch": 0.05553698905982804,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009943471735575672,
      "loss": 2.2731,
      "step": 8174
    },
    {
      "epoch": 0.055543783406422104,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009943455602477614,
      "loss": 2.4083,
      "step": 8175
    },
    {
      "epoch": 0.05555057775301616,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009943439467090791,
      "loss": 2.2911,
      "step": 8176
    },
    {
      "epoch": 0.05555737209961022,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009943423329415214,
      "loss": 2.2324,
      "step": 8177
    },
    {
      "epoch": 0.05556416644620427,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009943407189450888,
      "loss": 2.2626,
      "step": 8178
    },
    {
      "epoch": 0.05557096079279833,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000994339104719782,
      "loss": 2.2944,
      "step": 8179
    },
    {
      "epoch": 0.05557775513939239,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009943374902656019,
      "loss": 2.2259,
      "step": 8180
    },
    {
      "epoch": 0.05558454948598645,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009943358755825494,
      "loss": 2.2426,
      "step": 8181
    },
    {
      "epoch": 0.055591343832580506,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994334260670625,
      "loss": 2.3194,
      "step": 8182
    },
    {
      "epoch": 0.05559813817917456,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009943326455298294,
      "loss": 2.1882,
      "step": 8183
    },
    {
      "epoch": 0.05560493252576862,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009943310301601634,
      "loss": 2.3641,
      "step": 8184
    },
    {
      "epoch": 0.055611726872362675,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000994329414561628,
      "loss": 2.3389,
      "step": 8185
    },
    {
      "epoch": 0.05561852121895674,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009943277987342237,
      "loss": 2.2319,
      "step": 8186
    },
    {
      "epoch": 0.055625315565550795,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009943261826779511,
      "loss": 2.2833,
      "step": 8187
    },
    {
      "epoch": 0.05563210991214485,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009943245663928113,
      "loss": 2.2471,
      "step": 8188
    },
    {
      "epoch": 0.05563890425873891,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000994322949878805,
      "loss": 2.3424,
      "step": 8189
    },
    {
      "epoch": 0.055645698605332965,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009943213331359327,
      "loss": 2.3585,
      "step": 8190
    },
    {
      "epoch": 0.05565249295192703,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009943197161641952,
      "loss": 2.3235,
      "step": 8191
    },
    {
      "epoch": 0.055659287298521085,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009943180989635933,
      "loss": 2.3163,
      "step": 8192
    },
    {
      "epoch": 0.05566608164511514,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009943164815341282,
      "loss": 2.3897,
      "step": 8193
    },
    {
      "epoch": 0.0556728759917092,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009943148638758,
      "loss": 2.2282,
      "step": 8194
    },
    {
      "epoch": 0.055679670338303254,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009943132459886094,
      "loss": 2.1213,
      "step": 8195
    },
    {
      "epoch": 0.05568646468489732,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009943116278725577,
      "loss": 2.325,
      "step": 8196
    },
    {
      "epoch": 0.055693259031491374,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009943100095276454,
      "loss": 2.3231,
      "step": 8197
    },
    {
      "epoch": 0.05570005337808543,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009943083909538732,
      "loss": 2.1537,
      "step": 8198
    },
    {
      "epoch": 0.05570684772467949,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994306772151242,
      "loss": 2.3764,
      "step": 8199
    },
    {
      "epoch": 0.05571364207127354,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009943051531197521,
      "loss": 2.2371,
      "step": 8200
    },
    {
      "epoch": 0.0557204364178676,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009943035338594048,
      "loss": 2.2415,
      "step": 8201
    },
    {
      "epoch": 0.05572723076446166,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009943019143702008,
      "loss": 2.2586,
      "step": 8202
    },
    {
      "epoch": 0.05573402511105572,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009943002946521404,
      "loss": 2.3743,
      "step": 8203
    },
    {
      "epoch": 0.055740819457649776,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009942986747052246,
      "loss": 2.3295,
      "step": 8204
    },
    {
      "epoch": 0.05574761380424383,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009942970545294544,
      "loss": 2.3191,
      "step": 8205
    },
    {
      "epoch": 0.05575440815083789,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009942954341248303,
      "loss": 2.3605,
      "step": 8206
    },
    {
      "epoch": 0.05576120249743195,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000994293813491353,
      "loss": 2.2432,
      "step": 8207
    },
    {
      "epoch": 0.05576799684402601,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009942921926290233,
      "loss": 2.2819,
      "step": 8208
    },
    {
      "epoch": 0.055774791190620066,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009942905715378421,
      "loss": 2.3259,
      "step": 8209
    },
    {
      "epoch": 0.05578158553721412,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00099428895021781,
      "loss": 2.3024,
      "step": 8210
    },
    {
      "epoch": 0.05578837988380818,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009942873286689277,
      "loss": 2.5075,
      "step": 8211
    },
    {
      "epoch": 0.055795174230402235,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994285706891196,
      "loss": 2.3517,
      "step": 8212
    },
    {
      "epoch": 0.0558019685769963,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000994284084884616,
      "loss": 2.2848,
      "step": 8213
    },
    {
      "epoch": 0.055808762923590355,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009942824626491878,
      "loss": 2.329,
      "step": 8214
    },
    {
      "epoch": 0.05581555727018441,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009942808401849127,
      "loss": 2.3388,
      "step": 8215
    },
    {
      "epoch": 0.05582235161677847,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009942792174917911,
      "loss": 2.3967,
      "step": 8216
    },
    {
      "epoch": 0.055829145963372524,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000994277594569824,
      "loss": 2.1858,
      "step": 8217
    },
    {
      "epoch": 0.05583594030996659,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009942759714190118,
      "loss": 2.3992,
      "step": 8218
    },
    {
      "epoch": 0.055842734656560644,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009942743480393557,
      "loss": 2.5609,
      "step": 8219
    },
    {
      "epoch": 0.0558495290031547,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009942727244308563,
      "loss": 2.1838,
      "step": 8220
    },
    {
      "epoch": 0.05585632334974876,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009942711005935141,
      "loss": 2.3544,
      "step": 8221
    },
    {
      "epoch": 0.055863117696342814,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00099426947652733,
      "loss": 2.2459,
      "step": 8222
    },
    {
      "epoch": 0.05586991204293687,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009942678522323052,
      "loss": 2.1709,
      "step": 8223
    },
    {
      "epoch": 0.055876706389530933,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009942662277084397,
      "loss": 2.3232,
      "step": 8224
    },
    {
      "epoch": 0.05588350073612499,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009942646029557347,
      "loss": 2.3826,
      "step": 8225
    },
    {
      "epoch": 0.055890295082719046,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009942629779741909,
      "loss": 2.2918,
      "step": 8226
    },
    {
      "epoch": 0.0558970894293131,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009942613527638088,
      "loss": 2.2847,
      "step": 8227
    },
    {
      "epoch": 0.05590388377590716,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009942597273245896,
      "loss": 2.2375,
      "step": 8228
    },
    {
      "epoch": 0.05591067812250122,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009942581016565337,
      "loss": 2.3218,
      "step": 8229
    },
    {
      "epoch": 0.05591747246909528,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994256475759642,
      "loss": 2.2764,
      "step": 8230
    },
    {
      "epoch": 0.055924266815689336,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009942548496339153,
      "loss": 2.2797,
      "step": 8231
    },
    {
      "epoch": 0.05593106116228339,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009942532232793543,
      "loss": 2.3344,
      "step": 8232
    },
    {
      "epoch": 0.05593785550887745,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009942515966959594,
      "loss": 2.3037,
      "step": 8233
    },
    {
      "epoch": 0.055944649855471505,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994249969883732,
      "loss": 2.4249,
      "step": 8234
    },
    {
      "epoch": 0.05595144420206557,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009942483428426723,
      "loss": 2.2374,
      "step": 8235
    },
    {
      "epoch": 0.055958238548659625,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009942467155727814,
      "loss": 2.3547,
      "step": 8236
    },
    {
      "epoch": 0.05596503289525368,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00099424508807406,
      "loss": 2.2399,
      "step": 8237
    },
    {
      "epoch": 0.05597182724184774,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009942434603465086,
      "loss": 2.3271,
      "step": 8238
    },
    {
      "epoch": 0.055978621588441795,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009942418323901283,
      "loss": 2.3142,
      "step": 8239
    },
    {
      "epoch": 0.05598541593503586,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009942402042049197,
      "loss": 2.3077,
      "step": 8240
    },
    {
      "epoch": 0.055992210281629914,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009942385757908837,
      "loss": 2.3623,
      "step": 8241
    },
    {
      "epoch": 0.05599900462822397,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009942369471480205,
      "loss": 2.3827,
      "step": 8242
    },
    {
      "epoch": 0.05600579897481803,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009942353182763315,
      "loss": 2.169,
      "step": 8243
    },
    {
      "epoch": 0.056012593321412084,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009942336891758173,
      "loss": 2.2343,
      "step": 8244
    },
    {
      "epoch": 0.05601938766800614,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009942320598464785,
      "loss": 2.2756,
      "step": 8245
    },
    {
      "epoch": 0.056026182014600204,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009942304302883159,
      "loss": 2.3448,
      "step": 8246
    },
    {
      "epoch": 0.05603297636119426,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009942288005013301,
      "loss": 2.1864,
      "step": 8247
    },
    {
      "epoch": 0.05603977070778832,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009942271704855223,
      "loss": 2.2288,
      "step": 8248
    },
    {
      "epoch": 0.05604656505438237,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009942255402408929,
      "loss": 2.2407,
      "step": 8249
    },
    {
      "epoch": 0.05605335940097643,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009942239097674426,
      "loss": 2.344,
      "step": 8250
    },
    {
      "epoch": 0.05606015374757049,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009942222790651726,
      "loss": 2.1823,
      "step": 8251
    },
    {
      "epoch": 0.05606694809416455,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009942206481340832,
      "loss": 2.2178,
      "step": 8252
    },
    {
      "epoch": 0.056073742440758606,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000994219016974175,
      "loss": 2.2304,
      "step": 8253
    },
    {
      "epoch": 0.05608053678735266,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009942173855854496,
      "loss": 2.2073,
      "step": 8254
    },
    {
      "epoch": 0.05608733113394672,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009942157539679069,
      "loss": 2.3637,
      "step": 8255
    },
    {
      "epoch": 0.05609412548054078,
      "grad_norm": 0.140625,
      "learning_rate": 0.000994214122121548,
      "loss": 2.3025,
      "step": 8256
    },
    {
      "epoch": 0.05610091982713484,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009942124900463737,
      "loss": 2.3049,
      "step": 8257
    },
    {
      "epoch": 0.056107714173728895,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009942108577423844,
      "loss": 2.2314,
      "step": 8258
    },
    {
      "epoch": 0.05611450852032295,
      "grad_norm": 0.125,
      "learning_rate": 0.0009942092252095815,
      "loss": 2.3181,
      "step": 8259
    },
    {
      "epoch": 0.05612130286691701,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009942075924479652,
      "loss": 2.3279,
      "step": 8260
    },
    {
      "epoch": 0.056128097213511065,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009942059594575366,
      "loss": 2.3121,
      "step": 8261
    },
    {
      "epoch": 0.05613489156010513,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009942043262382962,
      "loss": 2.2882,
      "step": 8262
    },
    {
      "epoch": 0.056141685906699185,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000994202692790245,
      "loss": 2.2522,
      "step": 8263
    },
    {
      "epoch": 0.05614848025329324,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009942010591133834,
      "loss": 2.2415,
      "step": 8264
    },
    {
      "epoch": 0.0561552745998873,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009941994252077127,
      "loss": 2.3198,
      "step": 8265
    },
    {
      "epoch": 0.056162068946481354,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000994197791073233,
      "loss": 2.3065,
      "step": 8266
    },
    {
      "epoch": 0.05616886329307542,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009941961567099455,
      "loss": 2.2303,
      "step": 8267
    },
    {
      "epoch": 0.056175657639669474,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009941945221178508,
      "loss": 2.3911,
      "step": 8268
    },
    {
      "epoch": 0.05618245198626353,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009941928872969498,
      "loss": 2.2998,
      "step": 8269
    },
    {
      "epoch": 0.05618924633285759,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009941912522472432,
      "loss": 2.2872,
      "step": 8270
    },
    {
      "epoch": 0.05619604067945164,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009941896169687315,
      "loss": 2.2064,
      "step": 8271
    },
    {
      "epoch": 0.0562028350260457,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000994187981461416,
      "loss": 2.3352,
      "step": 8272
    },
    {
      "epoch": 0.05620962937263976,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009941863457252967,
      "loss": 2.2174,
      "step": 8273
    },
    {
      "epoch": 0.05621642371923382,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009941847097603752,
      "loss": 2.3396,
      "step": 8274
    },
    {
      "epoch": 0.056223218065827876,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009941830735666515,
      "loss": 2.3865,
      "step": 8275
    },
    {
      "epoch": 0.05623001241242193,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000994181437144127,
      "loss": 2.1792,
      "step": 8276
    },
    {
      "epoch": 0.05623680675901599,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000994179800492802,
      "loss": 2.4217,
      "step": 8277
    },
    {
      "epoch": 0.05624360110561005,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009941781636126775,
      "loss": 2.2443,
      "step": 8278
    },
    {
      "epoch": 0.05625039545220411,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009941765265037538,
      "loss": 2.3346,
      "step": 8279
    },
    {
      "epoch": 0.056257189798798166,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009941748891660325,
      "loss": 2.2925,
      "step": 8280
    },
    {
      "epoch": 0.05626398414539222,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009941732515995137,
      "loss": 2.2656,
      "step": 8281
    },
    {
      "epoch": 0.05627077849198628,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009941716138041985,
      "loss": 2.2546,
      "step": 8282
    },
    {
      "epoch": 0.056277572838580335,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009941699757800874,
      "loss": 2.2657,
      "step": 8283
    },
    {
      "epoch": 0.0562843671851744,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009941683375271813,
      "loss": 2.2541,
      "step": 8284
    },
    {
      "epoch": 0.056291161531768455,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000994166699045481,
      "loss": 2.2922,
      "step": 8285
    },
    {
      "epoch": 0.05629795587836251,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000994165060334987,
      "loss": 2.2139,
      "step": 8286
    },
    {
      "epoch": 0.05630475022495657,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009941634213957002,
      "loss": 2.2782,
      "step": 8287
    },
    {
      "epoch": 0.056311544571550624,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009941617822276216,
      "loss": 2.2121,
      "step": 8288
    },
    {
      "epoch": 0.05631833891814469,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009941601428307516,
      "loss": 2.2897,
      "step": 8289
    },
    {
      "epoch": 0.056325133264738744,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009941585032050913,
      "loss": 2.2901,
      "step": 8290
    },
    {
      "epoch": 0.0563319276113328,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009941568633506411,
      "loss": 2.3535,
      "step": 8291
    },
    {
      "epoch": 0.05633872195792686,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000994155223267402,
      "loss": 2.3303,
      "step": 8292
    },
    {
      "epoch": 0.056345516304520914,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009941535829553747,
      "loss": 2.1251,
      "step": 8293
    },
    {
      "epoch": 0.05635231065111497,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00099415194241456,
      "loss": 2.3549,
      "step": 8294
    },
    {
      "epoch": 0.056359104997709034,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009941503016449586,
      "loss": 2.3078,
      "step": 8295
    },
    {
      "epoch": 0.05636589934430309,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009941486606465713,
      "loss": 2.2883,
      "step": 8296
    },
    {
      "epoch": 0.05637269369089715,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009941470194193987,
      "loss": 2.2344,
      "step": 8297
    },
    {
      "epoch": 0.0563794880374912,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000994145377963442,
      "loss": 2.2392,
      "step": 8298
    },
    {
      "epoch": 0.05638628238408526,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009941437362787013,
      "loss": 2.3063,
      "step": 8299
    },
    {
      "epoch": 0.05639307673067932,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000994142094365178,
      "loss": 2.2852,
      "step": 8300
    },
    {
      "epoch": 0.05639987107727338,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009941404522228723,
      "loss": 2.4367,
      "step": 8301
    },
    {
      "epoch": 0.056406665423867436,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009941388098517853,
      "loss": 2.143,
      "step": 8302
    },
    {
      "epoch": 0.05641345977046149,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009941371672519178,
      "loss": 2.3296,
      "step": 8303
    },
    {
      "epoch": 0.05642025411705555,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009941355244232706,
      "loss": 2.3137,
      "step": 8304
    },
    {
      "epoch": 0.056427048463649605,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000994133881365844,
      "loss": 2.3735,
      "step": 8305
    },
    {
      "epoch": 0.05643384281024367,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009941322380796393,
      "loss": 2.269,
      "step": 8306
    },
    {
      "epoch": 0.056440637156837725,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009941305945646568,
      "loss": 2.2493,
      "step": 8307
    },
    {
      "epoch": 0.05644743150343178,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009941289508208976,
      "loss": 2.2041,
      "step": 8308
    },
    {
      "epoch": 0.05645422585002584,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009941273068483625,
      "loss": 2.3441,
      "step": 8309
    },
    {
      "epoch": 0.056461020196619895,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000994125662647052,
      "loss": 2.1794,
      "step": 8310
    },
    {
      "epoch": 0.05646781454321396,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000994124018216967,
      "loss": 2.147,
      "step": 8311
    },
    {
      "epoch": 0.056474608889808015,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009941223735581083,
      "loss": 2.3301,
      "step": 8312
    },
    {
      "epoch": 0.05648140323640207,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009941207286704769,
      "loss": 2.2789,
      "step": 8313
    },
    {
      "epoch": 0.05648819758299613,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009941190835540729,
      "loss": 2.1995,
      "step": 8314
    },
    {
      "epoch": 0.056494991929590184,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009941174382088974,
      "loss": 2.2853,
      "step": 8315
    },
    {
      "epoch": 0.05650178627618425,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994115792634951,
      "loss": 2.3284,
      "step": 8316
    },
    {
      "epoch": 0.056508580622778304,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000994114146832235,
      "loss": 2.1391,
      "step": 8317
    },
    {
      "epoch": 0.05651537496937236,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099411250080075,
      "loss": 2.2497,
      "step": 8318
    },
    {
      "epoch": 0.05652216931596642,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009941108545404962,
      "loss": 2.3124,
      "step": 8319
    },
    {
      "epoch": 0.05652896366256047,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009941092080514747,
      "loss": 2.1414,
      "step": 8320
    },
    {
      "epoch": 0.05653575800915453,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009941075613336865,
      "loss": 2.2921,
      "step": 8321
    },
    {
      "epoch": 0.05654255235574859,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994105914387132,
      "loss": 2.2984,
      "step": 8322
    },
    {
      "epoch": 0.05654934670234265,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009941042672118124,
      "loss": 2.3557,
      "step": 8323
    },
    {
      "epoch": 0.056556141048936706,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009941026198077279,
      "loss": 2.4017,
      "step": 8324
    },
    {
      "epoch": 0.05656293539553076,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009941009721748798,
      "loss": 2.2564,
      "step": 8325
    },
    {
      "epoch": 0.05656972974212482,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009940993243132685,
      "loss": 2.1682,
      "step": 8326
    },
    {
      "epoch": 0.05657652408871888,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009940976762228947,
      "loss": 2.0954,
      "step": 8327
    },
    {
      "epoch": 0.05658331843531294,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009940960279037597,
      "loss": 2.1953,
      "step": 8328
    },
    {
      "epoch": 0.056590112781906995,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009940943793558636,
      "loss": 2.225,
      "step": 8329
    },
    {
      "epoch": 0.05659690712850105,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009940927305792076,
      "loss": 2.3813,
      "step": 8330
    },
    {
      "epoch": 0.05660370147509511,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009940910815737923,
      "loss": 2.4429,
      "step": 8331
    },
    {
      "epoch": 0.056610495821689165,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009940894323396186,
      "loss": 2.2799,
      "step": 8332
    },
    {
      "epoch": 0.05661729016828323,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000994087782876687,
      "loss": 2.3846,
      "step": 8333
    },
    {
      "epoch": 0.056624084514877285,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009940861331849985,
      "loss": 2.2822,
      "step": 8334
    },
    {
      "epoch": 0.05663087886147134,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009940844832645537,
      "loss": 2.399,
      "step": 8335
    },
    {
      "epoch": 0.0566376732080654,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009940828331153535,
      "loss": 2.419,
      "step": 8336
    },
    {
      "epoch": 0.056644467554659454,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009940811827373987,
      "loss": 2.3107,
      "step": 8337
    },
    {
      "epoch": 0.05665126190125352,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.00099407953213069,
      "loss": 2.2767,
      "step": 8338
    },
    {
      "epoch": 0.056658056247847574,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000994077881295228,
      "loss": 2.368,
      "step": 8339
    },
    {
      "epoch": 0.05666485059444163,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009940762302310135,
      "loss": 2.1097,
      "step": 8340
    },
    {
      "epoch": 0.05667164494103569,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009940745789380476,
      "loss": 2.341,
      "step": 8341
    },
    {
      "epoch": 0.056678439287629744,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009940729274163307,
      "loss": 2.2232,
      "step": 8342
    },
    {
      "epoch": 0.0566852336342238,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009940712756658639,
      "loss": 2.1848,
      "step": 8343
    },
    {
      "epoch": 0.05669202798081786,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009940696236866475,
      "loss": 2.2937,
      "step": 8344
    },
    {
      "epoch": 0.05669882232741192,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009940679714786826,
      "loss": 2.3383,
      "step": 8345
    },
    {
      "epoch": 0.056705616674005976,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009940663190419702,
      "loss": 2.2989,
      "step": 8346
    },
    {
      "epoch": 0.05671241102060003,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009940646663765104,
      "loss": 2.3006,
      "step": 8347
    },
    {
      "epoch": 0.05671920536719409,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009940630134823044,
      "loss": 2.3431,
      "step": 8348
    },
    {
      "epoch": 0.05672599971378815,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000994061360359353,
      "loss": 2.2382,
      "step": 8349
    },
    {
      "epoch": 0.05673279406038221,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009940597070076567,
      "loss": 2.236,
      "step": 8350
    },
    {
      "epoch": 0.056739588406976266,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009940580534272163,
      "loss": 2.2824,
      "step": 8351
    },
    {
      "epoch": 0.05674638275357032,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000994056399618033,
      "loss": 2.3034,
      "step": 8352
    },
    {
      "epoch": 0.05675317710016438,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009940547455801073,
      "loss": 2.2751,
      "step": 8353
    },
    {
      "epoch": 0.056759971446758435,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009940530913134396,
      "loss": 2.2644,
      "step": 8354
    },
    {
      "epoch": 0.0567667657933525,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000994051436818031,
      "loss": 2.233,
      "step": 8355
    },
    {
      "epoch": 0.056773560139946555,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009940497820938826,
      "loss": 2.2476,
      "step": 8356
    },
    {
      "epoch": 0.05678035448654061,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009940481271409945,
      "loss": 2.2027,
      "step": 8357
    },
    {
      "epoch": 0.05678714883313467,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009940464719593678,
      "loss": 2.3229,
      "step": 8358
    },
    {
      "epoch": 0.056793943179728724,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009940448165490035,
      "loss": 2.1725,
      "step": 8359
    },
    {
      "epoch": 0.05680073752632279,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000994043160909902,
      "loss": 2.2907,
      "step": 8360
    },
    {
      "epoch": 0.056807531872916844,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000994041505042064,
      "loss": 2.126,
      "step": 8361
    },
    {
      "epoch": 0.0568143262195109,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009940398489454905,
      "loss": 2.2604,
      "step": 8362
    },
    {
      "epoch": 0.05682112056610496,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009940381926201822,
      "loss": 2.3547,
      "step": 8363
    },
    {
      "epoch": 0.056827914912699014,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099403653606614,
      "loss": 2.3396,
      "step": 8364
    },
    {
      "epoch": 0.05683470925929307,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009940348792833644,
      "loss": 2.2191,
      "step": 8365
    },
    {
      "epoch": 0.056841503605887134,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009940332222718564,
      "loss": 2.2719,
      "step": 8366
    },
    {
      "epoch": 0.05684829795248119,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009940315650316168,
      "loss": 2.1241,
      "step": 8367
    },
    {
      "epoch": 0.05685509229907525,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009940299075626461,
      "loss": 2.3422,
      "step": 8368
    },
    {
      "epoch": 0.0568618866456693,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009940282498649452,
      "loss": 2.2417,
      "step": 8369
    },
    {
      "epoch": 0.05686868099226336,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000994026591938515,
      "loss": 2.4027,
      "step": 8370
    },
    {
      "epoch": 0.05687547533885742,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009940249337833562,
      "loss": 2.4815,
      "step": 8371
    },
    {
      "epoch": 0.05688226968545148,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009940232753994694,
      "loss": 2.3692,
      "step": 8372
    },
    {
      "epoch": 0.056889064032045536,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009940216167868554,
      "loss": 2.2989,
      "step": 8373
    },
    {
      "epoch": 0.05689585837863959,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009940199579455153,
      "loss": 2.2653,
      "step": 8374
    },
    {
      "epoch": 0.05690265272523365,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009940182988754493,
      "loss": 2.2003,
      "step": 8375
    },
    {
      "epoch": 0.05690944707182771,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009940166395766587,
      "loss": 2.1182,
      "step": 8376
    },
    {
      "epoch": 0.05691624141842177,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009940149800491441,
      "loss": 2.1193,
      "step": 8377
    },
    {
      "epoch": 0.056923035765015825,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009940133202929063,
      "loss": 2.1404,
      "step": 8378
    },
    {
      "epoch": 0.05692983011160988,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009940116603079458,
      "loss": 2.1494,
      "step": 8379
    },
    {
      "epoch": 0.05693662445820394,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009940100000942636,
      "loss": 2.3272,
      "step": 8380
    },
    {
      "epoch": 0.056943418804797995,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009940083396518605,
      "loss": 2.3544,
      "step": 8381
    },
    {
      "epoch": 0.05695021315139206,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000994006678980737,
      "loss": 2.2236,
      "step": 8382
    },
    {
      "epoch": 0.056957007497986115,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009940050180808945,
      "loss": 2.253,
      "step": 8383
    },
    {
      "epoch": 0.05696380184458017,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000994003356952333,
      "loss": 2.2305,
      "step": 8384
    },
    {
      "epoch": 0.05697059619117423,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009940016955950536,
      "loss": 2.3665,
      "step": 8385
    },
    {
      "epoch": 0.056977390537768284,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009940000340090572,
      "loss": 2.2186,
      "step": 8386
    },
    {
      "epoch": 0.05698418488436235,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009939983721943442,
      "loss": 2.3976,
      "step": 8387
    },
    {
      "epoch": 0.056990979230956404,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000993996710150916,
      "loss": 2.4147,
      "step": 8388
    },
    {
      "epoch": 0.05699777357755046,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009939950478787727,
      "loss": 2.2541,
      "step": 8389
    },
    {
      "epoch": 0.05700456792414452,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009939933853779154,
      "loss": 2.2165,
      "step": 8390
    },
    {
      "epoch": 0.05701136227073857,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000993991722648345,
      "loss": 2.2331,
      "step": 8391
    },
    {
      "epoch": 0.05701815661733263,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009939900596900619,
      "loss": 2.116,
      "step": 8392
    },
    {
      "epoch": 0.05702495096392669,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009939883965030673,
      "loss": 2.3991,
      "step": 8393
    },
    {
      "epoch": 0.05703174531052075,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009939867330873615,
      "loss": 2.3195,
      "step": 8394
    },
    {
      "epoch": 0.057038539657114806,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009939850694429457,
      "loss": 2.3293,
      "step": 8395
    },
    {
      "epoch": 0.05704533400370886,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009939834055698203,
      "loss": 2.2366,
      "step": 8396
    },
    {
      "epoch": 0.05705212835030292,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009939817414679863,
      "loss": 2.2148,
      "step": 8397
    },
    {
      "epoch": 0.05705892269689698,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009939800771374444,
      "loss": 2.3639,
      "step": 8398
    },
    {
      "epoch": 0.05706571704349104,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009939784125781955,
      "loss": 2.305,
      "step": 8399
    },
    {
      "epoch": 0.057072511390085096,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009939767477902403,
      "loss": 2.2575,
      "step": 8400
    },
    {
      "epoch": 0.05707930573667915,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009939750827735792,
      "loss": 2.3485,
      "step": 8401
    },
    {
      "epoch": 0.05708610008327321,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009939734175282136,
      "loss": 2.2586,
      "step": 8402
    },
    {
      "epoch": 0.057092894429867265,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009939717520541437,
      "loss": 2.1325,
      "step": 8403
    },
    {
      "epoch": 0.05709968877646133,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009939700863513708,
      "loss": 2.2174,
      "step": 8404
    },
    {
      "epoch": 0.057106483123055385,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009939684204198953,
      "loss": 2.2818,
      "step": 8405
    },
    {
      "epoch": 0.05711327746964944,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009939667542597183,
      "loss": 2.2513,
      "step": 8406
    },
    {
      "epoch": 0.0571200718162435,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009939650878708402,
      "loss": 2.1887,
      "step": 8407
    },
    {
      "epoch": 0.057126866162837554,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009939634212532617,
      "loss": 2.1714,
      "step": 8408
    },
    {
      "epoch": 0.05713366050943162,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009939617544069842,
      "loss": 2.2701,
      "step": 8409
    },
    {
      "epoch": 0.057140454856025674,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009939600873320078,
      "loss": 2.368,
      "step": 8410
    },
    {
      "epoch": 0.05714724920261973,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009939584200283336,
      "loss": 2.0755,
      "step": 8411
    },
    {
      "epoch": 0.05715404354921379,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009939567524959623,
      "loss": 2.2511,
      "step": 8412
    },
    {
      "epoch": 0.057160837895807844,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009939550847348948,
      "loss": 2.3034,
      "step": 8413
    },
    {
      "epoch": 0.0571676322424019,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009939534167451316,
      "loss": 2.3952,
      "step": 8414
    },
    {
      "epoch": 0.057174426588995964,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009939517485266737,
      "loss": 2.2385,
      "step": 8415
    },
    {
      "epoch": 0.05718122093559002,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009939500800795217,
      "loss": 2.3178,
      "step": 8416
    },
    {
      "epoch": 0.057188015282184076,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009939484114036767,
      "loss": 2.1753,
      "step": 8417
    },
    {
      "epoch": 0.05719480962877813,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000993946742499139,
      "loss": 2.1934,
      "step": 8418
    },
    {
      "epoch": 0.05720160397537219,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099394507336591,
      "loss": 2.3282,
      "step": 8419
    },
    {
      "epoch": 0.05720839832196625,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009939434040039897,
      "loss": 2.2606,
      "step": 8420
    },
    {
      "epoch": 0.05721519266856031,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009939417344133794,
      "loss": 2.4864,
      "step": 8421
    },
    {
      "epoch": 0.057221987015154366,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009939400645940799,
      "loss": 2.1764,
      "step": 8422
    },
    {
      "epoch": 0.05722878136174842,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009939383945460915,
      "loss": 2.3227,
      "step": 8423
    },
    {
      "epoch": 0.05723557570834248,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009939367242694154,
      "loss": 2.4567,
      "step": 8424
    },
    {
      "epoch": 0.057242370054936535,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009939350537640523,
      "loss": 2.1697,
      "step": 8425
    },
    {
      "epoch": 0.0572491644015306,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009939333830300031,
      "loss": 2.3151,
      "step": 8426
    },
    {
      "epoch": 0.057255958748124655,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009939317120672682,
      "loss": 2.2748,
      "step": 8427
    },
    {
      "epoch": 0.05726275309471871,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009939300408758488,
      "loss": 2.2848,
      "step": 8428
    },
    {
      "epoch": 0.05726954744131277,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009939283694557451,
      "loss": 2.2174,
      "step": 8429
    },
    {
      "epoch": 0.057276341787906825,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009939266978069585,
      "loss": 2.1627,
      "step": 8430
    },
    {
      "epoch": 0.05728313613450089,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009939250259294893,
      "loss": 2.1691,
      "step": 8431
    },
    {
      "epoch": 0.057289930481094944,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009939233538233388,
      "loss": 2.167,
      "step": 8432
    },
    {
      "epoch": 0.057296724827689,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000993921681488507,
      "loss": 2.1744,
      "step": 8433
    },
    {
      "epoch": 0.05730351917428306,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009939200089249954,
      "loss": 2.3773,
      "step": 8434
    },
    {
      "epoch": 0.057310313520877114,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009939183361328046,
      "loss": 2.403,
      "step": 8435
    },
    {
      "epoch": 0.05731710786747118,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000993916663111935,
      "loss": 2.2845,
      "step": 8436
    },
    {
      "epoch": 0.057323902214065234,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009939149898623878,
      "loss": 2.2854,
      "step": 8437
    },
    {
      "epoch": 0.05733069656065929,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009939133163841636,
      "loss": 2.1325,
      "step": 8438
    },
    {
      "epoch": 0.05733749090725335,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009939116426772633,
      "loss": 2.1293,
      "step": 8439
    },
    {
      "epoch": 0.0573442852538474,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009939099687416874,
      "loss": 2.1776,
      "step": 8440
    },
    {
      "epoch": 0.05735107960044146,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009939082945774368,
      "loss": 2.2662,
      "step": 8441
    },
    {
      "epoch": 0.05735787394703552,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009939066201845126,
      "loss": 2.3114,
      "step": 8442
    },
    {
      "epoch": 0.05736466829362958,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009939049455629152,
      "loss": 2.3959,
      "step": 8443
    },
    {
      "epoch": 0.057371462640223636,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009939032707126454,
      "loss": 2.2144,
      "step": 8444
    },
    {
      "epoch": 0.05737825698681769,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000993901595633704,
      "loss": 2.1952,
      "step": 8445
    },
    {
      "epoch": 0.05738505133341175,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000993899920326092,
      "loss": 2.3389,
      "step": 8446
    },
    {
      "epoch": 0.05739184568000581,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009938982447898097,
      "loss": 2.2633,
      "step": 8447
    },
    {
      "epoch": 0.05739864002659987,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009938965690248585,
      "loss": 2.2372,
      "step": 8448
    },
    {
      "epoch": 0.057405434373193925,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009938948930312385,
      "loss": 2.3373,
      "step": 8449
    },
    {
      "epoch": 0.05741222871978798,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009938932168089513,
      "loss": 2.2409,
      "step": 8450
    },
    {
      "epoch": 0.05741902306638204,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009938915403579968,
      "loss": 2.2687,
      "step": 8451
    },
    {
      "epoch": 0.057425817412976095,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009938898636783764,
      "loss": 2.2391,
      "step": 8452
    },
    {
      "epoch": 0.05743261175957016,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009938881867700906,
      "loss": 2.2717,
      "step": 8453
    },
    {
      "epoch": 0.057439406106164215,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009938865096331402,
      "loss": 2.294,
      "step": 8454
    },
    {
      "epoch": 0.05744620045275827,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000993884832267526,
      "loss": 2.3933,
      "step": 8455
    },
    {
      "epoch": 0.05745299479935233,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009938831546732487,
      "loss": 2.3762,
      "step": 8456
    },
    {
      "epoch": 0.057459789145946384,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009938814768503093,
      "loss": 2.2133,
      "step": 8457
    },
    {
      "epoch": 0.05746658349254045,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009938797987987084,
      "loss": 2.3635,
      "step": 8458
    },
    {
      "epoch": 0.057473377839134504,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000993878120518447,
      "loss": 2.2893,
      "step": 8459
    },
    {
      "epoch": 0.05748017218572856,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009938764420095252,
      "loss": 2.2634,
      "step": 8460
    },
    {
      "epoch": 0.05748696653232262,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009938747632719446,
      "loss": 2.3201,
      "step": 8461
    },
    {
      "epoch": 0.057493760878916673,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009938730843057056,
      "loss": 2.2507,
      "step": 8462
    },
    {
      "epoch": 0.05750055522551073,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009938714051108092,
      "loss": 2.2971,
      "step": 8463
    },
    {
      "epoch": 0.05750734957210479,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009938697256872558,
      "loss": 2.2828,
      "step": 8464
    },
    {
      "epoch": 0.05751414391869885,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009938680460350464,
      "loss": 2.2614,
      "step": 8465
    },
    {
      "epoch": 0.057520938265292906,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009938663661541818,
      "loss": 2.307,
      "step": 8466
    },
    {
      "epoch": 0.05752773261188696,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009938646860446628,
      "loss": 2.2816,
      "step": 8467
    },
    {
      "epoch": 0.05753452695848102,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099386300570649,
      "loss": 2.2619,
      "step": 8468
    },
    {
      "epoch": 0.05754132130507508,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009938613251396643,
      "loss": 2.4031,
      "step": 8469
    },
    {
      "epoch": 0.05754811565166914,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009938596443441864,
      "loss": 2.4844,
      "step": 8470
    },
    {
      "epoch": 0.057554909998263196,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009938579633200573,
      "loss": 2.3589,
      "step": 8471
    },
    {
      "epoch": 0.05756170434485725,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009938562820672775,
      "loss": 2.3747,
      "step": 8472
    },
    {
      "epoch": 0.05756849869145131,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000993854600585848,
      "loss": 2.3316,
      "step": 8473
    },
    {
      "epoch": 0.057575293038045365,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009938529188757695,
      "loss": 2.3293,
      "step": 8474
    },
    {
      "epoch": 0.05758208738463943,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009938512369370427,
      "loss": 2.3605,
      "step": 8475
    },
    {
      "epoch": 0.057588881731233485,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009938495547696683,
      "loss": 2.2864,
      "step": 8476
    },
    {
      "epoch": 0.05759567607782754,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009938478723736475,
      "loss": 2.333,
      "step": 8477
    },
    {
      "epoch": 0.0576024704244216,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009938461897489805,
      "loss": 2.2312,
      "step": 8478
    },
    {
      "epoch": 0.057609264771015654,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009938445068956687,
      "loss": 2.3713,
      "step": 8479
    },
    {
      "epoch": 0.05761605911760972,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009938428238137123,
      "loss": 2.2677,
      "step": 8480
    },
    {
      "epoch": 0.057622853464203774,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009938411405031125,
      "loss": 2.2274,
      "step": 8481
    },
    {
      "epoch": 0.05762964781079783,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009938394569638698,
      "loss": 2.2526,
      "step": 8482
    },
    {
      "epoch": 0.05763644215739189,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009938377731959851,
      "loss": 2.2912,
      "step": 8483
    },
    {
      "epoch": 0.057643236503985944,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009938360891994591,
      "loss": 2.1006,
      "step": 8484
    },
    {
      "epoch": 0.05765003085058,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000993834404974293,
      "loss": 2.4055,
      "step": 8485
    },
    {
      "epoch": 0.057656825197174064,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009938327205204869,
      "loss": 2.1654,
      "step": 8486
    },
    {
      "epoch": 0.05766361954376812,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000993831035838042,
      "loss": 2.2759,
      "step": 8487
    },
    {
      "epoch": 0.05767041389036218,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009938293509269588,
      "loss": 2.2364,
      "step": 8488
    },
    {
      "epoch": 0.05767720823695623,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009938276657872386,
      "loss": 2.219,
      "step": 8489
    },
    {
      "epoch": 0.05768400258355029,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009938259804188818,
      "loss": 2.1904,
      "step": 8490
    },
    {
      "epoch": 0.05769079693014435,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000993824294821889,
      "loss": 2.2906,
      "step": 8491
    },
    {
      "epoch": 0.05769759127673841,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009938226089962615,
      "loss": 2.2861,
      "step": 8492
    },
    {
      "epoch": 0.057704385623332466,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009938209229419996,
      "loss": 2.2937,
      "step": 8493
    },
    {
      "epoch": 0.05771117996992652,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009938192366591044,
      "loss": 2.1901,
      "step": 8494
    },
    {
      "epoch": 0.05771797431652058,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009938175501475763,
      "loss": 2.2091,
      "step": 8495
    },
    {
      "epoch": 0.05772476866311464,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009938158634074166,
      "loss": 2.253,
      "step": 8496
    },
    {
      "epoch": 0.0577315630097087,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009938141764386256,
      "loss": 2.2569,
      "step": 8497
    },
    {
      "epoch": 0.057738357356302755,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009938124892412046,
      "loss": 2.2782,
      "step": 8498
    },
    {
      "epoch": 0.05774515170289681,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000993810801815154,
      "loss": 2.448,
      "step": 8499
    },
    {
      "epoch": 0.05775194604949087,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009938091141604743,
      "loss": 2.2275,
      "step": 8500
    },
    {
      "epoch": 0.057758740396084925,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993807426277167,
      "loss": 2.3035,
      "step": 8501
    },
    {
      "epoch": 0.05776553474267899,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009938057381652327,
      "loss": 2.2404,
      "step": 8502
    },
    {
      "epoch": 0.057772329089273045,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009938040498246716,
      "loss": 2.1755,
      "step": 8503
    },
    {
      "epoch": 0.0577791234358671,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000993802361255485,
      "loss": 2.2928,
      "step": 8504
    },
    {
      "epoch": 0.05778591778246116,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009938006724576735,
      "loss": 2.3239,
      "step": 8505
    },
    {
      "epoch": 0.057792712129055214,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000993798983431238,
      "loss": 2.1814,
      "step": 8506
    },
    {
      "epoch": 0.05779950647564928,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009937972941761793,
      "loss": 2.3639,
      "step": 8507
    },
    {
      "epoch": 0.057806300822243334,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009937956046924981,
      "loss": 2.2262,
      "step": 8508
    },
    {
      "epoch": 0.05781309516883739,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009937939149801953,
      "loss": 2.2018,
      "step": 8509
    },
    {
      "epoch": 0.05781988951543145,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009937922250392716,
      "loss": 2.2024,
      "step": 8510
    },
    {
      "epoch": 0.0578266838620255,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009937905348697273,
      "loss": 2.1834,
      "step": 8511
    },
    {
      "epoch": 0.05783347820861956,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000993788844471564,
      "loss": 2.2601,
      "step": 8512
    },
    {
      "epoch": 0.05784027255521362,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009937871538447823,
      "loss": 2.3219,
      "step": 8513
    },
    {
      "epoch": 0.05784706690180768,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009937854629893826,
      "loss": 2.3828,
      "step": 8514
    },
    {
      "epoch": 0.057853861248401736,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009937837719053658,
      "loss": 2.2751,
      "step": 8515
    },
    {
      "epoch": 0.05786065559499579,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000993782080592733,
      "loss": 2.41,
      "step": 8516
    },
    {
      "epoch": 0.05786744994158985,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009937803890514845,
      "loss": 2.3035,
      "step": 8517
    },
    {
      "epoch": 0.05787424428818391,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009937786972816214,
      "loss": 2.2525,
      "step": 8518
    },
    {
      "epoch": 0.05788103863477797,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009937770052831445,
      "loss": 2.2119,
      "step": 8519
    },
    {
      "epoch": 0.057887832981372025,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009937753130560545,
      "loss": 2.3057,
      "step": 8520
    },
    {
      "epoch": 0.05789462732796608,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009937736206003522,
      "loss": 2.1341,
      "step": 8521
    },
    {
      "epoch": 0.05790142167456014,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009937719279160383,
      "loss": 2.2587,
      "step": 8522
    },
    {
      "epoch": 0.057908216021154195,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993770235003114,
      "loss": 2.1654,
      "step": 8523
    },
    {
      "epoch": 0.05791501036774826,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009937685418615794,
      "loss": 2.3769,
      "step": 8524
    },
    {
      "epoch": 0.057921804714342315,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009937668484914354,
      "loss": 2.3667,
      "step": 8525
    },
    {
      "epoch": 0.05792859906093637,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009937651548926835,
      "loss": 2.3134,
      "step": 8526
    },
    {
      "epoch": 0.05793539340753043,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009937634610653239,
      "loss": 2.4556,
      "step": 8527
    },
    {
      "epoch": 0.057942187754124484,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009937617670093572,
      "loss": 2.2816,
      "step": 8528
    },
    {
      "epoch": 0.05794898210071855,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009937600727247847,
      "loss": 2.2002,
      "step": 8529
    },
    {
      "epoch": 0.057955776447312604,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993758378211607,
      "loss": 2.2411,
      "step": 8530
    },
    {
      "epoch": 0.05796257079390666,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009937566834698247,
      "loss": 2.186,
      "step": 8531
    },
    {
      "epoch": 0.05796936514050072,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009937549884994387,
      "loss": 2.3419,
      "step": 8532
    },
    {
      "epoch": 0.057976159487094774,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009937532933004497,
      "loss": 2.1695,
      "step": 8533
    },
    {
      "epoch": 0.05798295383368883,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009937515978728588,
      "loss": 2.1735,
      "step": 8534
    },
    {
      "epoch": 0.05798974818028289,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009937499022166665,
      "loss": 2.2632,
      "step": 8535
    },
    {
      "epoch": 0.05799654252687695,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009937482063318737,
      "loss": 2.3929,
      "step": 8536
    },
    {
      "epoch": 0.058003336873471006,
      "grad_norm": 0.15625,
      "learning_rate": 0.000993746510218481,
      "loss": 2.213,
      "step": 8537
    },
    {
      "epoch": 0.05801013122006506,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009937448138764894,
      "loss": 2.2019,
      "step": 8538
    },
    {
      "epoch": 0.05801692556665912,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009937431173058996,
      "loss": 2.3128,
      "step": 8539
    },
    {
      "epoch": 0.05802371991325318,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009937414205067124,
      "loss": 2.358,
      "step": 8540
    },
    {
      "epoch": 0.05803051425984724,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009937397234789287,
      "loss": 2.3542,
      "step": 8541
    },
    {
      "epoch": 0.058037308606441296,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000993738026222549,
      "loss": 2.5027,
      "step": 8542
    },
    {
      "epoch": 0.05804410295303535,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009937363287375743,
      "loss": 2.1494,
      "step": 8543
    },
    {
      "epoch": 0.05805089729962941,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009937346310240054,
      "loss": 2.46,
      "step": 8544
    },
    {
      "epoch": 0.058057691646223465,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000993732933081843,
      "loss": 2.327,
      "step": 8545
    },
    {
      "epoch": 0.05806448599281753,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009937312349110878,
      "loss": 2.5173,
      "step": 8546
    },
    {
      "epoch": 0.058071280339411585,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000993729536511741,
      "loss": 2.2314,
      "step": 8547
    },
    {
      "epoch": 0.05807807468600564,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009937278378838027,
      "loss": 2.1421,
      "step": 8548
    },
    {
      "epoch": 0.0580848690325997,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009937261390272744,
      "loss": 2.2114,
      "step": 8549
    },
    {
      "epoch": 0.058091663379193755,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009937244399421564,
      "loss": 2.2551,
      "step": 8550
    },
    {
      "epoch": 0.05809845772578782,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009937227406284497,
      "loss": 2.3518,
      "step": 8551
    },
    {
      "epoch": 0.058105252072381874,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009937210410861548,
      "loss": 2.2957,
      "step": 8552
    },
    {
      "epoch": 0.05811204641897593,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000993719341315273,
      "loss": 2.3271,
      "step": 8553
    },
    {
      "epoch": 0.05811884076556999,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009937176413158048,
      "loss": 2.304,
      "step": 8554
    },
    {
      "epoch": 0.058125635112164044,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000993715941087751,
      "loss": 2.3384,
      "step": 8555
    },
    {
      "epoch": 0.05813242945875811,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009937142406311122,
      "loss": 2.2842,
      "step": 8556
    },
    {
      "epoch": 0.058139223805352164,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009937125399458893,
      "loss": 2.306,
      "step": 8557
    },
    {
      "epoch": 0.05814601815194622,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009937108390320835,
      "loss": 2.1689,
      "step": 8558
    },
    {
      "epoch": 0.05815281249854028,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000993709137889695,
      "loss": 2.2092,
      "step": 8559
    },
    {
      "epoch": 0.05815960684513433,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000993707436518725,
      "loss": 2.2536,
      "step": 8560
    },
    {
      "epoch": 0.05816640119172839,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009937057349191738,
      "loss": 2.3566,
      "step": 8561
    },
    {
      "epoch": 0.05817319553832245,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009937040330910427,
      "loss": 2.247,
      "step": 8562
    },
    {
      "epoch": 0.05817998988491651,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009937023310343323,
      "loss": 2.2627,
      "step": 8563
    },
    {
      "epoch": 0.058186784231510566,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009937006287490434,
      "loss": 2.4716,
      "step": 8564
    },
    {
      "epoch": 0.05819357857810462,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009936989262351768,
      "loss": 2.2029,
      "step": 8565
    },
    {
      "epoch": 0.05820037292469868,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009936972234927332,
      "loss": 2.2713,
      "step": 8566
    },
    {
      "epoch": 0.05820716727129274,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009936955205217134,
      "loss": 2.2959,
      "step": 8567
    },
    {
      "epoch": 0.0582139616178868,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009936938173221183,
      "loss": 2.3111,
      "step": 8568
    },
    {
      "epoch": 0.058220755964480855,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009936921138939485,
      "loss": 2.2705,
      "step": 8569
    },
    {
      "epoch": 0.05822755031107491,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000993690410237205,
      "loss": 2.1497,
      "step": 8570
    },
    {
      "epoch": 0.05823434465766897,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009936887063518886,
      "loss": 2.3256,
      "step": 8571
    },
    {
      "epoch": 0.058241139004263025,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993687002238,
      "loss": 2.2163,
      "step": 8572
    },
    {
      "epoch": 0.05824793335085709,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009936852978955398,
      "loss": 2.4289,
      "step": 8573
    },
    {
      "epoch": 0.058254727697451145,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000993683593324509,
      "loss": 2.2857,
      "step": 8574
    },
    {
      "epoch": 0.0582615220440452,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009936818885249085,
      "loss": 2.2551,
      "step": 8575
    },
    {
      "epoch": 0.05826831639063926,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000993680183496739,
      "loss": 2.3013,
      "step": 8576
    },
    {
      "epoch": 0.058275110737233314,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000993678478240001,
      "loss": 2.2909,
      "step": 8577
    },
    {
      "epoch": 0.05828190508382738,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009936767727546955,
      "loss": 2.3627,
      "step": 8578
    },
    {
      "epoch": 0.058288699430421434,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009936750670408235,
      "loss": 2.3769,
      "step": 8579
    },
    {
      "epoch": 0.05829549377701549,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009936733610983854,
      "loss": 2.3766,
      "step": 8580
    },
    {
      "epoch": 0.05830228812360955,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009936716549273825,
      "loss": 2.4213,
      "step": 8581
    },
    {
      "epoch": 0.0583090824702036,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000993669948527815,
      "loss": 2.2391,
      "step": 8582
    },
    {
      "epoch": 0.05831587681679766,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009936682418996842,
      "loss": 2.3251,
      "step": 8583
    },
    {
      "epoch": 0.05832267116339172,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009936665350429904,
      "loss": 2.3773,
      "step": 8584
    },
    {
      "epoch": 0.05832946550998578,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000993664827957735,
      "loss": 2.2794,
      "step": 8585
    },
    {
      "epoch": 0.058336259856579836,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009936631206439181,
      "loss": 2.3116,
      "step": 8586
    },
    {
      "epoch": 0.05834305420317389,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009936614131015409,
      "loss": 2.3509,
      "step": 8587
    },
    {
      "epoch": 0.05834984854976795,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009936597053306043,
      "loss": 2.2807,
      "step": 8588
    },
    {
      "epoch": 0.05835664289636201,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009936579973311088,
      "loss": 2.219,
      "step": 8589
    },
    {
      "epoch": 0.05836343724295607,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009936562891030552,
      "loss": 2.2739,
      "step": 8590
    },
    {
      "epoch": 0.058370231589550126,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009936545806464446,
      "loss": 2.2821,
      "step": 8591
    },
    {
      "epoch": 0.05837702593614418,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009936528719612775,
      "loss": 2.3219,
      "step": 8592
    },
    {
      "epoch": 0.05838382028273824,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993651163047555,
      "loss": 2.2862,
      "step": 8593
    },
    {
      "epoch": 0.058390614629332295,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009936494539052772,
      "loss": 2.304,
      "step": 8594
    },
    {
      "epoch": 0.05839740897592636,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009936477445344457,
      "loss": 2.2208,
      "step": 8595
    },
    {
      "epoch": 0.058404203322520415,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009936460349350609,
      "loss": 2.2157,
      "step": 8596
    },
    {
      "epoch": 0.05841099766911447,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009936443251071236,
      "loss": 2.2299,
      "step": 8597
    },
    {
      "epoch": 0.05841779201570853,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009936426150506346,
      "loss": 2.2009,
      "step": 8598
    },
    {
      "epoch": 0.058424586362302584,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009936409047655948,
      "loss": 2.2987,
      "step": 8599
    },
    {
      "epoch": 0.05843138070889665,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009936391942520048,
      "loss": 2.3713,
      "step": 8600
    },
    {
      "epoch": 0.058438175055490704,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009936374835098657,
      "loss": 2.4027,
      "step": 8601
    },
    {
      "epoch": 0.05844496940208476,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009936357725391781,
      "loss": 2.1059,
      "step": 8602
    },
    {
      "epoch": 0.05845176374867882,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009936340613399426,
      "loss": 2.2916,
      "step": 8603
    },
    {
      "epoch": 0.058458558095272874,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009936323499121604,
      "loss": 2.2787,
      "step": 8604
    },
    {
      "epoch": 0.05846535244186693,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000993630638255832,
      "loss": 2.2267,
      "step": 8605
    },
    {
      "epoch": 0.058472146788460994,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009936289263709583,
      "loss": 2.3164,
      "step": 8606
    },
    {
      "epoch": 0.05847894113505505,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00099362721425754,
      "loss": 2.403,
      "step": 8607
    },
    {
      "epoch": 0.058485735481649107,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000993625501915578,
      "loss": 2.2233,
      "step": 8608
    },
    {
      "epoch": 0.05849252982824316,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993623789345073,
      "loss": 2.1283,
      "step": 8609
    },
    {
      "epoch": 0.05849932417483722,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993622076546026,
      "loss": 2.282,
      "step": 8610
    },
    {
      "epoch": 0.05850611852143128,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009936203635184375,
      "loss": 2.3481,
      "step": 8611
    },
    {
      "epoch": 0.05851291286802534,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009936186502623083,
      "loss": 2.5063,
      "step": 8612
    },
    {
      "epoch": 0.058519707214619396,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009936169367776395,
      "loss": 2.1753,
      "step": 8613
    },
    {
      "epoch": 0.05852650156121345,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009936152230644317,
      "loss": 2.3281,
      "step": 8614
    },
    {
      "epoch": 0.05853329590780751,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009936135091226858,
      "loss": 2.2607,
      "step": 8615
    },
    {
      "epoch": 0.05854009025440157,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009936117949524024,
      "loss": 2.2988,
      "step": 8616
    },
    {
      "epoch": 0.05854688460099563,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009936100805535823,
      "loss": 2.377,
      "step": 8617
    },
    {
      "epoch": 0.058553678947589685,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009936083659262266,
      "loss": 2.231,
      "step": 8618
    },
    {
      "epoch": 0.05856047329418374,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009936066510703358,
      "loss": 2.2709,
      "step": 8619
    },
    {
      "epoch": 0.0585672676407778,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009936049359859107,
      "loss": 2.2898,
      "step": 8620
    },
    {
      "epoch": 0.058574061987371855,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009936032206729524,
      "loss": 2.243,
      "step": 8621
    },
    {
      "epoch": 0.05858085633396592,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009936015051314612,
      "loss": 2.0555,
      "step": 8622
    },
    {
      "epoch": 0.058587650680559974,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009935997893614382,
      "loss": 2.3659,
      "step": 8623
    },
    {
      "epoch": 0.05859444502715403,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009935980733628842,
      "loss": 2.3578,
      "step": 8624
    },
    {
      "epoch": 0.05860123937374809,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009935963571358,
      "loss": 2.3814,
      "step": 8625
    },
    {
      "epoch": 0.058608033720342144,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009935946406801864,
      "loss": 2.2345,
      "step": 8626
    },
    {
      "epoch": 0.05861482806693621,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009935929239960441,
      "loss": 2.271,
      "step": 8627
    },
    {
      "epoch": 0.058621622413530264,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993591207083374,
      "loss": 2.164,
      "step": 8628
    },
    {
      "epoch": 0.05862841676012432,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009935894899421765,
      "loss": 2.2586,
      "step": 8629
    },
    {
      "epoch": 0.05863521110671838,
      "grad_norm": 0.15625,
      "learning_rate": 0.000993587772572453,
      "loss": 2.3654,
      "step": 8630
    },
    {
      "epoch": 0.05864200545331243,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009935860549742039,
      "loss": 2.1786,
      "step": 8631
    },
    {
      "epoch": 0.05864879979990649,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009935843371474302,
      "loss": 2.3357,
      "step": 8632
    },
    {
      "epoch": 0.05865559414650055,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009935826190921326,
      "loss": 2.2504,
      "step": 8633
    },
    {
      "epoch": 0.05866238849309461,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000993580900808312,
      "loss": 2.1446,
      "step": 8634
    },
    {
      "epoch": 0.058669182839688666,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000993579182295969,
      "loss": 2.2533,
      "step": 8635
    },
    {
      "epoch": 0.05867597718628272,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009935774635551045,
      "loss": 2.2608,
      "step": 8636
    },
    {
      "epoch": 0.05868277153287678,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009935757445857193,
      "loss": 2.3168,
      "step": 8637
    },
    {
      "epoch": 0.05868956587947084,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009935740253878141,
      "loss": 2.3002,
      "step": 8638
    },
    {
      "epoch": 0.0586963602260649,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00099357230596139,
      "loss": 2.1882,
      "step": 8639
    },
    {
      "epoch": 0.058703154572658955,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009935705863064474,
      "loss": 2.1134,
      "step": 8640
    },
    {
      "epoch": 0.05870994891925301,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009935688664229874,
      "loss": 2.1928,
      "step": 8641
    },
    {
      "epoch": 0.05871674326584707,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009935671463110105,
      "loss": 2.3002,
      "step": 8642
    },
    {
      "epoch": 0.058723537612441125,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009935654259705178,
      "loss": 2.1683,
      "step": 8643
    },
    {
      "epoch": 0.05873033195903519,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009935637054015098,
      "loss": 2.1941,
      "step": 8644
    },
    {
      "epoch": 0.058737126305629245,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009935619846039878,
      "loss": 2.2255,
      "step": 8645
    },
    {
      "epoch": 0.0587439206522233,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009935602635779519,
      "loss": 2.3638,
      "step": 8646
    },
    {
      "epoch": 0.05875071499881736,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009935585423234035,
      "loss": 2.2593,
      "step": 8647
    },
    {
      "epoch": 0.058757509345411414,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009935568208403432,
      "loss": 2.4317,
      "step": 8648
    },
    {
      "epoch": 0.05876430369200548,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009935550991287715,
      "loss": 2.1996,
      "step": 8649
    },
    {
      "epoch": 0.058771098038599534,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009935533771886895,
      "loss": 2.3164,
      "step": 8650
    },
    {
      "epoch": 0.05877789238519359,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009935516550200981,
      "loss": 2.24,
      "step": 8651
    },
    {
      "epoch": 0.05878468673178765,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009935499326229977,
      "loss": 2.248,
      "step": 8652
    },
    {
      "epoch": 0.058791481078381704,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009935482099973896,
      "loss": 2.3906,
      "step": 8653
    },
    {
      "epoch": 0.05879827542497576,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009935464871432742,
      "loss": 2.2307,
      "step": 8654
    },
    {
      "epoch": 0.05880506977156982,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009935447640606525,
      "loss": 2.3583,
      "step": 8655
    },
    {
      "epoch": 0.05881186411816388,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009935430407495252,
      "loss": 2.3424,
      "step": 8656
    },
    {
      "epoch": 0.058818658464757936,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009935413172098932,
      "loss": 2.2982,
      "step": 8657
    },
    {
      "epoch": 0.05882545281135199,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993539593441757,
      "loss": 2.3012,
      "step": 8658
    },
    {
      "epoch": 0.05883224715794605,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009935378694451177,
      "loss": 2.2596,
      "step": 8659
    },
    {
      "epoch": 0.05883904150454011,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009935361452199762,
      "loss": 2.2616,
      "step": 8660
    },
    {
      "epoch": 0.05884583585113417,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009935344207663331,
      "loss": 2.2661,
      "step": 8661
    },
    {
      "epoch": 0.058852630197728226,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000993532696084189,
      "loss": 2.2152,
      "step": 8662
    },
    {
      "epoch": 0.05885942454432228,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000993530971173545,
      "loss": 2.1895,
      "step": 8663
    },
    {
      "epoch": 0.05886621889091634,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000993529246034402,
      "loss": 2.4561,
      "step": 8664
    },
    {
      "epoch": 0.058873013237510395,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009935275206667607,
      "loss": 2.3634,
      "step": 8665
    },
    {
      "epoch": 0.05887980758410446,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009935257950706216,
      "loss": 2.2647,
      "step": 8666
    },
    {
      "epoch": 0.058886601930698515,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009935240692459857,
      "loss": 2.221,
      "step": 8667
    },
    {
      "epoch": 0.05889339627729257,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009935223431928538,
      "loss": 2.2196,
      "step": 8668
    },
    {
      "epoch": 0.05890019062388663,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009935206169112267,
      "loss": 2.3776,
      "step": 8669
    },
    {
      "epoch": 0.058906984970480684,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009935188904011054,
      "loss": 2.3743,
      "step": 8670
    },
    {
      "epoch": 0.05891377931707475,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009935171636624904,
      "loss": 2.2976,
      "step": 8671
    },
    {
      "epoch": 0.058920573663668804,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009935154366953825,
      "loss": 2.182,
      "step": 8672
    },
    {
      "epoch": 0.05892736801026286,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009935137094997828,
      "loss": 2.2104,
      "step": 8673
    },
    {
      "epoch": 0.05893416235685692,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993511982075692,
      "loss": 2.2556,
      "step": 8674
    },
    {
      "epoch": 0.058940956703450974,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009935102544231106,
      "loss": 2.1589,
      "step": 8675
    },
    {
      "epoch": 0.05894775105004504,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009935085265420395,
      "loss": 2.3219,
      "step": 8676
    },
    {
      "epoch": 0.058954545396639094,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009935067984324798,
      "loss": 2.3351,
      "step": 8677
    },
    {
      "epoch": 0.05896133974323315,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000993505070094432,
      "loss": 2.3606,
      "step": 8678
    },
    {
      "epoch": 0.05896813408982721,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000993503341527897,
      "loss": 2.329,
      "step": 8679
    },
    {
      "epoch": 0.05897492843642126,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009935016127328757,
      "loss": 2.3594,
      "step": 8680
    },
    {
      "epoch": 0.05898172278301532,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009934998837093688,
      "loss": 2.2689,
      "step": 8681
    },
    {
      "epoch": 0.05898851712960938,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009934981544573772,
      "loss": 2.2226,
      "step": 8682
    },
    {
      "epoch": 0.05899531147620344,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009934964249769015,
      "loss": 2.3281,
      "step": 8683
    },
    {
      "epoch": 0.059002105822797496,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009934946952679428,
      "loss": 2.2198,
      "step": 8684
    },
    {
      "epoch": 0.05900890016939155,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009934929653305014,
      "loss": 2.2995,
      "step": 8685
    },
    {
      "epoch": 0.05901569451598561,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009934912351645786,
      "loss": 2.2319,
      "step": 8686
    },
    {
      "epoch": 0.05902248886257967,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0009934895047701748,
      "loss": 2.3292,
      "step": 8687
    },
    {
      "epoch": 0.05902928320917373,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009934877741472911,
      "loss": 2.311,
      "step": 8688
    },
    {
      "epoch": 0.059036077555767785,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009934860432959285,
      "loss": 2.4211,
      "step": 8689
    },
    {
      "epoch": 0.05904287190236184,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009934843122160872,
      "loss": 2.2819,
      "step": 8690
    },
    {
      "epoch": 0.0590496662489559,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009934825809077685,
      "loss": 2.3432,
      "step": 8691
    },
    {
      "epoch": 0.059056460595549955,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993480849370973,
      "loss": 2.1411,
      "step": 8692
    },
    {
      "epoch": 0.05906325494214402,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009934791176057015,
      "loss": 2.295,
      "step": 8693
    },
    {
      "epoch": 0.059070049288738075,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009934773856119547,
      "loss": 2.2168,
      "step": 8694
    },
    {
      "epoch": 0.05907684363533213,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009934756533897338,
      "loss": 2.2803,
      "step": 8695
    },
    {
      "epoch": 0.05908363798192619,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009934739209390391,
      "loss": 2.2453,
      "step": 8696
    },
    {
      "epoch": 0.059090432328520244,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993472188259872,
      "loss": 2.2442,
      "step": 8697
    },
    {
      "epoch": 0.05909722667511431,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009934704553522325,
      "loss": 2.1054,
      "step": 8698
    },
    {
      "epoch": 0.059104021021708364,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000993468722216122,
      "loss": 2.1391,
      "step": 8699
    },
    {
      "epoch": 0.05911081536830242,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009934669888515412,
      "loss": 2.1809,
      "step": 8700
    },
    {
      "epoch": 0.05911760971489648,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009934652552584906,
      "loss": 2.1941,
      "step": 8701
    },
    {
      "epoch": 0.05912440406149053,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009934635214369713,
      "loss": 2.2037,
      "step": 8702
    },
    {
      "epoch": 0.05913119840808459,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009934617873869841,
      "loss": 2.3504,
      "step": 8703
    },
    {
      "epoch": 0.05913799275467865,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00099346005310853,
      "loss": 2.182,
      "step": 8704
    },
    {
      "epoch": 0.05914478710127271,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009934583186016093,
      "loss": 2.3124,
      "step": 8705
    },
    {
      "epoch": 0.059151581447866766,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009934565838662231,
      "loss": 2.321,
      "step": 8706
    },
    {
      "epoch": 0.05915837579446082,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009934548489023722,
      "loss": 2.1958,
      "step": 8707
    },
    {
      "epoch": 0.05916517014105488,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009934531137100573,
      "loss": 2.3603,
      "step": 8708
    },
    {
      "epoch": 0.05917196448764894,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009934513782892793,
      "loss": 2.3648,
      "step": 8709
    },
    {
      "epoch": 0.059178758834243,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000993449642640039,
      "loss": 2.1686,
      "step": 8710
    },
    {
      "epoch": 0.059185553180837056,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009934479067623373,
      "loss": 2.3147,
      "step": 8711
    },
    {
      "epoch": 0.05919234752743111,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009934461706561746,
      "loss": 2.1638,
      "step": 8712
    },
    {
      "epoch": 0.05919914187402517,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993444434321552,
      "loss": 2.2404,
      "step": 8713
    },
    {
      "epoch": 0.059205936220619225,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009934426977584706,
      "loss": 2.3675,
      "step": 8714
    },
    {
      "epoch": 0.05921273056721329,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009934409609669305,
      "loss": 2.2547,
      "step": 8715
    },
    {
      "epoch": 0.059219524913807345,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000993439223946933,
      "loss": 2.3064,
      "step": 8716
    },
    {
      "epoch": 0.0592263192604014,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993437486698479,
      "loss": 2.2495,
      "step": 8717
    },
    {
      "epoch": 0.05923311360699546,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009934357492215689,
      "loss": 2.3147,
      "step": 8718
    },
    {
      "epoch": 0.059239907953589514,
      "grad_norm": 0.125,
      "learning_rate": 0.0009934340115162036,
      "loss": 2.041,
      "step": 8719
    },
    {
      "epoch": 0.05924670230018358,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009934322735823842,
      "loss": 2.2528,
      "step": 8720
    },
    {
      "epoch": 0.059253496646777634,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009934305354201115,
      "loss": 2.382,
      "step": 8721
    },
    {
      "epoch": 0.05926029099337169,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009934287970293858,
      "loss": 2.1443,
      "step": 8722
    },
    {
      "epoch": 0.05926708533996575,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009934270584102084,
      "loss": 2.2562,
      "step": 8723
    },
    {
      "epoch": 0.059273879686559804,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009934253195625798,
      "loss": 2.1737,
      "step": 8724
    },
    {
      "epoch": 0.05928067403315386,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000993423580486501,
      "loss": 2.2777,
      "step": 8725
    },
    {
      "epoch": 0.059287468379747923,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009934218411819728,
      "loss": 2.257,
      "step": 8726
    },
    {
      "epoch": 0.05929426272634198,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000993420101648996,
      "loss": 2.3222,
      "step": 8727
    },
    {
      "epoch": 0.059301057072936036,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009934183618875712,
      "loss": 2.2407,
      "step": 8728
    },
    {
      "epoch": 0.05930785141953009,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009934166218976991,
      "loss": 2.2874,
      "step": 8729
    },
    {
      "epoch": 0.05931464576612415,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009934148816793812,
      "loss": 2.397,
      "step": 8730
    },
    {
      "epoch": 0.05932144011271821,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009934131412326177,
      "loss": 2.3032,
      "step": 8731
    },
    {
      "epoch": 0.05932823445931227,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009934114005574095,
      "loss": 2.1601,
      "step": 8732
    },
    {
      "epoch": 0.059335028805906326,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009934096596537576,
      "loss": 2.3476,
      "step": 8733
    },
    {
      "epoch": 0.05934182315250038,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009934079185216628,
      "loss": 2.4081,
      "step": 8734
    },
    {
      "epoch": 0.05934861749909444,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009934061771611257,
      "loss": 2.2268,
      "step": 8735
    },
    {
      "epoch": 0.0593554118456885,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000993404435572147,
      "loss": 2.2219,
      "step": 8736
    },
    {
      "epoch": 0.05936220619228256,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009934026937547277,
      "loss": 2.4447,
      "step": 8737
    },
    {
      "epoch": 0.059369000538876615,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009934009517088688,
      "loss": 2.4829,
      "step": 8738
    },
    {
      "epoch": 0.05937579488547067,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009933992094345709,
      "loss": 2.2166,
      "step": 8739
    },
    {
      "epoch": 0.05938258923206473,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009933974669318348,
      "loss": 2.3654,
      "step": 8740
    },
    {
      "epoch": 0.059389383578658785,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009933957242006612,
      "loss": 2.2637,
      "step": 8741
    },
    {
      "epoch": 0.05939617792525285,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009933939812410513,
      "loss": 2.2402,
      "step": 8742
    },
    {
      "epoch": 0.059402972271846904,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009933922380530053,
      "loss": 2.4013,
      "step": 8743
    },
    {
      "epoch": 0.05940976661844096,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009933904946365247,
      "loss": 2.235,
      "step": 8744
    },
    {
      "epoch": 0.05941656096503502,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009933887509916098,
      "loss": 2.1187,
      "step": 8745
    },
    {
      "epoch": 0.059423355311629074,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009933870071182616,
      "loss": 2.3678,
      "step": 8746
    },
    {
      "epoch": 0.05943014965822314,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009933852630164808,
      "loss": 2.4043,
      "step": 8747
    },
    {
      "epoch": 0.059436944004817194,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009933835186862682,
      "loss": 2.3774,
      "step": 8748
    },
    {
      "epoch": 0.05944373835141125,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009933817741276248,
      "loss": 2.2078,
      "step": 8749
    },
    {
      "epoch": 0.05945053269800531,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009933800293405513,
      "loss": 2.1322,
      "step": 8750
    },
    {
      "epoch": 0.05945732704459936,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009933782843250486,
      "loss": 2.2948,
      "step": 8751
    },
    {
      "epoch": 0.05946412139119342,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009933765390811172,
      "loss": 2.0933,
      "step": 8752
    },
    {
      "epoch": 0.05947091573778748,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009933747936087583,
      "loss": 2.2035,
      "step": 8753
    },
    {
      "epoch": 0.05947771008438154,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009933730479079724,
      "loss": 2.3179,
      "step": 8754
    },
    {
      "epoch": 0.059484504430975596,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009933713019787604,
      "loss": 2.1569,
      "step": 8755
    },
    {
      "epoch": 0.05949129877756965,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009933695558211233,
      "loss": 2.3008,
      "step": 8756
    },
    {
      "epoch": 0.05949809312416371,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009933678094350617,
      "loss": 2.3526,
      "step": 8757
    },
    {
      "epoch": 0.05950488747075777,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009933660628205765,
      "loss": 2.2457,
      "step": 8758
    },
    {
      "epoch": 0.05951168181735183,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009933643159776683,
      "loss": 2.2921,
      "step": 8759
    },
    {
      "epoch": 0.059518476163945885,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009933625689063383,
      "loss": 2.2382,
      "step": 8760
    },
    {
      "epoch": 0.05952527051053994,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000993360821606587,
      "loss": 2.3044,
      "step": 8761
    },
    {
      "epoch": 0.059532064857134,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009933590740784154,
      "loss": 2.2998,
      "step": 8762
    },
    {
      "epoch": 0.059538859203728055,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000993357326321824,
      "loss": 2.312,
      "step": 8763
    },
    {
      "epoch": 0.05954565355032212,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009933555783368138,
      "loss": 2.2592,
      "step": 8764
    },
    {
      "epoch": 0.059552447896916175,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009933538301233858,
      "loss": 2.1878,
      "step": 8765
    },
    {
      "epoch": 0.05955924224351023,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009933520816815405,
      "loss": 2.3083,
      "step": 8766
    },
    {
      "epoch": 0.05956603659010429,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000993350333011279,
      "loss": 2.3193,
      "step": 8767
    },
    {
      "epoch": 0.059572830936698344,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009933485841126018,
      "loss": 2.1953,
      "step": 8768
    },
    {
      "epoch": 0.05957962528329241,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00099334683498551,
      "loss": 2.4231,
      "step": 8769
    },
    {
      "epoch": 0.059586419629886464,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000993345085630004,
      "loss": 2.268,
      "step": 8770
    },
    {
      "epoch": 0.05959321397648052,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993343336046085,
      "loss": 2.2954,
      "step": 8771
    },
    {
      "epoch": 0.05960000832307458,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009933415862337538,
      "loss": 2.4927,
      "step": 8772
    },
    {
      "epoch": 0.05960680266966863,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000993339836193011,
      "loss": 2.4161,
      "step": 8773
    },
    {
      "epoch": 0.05961359701626269,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009933380859238577,
      "loss": 2.3856,
      "step": 8774
    },
    {
      "epoch": 0.05962039136285675,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009933363354262942,
      "loss": 2.2507,
      "step": 8775
    },
    {
      "epoch": 0.05962718570945081,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000993334584700322,
      "loss": 2.267,
      "step": 8776
    },
    {
      "epoch": 0.059633980056044866,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009933328337459412,
      "loss": 2.2644,
      "step": 8777
    },
    {
      "epoch": 0.05964077440263892,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009933310825631532,
      "loss": 2.1801,
      "step": 8778
    },
    {
      "epoch": 0.05964756874923298,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009933293311519585,
      "loss": 2.3423,
      "step": 8779
    },
    {
      "epoch": 0.05965436309582704,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000993327579512358,
      "loss": 2.202,
      "step": 8780
    },
    {
      "epoch": 0.0596611574424211,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009933258276443523,
      "loss": 2.133,
      "step": 8781
    },
    {
      "epoch": 0.059667951789015156,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009933240755479428,
      "loss": 2.3541,
      "step": 8782
    },
    {
      "epoch": 0.05967474613560921,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009933223232231295,
      "loss": 2.1798,
      "step": 8783
    },
    {
      "epoch": 0.05968154048220327,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009933205706699137,
      "loss": 2.2205,
      "step": 8784
    },
    {
      "epoch": 0.059688334828797325,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009933188178882963,
      "loss": 2.1898,
      "step": 8785
    },
    {
      "epoch": 0.05969512917539139,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000993317064878278,
      "loss": 2.3455,
      "step": 8786
    },
    {
      "epoch": 0.059701923521985445,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009933153116398591,
      "loss": 2.2677,
      "step": 8787
    },
    {
      "epoch": 0.0597087178685795,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009933135581730412,
      "loss": 2.3567,
      "step": 8788
    },
    {
      "epoch": 0.05971551221517356,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009933118044778248,
      "loss": 2.2753,
      "step": 8789
    },
    {
      "epoch": 0.059722306561767614,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009933100505542108,
      "loss": 2.3541,
      "step": 8790
    },
    {
      "epoch": 0.05972910090836168,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009933082964021995,
      "loss": 2.2432,
      "step": 8791
    },
    {
      "epoch": 0.059735895254955734,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009933065420217925,
      "loss": 2.1344,
      "step": 8792
    },
    {
      "epoch": 0.05974268960154979,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00099330478741299,
      "loss": 2.3117,
      "step": 8793
    },
    {
      "epoch": 0.05974948394814385,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009933030325757933,
      "loss": 2.1845,
      "step": 8794
    },
    {
      "epoch": 0.059756278294737904,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009933012775102026,
      "loss": 2.3671,
      "step": 8795
    },
    {
      "epoch": 0.05976307264133197,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009932995222162193,
      "loss": 2.3692,
      "step": 8796
    },
    {
      "epoch": 0.059769866987926024,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009932977666938438,
      "loss": 2.3759,
      "step": 8797
    },
    {
      "epoch": 0.05977666133452008,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009932960109430772,
      "loss": 2.1543,
      "step": 8798
    },
    {
      "epoch": 0.05978345568111414,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009932942549639202,
      "loss": 2.2235,
      "step": 8799
    },
    {
      "epoch": 0.05979025002770819,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009932924987563737,
      "loss": 2.3487,
      "step": 8800
    },
    {
      "epoch": 0.05979704437430225,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009932907423204382,
      "loss": 2.2305,
      "step": 8801
    },
    {
      "epoch": 0.05980383872089631,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009932889856561148,
      "loss": 2.1945,
      "step": 8802
    },
    {
      "epoch": 0.05981063306749037,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009932872287634045,
      "loss": 2.4202,
      "step": 8803
    },
    {
      "epoch": 0.059817427414084426,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009932854716423077,
      "loss": 2.3979,
      "step": 8804
    },
    {
      "epoch": 0.05982422176067848,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009932837142928256,
      "loss": 2.1974,
      "step": 8805
    },
    {
      "epoch": 0.05983101610727254,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009932819567149585,
      "loss": 2.28,
      "step": 8806
    },
    {
      "epoch": 0.0598378104538666,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009932801989087075,
      "loss": 2.3126,
      "step": 8807
    },
    {
      "epoch": 0.05984460480046066,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009932784408740736,
      "loss": 2.2724,
      "step": 8808
    },
    {
      "epoch": 0.059851399147054715,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009932766826110573,
      "loss": 2.2748,
      "step": 8809
    },
    {
      "epoch": 0.05985819349364877,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009932749241196596,
      "loss": 2.3648,
      "step": 8810
    },
    {
      "epoch": 0.05986498784024283,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009932731653998813,
      "loss": 2.1071,
      "step": 8811
    },
    {
      "epoch": 0.059871782186836885,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009932714064517232,
      "loss": 2.1165,
      "step": 8812
    },
    {
      "epoch": 0.05987857653343095,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009932696472751859,
      "loss": 2.3879,
      "step": 8813
    },
    {
      "epoch": 0.059885370880025005,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009932678878702705,
      "loss": 2.2092,
      "step": 8814
    },
    {
      "epoch": 0.05989216522661906,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000993266128236978,
      "loss": 2.3761,
      "step": 8815
    },
    {
      "epoch": 0.05989895957321312,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009932643683753086,
      "loss": 2.2364,
      "step": 8816
    },
    {
      "epoch": 0.059905753919807174,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009932626082852637,
      "loss": 2.1948,
      "step": 8817
    },
    {
      "epoch": 0.05991254826640124,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009932608479668436,
      "loss": 2.2074,
      "step": 8818
    },
    {
      "epoch": 0.059919342612995294,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009932590874200495,
      "loss": 2.3041,
      "step": 8819
    },
    {
      "epoch": 0.05992613695958935,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009932573266448821,
      "loss": 2.3335,
      "step": 8820
    },
    {
      "epoch": 0.05993293130618341,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009932555656413423,
      "loss": 2.3043,
      "step": 8821
    },
    {
      "epoch": 0.05993972565277746,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009932538044094307,
      "loss": 2.4121,
      "step": 8822
    },
    {
      "epoch": 0.05994651999937152,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009932520429491484,
      "loss": 2.1311,
      "step": 8823
    },
    {
      "epoch": 0.05995331434596558,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009932502812604958,
      "loss": 2.2979,
      "step": 8824
    },
    {
      "epoch": 0.05996010869255964,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009932485193434742,
      "loss": 2.3614,
      "step": 8825
    },
    {
      "epoch": 0.059966903039153696,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009932467571980841,
      "loss": 2.1746,
      "step": 8826
    },
    {
      "epoch": 0.05997369738574775,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009932449948243263,
      "loss": 2.2411,
      "step": 8827
    },
    {
      "epoch": 0.05998049173234181,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000993243232222202,
      "loss": 2.2329,
      "step": 8828
    },
    {
      "epoch": 0.05998728607893587,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009932414693917117,
      "loss": 2.27,
      "step": 8829
    },
    {
      "epoch": 0.05999408042552993,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000993239706332856,
      "loss": 2.1664,
      "step": 8830
    },
    {
      "epoch": 0.060000874772123985,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009932379430456363,
      "loss": 2.2075,
      "step": 8831
    },
    {
      "epoch": 0.06000766911871804,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009932361795300527,
      "loss": 2.4007,
      "step": 8832
    },
    {
      "epoch": 0.0600144634653121,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009932344157861067,
      "loss": 2.3003,
      "step": 8833
    },
    {
      "epoch": 0.060021257811906155,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009932326518137986,
      "loss": 2.3087,
      "step": 8834
    },
    {
      "epoch": 0.06002805215850022,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009932308876131296,
      "loss": 2.3029,
      "step": 8835
    },
    {
      "epoch": 0.060034846505094275,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009932291231841003,
      "loss": 2.2091,
      "step": 8836
    },
    {
      "epoch": 0.06004164085168833,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009932273585267115,
      "loss": 2.3308,
      "step": 8837
    },
    {
      "epoch": 0.06004843519828239,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009932255936409642,
      "loss": 2.2635,
      "step": 8838
    },
    {
      "epoch": 0.060055229544876444,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009932238285268592,
      "loss": 2.3695,
      "step": 8839
    },
    {
      "epoch": 0.06006202389147051,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009932220631843972,
      "loss": 2.3033,
      "step": 8840
    },
    {
      "epoch": 0.060068818238064564,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009932202976135787,
      "loss": 2.002,
      "step": 8841
    },
    {
      "epoch": 0.06007561258465862,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009932185318144052,
      "loss": 2.4174,
      "step": 8842
    },
    {
      "epoch": 0.06008240693125268,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000993216765786877,
      "loss": 2.2441,
      "step": 8843
    },
    {
      "epoch": 0.060089201277846734,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009932149995309952,
      "loss": 2.2979,
      "step": 8844
    },
    {
      "epoch": 0.06009599562444079,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009932132330467604,
      "loss": 2.3583,
      "step": 8845
    },
    {
      "epoch": 0.06010278997103485,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009932114663341737,
      "loss": 2.2233,
      "step": 8846
    },
    {
      "epoch": 0.06010958431762891,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009932096993932355,
      "loss": 2.2616,
      "step": 8847
    },
    {
      "epoch": 0.060116378664222966,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009932079322239472,
      "loss": 2.3223,
      "step": 8848
    },
    {
      "epoch": 0.06012317301081702,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000993206164826309,
      "loss": 2.3652,
      "step": 8849
    },
    {
      "epoch": 0.06012996735741108,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009932043972003222,
      "loss": 2.1456,
      "step": 8850
    },
    {
      "epoch": 0.06013676170400514,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009932026293459873,
      "loss": 2.3491,
      "step": 8851
    },
    {
      "epoch": 0.0601435560505992,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009932008612633053,
      "loss": 2.2503,
      "step": 8852
    },
    {
      "epoch": 0.060150350397193256,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009931990929522767,
      "loss": 2.2489,
      "step": 8853
    },
    {
      "epoch": 0.06015714474378731,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000993197324412903,
      "loss": 2.2375,
      "step": 8854
    },
    {
      "epoch": 0.06016393909038137,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009931955556451843,
      "loss": 2.1091,
      "step": 8855
    },
    {
      "epoch": 0.06017073343697543,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009931937866491216,
      "loss": 2.4117,
      "step": 8856
    },
    {
      "epoch": 0.06017752778356949,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009931920174247161,
      "loss": 2.4889,
      "step": 8857
    },
    {
      "epoch": 0.060184322130163545,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009931902479719684,
      "loss": 2.1431,
      "step": 8858
    },
    {
      "epoch": 0.0601911164767576,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993188478290879,
      "loss": 2.3106,
      "step": 8859
    },
    {
      "epoch": 0.06019791082335166,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000993186708381449,
      "loss": 2.3001,
      "step": 8860
    },
    {
      "epoch": 0.060204705169945714,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009931849382436794,
      "loss": 2.348,
      "step": 8861
    },
    {
      "epoch": 0.06021149951653978,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009931831678775707,
      "loss": 2.2743,
      "step": 8862
    },
    {
      "epoch": 0.060218293863133834,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993181397283124,
      "loss": 2.2183,
      "step": 8863
    },
    {
      "epoch": 0.06022508820972789,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009931796264603398,
      "loss": 2.1542,
      "step": 8864
    },
    {
      "epoch": 0.06023188255632195,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009931778554092192,
      "loss": 2.2237,
      "step": 8865
    },
    {
      "epoch": 0.060238676902916004,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009931760841297627,
      "loss": 2.4091,
      "step": 8866
    },
    {
      "epoch": 0.06024547124951007,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009931743126219716,
      "loss": 2.3073,
      "step": 8867
    },
    {
      "epoch": 0.060252265596104124,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009931725408858464,
      "loss": 2.2479,
      "step": 8868
    },
    {
      "epoch": 0.06025905994269818,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000993170768921388,
      "loss": 2.1512,
      "step": 8869
    },
    {
      "epoch": 0.06026585428929224,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000993168996728597,
      "loss": 2.3015,
      "step": 8870
    },
    {
      "epoch": 0.06027264863588629,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009931672243074746,
      "loss": 2.4486,
      "step": 8871
    },
    {
      "epoch": 0.06027944298248035,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009931654516580214,
      "loss": 2.4275,
      "step": 8872
    },
    {
      "epoch": 0.06028623732907441,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000993163678780238,
      "loss": 2.2386,
      "step": 8873
    },
    {
      "epoch": 0.06029303167566847,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009931619056741258,
      "loss": 2.4374,
      "step": 8874
    },
    {
      "epoch": 0.060299826022262526,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000993160132339685,
      "loss": 2.2689,
      "step": 8875
    },
    {
      "epoch": 0.06030662036885658,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009931583587769169,
      "loss": 2.0973,
      "step": 8876
    },
    {
      "epoch": 0.06031341471545064,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000993156584985822,
      "loss": 2.3892,
      "step": 8877
    },
    {
      "epoch": 0.0603202090620447,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009931548109664015,
      "loss": 2.3664,
      "step": 8878
    },
    {
      "epoch": 0.06032700340863876,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009931530367186556,
      "loss": 2.2904,
      "step": 8879
    },
    {
      "epoch": 0.060333797755232815,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009931512622425858,
      "loss": 2.4228,
      "step": 8880
    },
    {
      "epoch": 0.06034059210182687,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009931494875381926,
      "loss": 2.233,
      "step": 8881
    },
    {
      "epoch": 0.06034738644842093,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009931477126054768,
      "loss": 2.2251,
      "step": 8882
    },
    {
      "epoch": 0.060354180795014985,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000993145937444439,
      "loss": 2.2857,
      "step": 8883
    },
    {
      "epoch": 0.06036097514160905,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009931441620550808,
      "loss": 2.2977,
      "step": 8884
    },
    {
      "epoch": 0.060367769488203105,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000993142386437402,
      "loss": 2.2109,
      "step": 8885
    },
    {
      "epoch": 0.06037456383479716,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009931406105914043,
      "loss": 2.3872,
      "step": 8886
    },
    {
      "epoch": 0.06038135818139122,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000993138834517088,
      "loss": 2.3022,
      "step": 8887
    },
    {
      "epoch": 0.060388152527985274,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009931370582144541,
      "loss": 2.1925,
      "step": 8888
    },
    {
      "epoch": 0.06039494687457934,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009931352816835034,
      "loss": 2.1866,
      "step": 8889
    },
    {
      "epoch": 0.060401741221173394,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009931335049242367,
      "loss": 2.2412,
      "step": 8890
    },
    {
      "epoch": 0.06040853556776745,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009931317279366547,
      "loss": 2.381,
      "step": 8891
    },
    {
      "epoch": 0.06041532991436151,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009931299507207586,
      "loss": 2.345,
      "step": 8892
    },
    {
      "epoch": 0.06042212426095556,
      "grad_norm": 0.140625,
      "learning_rate": 0.000993128173276549,
      "loss": 2.1439,
      "step": 8893
    },
    {
      "epoch": 0.06042891860754962,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009931263956040265,
      "loss": 2.2167,
      "step": 8894
    },
    {
      "epoch": 0.06043571295414368,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009931246177031924,
      "loss": 2.2982,
      "step": 8895
    },
    {
      "epoch": 0.06044250730073774,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000993122839574047,
      "loss": 2.3741,
      "step": 8896
    },
    {
      "epoch": 0.060449301647331796,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009931210612165915,
      "loss": 2.3926,
      "step": 8897
    },
    {
      "epoch": 0.06045609599392585,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009931192826308265,
      "loss": 2.1596,
      "step": 8898
    },
    {
      "epoch": 0.06046289034051991,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009931175038167528,
      "loss": 2.145,
      "step": 8899
    },
    {
      "epoch": 0.06046968468711397,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009931157247743715,
      "loss": 2.2488,
      "step": 8900
    },
    {
      "epoch": 0.06047647903370803,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009931139455036835,
      "loss": 2.3171,
      "step": 8901
    },
    {
      "epoch": 0.060483273380302086,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009931121660046891,
      "loss": 2.2878,
      "step": 8902
    },
    {
      "epoch": 0.06049006772689614,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009931103862773896,
      "loss": 2.2709,
      "step": 8903
    },
    {
      "epoch": 0.0604968620734902,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009931086063217857,
      "loss": 2.3233,
      "step": 8904
    },
    {
      "epoch": 0.060503656420084255,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000993106826137878,
      "loss": 2.2401,
      "step": 8905
    },
    {
      "epoch": 0.06051045076667832,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009931050457256676,
      "loss": 2.0238,
      "step": 8906
    },
    {
      "epoch": 0.060517245113272375,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000993103265085155,
      "loss": 2.2134,
      "step": 8907
    },
    {
      "epoch": 0.06052403945986643,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009931014842163414,
      "loss": 2.3992,
      "step": 8908
    },
    {
      "epoch": 0.06053083380646049,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009930997031192274,
      "loss": 2.1867,
      "step": 8909
    },
    {
      "epoch": 0.060537628153054544,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000993097921793814,
      "loss": 2.3222,
      "step": 8910
    },
    {
      "epoch": 0.06054442249964861,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000993096140240102,
      "loss": 2.2035,
      "step": 8911
    },
    {
      "epoch": 0.060551216846242664,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000993094358458092,
      "loss": 2.411,
      "step": 8912
    },
    {
      "epoch": 0.06055801119283672,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000993092576447785,
      "loss": 2.2638,
      "step": 8913
    },
    {
      "epoch": 0.06056480553943078,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009930907942091818,
      "loss": 2.1605,
      "step": 8914
    },
    {
      "epoch": 0.060571599886024834,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000993089011742283,
      "loss": 2.2931,
      "step": 8915
    },
    {
      "epoch": 0.0605783942326189,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009930872290470898,
      "loss": 2.0595,
      "step": 8916
    },
    {
      "epoch": 0.060585188579212954,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000993085446123603,
      "loss": 2.2447,
      "step": 8917
    },
    {
      "epoch": 0.06059198292580701,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009930836629718231,
      "loss": 2.3256,
      "step": 8918
    },
    {
      "epoch": 0.060598777272401066,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009930818795917513,
      "loss": 2.2118,
      "step": 8919
    },
    {
      "epoch": 0.06060557161899512,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000993080095983388,
      "loss": 2.4732,
      "step": 8920
    },
    {
      "epoch": 0.06061236596558918,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009930783121467343,
      "loss": 2.3142,
      "step": 8921
    },
    {
      "epoch": 0.06061916031218324,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009930765280817913,
      "loss": 2.0853,
      "step": 8922
    },
    {
      "epoch": 0.0606259546587773,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009930747437885592,
      "loss": 2.2661,
      "step": 8923
    },
    {
      "epoch": 0.060632749005371356,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009930729592670393,
      "loss": 2.2498,
      "step": 8924
    },
    {
      "epoch": 0.06063954335196541,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009930711745172324,
      "loss": 2.3961,
      "step": 8925
    },
    {
      "epoch": 0.06064633769855947,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009930693895391389,
      "loss": 2.1182,
      "step": 8926
    },
    {
      "epoch": 0.06065313204515353,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099306760433276,
      "loss": 2.2897,
      "step": 8927
    },
    {
      "epoch": 0.06065992639174759,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009930658188980965,
      "loss": 2.3005,
      "step": 8928
    },
    {
      "epoch": 0.060666720738341645,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009930640332351493,
      "loss": 2.4006,
      "step": 8929
    },
    {
      "epoch": 0.0606735150849357,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000993062247343919,
      "loss": 2.2623,
      "step": 8930
    },
    {
      "epoch": 0.06068030943152976,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009930604612244066,
      "loss": 2.1459,
      "step": 8931
    },
    {
      "epoch": 0.060687103778123815,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009930586748766127,
      "loss": 2.2416,
      "step": 8932
    },
    {
      "epoch": 0.06069389812471788,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009930568883005384,
      "loss": 2.3767,
      "step": 8933
    },
    {
      "epoch": 0.060700692471311934,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009930551014961844,
      "loss": 2.2969,
      "step": 8934
    },
    {
      "epoch": 0.06070748681790599,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009930533144635516,
      "loss": 2.2805,
      "step": 8935
    },
    {
      "epoch": 0.06071428116450005,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009930515272026405,
      "loss": 2.4419,
      "step": 8936
    },
    {
      "epoch": 0.060721075511094104,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009930497397134526,
      "loss": 2.4366,
      "step": 8937
    },
    {
      "epoch": 0.06072786985768817,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000993047951995988,
      "loss": 2.1624,
      "step": 8938
    },
    {
      "epoch": 0.060734664204282224,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000993046164050248,
      "loss": 2.3185,
      "step": 8939
    },
    {
      "epoch": 0.06074145855087628,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009930443758762331,
      "loss": 2.216,
      "step": 8940
    },
    {
      "epoch": 0.06074825289747034,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009930425874739447,
      "loss": 2.2935,
      "step": 8941
    },
    {
      "epoch": 0.06075504724406439,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009930407988433829,
      "loss": 2.1504,
      "step": 8942
    },
    {
      "epoch": 0.06076184159065845,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009930390099845488,
      "loss": 2.3278,
      "step": 8943
    },
    {
      "epoch": 0.06076863593725251,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009930372208974436,
      "loss": 2.2189,
      "step": 8944
    },
    {
      "epoch": 0.06077543028384657,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009930354315820676,
      "loss": 2.2524,
      "step": 8945
    },
    {
      "epoch": 0.060782224630440626,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000993033642038422,
      "loss": 2.3113,
      "step": 8946
    },
    {
      "epoch": 0.06078901897703468,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009930318522665073,
      "loss": 2.2439,
      "step": 8947
    },
    {
      "epoch": 0.06079581332362874,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009930300622663247,
      "loss": 2.2327,
      "step": 8948
    },
    {
      "epoch": 0.0608026076702228,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009930282720378746,
      "loss": 2.3132,
      "step": 8949
    },
    {
      "epoch": 0.06080940201681686,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009930264815811581,
      "loss": 2.2073,
      "step": 8950
    },
    {
      "epoch": 0.060816196363410915,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009930246908961761,
      "loss": 2.1978,
      "step": 8951
    },
    {
      "epoch": 0.06082299071000497,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009930228999829294,
      "loss": 2.3857,
      "step": 8952
    },
    {
      "epoch": 0.06082978505659903,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009930211088414187,
      "loss": 2.3116,
      "step": 8953
    },
    {
      "epoch": 0.060836579403193085,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009930193174716446,
      "loss": 2.345,
      "step": 8954
    },
    {
      "epoch": 0.06084337374978715,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009930175258736084,
      "loss": 2.2688,
      "step": 8955
    },
    {
      "epoch": 0.060850168096381205,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009930157340473109,
      "loss": 2.2612,
      "step": 8956
    },
    {
      "epoch": 0.06085696244297526,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009930139419927527,
      "loss": 2.3512,
      "step": 8957
    },
    {
      "epoch": 0.06086375678956932,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009930121497099345,
      "loss": 2.257,
      "step": 8958
    },
    {
      "epoch": 0.060870551136163374,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009930103571988576,
      "loss": 2.3387,
      "step": 8959
    },
    {
      "epoch": 0.06087734548275744,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009930085644595224,
      "loss": 2.1956,
      "step": 8960
    },
    {
      "epoch": 0.060884139829351494,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009930067714919299,
      "loss": 2.4737,
      "step": 8961
    },
    {
      "epoch": 0.06089093417594555,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009930049782960809,
      "loss": 2.1781,
      "step": 8962
    },
    {
      "epoch": 0.06089772852253961,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009930031848719763,
      "loss": 2.2592,
      "step": 8963
    },
    {
      "epoch": 0.060904522869133663,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009930013912196167,
      "loss": 2.2771,
      "step": 8964
    },
    {
      "epoch": 0.06091131721572772,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009929995973390034,
      "loss": 2.2798,
      "step": 8965
    },
    {
      "epoch": 0.06091811156232178,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009929978032301367,
      "loss": 2.1511,
      "step": 8966
    },
    {
      "epoch": 0.06092490590891584,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000992996008893018,
      "loss": 2.3877,
      "step": 8967
    },
    {
      "epoch": 0.060931700255509896,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0009929942143276473,
      "loss": 2.3411,
      "step": 8968
    },
    {
      "epoch": 0.06093849460210395,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009929924195340263,
      "loss": 2.3367,
      "step": 8969
    },
    {
      "epoch": 0.06094528894869801,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009929906245121553,
      "loss": 2.3721,
      "step": 8970
    },
    {
      "epoch": 0.06095208329529207,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009929888292620353,
      "loss": 2.2965,
      "step": 8971
    },
    {
      "epoch": 0.06095887764188613,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009929870337836673,
      "loss": 2.2413,
      "step": 8972
    },
    {
      "epoch": 0.060965671988480186,
      "grad_norm": 0.224609375,
      "learning_rate": 0.0009929852380770518,
      "loss": 2.3557,
      "step": 8973
    },
    {
      "epoch": 0.06097246633507424,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009929834421421896,
      "loss": 2.1353,
      "step": 8974
    },
    {
      "epoch": 0.0609792606816683,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000992981645979082,
      "loss": 2.2583,
      "step": 8975
    },
    {
      "epoch": 0.06098605502826236,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009929798495877295,
      "loss": 2.3911,
      "step": 8976
    },
    {
      "epoch": 0.06099284937485642,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000992978052968133,
      "loss": 2.2896,
      "step": 8977
    },
    {
      "epoch": 0.060999643721450475,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009929762561202934,
      "loss": 2.2692,
      "step": 8978
    },
    {
      "epoch": 0.06100643806804453,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009929744590442113,
      "loss": 2.2058,
      "step": 8979
    },
    {
      "epoch": 0.06101323241463859,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009929726617398877,
      "loss": 2.1965,
      "step": 8980
    },
    {
      "epoch": 0.061020026761232644,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009929708642073235,
      "loss": 2.421,
      "step": 8981
    },
    {
      "epoch": 0.06102682110782671,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000992969066446519,
      "loss": 2.43,
      "step": 8982
    },
    {
      "epoch": 0.061033615454420764,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000992967268457476,
      "loss": 2.3143,
      "step": 8983
    },
    {
      "epoch": 0.06104040980101482,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009929654702401946,
      "loss": 2.1835,
      "step": 8984
    },
    {
      "epoch": 0.06104720414760888,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992963671794676,
      "loss": 2.281,
      "step": 8985
    },
    {
      "epoch": 0.061053998494202934,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009929618731209207,
      "loss": 2.3586,
      "step": 8986
    },
    {
      "epoch": 0.061060792840797,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009929600742189298,
      "loss": 2.0868,
      "step": 8987
    },
    {
      "epoch": 0.061067587187391054,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000992958275088704,
      "loss": 2.302,
      "step": 8988
    },
    {
      "epoch": 0.06107438153398511,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000992956475730244,
      "loss": 2.3626,
      "step": 8989
    },
    {
      "epoch": 0.06108117588057917,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992954676143551,
      "loss": 2.2872,
      "step": 8990
    },
    {
      "epoch": 0.06108797022717322,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009929528763286256,
      "loss": 2.2381,
      "step": 8991
    },
    {
      "epoch": 0.06109476457376728,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009929510762854686,
      "loss": 2.2072,
      "step": 8992
    },
    {
      "epoch": 0.06110155892036134,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992949276014081,
      "loss": 2.2345,
      "step": 8993
    },
    {
      "epoch": 0.0611083532669554,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009929474755144634,
      "loss": 2.2835,
      "step": 8994
    },
    {
      "epoch": 0.061115147613549456,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000992945674786617,
      "loss": 2.2173,
      "step": 8995
    },
    {
      "epoch": 0.06112194196014351,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009929438738305423,
      "loss": 2.2455,
      "step": 8996
    },
    {
      "epoch": 0.06112873630673757,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00099294207264624,
      "loss": 2.313,
      "step": 8997
    },
    {
      "epoch": 0.06113553065333163,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009929402712337114,
      "loss": 2.2323,
      "step": 8998
    },
    {
      "epoch": 0.06114232499992569,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009929384695929573,
      "loss": 2.2935,
      "step": 8999
    },
    {
      "epoch": 0.061149119346519745,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000992936667723978,
      "loss": 2.3469,
      "step": 9000
    },
    {
      "epoch": 0.0611559136931138,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992934865626775,
      "loss": 2.0333,
      "step": 9001
    },
    {
      "epoch": 0.06116270803970786,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009929330633013484,
      "loss": 2.2006,
      "step": 9002
    },
    {
      "epoch": 0.061169502386301915,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009929312607476998,
      "loss": 2.3207,
      "step": 9003
    },
    {
      "epoch": 0.06117629673289598,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009929294579658294,
      "loss": 2.2327,
      "step": 9004
    },
    {
      "epoch": 0.061183091079490035,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009929276549557385,
      "loss": 2.4326,
      "step": 9005
    },
    {
      "epoch": 0.06118988542608409,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009929258517174277,
      "loss": 2.1102,
      "step": 9006
    },
    {
      "epoch": 0.06119667977267815,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000992924048250898,
      "loss": 2.2861,
      "step": 9007
    },
    {
      "epoch": 0.061203474119272204,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009929222445561499,
      "loss": 2.419,
      "step": 9008
    },
    {
      "epoch": 0.06121026846586627,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009929204406331845,
      "loss": 2.1337,
      "step": 9009
    },
    {
      "epoch": 0.061217062812460324,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009929186364820027,
      "loss": 2.3102,
      "step": 9010
    },
    {
      "epoch": 0.06122385715905438,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000992916832102605,
      "loss": 2.2746,
      "step": 9011
    },
    {
      "epoch": 0.06123065150564844,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009929150274949926,
      "loss": 2.3341,
      "step": 9012
    },
    {
      "epoch": 0.06123744585224249,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009929132226591663,
      "loss": 2.3335,
      "step": 9013
    },
    {
      "epoch": 0.06124424019883655,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009929114175951266,
      "loss": 2.254,
      "step": 9014
    },
    {
      "epoch": 0.06125103454543061,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009929096123028748,
      "loss": 2.3346,
      "step": 9015
    },
    {
      "epoch": 0.06125782889202467,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009929078067824114,
      "loss": 2.0902,
      "step": 9016
    },
    {
      "epoch": 0.061264623238618726,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009929060010337373,
      "loss": 2.2473,
      "step": 9017
    },
    {
      "epoch": 0.06127141758521278,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009929041950568534,
      "loss": 2.1588,
      "step": 9018
    },
    {
      "epoch": 0.06127821193180684,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009929023888517606,
      "loss": 2.2613,
      "step": 9019
    },
    {
      "epoch": 0.0612850062784009,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009929005824184595,
      "loss": 2.3268,
      "step": 9020
    },
    {
      "epoch": 0.06129180062499496,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009928987757569512,
      "loss": 2.3604,
      "step": 9021
    },
    {
      "epoch": 0.061298594971589015,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009928969688672363,
      "loss": 2.2566,
      "step": 9022
    },
    {
      "epoch": 0.06130538931818307,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992895161749316,
      "loss": 2.2663,
      "step": 9023
    },
    {
      "epoch": 0.06131218366477713,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009928933544031903,
      "loss": 2.3536,
      "step": 9024
    },
    {
      "epoch": 0.061318978011371185,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009928915468288611,
      "loss": 2.0744,
      "step": 9025
    },
    {
      "epoch": 0.06132577235796525,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009928897390263287,
      "loss": 2.3033,
      "step": 9026
    },
    {
      "epoch": 0.061332566704559305,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000992887930995594,
      "loss": 2.3444,
      "step": 9027
    },
    {
      "epoch": 0.06133936105115336,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000992886122736658,
      "loss": 2.2221,
      "step": 9028
    },
    {
      "epoch": 0.06134615539774742,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009928843142495212,
      "loss": 2.5119,
      "step": 9029
    },
    {
      "epoch": 0.061352949744341474,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009928825055341847,
      "loss": 2.1694,
      "step": 9030
    },
    {
      "epoch": 0.06135974409093554,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000992880696590649,
      "loss": 2.1937,
      "step": 9031
    },
    {
      "epoch": 0.061366538437529594,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009928788874189155,
      "loss": 2.2823,
      "step": 9032
    },
    {
      "epoch": 0.06137333278412365,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009928770780189844,
      "loss": 2.109,
      "step": 9033
    },
    {
      "epoch": 0.06138012713071771,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009928752683908571,
      "loss": 2.206,
      "step": 9034
    },
    {
      "epoch": 0.061386921477311764,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009928734585345344,
      "loss": 2.4411,
      "step": 9035
    },
    {
      "epoch": 0.06139371582390583,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009928716484500163,
      "loss": 2.1287,
      "step": 9036
    },
    {
      "epoch": 0.06140051017049988,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992869838137305,
      "loss": 2.3327,
      "step": 9037
    },
    {
      "epoch": 0.06140730451709394,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009928680275964003,
      "loss": 2.3638,
      "step": 9038
    },
    {
      "epoch": 0.061414098863687996,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992866216827303,
      "loss": 2.0355,
      "step": 9039
    },
    {
      "epoch": 0.06142089321028205,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009928644058300148,
      "loss": 2.2477,
      "step": 9040
    },
    {
      "epoch": 0.06142768755687611,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000992862594604536,
      "loss": 2.1989,
      "step": 9041
    },
    {
      "epoch": 0.06143448190347017,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009928607831508672,
      "loss": 2.2278,
      "step": 9042
    },
    {
      "epoch": 0.06144127625006423,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009928589714690095,
      "loss": 2.4503,
      "step": 9043
    },
    {
      "epoch": 0.061448070596658286,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000992857159558964,
      "loss": 2.0963,
      "step": 9044
    },
    {
      "epoch": 0.06145486494325234,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009928553474207312,
      "loss": 2.4171,
      "step": 9045
    },
    {
      "epoch": 0.0614616592898464,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000992853535054312,
      "loss": 2.2545,
      "step": 9046
    },
    {
      "epoch": 0.06146845363644046,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009928517224597072,
      "loss": 2.1924,
      "step": 9047
    },
    {
      "epoch": 0.06147524798303452,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009928499096369177,
      "loss": 2.3163,
      "step": 9048
    },
    {
      "epoch": 0.061482042329628575,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009928480965859445,
      "loss": 2.2711,
      "step": 9049
    },
    {
      "epoch": 0.06148883667622263,
      "grad_norm": 0.15625,
      "learning_rate": 0.000992846283306788,
      "loss": 2.2913,
      "step": 9050
    },
    {
      "epoch": 0.06149563102281669,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009928444697994497,
      "loss": 2.4898,
      "step": 9051
    },
    {
      "epoch": 0.061502425369410745,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009928426560639298,
      "loss": 2.5046,
      "step": 9052
    },
    {
      "epoch": 0.06150921971600481,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009928408421002294,
      "loss": 2.2481,
      "step": 9053
    },
    {
      "epoch": 0.061516014062598864,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009928390279083494,
      "loss": 2.1487,
      "step": 9054
    },
    {
      "epoch": 0.06152280840919292,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009928372134882907,
      "loss": 2.1995,
      "step": 9055
    },
    {
      "epoch": 0.06152960275578698,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992835398840054,
      "loss": 2.3074,
      "step": 9056
    },
    {
      "epoch": 0.061536397102381034,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099283358396364,
      "loss": 2.2031,
      "step": 9057
    },
    {
      "epoch": 0.0615431914489751,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009928317688590498,
      "loss": 2.199,
      "step": 9058
    },
    {
      "epoch": 0.061549985795569154,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000992829953526284,
      "loss": 2.1281,
      "step": 9059
    },
    {
      "epoch": 0.06155678014216321,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009928281379653439,
      "loss": 2.1548,
      "step": 9060
    },
    {
      "epoch": 0.06156357448875727,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009928263221762297,
      "loss": 2.2074,
      "step": 9061
    },
    {
      "epoch": 0.06157036883535132,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009928245061589427,
      "loss": 2.2675,
      "step": 9062
    },
    {
      "epoch": 0.06157716318194538,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009928226899134835,
      "loss": 2.3102,
      "step": 9063
    },
    {
      "epoch": 0.06158395752853944,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009928208734398532,
      "loss": 2.223,
      "step": 9064
    },
    {
      "epoch": 0.0615907518751335,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009928190567380524,
      "loss": 2.3669,
      "step": 9065
    },
    {
      "epoch": 0.061597546221727556,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009928172398080818,
      "loss": 2.2366,
      "step": 9066
    },
    {
      "epoch": 0.06160434056832161,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009928154226499428,
      "loss": 2.2188,
      "step": 9067
    },
    {
      "epoch": 0.06161113491491567,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009928136052636358,
      "loss": 2.2473,
      "step": 9068
    },
    {
      "epoch": 0.06161792926150973,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009928117876491617,
      "loss": 2.2552,
      "step": 9069
    },
    {
      "epoch": 0.06162472360810379,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009928099698065216,
      "loss": 2.3135,
      "step": 9070
    },
    {
      "epoch": 0.061631517954697845,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992808151735716,
      "loss": 2.1686,
      "step": 9071
    },
    {
      "epoch": 0.0616383123012919,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009928063334367458,
      "loss": 2.2511,
      "step": 9072
    },
    {
      "epoch": 0.06164510664788596,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000992804514909612,
      "loss": 2.2743,
      "step": 9073
    },
    {
      "epoch": 0.061651900994480015,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009928026961543155,
      "loss": 2.2801,
      "step": 9074
    },
    {
      "epoch": 0.06165869534107408,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009928008771708568,
      "loss": 2.2019,
      "step": 9075
    },
    {
      "epoch": 0.061665489687668135,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000992799057959237,
      "loss": 2.4363,
      "step": 9076
    },
    {
      "epoch": 0.06167228403426219,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009927972385194567,
      "loss": 2.2914,
      "step": 9077
    },
    {
      "epoch": 0.06167907838085625,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009927954188515171,
      "loss": 2.1629,
      "step": 9078
    },
    {
      "epoch": 0.061685872727450304,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000992793598955419,
      "loss": 2.1386,
      "step": 9079
    },
    {
      "epoch": 0.06169266707404437,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000992791778831163,
      "loss": 2.2179,
      "step": 9080
    },
    {
      "epoch": 0.061699461420638424,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00099278995847875,
      "loss": 2.2853,
      "step": 9081
    },
    {
      "epoch": 0.06170625576723248,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009927881378981808,
      "loss": 2.3399,
      "step": 9082
    },
    {
      "epoch": 0.06171305011382654,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009927863170894567,
      "loss": 2.2544,
      "step": 9083
    },
    {
      "epoch": 0.06171984446042059,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009927844960525777,
      "loss": 2.238,
      "step": 9084
    },
    {
      "epoch": 0.06172663880701465,
      "grad_norm": 0.21875,
      "learning_rate": 0.0009927826747875455,
      "loss": 2.1852,
      "step": 9085
    },
    {
      "epoch": 0.06173343315360871,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009927808532943605,
      "loss": 2.19,
      "step": 9086
    },
    {
      "epoch": 0.06174022750020277,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009927790315730235,
      "loss": 2.1977,
      "step": 9087
    },
    {
      "epoch": 0.061747021846796826,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009927772096235358,
      "loss": 2.0832,
      "step": 9088
    },
    {
      "epoch": 0.06175381619339088,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009927753874458976,
      "loss": 2.3967,
      "step": 9089
    },
    {
      "epoch": 0.06176061053998494,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099277356504011,
      "loss": 2.1795,
      "step": 9090
    },
    {
      "epoch": 0.061767404886579,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000992771742406174,
      "loss": 2.2727,
      "step": 9091
    },
    {
      "epoch": 0.06177419923317306,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009927699195440903,
      "loss": 2.3834,
      "step": 9092
    },
    {
      "epoch": 0.061780993579767116,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009927680964538596,
      "loss": 2.2142,
      "step": 9093
    },
    {
      "epoch": 0.06178778792636117,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009927662731354832,
      "loss": 2.181,
      "step": 9094
    },
    {
      "epoch": 0.06179458227295523,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009927644495889617,
      "loss": 2.2842,
      "step": 9095
    },
    {
      "epoch": 0.06180137661954929,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009927626258142957,
      "loss": 2.2642,
      "step": 9096
    },
    {
      "epoch": 0.06180817096614335,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009927608018114862,
      "loss": 2.1921,
      "step": 9097
    },
    {
      "epoch": 0.061814965312737405,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009927589775805344,
      "loss": 2.3061,
      "step": 9098
    },
    {
      "epoch": 0.06182175965933146,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009927571531214406,
      "loss": 2.2546,
      "step": 9099
    },
    {
      "epoch": 0.06182855400592552,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000992755328434206,
      "loss": 2.327,
      "step": 9100
    },
    {
      "epoch": 0.061835348352519574,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009927535035188313,
      "loss": 2.3008,
      "step": 9101
    },
    {
      "epoch": 0.06184214269911364,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009927516783753173,
      "loss": 2.2642,
      "step": 9102
    },
    {
      "epoch": 0.061848937045707694,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009927498530036651,
      "loss": 2.181,
      "step": 9103
    },
    {
      "epoch": 0.06185573139230175,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000992748027403875,
      "loss": 2.2673,
      "step": 9104
    },
    {
      "epoch": 0.06186252573889581,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009927462015759487,
      "loss": 2.194,
      "step": 9105
    },
    {
      "epoch": 0.061869320085489864,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009927443755198863,
      "loss": 2.1973,
      "step": 9106
    },
    {
      "epoch": 0.06187611443208393,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992742549235689,
      "loss": 2.2451,
      "step": 9107
    },
    {
      "epoch": 0.061882908778677984,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009927407227233573,
      "loss": 2.3595,
      "step": 9108
    },
    {
      "epoch": 0.06188970312527204,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009927388959828927,
      "loss": 2.2498,
      "step": 9109
    },
    {
      "epoch": 0.061896497471866097,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009927370690142952,
      "loss": 2.253,
      "step": 9110
    },
    {
      "epoch": 0.06190329181846015,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009927352418175664,
      "loss": 2.4145,
      "step": 9111
    },
    {
      "epoch": 0.06191008616505421,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009927334143927068,
      "loss": 2.3174,
      "step": 9112
    },
    {
      "epoch": 0.06191688051164827,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000992731586739717,
      "loss": 2.2131,
      "step": 9113
    },
    {
      "epoch": 0.06192367485824233,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009927297588585985,
      "loss": 2.3376,
      "step": 9114
    },
    {
      "epoch": 0.061930469204836386,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009927279307493516,
      "loss": 2.3082,
      "step": 9115
    },
    {
      "epoch": 0.06193726355143044,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009927261024119773,
      "loss": 2.3777,
      "step": 9116
    },
    {
      "epoch": 0.0619440578980245,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009927242738464765,
      "loss": 2.156,
      "step": 9117
    },
    {
      "epoch": 0.06195085224461856,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099272244505285,
      "loss": 2.4051,
      "step": 9118
    },
    {
      "epoch": 0.06195764659121262,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009927206160310986,
      "loss": 2.3297,
      "step": 9119
    },
    {
      "epoch": 0.061964440937806675,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009927187867812234,
      "loss": 2.2215,
      "step": 9120
    },
    {
      "epoch": 0.06197123528440073,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009927169573032248,
      "loss": 2.3125,
      "step": 9121
    },
    {
      "epoch": 0.06197802963099479,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000992715127597104,
      "loss": 2.1865,
      "step": 9122
    },
    {
      "epoch": 0.061984823977588845,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009927132976628617,
      "loss": 2.2086,
      "step": 9123
    },
    {
      "epoch": 0.06199161832418291,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009927114675004989,
      "loss": 2.1951,
      "step": 9124
    },
    {
      "epoch": 0.061998412670776964,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009927096371100163,
      "loss": 2.221,
      "step": 9125
    },
    {
      "epoch": 0.06200520701737102,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009927078064914149,
      "loss": 2.1967,
      "step": 9126
    },
    {
      "epoch": 0.06201200136396508,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009927059756446952,
      "loss": 2.0613,
      "step": 9127
    },
    {
      "epoch": 0.062018795710559134,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009927041445698583,
      "loss": 2.1145,
      "step": 9128
    },
    {
      "epoch": 0.0620255900571532,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000992702313266905,
      "loss": 2.408,
      "step": 9129
    },
    {
      "epoch": 0.062032384403747254,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009927004817358363,
      "loss": 2.3121,
      "step": 9130
    },
    {
      "epoch": 0.06203917875034131,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000992698649976653,
      "loss": 2.2202,
      "step": 9131
    },
    {
      "epoch": 0.06204597309693537,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009926968179893559,
      "loss": 2.1599,
      "step": 9132
    },
    {
      "epoch": 0.06205276744352942,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009926949857739457,
      "loss": 2.2342,
      "step": 9133
    },
    {
      "epoch": 0.06205956179012348,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009926931533304232,
      "loss": 2.4064,
      "step": 9134
    },
    {
      "epoch": 0.06206635613671754,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009926913206587899,
      "loss": 2.3325,
      "step": 9135
    },
    {
      "epoch": 0.0620731504833116,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009926894877590458,
      "loss": 2.2531,
      "step": 9136
    },
    {
      "epoch": 0.062079944829905656,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000992687654631192,
      "loss": 2.3599,
      "step": 9137
    },
    {
      "epoch": 0.06208673917649971,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009926858212752296,
      "loss": 2.3213,
      "step": 9138
    },
    {
      "epoch": 0.06209353352309377,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009926839876911594,
      "loss": 2.3485,
      "step": 9139
    },
    {
      "epoch": 0.06210032786968783,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000992682153878982,
      "loss": 2.2773,
      "step": 9140
    },
    {
      "epoch": 0.06210712221628189,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009926803198386986,
      "loss": 2.1816,
      "step": 9141
    },
    {
      "epoch": 0.062113916562875945,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099267848557031,
      "loss": 2.3547,
      "step": 9142
    },
    {
      "epoch": 0.06212071090947,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009926766510738164,
      "loss": 2.3673,
      "step": 9143
    },
    {
      "epoch": 0.06212750525606406,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009926748163492196,
      "loss": 2.261,
      "step": 9144
    },
    {
      "epoch": 0.062134299602658115,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009926729813965198,
      "loss": 2.267,
      "step": 9145
    },
    {
      "epoch": 0.06214109394925218,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009926711462157182,
      "loss": 2.2035,
      "step": 9146
    },
    {
      "epoch": 0.062147888295846235,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009926693108068155,
      "loss": 2.21,
      "step": 9147
    },
    {
      "epoch": 0.06215468264244029,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009926674751698124,
      "loss": 2.3576,
      "step": 9148
    },
    {
      "epoch": 0.06216147698903435,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00099266563930471,
      "loss": 2.1029,
      "step": 9149
    },
    {
      "epoch": 0.062168271335628404,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009926638032115092,
      "loss": 2.1447,
      "step": 9150
    },
    {
      "epoch": 0.06217506568222247,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009926619668902104,
      "loss": 2.2333,
      "step": 9151
    },
    {
      "epoch": 0.062181860028816524,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000992660130340815,
      "loss": 2.1349,
      "step": 9152
    },
    {
      "epoch": 0.06218865437541058,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009926582935633234,
      "loss": 2.2374,
      "step": 9153
    },
    {
      "epoch": 0.06219544872200464,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000992656456557737,
      "loss": 2.2703,
      "step": 9154
    },
    {
      "epoch": 0.062202243068598694,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000992654619324056,
      "loss": 2.3405,
      "step": 9155
    },
    {
      "epoch": 0.06220903741519276,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009926527818622816,
      "loss": 2.2567,
      "step": 9156
    },
    {
      "epoch": 0.06221583176178681,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009926509441724148,
      "loss": 2.385,
      "step": 9157
    },
    {
      "epoch": 0.06222262610838087,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000992649106254456,
      "loss": 2.4748,
      "step": 9158
    },
    {
      "epoch": 0.062229420454974926,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009926472681084065,
      "loss": 2.3308,
      "step": 9159
    },
    {
      "epoch": 0.06223621480156898,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000992645429734267,
      "loss": 2.3179,
      "step": 9160
    },
    {
      "epoch": 0.06224300914816304,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009926435911320382,
      "loss": 2.2494,
      "step": 9161
    },
    {
      "epoch": 0.0622498034947571,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992641752301721,
      "loss": 2.2963,
      "step": 9162
    },
    {
      "epoch": 0.06225659784135116,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009926399132433163,
      "loss": 2.1172,
      "step": 9163
    },
    {
      "epoch": 0.062263392187945216,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009926380739568252,
      "loss": 2.2416,
      "step": 9164
    },
    {
      "epoch": 0.06227018653453927,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009926362344422482,
      "loss": 2.1543,
      "step": 9165
    },
    {
      "epoch": 0.06227698088113333,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009926343946995862,
      "loss": 2.2967,
      "step": 9166
    },
    {
      "epoch": 0.06228377522772739,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009926325547288401,
      "loss": 2.314,
      "step": 9167
    },
    {
      "epoch": 0.06229056957432145,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009926307145300109,
      "loss": 2.2163,
      "step": 9168
    },
    {
      "epoch": 0.062297363920915505,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009926288741030993,
      "loss": 2.3171,
      "step": 9169
    },
    {
      "epoch": 0.06230415826750956,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992627033448106,
      "loss": 2.2867,
      "step": 9170
    },
    {
      "epoch": 0.06231095261410362,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009926251925650323,
      "loss": 2.3039,
      "step": 9171
    },
    {
      "epoch": 0.062317746960697674,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009926233514538788,
      "loss": 2.225,
      "step": 9172
    },
    {
      "epoch": 0.06232454130729174,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000992621510114646,
      "loss": 2.2428,
      "step": 9173
    },
    {
      "epoch": 0.062331335653885794,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009926196685473355,
      "loss": 2.307,
      "step": 9174
    },
    {
      "epoch": 0.06233813000047985,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009926178267519474,
      "loss": 2.1939,
      "step": 9175
    },
    {
      "epoch": 0.06234492434707391,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000992615984728483,
      "loss": 2.1705,
      "step": 9176
    },
    {
      "epoch": 0.062351718693667964,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0009926141424769432,
      "loss": 2.3985,
      "step": 9177
    },
    {
      "epoch": 0.06235851304026203,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009926122999973285,
      "loss": 2.3034,
      "step": 9178
    },
    {
      "epoch": 0.062365307386856084,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099261045728964,
      "loss": 2.2615,
      "step": 9179
    },
    {
      "epoch": 0.06237210173345014,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009926086143538786,
      "loss": 2.338,
      "step": 9180
    },
    {
      "epoch": 0.0623788960800442,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000992606771190045,
      "loss": 2.0889,
      "step": 9181
    },
    {
      "epoch": 0.06238569042663825,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009926049277981402,
      "loss": 2.3822,
      "step": 9182
    },
    {
      "epoch": 0.06239248477323231,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009926030841781648,
      "loss": 2.4195,
      "step": 9183
    },
    {
      "epoch": 0.06239927911982637,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009926012403301198,
      "loss": 2.2844,
      "step": 9184
    },
    {
      "epoch": 0.06240607346642043,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009925993962540062,
      "loss": 2.4162,
      "step": 9185
    },
    {
      "epoch": 0.062412867813014486,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009925975519498246,
      "loss": 2.358,
      "step": 9186
    },
    {
      "epoch": 0.06241966215960854,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000992595707417576,
      "loss": 2.4854,
      "step": 9187
    },
    {
      "epoch": 0.0624264565062026,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009925938626572615,
      "loss": 2.2656,
      "step": 9188
    },
    {
      "epoch": 0.06243325085279666,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009925920176688816,
      "loss": 2.0336,
      "step": 9189
    },
    {
      "epoch": 0.06244004519939072,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000992590172452437,
      "loss": 2.1925,
      "step": 9190
    },
    {
      "epoch": 0.062446839545984775,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000992588327007929,
      "loss": 2.3081,
      "step": 9191
    },
    {
      "epoch": 0.06245363389257883,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009925864813353582,
      "loss": 2.2403,
      "step": 9192
    },
    {
      "epoch": 0.06246042823917289,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009925846354347254,
      "loss": 2.3152,
      "step": 9193
    },
    {
      "epoch": 0.062467222585766945,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009925827893060317,
      "loss": 2.3271,
      "step": 9194
    },
    {
      "epoch": 0.06247401693236101,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009925809429492778,
      "loss": 2.2294,
      "step": 9195
    },
    {
      "epoch": 0.062480811278955065,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009925790963644645,
      "loss": 2.2385,
      "step": 9196
    },
    {
      "epoch": 0.06248760562554912,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009925772495515928,
      "loss": 2.2456,
      "step": 9197
    },
    {
      "epoch": 0.06249439997214318,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009925754025106636,
      "loss": 2.3041,
      "step": 9198
    },
    {
      "epoch": 0.06250119431873724,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009925735552416776,
      "loss": 2.3245,
      "step": 9199
    },
    {
      "epoch": 0.06250798866533129,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009925717077446353,
      "loss": 2.4025,
      "step": 9200
    },
    {
      "epoch": 0.06251478301192535,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009925698600195382,
      "loss": 2.3313,
      "step": 9201
    },
    {
      "epoch": 0.0625215773585194,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000992568012066387,
      "loss": 2.3307,
      "step": 9202
    },
    {
      "epoch": 0.06252837170511347,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009925661638851824,
      "loss": 2.1888,
      "step": 9203
    },
    {
      "epoch": 0.06253516605170753,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009925643154759254,
      "loss": 2.3723,
      "step": 9204
    },
    {
      "epoch": 0.06254196039830158,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009925624668386166,
      "loss": 2.1762,
      "step": 9205
    },
    {
      "epoch": 0.06254875474489564,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000992560617973257,
      "loss": 2.3581,
      "step": 9206
    },
    {
      "epoch": 0.06255554909148969,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009925587688798475,
      "loss": 2.208,
      "step": 9207
    },
    {
      "epoch": 0.06256234343808376,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009925569195583892,
      "loss": 2.3413,
      "step": 9208
    },
    {
      "epoch": 0.06256913778467782,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009925550700088824,
      "loss": 2.2174,
      "step": 9209
    },
    {
      "epoch": 0.06257593213127187,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009925532202313284,
      "loss": 2.2672,
      "step": 9210
    },
    {
      "epoch": 0.06258272647786593,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009925513702257278,
      "loss": 2.2976,
      "step": 9211
    },
    {
      "epoch": 0.06258952082445998,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009925495199920816,
      "loss": 2.3538,
      "step": 9212
    },
    {
      "epoch": 0.06259631517105405,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009925476695303909,
      "loss": 2.3192,
      "step": 9213
    },
    {
      "epoch": 0.06260310951764811,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009925458188406557,
      "loss": 2.2885,
      "step": 9214
    },
    {
      "epoch": 0.06260990386424216,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009925439679228777,
      "loss": 2.2713,
      "step": 9215
    },
    {
      "epoch": 0.06261669821083622,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009925421167770575,
      "loss": 2.4741,
      "step": 9216
    },
    {
      "epoch": 0.06262349255743027,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009925402654031963,
      "loss": 2.2038,
      "step": 9217
    },
    {
      "epoch": 0.06263028690402433,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992538413801294,
      "loss": 2.1687,
      "step": 9218
    },
    {
      "epoch": 0.0626370812506184,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009925365619713525,
      "loss": 2.4149,
      "step": 9219
    },
    {
      "epoch": 0.06264387559721245,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009925347099133722,
      "loss": 2.1522,
      "step": 9220
    },
    {
      "epoch": 0.06265066994380651,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009925328576273539,
      "loss": 2.1977,
      "step": 9221
    },
    {
      "epoch": 0.06265746429040056,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009925310051132983,
      "loss": 2.2287,
      "step": 9222
    },
    {
      "epoch": 0.06266425863699462,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009925291523712066,
      "loss": 2.3875,
      "step": 9223
    },
    {
      "epoch": 0.06267105298358867,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009925272994010797,
      "loss": 2.1887,
      "step": 9224
    },
    {
      "epoch": 0.06267784733018274,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009925254462029183,
      "loss": 2.2791,
      "step": 9225
    },
    {
      "epoch": 0.0626846416767768,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000992523592776723,
      "loss": 2.2709,
      "step": 9226
    },
    {
      "epoch": 0.06269143602337085,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009925217391224953,
      "loss": 2.2409,
      "step": 9227
    },
    {
      "epoch": 0.06269823036996491,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009925198852402354,
      "loss": 2.306,
      "step": 9228
    },
    {
      "epoch": 0.06270502471655896,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009925180311299445,
      "loss": 2.1636,
      "step": 9229
    },
    {
      "epoch": 0.06271181906315303,
      "grad_norm": 0.125,
      "learning_rate": 0.0009925161767916236,
      "loss": 2.2821,
      "step": 9230
    },
    {
      "epoch": 0.06271861340974709,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000992514322225273,
      "loss": 2.2916,
      "step": 9231
    },
    {
      "epoch": 0.06272540775634114,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009925124674308942,
      "loss": 2.398,
      "step": 9232
    },
    {
      "epoch": 0.0627322021029352,
      "grad_norm": 0.125,
      "learning_rate": 0.0009925106124084877,
      "loss": 2.2388,
      "step": 9233
    },
    {
      "epoch": 0.06273899644952925,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009925087571580544,
      "loss": 2.3092,
      "step": 9234
    },
    {
      "epoch": 0.06274579079612332,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009925069016795951,
      "loss": 2.2778,
      "step": 9235
    },
    {
      "epoch": 0.06275258514271738,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009925050459731108,
      "loss": 2.1727,
      "step": 9236
    },
    {
      "epoch": 0.06275937948931143,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009925031900386025,
      "loss": 2.2918,
      "step": 9237
    },
    {
      "epoch": 0.06276617383590549,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009925013338760707,
      "loss": 2.2666,
      "step": 9238
    },
    {
      "epoch": 0.06277296818249954,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009924994774855166,
      "loss": 2.235,
      "step": 9239
    },
    {
      "epoch": 0.0627797625290936,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009924976208669407,
      "loss": 2.2593,
      "step": 9240
    },
    {
      "epoch": 0.06278655687568767,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000992495764020344,
      "loss": 2.239,
      "step": 9241
    },
    {
      "epoch": 0.06279335122228172,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009924939069457274,
      "loss": 2.2282,
      "step": 9242
    },
    {
      "epoch": 0.06280014556887578,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009924920496430921,
      "loss": 2.1568,
      "step": 9243
    },
    {
      "epoch": 0.06280693991546983,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009924901921124383,
      "loss": 2.085,
      "step": 9244
    },
    {
      "epoch": 0.0628137342620639,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009924883343537672,
      "loss": 2.1937,
      "step": 9245
    },
    {
      "epoch": 0.06282052860865796,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009924864763670798,
      "loss": 2.3055,
      "step": 9246
    },
    {
      "epoch": 0.06282732295525201,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009924846181523768,
      "loss": 2.356,
      "step": 9247
    },
    {
      "epoch": 0.06283411730184607,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992482759709659,
      "loss": 2.1969,
      "step": 9248
    },
    {
      "epoch": 0.06284091164844012,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009924809010389273,
      "loss": 2.2647,
      "step": 9249
    },
    {
      "epoch": 0.06284770599503418,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009924790421401825,
      "loss": 2.2553,
      "step": 9250
    },
    {
      "epoch": 0.06285450034162823,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009924771830134259,
      "loss": 2.305,
      "step": 9251
    },
    {
      "epoch": 0.0628612946882223,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009924753236586576,
      "loss": 2.3108,
      "step": 9252
    },
    {
      "epoch": 0.06286808903481636,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000992473464075879,
      "loss": 2.3147,
      "step": 9253
    },
    {
      "epoch": 0.06287488338141041,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000992471604265091,
      "loss": 2.3256,
      "step": 9254
    },
    {
      "epoch": 0.06288167772800447,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009924697442262943,
      "loss": 2.2273,
      "step": 9255
    },
    {
      "epoch": 0.06288847207459852,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009924678839594895,
      "loss": 2.2883,
      "step": 9256
    },
    {
      "epoch": 0.06289526642119259,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000992466023464678,
      "loss": 2.3277,
      "step": 9257
    },
    {
      "epoch": 0.06290206076778665,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009924641627418601,
      "loss": 2.2257,
      "step": 9258
    },
    {
      "epoch": 0.0629088551143807,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000992462301791037,
      "loss": 2.2528,
      "step": 9259
    },
    {
      "epoch": 0.06291564946097476,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009924604406122096,
      "loss": 2.358,
      "step": 9260
    },
    {
      "epoch": 0.06292244380756881,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009924585792053786,
      "loss": 2.0858,
      "step": 9261
    },
    {
      "epoch": 0.06292923815416288,
      "grad_norm": 0.15625,
      "learning_rate": 0.000992456717570545,
      "loss": 2.462,
      "step": 9262
    },
    {
      "epoch": 0.06293603250075694,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009924548557077096,
      "loss": 2.3492,
      "step": 9263
    },
    {
      "epoch": 0.06294282684735099,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000992452993616873,
      "loss": 2.2292,
      "step": 9264
    },
    {
      "epoch": 0.06294962119394505,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009924511312980366,
      "loss": 2.138,
      "step": 9265
    },
    {
      "epoch": 0.0629564155405391,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009924492687512008,
      "loss": 2.1546,
      "step": 9266
    },
    {
      "epoch": 0.06296320988713316,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992447405976367,
      "loss": 2.2548,
      "step": 9267
    },
    {
      "epoch": 0.06297000423372723,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009924455429735353,
      "loss": 2.3513,
      "step": 9268
    },
    {
      "epoch": 0.06297679858032128,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000992443679742707,
      "loss": 2.2063,
      "step": 9269
    },
    {
      "epoch": 0.06298359292691534,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009924418162838829,
      "loss": 2.19,
      "step": 9270
    },
    {
      "epoch": 0.06299038727350939,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000992439952597064,
      "loss": 2.2903,
      "step": 9271
    },
    {
      "epoch": 0.06299718162010345,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992438088682251,
      "loss": 2.1403,
      "step": 9272
    },
    {
      "epoch": 0.0630039759666975,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009924362245394447,
      "loss": 2.2845,
      "step": 9273
    },
    {
      "epoch": 0.06301077031329157,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009924343601686464,
      "loss": 2.1905,
      "step": 9274
    },
    {
      "epoch": 0.06301756465988563,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009924324955698563,
      "loss": 2.2493,
      "step": 9275
    },
    {
      "epoch": 0.06302435900647968,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992430630743076,
      "loss": 2.3442,
      "step": 9276
    },
    {
      "epoch": 0.06303115335307374,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009924287656883056,
      "loss": 2.2796,
      "step": 9277
    },
    {
      "epoch": 0.06303794769966779,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009924269004055464,
      "loss": 2.2233,
      "step": 9278
    },
    {
      "epoch": 0.06304474204626186,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009924250348947992,
      "loss": 2.3616,
      "step": 9279
    },
    {
      "epoch": 0.06305153639285592,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009924231691560647,
      "loss": 2.123,
      "step": 9280
    },
    {
      "epoch": 0.06305833073944997,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009924213031893441,
      "loss": 2.1966,
      "step": 9281
    },
    {
      "epoch": 0.06306512508604403,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009924194369946382,
      "loss": 2.2355,
      "step": 9282
    },
    {
      "epoch": 0.06307191943263808,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009924175705719475,
      "loss": 2.2826,
      "step": 9283
    },
    {
      "epoch": 0.06307871377923215,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009924157039212732,
      "loss": 2.2833,
      "step": 9284
    },
    {
      "epoch": 0.06308550812582621,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009924138370426161,
      "loss": 2.3245,
      "step": 9285
    },
    {
      "epoch": 0.06309230247242026,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009924119699359771,
      "loss": 2.2362,
      "step": 9286
    },
    {
      "epoch": 0.06309909681901432,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000992410102601357,
      "loss": 2.1525,
      "step": 9287
    },
    {
      "epoch": 0.06310589116560837,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009924082350387566,
      "loss": 2.1863,
      "step": 9288
    },
    {
      "epoch": 0.06311268551220243,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000992406367248177,
      "loss": 2.2238,
      "step": 9289
    },
    {
      "epoch": 0.0631194798587965,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009924044992296187,
      "loss": 2.3584,
      "step": 9290
    },
    {
      "epoch": 0.06312627420539055,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009924026309830827,
      "loss": 2.2242,
      "step": 9291
    },
    {
      "epoch": 0.06313306855198461,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099240076250857,
      "loss": 2.3358,
      "step": 9292
    },
    {
      "epoch": 0.06313986289857866,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009923988938060815,
      "loss": 2.2233,
      "step": 9293
    },
    {
      "epoch": 0.06314665724517272,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009923970248756177,
      "loss": 2.2046,
      "step": 9294
    },
    {
      "epoch": 0.06315345159176677,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00099239515571718,
      "loss": 2.2952,
      "step": 9295
    },
    {
      "epoch": 0.06316024593836084,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009923932863307688,
      "loss": 2.3141,
      "step": 9296
    },
    {
      "epoch": 0.0631670402849549,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009923914167163852,
      "loss": 2.2625,
      "step": 9297
    },
    {
      "epoch": 0.06317383463154895,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009923895468740302,
      "loss": 2.163,
      "step": 9298
    },
    {
      "epoch": 0.06318062897814301,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009923876768037044,
      "loss": 2.2578,
      "step": 9299
    },
    {
      "epoch": 0.06318742332473706,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009923858065054086,
      "loss": 2.1233,
      "step": 9300
    },
    {
      "epoch": 0.06319421767133113,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000992383935979144,
      "loss": 2.2715,
      "step": 9301
    },
    {
      "epoch": 0.06320101201792519,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009923820652249112,
      "loss": 2.3303,
      "step": 9302
    },
    {
      "epoch": 0.06320780636451924,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000992380194242711,
      "loss": 2.112,
      "step": 9303
    },
    {
      "epoch": 0.0632146007111133,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009923783230325446,
      "loss": 2.4436,
      "step": 9304
    },
    {
      "epoch": 0.06322139505770735,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009923764515944128,
      "loss": 2.3277,
      "step": 9305
    },
    {
      "epoch": 0.06322818940430142,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000992374579928316,
      "loss": 2.2832,
      "step": 9306
    },
    {
      "epoch": 0.06323498375089548,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009923727080342556,
      "loss": 2.2985,
      "step": 9307
    },
    {
      "epoch": 0.06324177809748953,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009923708359122322,
      "loss": 2.2358,
      "step": 9308
    },
    {
      "epoch": 0.06324857244408359,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000992368963562247,
      "loss": 2.1285,
      "step": 9309
    },
    {
      "epoch": 0.06325536679067764,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009923670909843002,
      "loss": 1.9769,
      "step": 9310
    },
    {
      "epoch": 0.0632621611372717,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009923652181783936,
      "loss": 2.2487,
      "step": 9311
    },
    {
      "epoch": 0.06326895548386577,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009923633451445271,
      "loss": 2.0849,
      "step": 9312
    },
    {
      "epoch": 0.06327574983045982,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009923614718827024,
      "loss": 2.2889,
      "step": 9313
    },
    {
      "epoch": 0.06328254417705388,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009923595983929196,
      "loss": 2.2598,
      "step": 9314
    },
    {
      "epoch": 0.06328933852364793,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009923577246751802,
      "loss": 2.2662,
      "step": 9315
    },
    {
      "epoch": 0.063296132870242,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009923558507294848,
      "loss": 2.2972,
      "step": 9316
    },
    {
      "epoch": 0.06330292721683606,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000992353976555834,
      "loss": 2.3509,
      "step": 9317
    },
    {
      "epoch": 0.06330972156343011,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009923521021542293,
      "loss": 2.2824,
      "step": 9318
    },
    {
      "epoch": 0.06331651591002417,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009923502275246711,
      "loss": 2.1845,
      "step": 9319
    },
    {
      "epoch": 0.06332331025661822,
      "grad_norm": 0.125,
      "learning_rate": 0.0009923483526671603,
      "loss": 2.2565,
      "step": 9320
    },
    {
      "epoch": 0.06333010460321228,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000992346477581698,
      "loss": 2.2305,
      "step": 9321
    },
    {
      "epoch": 0.06333689894980633,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000992344602268285,
      "loss": 2.2964,
      "step": 9322
    },
    {
      "epoch": 0.0633436932964004,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000992342726726922,
      "loss": 2.372,
      "step": 9323
    },
    {
      "epoch": 0.06335048764299446,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099234085095761,
      "loss": 2.327,
      "step": 9324
    },
    {
      "epoch": 0.06335728198958851,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009923389749603496,
      "loss": 2.276,
      "step": 9325
    },
    {
      "epoch": 0.06336407633618257,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000992337098735142,
      "loss": 2.2073,
      "step": 9326
    },
    {
      "epoch": 0.06337087068277662,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009923352222819882,
      "loss": 2.425,
      "step": 9327
    },
    {
      "epoch": 0.06337766502937069,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009923333456008886,
      "loss": 2.2028,
      "step": 9328
    },
    {
      "epoch": 0.06338445937596475,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009923314686918444,
      "loss": 2.273,
      "step": 9329
    },
    {
      "epoch": 0.0633912537225588,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009923295915548562,
      "loss": 2.3749,
      "step": 9330
    },
    {
      "epoch": 0.06339804806915286,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009923277141899252,
      "loss": 2.3393,
      "step": 9331
    },
    {
      "epoch": 0.06340484241574691,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009923258365970521,
      "loss": 2.2172,
      "step": 9332
    },
    {
      "epoch": 0.06341163676234098,
      "grad_norm": 0.173828125,
      "learning_rate": 0.000992323958776238,
      "loss": 2.3851,
      "step": 9333
    },
    {
      "epoch": 0.06341843110893504,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992322080727483,
      "loss": 2.3499,
      "step": 9334
    },
    {
      "epoch": 0.06342522545552909,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009923202024507889,
      "loss": 2.2409,
      "step": 9335
    },
    {
      "epoch": 0.06343201980212315,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009923183239461562,
      "loss": 2.3972,
      "step": 9336
    },
    {
      "epoch": 0.0634388141487172,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009923164452135859,
      "loss": 2.2175,
      "step": 9337
    },
    {
      "epoch": 0.06344560849531126,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009923145662530783,
      "loss": 2.1106,
      "step": 9338
    },
    {
      "epoch": 0.06345240284190533,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000992312687064635,
      "loss": 2.2949,
      "step": 9339
    },
    {
      "epoch": 0.06345919718849938,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009923108076482566,
      "loss": 2.2244,
      "step": 9340
    },
    {
      "epoch": 0.06346599153509344,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009923089280039438,
      "loss": 2.1807,
      "step": 9341
    },
    {
      "epoch": 0.06347278588168749,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009923070481316976,
      "loss": 2.3806,
      "step": 9342
    },
    {
      "epoch": 0.06347958022828155,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000992305168031519,
      "loss": 2.3082,
      "step": 9343
    },
    {
      "epoch": 0.0634863745748756,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009923032877034086,
      "loss": 2.2575,
      "step": 9344
    },
    {
      "epoch": 0.06349316892146967,
      "grad_norm": 0.125,
      "learning_rate": 0.0009923014071473676,
      "loss": 2.1307,
      "step": 9345
    },
    {
      "epoch": 0.06349996326806373,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009922995263633965,
      "loss": 2.266,
      "step": 9346
    },
    {
      "epoch": 0.06350675761465778,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009922976453514965,
      "loss": 2.1222,
      "step": 9347
    },
    {
      "epoch": 0.06351355196125184,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009922957641116683,
      "loss": 2.3727,
      "step": 9348
    },
    {
      "epoch": 0.06352034630784589,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009922938826439128,
      "loss": 2.3032,
      "step": 9349
    },
    {
      "epoch": 0.06352714065443996,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009922920009482307,
      "loss": 2.2206,
      "step": 9350
    },
    {
      "epoch": 0.06353393500103402,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009922901190246233,
      "loss": 2.1566,
      "step": 9351
    },
    {
      "epoch": 0.06354072934762807,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009922882368730913,
      "loss": 2.3247,
      "step": 9352
    },
    {
      "epoch": 0.06354752369422213,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009922863544936354,
      "loss": 2.2292,
      "step": 9353
    },
    {
      "epoch": 0.06355431804081618,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009922844718862564,
      "loss": 2.2635,
      "step": 9354
    },
    {
      "epoch": 0.06356111238741025,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009922825890509555,
      "loss": 2.1656,
      "step": 9355
    },
    {
      "epoch": 0.06356790673400431,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009922807059877332,
      "loss": 2.3407,
      "step": 9356
    },
    {
      "epoch": 0.06357470108059836,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009922788226965908,
      "loss": 2.1295,
      "step": 9357
    },
    {
      "epoch": 0.06358149542719242,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009922769391775288,
      "loss": 2.2941,
      "step": 9358
    },
    {
      "epoch": 0.06358828977378647,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009922750554305483,
      "loss": 2.3131,
      "step": 9359
    },
    {
      "epoch": 0.06359508412038054,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009922731714556501,
      "loss": 2.1848,
      "step": 9360
    },
    {
      "epoch": 0.0636018784669746,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000992271287252835,
      "loss": 2.3505,
      "step": 9361
    },
    {
      "epoch": 0.06360867281356865,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000992269402822104,
      "loss": 2.262,
      "step": 9362
    },
    {
      "epoch": 0.06361546716016271,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000992267518163458,
      "loss": 2.313,
      "step": 9363
    },
    {
      "epoch": 0.06362226150675676,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009922656332768975,
      "loss": 2.2083,
      "step": 9364
    },
    {
      "epoch": 0.06362905585335082,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000992263748162424,
      "loss": 2.347,
      "step": 9365
    },
    {
      "epoch": 0.06363585019994489,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009922618628200378,
      "loss": 2.2272,
      "step": 9366
    },
    {
      "epoch": 0.06364264454653894,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00099225997724974,
      "loss": 2.3803,
      "step": 9367
    },
    {
      "epoch": 0.063649438893133,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009922580914515317,
      "loss": 2.3893,
      "step": 9368
    },
    {
      "epoch": 0.06365623323972705,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009922562054254132,
      "loss": 2.3251,
      "step": 9369
    },
    {
      "epoch": 0.06366302758632111,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992254319171386,
      "loss": 2.3195,
      "step": 9370
    },
    {
      "epoch": 0.06366982193291516,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009922524326894503,
      "loss": 2.2237,
      "step": 9371
    },
    {
      "epoch": 0.06367661627950923,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009922505459796077,
      "loss": 2.1105,
      "step": 9372
    },
    {
      "epoch": 0.06368341062610329,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009922486590418589,
      "loss": 2.1931,
      "step": 9373
    },
    {
      "epoch": 0.06369020497269734,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009922467718762045,
      "loss": 2.2832,
      "step": 9374
    },
    {
      "epoch": 0.0636969993192914,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009922448844826453,
      "loss": 2.391,
      "step": 9375
    },
    {
      "epoch": 0.06370379366588545,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009922429968611823,
      "loss": 2.1893,
      "step": 9376
    },
    {
      "epoch": 0.06371058801247952,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009922411090118168,
      "loss": 2.2533,
      "step": 9377
    },
    {
      "epoch": 0.06371738235907358,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992239220934549,
      "loss": 2.2856,
      "step": 9378
    },
    {
      "epoch": 0.06372417670566763,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009922373326293802,
      "loss": 2.1937,
      "step": 9379
    },
    {
      "epoch": 0.06373097105226169,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000992235444096311,
      "loss": 2.4007,
      "step": 9380
    },
    {
      "epoch": 0.06373776539885574,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009922335553353427,
      "loss": 2.0757,
      "step": 9381
    },
    {
      "epoch": 0.0637445597454498,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000992231666346476,
      "loss": 2.3732,
      "step": 9382
    },
    {
      "epoch": 0.06375135409204387,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009922297771297113,
      "loss": 2.3588,
      "step": 9383
    },
    {
      "epoch": 0.06375814843863792,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099222788768505,
      "loss": 2.195,
      "step": 9384
    },
    {
      "epoch": 0.06376494278523198,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000992225998012493,
      "loss": 2.3293,
      "step": 9385
    },
    {
      "epoch": 0.06377173713182603,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009922241081120407,
      "loss": 2.2606,
      "step": 9386
    },
    {
      "epoch": 0.0637785314784201,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009922222179836945,
      "loss": 2.4171,
      "step": 9387
    },
    {
      "epoch": 0.06378532582501416,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009922203276274552,
      "loss": 2.1618,
      "step": 9388
    },
    {
      "epoch": 0.06379212017160821,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009922184370433232,
      "loss": 2.3256,
      "step": 9389
    },
    {
      "epoch": 0.06379891451820227,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009922165462312997,
      "loss": 2.3056,
      "step": 9390
    },
    {
      "epoch": 0.06380570886479632,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009922146551913857,
      "loss": 2.3584,
      "step": 9391
    },
    {
      "epoch": 0.06381250321139038,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000992212763923582,
      "loss": 2.4336,
      "step": 9392
    },
    {
      "epoch": 0.06381929755798443,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009922108724278896,
      "loss": 2.2096,
      "step": 9393
    },
    {
      "epoch": 0.0638260919045785,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009922089807043088,
      "loss": 2.1162,
      "step": 9394
    },
    {
      "epoch": 0.06383288625117256,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009922070887528412,
      "loss": 2.3615,
      "step": 9395
    },
    {
      "epoch": 0.06383968059776661,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009922051965734872,
      "loss": 2.2201,
      "step": 9396
    },
    {
      "epoch": 0.06384647494436067,
      "grad_norm": 0.140625,
      "learning_rate": 0.000992203304166248,
      "loss": 2.2464,
      "step": 9397
    },
    {
      "epoch": 0.06385326929095472,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009922014115311241,
      "loss": 2.1753,
      "step": 9398
    },
    {
      "epoch": 0.06386006363754879,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000992199518668117,
      "loss": 2.2012,
      "step": 9399
    },
    {
      "epoch": 0.06386685798414285,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009921976255772267,
      "loss": 2.1813,
      "step": 9400
    },
    {
      "epoch": 0.0638736523307369,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009921957322584548,
      "loss": 2.3352,
      "step": 9401
    },
    {
      "epoch": 0.06388044667733096,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009921938387118018,
      "loss": 2.1625,
      "step": 9402
    },
    {
      "epoch": 0.06388724102392501,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009921919449372687,
      "loss": 2.4347,
      "step": 9403
    },
    {
      "epoch": 0.06389403537051908,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009921900509348564,
      "loss": 2.2499,
      "step": 9404
    },
    {
      "epoch": 0.06390082971711314,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009921881567045657,
      "loss": 2.3995,
      "step": 9405
    },
    {
      "epoch": 0.06390762406370719,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009921862622463977,
      "loss": 2.3783,
      "step": 9406
    },
    {
      "epoch": 0.06391441841030125,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992184367560353,
      "loss": 2.386,
      "step": 9407
    },
    {
      "epoch": 0.0639212127568953,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009921824726464325,
      "loss": 2.2367,
      "step": 9408
    },
    {
      "epoch": 0.06392800710348936,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009921805775046373,
      "loss": 2.3249,
      "step": 9409
    },
    {
      "epoch": 0.06393480145008343,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992178682134968,
      "loss": 2.3614,
      "step": 9410
    },
    {
      "epoch": 0.06394159579667748,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009921767865374257,
      "loss": 2.2596,
      "step": 9411
    },
    {
      "epoch": 0.06394839014327154,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000992174890712011,
      "loss": 2.064,
      "step": 9412
    },
    {
      "epoch": 0.06395518448986559,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009921729946587253,
      "loss": 2.3607,
      "step": 9413
    },
    {
      "epoch": 0.06396197883645965,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000992171098377569,
      "loss": 2.1683,
      "step": 9414
    },
    {
      "epoch": 0.0639687731830537,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000992169201868543,
      "loss": 2.2236,
      "step": 9415
    },
    {
      "epoch": 0.06397556752964777,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009921673051316484,
      "loss": 2.2483,
      "step": 9416
    },
    {
      "epoch": 0.06398236187624183,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009921654081668859,
      "loss": 2.2051,
      "step": 9417
    },
    {
      "epoch": 0.06398915622283588,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009921635109742565,
      "loss": 2.3662,
      "step": 9418
    },
    {
      "epoch": 0.06399595056942994,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009921616135537613,
      "loss": 2.4321,
      "step": 9419
    },
    {
      "epoch": 0.064002744916024,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009921597159054006,
      "loss": 2.2704,
      "step": 9420
    },
    {
      "epoch": 0.06400953926261806,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009921578180291757,
      "loss": 2.2085,
      "step": 9421
    },
    {
      "epoch": 0.06401633360921212,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009921559199250873,
      "loss": 2.2891,
      "step": 9422
    },
    {
      "epoch": 0.06402312795580617,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009921540215931364,
      "loss": 2.1346,
      "step": 9423
    },
    {
      "epoch": 0.06402992230240023,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009921521230333238,
      "loss": 2.2257,
      "step": 9424
    },
    {
      "epoch": 0.06403671664899428,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009921502242456505,
      "loss": 2.4318,
      "step": 9425
    },
    {
      "epoch": 0.06404351099558835,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009921483252301171,
      "loss": 2.339,
      "step": 9426
    },
    {
      "epoch": 0.06405030534218241,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009921464259867248,
      "loss": 2.2934,
      "step": 9427
    },
    {
      "epoch": 0.06405709968877646,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009921445265154746,
      "loss": 2.2295,
      "step": 9428
    },
    {
      "epoch": 0.06406389403537052,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009921426268163667,
      "loss": 2.2481,
      "step": 9429
    },
    {
      "epoch": 0.06407068838196457,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009921407268894027,
      "loss": 2.3042,
      "step": 9430
    },
    {
      "epoch": 0.06407748272855864,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000992138826734583,
      "loss": 2.3732,
      "step": 9431
    },
    {
      "epoch": 0.0640842770751527,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009921369263519086,
      "loss": 2.1991,
      "step": 9432
    },
    {
      "epoch": 0.06409107142174675,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009921350257413807,
      "loss": 2.4574,
      "step": 9433
    },
    {
      "epoch": 0.06409786576834081,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009921331249029997,
      "loss": 2.2006,
      "step": 9434
    },
    {
      "epoch": 0.06410466011493486,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009921312238367669,
      "loss": 2.2175,
      "step": 9435
    },
    {
      "epoch": 0.06411145446152892,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000992129322542683,
      "loss": 2.1736,
      "step": 9436
    },
    {
      "epoch": 0.06411824880812299,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009921274210207486,
      "loss": 2.2916,
      "step": 9437
    },
    {
      "epoch": 0.06412504315471704,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009921255192709652,
      "loss": 2.3229,
      "step": 9438
    },
    {
      "epoch": 0.0641318375013111,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000992123617293333,
      "loss": 2.3256,
      "step": 9439
    },
    {
      "epoch": 0.06413863184790515,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009921217150878534,
      "loss": 2.1193,
      "step": 9440
    },
    {
      "epoch": 0.06414542619449921,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000992119812654527,
      "loss": 2.5016,
      "step": 9441
    },
    {
      "epoch": 0.06415222054109326,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009921179099933548,
      "loss": 2.2978,
      "step": 9442
    },
    {
      "epoch": 0.06415901488768733,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009921160071043377,
      "loss": 2.3253,
      "step": 9443
    },
    {
      "epoch": 0.06416580923428139,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009921141039874763,
      "loss": 2.3013,
      "step": 9444
    },
    {
      "epoch": 0.06417260358087544,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000992112200642772,
      "loss": 2.1856,
      "step": 9445
    },
    {
      "epoch": 0.0641793979274695,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009921102970702252,
      "loss": 2.2087,
      "step": 9446
    },
    {
      "epoch": 0.06418619227406355,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000992108393269837,
      "loss": 2.172,
      "step": 9447
    },
    {
      "epoch": 0.06419298662065762,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009921064892416084,
      "loss": 2.3572,
      "step": 9448
    },
    {
      "epoch": 0.06419978096725168,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099210458498554,
      "loss": 2.2278,
      "step": 9449
    },
    {
      "epoch": 0.06420657531384573,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009921026805016328,
      "loss": 2.2394,
      "step": 9450
    },
    {
      "epoch": 0.06421336966043979,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000992100775789888,
      "loss": 2.2827,
      "step": 9451
    },
    {
      "epoch": 0.06422016400703384,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009920988708503058,
      "loss": 2.2448,
      "step": 9452
    },
    {
      "epoch": 0.0642269583536279,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009920969656828876,
      "loss": 2.3068,
      "step": 9453
    },
    {
      "epoch": 0.06423375270022197,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009920950602876342,
      "loss": 2.1194,
      "step": 9454
    },
    {
      "epoch": 0.06424054704681602,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009920931546645465,
      "loss": 2.2904,
      "step": 9455
    },
    {
      "epoch": 0.06424734139341008,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009920912488136252,
      "loss": 2.1758,
      "step": 9456
    },
    {
      "epoch": 0.06425413574000413,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009920893427348711,
      "loss": 2.206,
      "step": 9457
    },
    {
      "epoch": 0.0642609300865982,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009920874364282854,
      "loss": 2.1788,
      "step": 9458
    },
    {
      "epoch": 0.06426772443319226,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009920855298938692,
      "loss": 2.3654,
      "step": 9459
    },
    {
      "epoch": 0.06427451877978631,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009920836231316226,
      "loss": 2.2179,
      "step": 9460
    },
    {
      "epoch": 0.06428131312638037,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009920817161415472,
      "loss": 2.3406,
      "step": 9461
    },
    {
      "epoch": 0.06428810747297442,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009920798089236434,
      "loss": 2.2666,
      "step": 9462
    },
    {
      "epoch": 0.06429490181956848,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009920779014779123,
      "loss": 2.3893,
      "step": 9463
    },
    {
      "epoch": 0.06430169616616253,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000992075993804355,
      "loss": 2.1553,
      "step": 9464
    },
    {
      "epoch": 0.0643084905127566,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000992074085902972,
      "loss": 2.3036,
      "step": 9465
    },
    {
      "epoch": 0.06431528485935066,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009920721777737643,
      "loss": 2.3545,
      "step": 9466
    },
    {
      "epoch": 0.06432207920594471,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000992070269416733,
      "loss": 2.2783,
      "step": 9467
    },
    {
      "epoch": 0.06432887355253877,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009920683608318785,
      "loss": 2.2398,
      "step": 9468
    },
    {
      "epoch": 0.06433566789913282,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009920664520192022,
      "loss": 2.1833,
      "step": 9469
    },
    {
      "epoch": 0.06434246224572689,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000992064542978705,
      "loss": 2.205,
      "step": 9470
    },
    {
      "epoch": 0.06434925659232095,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009920626337103873,
      "loss": 2.1696,
      "step": 9471
    },
    {
      "epoch": 0.064356050938915,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00099206072421425,
      "loss": 2.3307,
      "step": 9472
    },
    {
      "epoch": 0.06436284528550906,
      "grad_norm": 0.125,
      "learning_rate": 0.0009920588144902947,
      "loss": 2.1842,
      "step": 9473
    },
    {
      "epoch": 0.06436963963210311,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009920569045385214,
      "loss": 2.3659,
      "step": 9474
    },
    {
      "epoch": 0.06437643397869718,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009920549943589316,
      "loss": 2.2134,
      "step": 9475
    },
    {
      "epoch": 0.06438322832529124,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000992053083951526,
      "loss": 2.2294,
      "step": 9476
    },
    {
      "epoch": 0.06439002267188529,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009920511733163054,
      "loss": 2.3075,
      "step": 9477
    },
    {
      "epoch": 0.06439681701847935,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000992049262453271,
      "loss": 2.328,
      "step": 9478
    },
    {
      "epoch": 0.0644036113650734,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009920473513624231,
      "loss": 2.2807,
      "step": 9479
    },
    {
      "epoch": 0.06441040571166746,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000992045440043763,
      "loss": 2.172,
      "step": 9480
    },
    {
      "epoch": 0.06441720005826153,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009920435284972915,
      "loss": 2.3217,
      "step": 9481
    },
    {
      "epoch": 0.06442399440485558,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009920416167230096,
      "loss": 2.1982,
      "step": 9482
    },
    {
      "epoch": 0.06443078875144964,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009920397047209178,
      "loss": 2.3353,
      "step": 9483
    },
    {
      "epoch": 0.06443758309804369,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009920377924910176,
      "loss": 2.2069,
      "step": 9484
    },
    {
      "epoch": 0.06444437744463775,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009920358800333093,
      "loss": 2.3058,
      "step": 9485
    },
    {
      "epoch": 0.06445117179123182,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009920339673477941,
      "loss": 2.2057,
      "step": 9486
    },
    {
      "epoch": 0.06445796613782587,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009920320544344727,
      "loss": 2.2649,
      "step": 9487
    },
    {
      "epoch": 0.06446476048441993,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009920301412933462,
      "loss": 2.2453,
      "step": 9488
    },
    {
      "epoch": 0.06447155483101398,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009920282279244155,
      "loss": 2.1057,
      "step": 9489
    },
    {
      "epoch": 0.06447834917760804,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009920263143276813,
      "loss": 2.2091,
      "step": 9490
    },
    {
      "epoch": 0.0644851435242021,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009920244005031445,
      "loss": 2.3076,
      "step": 9491
    },
    {
      "epoch": 0.06449193787079616,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009920224864508062,
      "loss": 2.3408,
      "step": 9492
    },
    {
      "epoch": 0.06449873221739022,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009920205721706669,
      "loss": 2.3814,
      "step": 9493
    },
    {
      "epoch": 0.06450552656398427,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000992018657662728,
      "loss": 2.3569,
      "step": 9494
    },
    {
      "epoch": 0.06451232091057833,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009920167429269898,
      "loss": 2.3454,
      "step": 9495
    },
    {
      "epoch": 0.06451911525717238,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009920148279634537,
      "loss": 2.2063,
      "step": 9496
    },
    {
      "epoch": 0.06452590960376645,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009920129127721203,
      "loss": 2.1361,
      "step": 9497
    },
    {
      "epoch": 0.06453270395036051,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009920109973529905,
      "loss": 2.2511,
      "step": 9498
    },
    {
      "epoch": 0.06453949829695456,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009920090817060654,
      "loss": 2.2463,
      "step": 9499
    },
    {
      "epoch": 0.06454629264354862,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009920071658313457,
      "loss": 2.2628,
      "step": 9500
    },
    {
      "epoch": 0.06455308699014267,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000992005249728832,
      "loss": 2.3823,
      "step": 9501
    },
    {
      "epoch": 0.06455988133673674,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009920033333985258,
      "loss": 2.2433,
      "step": 9502
    },
    {
      "epoch": 0.0645666756833308,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009920014168404278,
      "loss": 2.2498,
      "step": 9503
    },
    {
      "epoch": 0.06457347002992485,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009919995000545387,
      "loss": 2.1889,
      "step": 9504
    },
    {
      "epoch": 0.06458026437651891,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009919975830408594,
      "loss": 2.2026,
      "step": 9505
    },
    {
      "epoch": 0.06458705872311296,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000991995665799391,
      "loss": 2.3189,
      "step": 9506
    },
    {
      "epoch": 0.06459385306970702,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000991993748330134,
      "loss": 2.2182,
      "step": 9507
    },
    {
      "epoch": 0.06460064741630109,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009919918306330897,
      "loss": 2.1409,
      "step": 9508
    },
    {
      "epoch": 0.06460744176289514,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009919899127082588,
      "loss": 2.4071,
      "step": 9509
    },
    {
      "epoch": 0.0646142361094892,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009919879945556422,
      "loss": 2.1056,
      "step": 9510
    },
    {
      "epoch": 0.06462103045608325,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009919860761752409,
      "loss": 2.2188,
      "step": 9511
    },
    {
      "epoch": 0.06462782480267731,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009919841575670555,
      "loss": 2.273,
      "step": 9512
    },
    {
      "epoch": 0.06463461914927136,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009919822387310872,
      "loss": 2.233,
      "step": 9513
    },
    {
      "epoch": 0.06464141349586543,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009919803196673366,
      "loss": 2.2181,
      "step": 9514
    },
    {
      "epoch": 0.06464820784245949,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991978400375805,
      "loss": 2.3339,
      "step": 9515
    },
    {
      "epoch": 0.06465500218905354,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009919764808564927,
      "loss": 2.3699,
      "step": 9516
    },
    {
      "epoch": 0.0646617965356476,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009919745611094014,
      "loss": 2.1991,
      "step": 9517
    },
    {
      "epoch": 0.06466859088224165,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009919726411345311,
      "loss": 2.336,
      "step": 9518
    },
    {
      "epoch": 0.06467538522883572,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009919707209318834,
      "loss": 2.1141,
      "step": 9519
    },
    {
      "epoch": 0.06468217957542978,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009919688005014587,
      "loss": 2.1757,
      "step": 9520
    },
    {
      "epoch": 0.06468897392202383,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991966879843258,
      "loss": 2.2227,
      "step": 9521
    },
    {
      "epoch": 0.06469576826861789,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009919649589572826,
      "loss": 2.245,
      "step": 9522
    },
    {
      "epoch": 0.06470256261521194,
      "grad_norm": 0.140625,
      "learning_rate": 0.000991963037843533,
      "loss": 2.4153,
      "step": 9523
    },
    {
      "epoch": 0.064709356961806,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00099196111650201,
      "loss": 2.1433,
      "step": 9524
    },
    {
      "epoch": 0.06471615130840007,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009919591949327147,
      "loss": 2.2301,
      "step": 9525
    },
    {
      "epoch": 0.06472294565499412,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000991957273135648,
      "loss": 2.2133,
      "step": 9526
    },
    {
      "epoch": 0.06472974000158818,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009919553511108107,
      "loss": 2.253,
      "step": 9527
    },
    {
      "epoch": 0.06473653434818223,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009919534288582036,
      "loss": 2.2355,
      "step": 9528
    },
    {
      "epoch": 0.0647433286947763,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009919515063778277,
      "loss": 2.1341,
      "step": 9529
    },
    {
      "epoch": 0.06475012304137036,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000991949583669684,
      "loss": 2.0948,
      "step": 9530
    },
    {
      "epoch": 0.06475691738796441,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009919476607337732,
      "loss": 2.4831,
      "step": 9531
    },
    {
      "epoch": 0.06476371173455847,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009919457375700965,
      "loss": 2.0454,
      "step": 9532
    },
    {
      "epoch": 0.06477050608115252,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009919438141786543,
      "loss": 2.2433,
      "step": 9533
    },
    {
      "epoch": 0.06477730042774658,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009919418905594477,
      "loss": 2.2815,
      "step": 9534
    },
    {
      "epoch": 0.06478409477434063,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009919399667124777,
      "loss": 2.276,
      "step": 9535
    },
    {
      "epoch": 0.0647908891209347,
      "grad_norm": 0.125,
      "learning_rate": 0.0009919380426377453,
      "loss": 2.1976,
      "step": 9536
    },
    {
      "epoch": 0.06479768346752876,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000991936118335251,
      "loss": 2.265,
      "step": 9537
    },
    {
      "epoch": 0.06480447781412281,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009919341938049962,
      "loss": 2.1066,
      "step": 9538
    },
    {
      "epoch": 0.06481127216071687,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009919322690469814,
      "loss": 2.1374,
      "step": 9539
    },
    {
      "epoch": 0.06481806650731092,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009919303440612075,
      "loss": 2.2795,
      "step": 9540
    },
    {
      "epoch": 0.06482486085390499,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009919284188476755,
      "loss": 2.3,
      "step": 9541
    },
    {
      "epoch": 0.06483165520049905,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009919264934063865,
      "loss": 2.3106,
      "step": 9542
    },
    {
      "epoch": 0.0648384495470931,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000991924567737341,
      "loss": 2.2442,
      "step": 9543
    },
    {
      "epoch": 0.06484524389368716,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099192264184054,
      "loss": 2.2265,
      "step": 9544
    },
    {
      "epoch": 0.06485203824028121,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009919207157159846,
      "loss": 2.3621,
      "step": 9545
    },
    {
      "epoch": 0.06485883258687528,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009919187893636754,
      "loss": 2.3348,
      "step": 9546
    },
    {
      "epoch": 0.06486562693346934,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009919168627836135,
      "loss": 2.3117,
      "step": 9547
    },
    {
      "epoch": 0.06487242128006339,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009919149359757997,
      "loss": 2.2025,
      "step": 9548
    },
    {
      "epoch": 0.06487921562665745,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009919130089402351,
      "loss": 2.2218,
      "step": 9549
    },
    {
      "epoch": 0.0648860099732515,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009919110816769203,
      "loss": 2.1613,
      "step": 9550
    },
    {
      "epoch": 0.06489280431984557,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009919091541858562,
      "loss": 2.1357,
      "step": 9551
    },
    {
      "epoch": 0.06489959866643963,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009919072264670439,
      "loss": 2.1056,
      "step": 9552
    },
    {
      "epoch": 0.06490639301303368,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009919052985204843,
      "loss": 2.2238,
      "step": 9553
    },
    {
      "epoch": 0.06491318735962774,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009919033703461781,
      "loss": 2.3461,
      "step": 9554
    },
    {
      "epoch": 0.06491998170622179,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009919014419441262,
      "loss": 2.2006,
      "step": 9555
    },
    {
      "epoch": 0.06492677605281585,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009918995133143295,
      "loss": 2.2149,
      "step": 9556
    },
    {
      "epoch": 0.06493357039940992,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009918975844567892,
      "loss": 2.3245,
      "step": 9557
    },
    {
      "epoch": 0.06494036474600397,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991895655371506,
      "loss": 2.2686,
      "step": 9558
    },
    {
      "epoch": 0.06494715909259803,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009918937260584805,
      "loss": 2.3615,
      "step": 9559
    },
    {
      "epoch": 0.06495395343919208,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000991891796517714,
      "loss": 2.195,
      "step": 9560
    },
    {
      "epoch": 0.06496074778578614,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000991889866749207,
      "loss": 2.1824,
      "step": 9561
    },
    {
      "epoch": 0.0649675421323802,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000991887936752961,
      "loss": 2.2214,
      "step": 9562
    },
    {
      "epoch": 0.06497433647897426,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009918860065289763,
      "loss": 2.3235,
      "step": 9563
    },
    {
      "epoch": 0.06498113082556832,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991884076077254,
      "loss": 2.3426,
      "step": 9564
    },
    {
      "epoch": 0.06498792517216237,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000991882145397795,
      "loss": 2.4367,
      "step": 9565
    },
    {
      "epoch": 0.06499471951875643,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009918802144906003,
      "loss": 2.2742,
      "step": 9566
    },
    {
      "epoch": 0.06500151386535048,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009918782833556708,
      "loss": 2.5383,
      "step": 9567
    },
    {
      "epoch": 0.06500830821194455,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009918763519930073,
      "loss": 2.3753,
      "step": 9568
    },
    {
      "epoch": 0.06501510255853861,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009918744204026105,
      "loss": 2.2864,
      "step": 9569
    },
    {
      "epoch": 0.06502189690513266,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009918724885844814,
      "loss": 2.2056,
      "step": 9570
    },
    {
      "epoch": 0.06502869125172672,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009918705565386213,
      "loss": 2.2684,
      "step": 9571
    },
    {
      "epoch": 0.06503548559832077,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009918686242650304,
      "loss": 2.2453,
      "step": 9572
    },
    {
      "epoch": 0.06504227994491484,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009918666917637102,
      "loss": 2.318,
      "step": 9573
    },
    {
      "epoch": 0.0650490742915089,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009918647590346614,
      "loss": 2.2622,
      "step": 9574
    },
    {
      "epoch": 0.06505586863810295,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009918628260778847,
      "loss": 2.1621,
      "step": 9575
    },
    {
      "epoch": 0.06506266298469701,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009918608928933812,
      "loss": 2.3507,
      "step": 9576
    },
    {
      "epoch": 0.06506945733129106,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009918589594811518,
      "loss": 2.3081,
      "step": 9577
    },
    {
      "epoch": 0.06507625167788512,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009918570258411973,
      "loss": 2.1499,
      "step": 9578
    },
    {
      "epoch": 0.06508304602447919,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009918550919735187,
      "loss": 2.1351,
      "step": 9579
    },
    {
      "epoch": 0.06508984037107324,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009918531578781166,
      "loss": 2.2614,
      "step": 9580
    },
    {
      "epoch": 0.0650966347176673,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009918512235549924,
      "loss": 2.1048,
      "step": 9581
    },
    {
      "epoch": 0.06510342906426135,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009918492890041465,
      "loss": 2.39,
      "step": 9582
    },
    {
      "epoch": 0.06511022341085541,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009918473542255804,
      "loss": 2.2493,
      "step": 9583
    },
    {
      "epoch": 0.06511701775744946,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009918454192192944,
      "loss": 2.1334,
      "step": 9584
    },
    {
      "epoch": 0.06512381210404353,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009918434839852894,
      "loss": 2.302,
      "step": 9585
    },
    {
      "epoch": 0.06513060645063759,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009918415485235667,
      "loss": 2.3706,
      "step": 9586
    },
    {
      "epoch": 0.06513740079723164,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000991839612834127,
      "loss": 2.2173,
      "step": 9587
    },
    {
      "epoch": 0.0651441951438257,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009918376769169712,
      "loss": 2.4765,
      "step": 9588
    },
    {
      "epoch": 0.06515098949041975,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009918357407721,
      "loss": 2.1285,
      "step": 9589
    },
    {
      "epoch": 0.06515778383701382,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009918338043995148,
      "loss": 2.2222,
      "step": 9590
    },
    {
      "epoch": 0.06516457818360788,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000991831867799216,
      "loss": 2.2034,
      "step": 9591
    },
    {
      "epoch": 0.06517137253020193,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009918299309712047,
      "loss": 2.1,
      "step": 9592
    },
    {
      "epoch": 0.06517816687679599,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009918279939154818,
      "loss": 2.386,
      "step": 9593
    },
    {
      "epoch": 0.06518496122339004,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009918260566320483,
      "loss": 2.4206,
      "step": 9594
    },
    {
      "epoch": 0.0651917555699841,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009918241191209049,
      "loss": 2.2099,
      "step": 9595
    },
    {
      "epoch": 0.06519854991657817,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009918221813820524,
      "loss": 2.2131,
      "step": 9596
    },
    {
      "epoch": 0.06520534426317222,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000991820243415492,
      "loss": 2.2335,
      "step": 9597
    },
    {
      "epoch": 0.06521213860976628,
      "grad_norm": 0.453125,
      "learning_rate": 0.0009918183052212246,
      "loss": 2.2327,
      "step": 9598
    },
    {
      "epoch": 0.06521893295636033,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009918163667992509,
      "loss": 2.2417,
      "step": 9599
    },
    {
      "epoch": 0.0652257273029544,
      "grad_norm": 0.18359375,
      "learning_rate": 0.0009918144281495715,
      "loss": 2.4119,
      "step": 9600
    },
    {
      "epoch": 0.06523252164954846,
      "grad_norm": 0.2265625,
      "learning_rate": 0.000991812489272188,
      "loss": 2.244,
      "step": 9601
    },
    {
      "epoch": 0.06523931599614251,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000991810550167101,
      "loss": 2.3129,
      "step": 9602
    },
    {
      "epoch": 0.06524611034273657,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009918086108343111,
      "loss": 2.1853,
      "step": 9603
    },
    {
      "epoch": 0.06525290468933062,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009918066712738198,
      "loss": 2.2267,
      "step": 9604
    },
    {
      "epoch": 0.06525969903592468,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009918047314856273,
      "loss": 2.0842,
      "step": 9605
    },
    {
      "epoch": 0.06526649338251875,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000991802791469735,
      "loss": 2.1613,
      "step": 9606
    },
    {
      "epoch": 0.0652732877291128,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009918008512261438,
      "loss": 2.1584,
      "step": 9607
    },
    {
      "epoch": 0.06528008207570686,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009917989107548542,
      "loss": 2.162,
      "step": 9608
    },
    {
      "epoch": 0.06528687642230091,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009917969700558675,
      "loss": 2.1778,
      "step": 9609
    },
    {
      "epoch": 0.06529367076889497,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009917950291291845,
      "loss": 2.3956,
      "step": 9610
    },
    {
      "epoch": 0.06530046511548902,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009917930879748058,
      "loss": 2.1555,
      "step": 9611
    },
    {
      "epoch": 0.06530725946208309,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009917911465927326,
      "loss": 2.3792,
      "step": 9612
    },
    {
      "epoch": 0.06531405380867715,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009917892049829658,
      "loss": 2.2315,
      "step": 9613
    },
    {
      "epoch": 0.0653208481552712,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009917872631455063,
      "loss": 2.2175,
      "step": 9614
    },
    {
      "epoch": 0.06532764250186526,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009917853210803548,
      "loss": 2.2815,
      "step": 9615
    },
    {
      "epoch": 0.06533443684845931,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009917833787875124,
      "loss": 2.1862,
      "step": 9616
    },
    {
      "epoch": 0.06534123119505338,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00099178143626698,
      "loss": 2.2172,
      "step": 9617
    },
    {
      "epoch": 0.06534802554164744,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009917794935187583,
      "loss": 2.1751,
      "step": 9618
    },
    {
      "epoch": 0.06535481988824149,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009917775505428484,
      "loss": 2.2417,
      "step": 9619
    },
    {
      "epoch": 0.06536161423483555,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009917756073392512,
      "loss": 2.2708,
      "step": 9620
    },
    {
      "epoch": 0.0653684085814296,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009917736639079675,
      "loss": 2.2841,
      "step": 9621
    },
    {
      "epoch": 0.06537520292802367,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009917717202489983,
      "loss": 2.2166,
      "step": 9622
    },
    {
      "epoch": 0.06538199727461773,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009917697763623442,
      "loss": 2.302,
      "step": 9623
    },
    {
      "epoch": 0.06538879162121178,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009917678322480064,
      "loss": 2.1668,
      "step": 9624
    },
    {
      "epoch": 0.06539558596780584,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009917658879059859,
      "loss": 2.2521,
      "step": 9625
    },
    {
      "epoch": 0.06540238031439989,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009917639433362833,
      "loss": 2.1511,
      "step": 9626
    },
    {
      "epoch": 0.06540917466099395,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009917619985388996,
      "loss": 2.3605,
      "step": 9627
    },
    {
      "epoch": 0.06541596900758802,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009917600535138359,
      "loss": 2.3592,
      "step": 9628
    },
    {
      "epoch": 0.06542276335418207,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009917581082610927,
      "loss": 2.4073,
      "step": 9629
    },
    {
      "epoch": 0.06542955770077613,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009917561627806712,
      "loss": 2.3061,
      "step": 9630
    },
    {
      "epoch": 0.06543635204737018,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009917542170725724,
      "loss": 2.3256,
      "step": 9631
    },
    {
      "epoch": 0.06544314639396424,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009917522711367969,
      "loss": 2.227,
      "step": 9632
    },
    {
      "epoch": 0.0654499407405583,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009917503249733458,
      "loss": 2.1849,
      "step": 9633
    },
    {
      "epoch": 0.06545673508715236,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009917483785822198,
      "loss": 2.2225,
      "step": 9634
    },
    {
      "epoch": 0.06546352943374642,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00099174643196342,
      "loss": 2.2714,
      "step": 9635
    },
    {
      "epoch": 0.06547032378034047,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009917444851169472,
      "loss": 2.2078,
      "step": 9636
    },
    {
      "epoch": 0.06547711812693453,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009917425380428025,
      "loss": 2.1578,
      "step": 9637
    },
    {
      "epoch": 0.06548391247352858,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009917405907409866,
      "loss": 2.2737,
      "step": 9638
    },
    {
      "epoch": 0.06549070682012265,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009917386432115004,
      "loss": 2.2974,
      "step": 9639
    },
    {
      "epoch": 0.06549750116671671,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009917366954543447,
      "loss": 2.1678,
      "step": 9640
    },
    {
      "epoch": 0.06550429551331076,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000991734747469521,
      "loss": 2.205,
      "step": 9641
    },
    {
      "epoch": 0.06551108985990482,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009917327992570292,
      "loss": 2.3818,
      "step": 9642
    },
    {
      "epoch": 0.06551788420649887,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000991730850816871,
      "loss": 2.2836,
      "step": 9643
    },
    {
      "epoch": 0.06552467855309294,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009917289021490472,
      "loss": 2.3789,
      "step": 9644
    },
    {
      "epoch": 0.065531472899687,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009917269532535584,
      "loss": 2.3065,
      "step": 9645
    },
    {
      "epoch": 0.06553826724628105,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009917250041304056,
      "loss": 2.3773,
      "step": 9646
    },
    {
      "epoch": 0.06554506159287511,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009917230547795898,
      "loss": 2.315,
      "step": 9647
    },
    {
      "epoch": 0.06555185593946916,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991721105201112,
      "loss": 2.243,
      "step": 9648
    },
    {
      "epoch": 0.06555865028606322,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009917191553949728,
      "loss": 2.2425,
      "step": 9649
    },
    {
      "epoch": 0.06556544463265729,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009917172053611732,
      "loss": 2.164,
      "step": 9650
    },
    {
      "epoch": 0.06557223897925134,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009917152550997144,
      "loss": 2.238,
      "step": 9651
    },
    {
      "epoch": 0.0655790333258454,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000991713304610597,
      "loss": 2.3477,
      "step": 9652
    },
    {
      "epoch": 0.06558582767243945,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009917113538938217,
      "loss": 2.1876,
      "step": 9653
    },
    {
      "epoch": 0.06559262201903351,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009917094029493901,
      "loss": 2.2218,
      "step": 9654
    },
    {
      "epoch": 0.06559941636562756,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009917074517773025,
      "loss": 2.0742,
      "step": 9655
    },
    {
      "epoch": 0.06560621071222163,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00099170550037756,
      "loss": 2.4384,
      "step": 9656
    },
    {
      "epoch": 0.06561300505881569,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009917035487501636,
      "loss": 2.2468,
      "step": 9657
    },
    {
      "epoch": 0.06561979940540974,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009917015968951138,
      "loss": 2.2228,
      "step": 9658
    },
    {
      "epoch": 0.0656265937520038,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000991699644812412,
      "loss": 2.2607,
      "step": 9659
    },
    {
      "epoch": 0.06563338809859785,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009916976925020589,
      "loss": 2.1071,
      "step": 9660
    },
    {
      "epoch": 0.06564018244519192,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009916957399640554,
      "loss": 2.3314,
      "step": 9661
    },
    {
      "epoch": 0.06564697679178598,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009916937871984025,
      "loss": 2.3402,
      "step": 9662
    },
    {
      "epoch": 0.06565377113838003,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009916918342051009,
      "loss": 2.3301,
      "step": 9663
    },
    {
      "epoch": 0.06566056548497409,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009916898809841515,
      "loss": 2.4058,
      "step": 9664
    },
    {
      "epoch": 0.06566735983156814,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009916879275355555,
      "loss": 2.1223,
      "step": 9665
    },
    {
      "epoch": 0.0656741541781622,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009916859738593137,
      "loss": 2.4022,
      "step": 9666
    },
    {
      "epoch": 0.06568094852475627,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009916840199554267,
      "loss": 2.1169,
      "step": 9667
    },
    {
      "epoch": 0.06568774287135032,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009916820658238956,
      "loss": 2.2464,
      "step": 9668
    },
    {
      "epoch": 0.06569453721794438,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009916801114647216,
      "loss": 2.2504,
      "step": 9669
    },
    {
      "epoch": 0.06570133156453843,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009916781568779052,
      "loss": 2.3499,
      "step": 9670
    },
    {
      "epoch": 0.0657081259111325,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009916762020634475,
      "loss": 2.3216,
      "step": 9671
    },
    {
      "epoch": 0.06571492025772656,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009916742470213492,
      "loss": 2.243,
      "step": 9672
    },
    {
      "epoch": 0.06572171460432061,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009916722917516115,
      "loss": 2.3542,
      "step": 9673
    },
    {
      "epoch": 0.06572850895091467,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009916703362542352,
      "loss": 2.2184,
      "step": 9674
    },
    {
      "epoch": 0.06573530329750872,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009916683805292212,
      "loss": 2.2261,
      "step": 9675
    },
    {
      "epoch": 0.06574209764410278,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009916664245765703,
      "loss": 2.2625,
      "step": 9676
    },
    {
      "epoch": 0.06574889199069685,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009916644683962835,
      "loss": 2.3144,
      "step": 9677
    },
    {
      "epoch": 0.0657556863372909,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009916625119883616,
      "loss": 2.3581,
      "step": 9678
    },
    {
      "epoch": 0.06576248068388496,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009916605553528057,
      "loss": 2.3282,
      "step": 9679
    },
    {
      "epoch": 0.06576927503047901,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009916585984896165,
      "loss": 2.2726,
      "step": 9680
    },
    {
      "epoch": 0.06577606937707307,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991656641398795,
      "loss": 2.2245,
      "step": 9681
    },
    {
      "epoch": 0.06578286372366712,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009916546840803423,
      "loss": 2.3425,
      "step": 9682
    },
    {
      "epoch": 0.06578965807026119,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000991652726534259,
      "loss": 2.2937,
      "step": 9683
    },
    {
      "epoch": 0.06579645241685525,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991650768760546,
      "loss": 2.4106,
      "step": 9684
    },
    {
      "epoch": 0.0658032467634493,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009916488107592045,
      "loss": 2.2654,
      "step": 9685
    },
    {
      "epoch": 0.06581004111004336,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009916468525302353,
      "loss": 2.1195,
      "step": 9686
    },
    {
      "epoch": 0.06581683545663741,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009916448940736392,
      "loss": 2.2444,
      "step": 9687
    },
    {
      "epoch": 0.06582362980323148,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000991642935389417,
      "loss": 2.3737,
      "step": 9688
    },
    {
      "epoch": 0.06583042414982554,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009916409764775698,
      "loss": 2.2379,
      "step": 9689
    },
    {
      "epoch": 0.06583721849641959,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009916390173380985,
      "loss": 2.1706,
      "step": 9690
    },
    {
      "epoch": 0.06584401284301365,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009916370579710042,
      "loss": 2.2062,
      "step": 9691
    },
    {
      "epoch": 0.0658508071896077,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009916350983762872,
      "loss": 2.4785,
      "step": 9692
    },
    {
      "epoch": 0.06585760153620177,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000991633138553949,
      "loss": 2.3889,
      "step": 9693
    },
    {
      "epoch": 0.06586439588279583,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009916311785039902,
      "loss": 2.2185,
      "step": 9694
    },
    {
      "epoch": 0.06587119022938988,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000991629218226412,
      "loss": 2.2887,
      "step": 9695
    },
    {
      "epoch": 0.06587798457598394,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000991627257721215,
      "loss": 2.3836,
      "step": 9696
    },
    {
      "epoch": 0.06588477892257799,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009916252969884,
      "loss": 2.3363,
      "step": 9697
    },
    {
      "epoch": 0.06589157326917205,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009916233360279683,
      "loss": 2.2346,
      "step": 9698
    },
    {
      "epoch": 0.06589836761576612,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009916213748399206,
      "loss": 2.222,
      "step": 9699
    },
    {
      "epoch": 0.06590516196236017,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009916194134242577,
      "loss": 2.372,
      "step": 9700
    },
    {
      "epoch": 0.06591195630895423,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009916174517809809,
      "loss": 2.2227,
      "step": 9701
    },
    {
      "epoch": 0.06591875065554828,
      "grad_norm": 0.140625,
      "learning_rate": 0.000991615489910091,
      "loss": 2.306,
      "step": 9702
    },
    {
      "epoch": 0.06592554500214234,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009916135278115884,
      "loss": 2.3018,
      "step": 9703
    },
    {
      "epoch": 0.0659323393487364,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009916115654854745,
      "loss": 2.1883,
      "step": 9704
    },
    {
      "epoch": 0.06593913369533046,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009916096029317502,
      "loss": 2.2957,
      "step": 9705
    },
    {
      "epoch": 0.06594592804192452,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991607640150416,
      "loss": 2.2231,
      "step": 9706
    },
    {
      "epoch": 0.06595272238851857,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000991605677141473,
      "loss": 2.2168,
      "step": 9707
    },
    {
      "epoch": 0.06595951673511263,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009916037139049228,
      "loss": 2.2235,
      "step": 9708
    },
    {
      "epoch": 0.06596631108170668,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009916017504407652,
      "loss": 2.4023,
      "step": 9709
    },
    {
      "epoch": 0.06597310542830075,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009915997867490018,
      "loss": 2.283,
      "step": 9710
    },
    {
      "epoch": 0.06597989977489481,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009915978228296333,
      "loss": 2.1658,
      "step": 9711
    },
    {
      "epoch": 0.06598669412148886,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009915958586826608,
      "loss": 2.1111,
      "step": 9712
    },
    {
      "epoch": 0.06599348846808292,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991593894308085,
      "loss": 2.3333,
      "step": 9713
    },
    {
      "epoch": 0.06600028281467697,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009915919297059069,
      "loss": 2.4827,
      "step": 9714
    },
    {
      "epoch": 0.06600707716127104,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009915899648761273,
      "loss": 2.2492,
      "step": 9715
    },
    {
      "epoch": 0.0660138715078651,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000991587999818747,
      "loss": 2.4179,
      "step": 9716
    },
    {
      "epoch": 0.06602066585445915,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009915860345337674,
      "loss": 2.1572,
      "step": 9717
    },
    {
      "epoch": 0.06602746020105321,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000991584069021189,
      "loss": 2.4232,
      "step": 9718
    },
    {
      "epoch": 0.06603425454764726,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009915821032810126,
      "loss": 2.2188,
      "step": 9719
    },
    {
      "epoch": 0.06604104889424132,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009915801373132395,
      "loss": 2.3012,
      "step": 9720
    },
    {
      "epoch": 0.06604784324083539,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009915781711178704,
      "loss": 2.2905,
      "step": 9721
    },
    {
      "epoch": 0.06605463758742944,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009915762046949063,
      "loss": 2.2119,
      "step": 9722
    },
    {
      "epoch": 0.0660614319340235,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009915742380443482,
      "loss": 2.2005,
      "step": 9723
    },
    {
      "epoch": 0.06606822628061755,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009915722711661965,
      "loss": 2.0991,
      "step": 9724
    },
    {
      "epoch": 0.06607502062721161,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009915703040604526,
      "loss": 2.346,
      "step": 9725
    },
    {
      "epoch": 0.06608181497380566,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009915683367271175,
      "loss": 2.2937,
      "step": 9726
    },
    {
      "epoch": 0.06608860932039973,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000991566369166192,
      "loss": 2.233,
      "step": 9727
    },
    {
      "epoch": 0.06609540366699379,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009915644013776764,
      "loss": 2.3416,
      "step": 9728
    },
    {
      "epoch": 0.06610219801358784,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009915624333615724,
      "loss": 2.4709,
      "step": 9729
    },
    {
      "epoch": 0.0661089923601819,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009915604651178805,
      "loss": 2.2748,
      "step": 9730
    },
    {
      "epoch": 0.06611578670677595,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000991558496646602,
      "loss": 2.3076,
      "step": 9731
    },
    {
      "epoch": 0.06612258105337002,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009915565279477372,
      "loss": 2.337,
      "step": 9732
    },
    {
      "epoch": 0.06612937539996408,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009915545590212876,
      "loss": 2.2384,
      "step": 9733
    },
    {
      "epoch": 0.06613616974655813,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000991552589867254,
      "loss": 2.3406,
      "step": 9734
    },
    {
      "epoch": 0.06614296409315219,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009915506204856367,
      "loss": 2.1299,
      "step": 9735
    },
    {
      "epoch": 0.06614975843974624,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009915486508764375,
      "loss": 2.1556,
      "step": 9736
    },
    {
      "epoch": 0.0661565527863403,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000991546681039657,
      "loss": 2.0902,
      "step": 9737
    },
    {
      "epoch": 0.06616334713293437,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009915447109752958,
      "loss": 2.2964,
      "step": 9738
    },
    {
      "epoch": 0.06617014147952842,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009915427406833551,
      "loss": 2.1062,
      "step": 9739
    },
    {
      "epoch": 0.06617693582612248,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009915407701638357,
      "loss": 2.3962,
      "step": 9740
    },
    {
      "epoch": 0.06618373017271653,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009915387994167385,
      "loss": 2.1254,
      "step": 9741
    },
    {
      "epoch": 0.0661905245193106,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009915368284420644,
      "loss": 2.1843,
      "step": 9742
    },
    {
      "epoch": 0.06619731886590466,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009915348572398146,
      "loss": 2.1423,
      "step": 9743
    },
    {
      "epoch": 0.06620411321249871,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009915328858099899,
      "loss": 2.1811,
      "step": 9744
    },
    {
      "epoch": 0.06621090755909277,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009915309141525909,
      "loss": 2.1926,
      "step": 9745
    },
    {
      "epoch": 0.06621770190568682,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009915289422676187,
      "loss": 2.2508,
      "step": 9746
    },
    {
      "epoch": 0.06622449625228088,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009915269701550743,
      "loss": 2.1999,
      "step": 9747
    },
    {
      "epoch": 0.06623129059887495,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009915249978149586,
      "loss": 2.232,
      "step": 9748
    },
    {
      "epoch": 0.066238084945469,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009915230252472724,
      "loss": 2.2932,
      "step": 9749
    },
    {
      "epoch": 0.06624487929206306,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009915210524520167,
      "loss": 2.2708,
      "step": 9750
    },
    {
      "epoch": 0.06625167363865711,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009915190794291923,
      "loss": 2.1907,
      "step": 9751
    },
    {
      "epoch": 0.06625846798525117,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009915171061788003,
      "loss": 2.307,
      "step": 9752
    },
    {
      "epoch": 0.06626526233184522,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009915151327008413,
      "loss": 2.2559,
      "step": 9753
    },
    {
      "epoch": 0.06627205667843929,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009915131589953168,
      "loss": 2.2309,
      "step": 9754
    },
    {
      "epoch": 0.06627885102503335,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000991511185062227,
      "loss": 2.1703,
      "step": 9755
    },
    {
      "epoch": 0.0662856453716274,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009915092109015733,
      "loss": 2.2093,
      "step": 9756
    },
    {
      "epoch": 0.06629243971822146,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009915072365133566,
      "loss": 2.2285,
      "step": 9757
    },
    {
      "epoch": 0.06629923406481551,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009915052618975776,
      "loss": 2.322,
      "step": 9758
    },
    {
      "epoch": 0.06630602841140958,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009915032870542373,
      "loss": 2.2989,
      "step": 9759
    },
    {
      "epoch": 0.06631282275800364,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009915013119833364,
      "loss": 2.2264,
      "step": 9760
    },
    {
      "epoch": 0.06631961710459769,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009914993366848764,
      "loss": 2.1394,
      "step": 9761
    },
    {
      "epoch": 0.06632641145119175,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009914973611588576,
      "loss": 2.2635,
      "step": 9762
    },
    {
      "epoch": 0.0663332057977858,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009914953854052813,
      "loss": 2.1155,
      "step": 9763
    },
    {
      "epoch": 0.06634000014437987,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009914934094241482,
      "loss": 2.2539,
      "step": 9764
    },
    {
      "epoch": 0.06634679449097393,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009914914332154591,
      "loss": 2.3794,
      "step": 9765
    },
    {
      "epoch": 0.06635358883756798,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009914894567792153,
      "loss": 2.3529,
      "step": 9766
    },
    {
      "epoch": 0.06636038318416204,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009914874801154174,
      "loss": 2.2808,
      "step": 9767
    },
    {
      "epoch": 0.06636717753075609,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009914855032240666,
      "loss": 2.1236,
      "step": 9768
    },
    {
      "epoch": 0.06637397187735015,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009914835261051636,
      "loss": 2.3615,
      "step": 9769
    },
    {
      "epoch": 0.06638076622394422,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009914815487587093,
      "loss": 2.2623,
      "step": 9770
    },
    {
      "epoch": 0.06638756057053827,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009914795711847048,
      "loss": 2.273,
      "step": 9771
    },
    {
      "epoch": 0.06639435491713233,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009914775933831505,
      "loss": 2.2179,
      "step": 9772
    },
    {
      "epoch": 0.06640114926372638,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000991475615354048,
      "loss": 2.2407,
      "step": 9773
    },
    {
      "epoch": 0.06640794361032044,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991473637097398,
      "loss": 2.2209,
      "step": 9774
    },
    {
      "epoch": 0.0664147379569145,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009914716586132014,
      "loss": 2.2833,
      "step": 9775
    },
    {
      "epoch": 0.06642153230350856,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000991469679901459,
      "loss": 2.1637,
      "step": 9776
    },
    {
      "epoch": 0.06642832665010262,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009914677009621716,
      "loss": 2.1758,
      "step": 9777
    },
    {
      "epoch": 0.06643512099669667,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009914657217953403,
      "loss": 2.3745,
      "step": 9778
    },
    {
      "epoch": 0.06644191534329073,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991463742400966,
      "loss": 2.2906,
      "step": 9779
    },
    {
      "epoch": 0.06644870968988478,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009914617627790497,
      "loss": 2.1754,
      "step": 9780
    },
    {
      "epoch": 0.06645550403647885,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009914597829295923,
      "loss": 2.251,
      "step": 9781
    },
    {
      "epoch": 0.06646229838307291,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009914578028525946,
      "loss": 2.3209,
      "step": 9782
    },
    {
      "epoch": 0.06646909272966696,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009914558225480577,
      "loss": 2.2531,
      "step": 9783
    },
    {
      "epoch": 0.06647588707626102,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000991453842015982,
      "loss": 2.262,
      "step": 9784
    },
    {
      "epoch": 0.06648268142285507,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009914518612563691,
      "loss": 2.1626,
      "step": 9785
    },
    {
      "epoch": 0.06648947576944914,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009914498802692196,
      "loss": 2.4664,
      "step": 9786
    },
    {
      "epoch": 0.0664962701160432,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009914478990545345,
      "loss": 2.2665,
      "step": 9787
    },
    {
      "epoch": 0.06650306446263725,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009914459176123146,
      "loss": 2.281,
      "step": 9788
    },
    {
      "epoch": 0.06650985880923131,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000991443935942561,
      "loss": 2.2686,
      "step": 9789
    },
    {
      "epoch": 0.06651665315582536,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009914419540452742,
      "loss": 2.2803,
      "step": 9790
    },
    {
      "epoch": 0.06652344750241942,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009914399719204556,
      "loss": 2.26,
      "step": 9791
    },
    {
      "epoch": 0.06653024184901349,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009914379895681058,
      "loss": 2.1668,
      "step": 9792
    },
    {
      "epoch": 0.06653703619560754,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000991436006988226,
      "loss": 2.1835,
      "step": 9793
    },
    {
      "epoch": 0.0665438305422016,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009914340241808169,
      "loss": 2.2782,
      "step": 9794
    },
    {
      "epoch": 0.06655062488879565,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009914320411458794,
      "loss": 2.3146,
      "step": 9795
    },
    {
      "epoch": 0.06655741923538971,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009914300578834146,
      "loss": 2.219,
      "step": 9796
    },
    {
      "epoch": 0.06656421358198378,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009914280743934232,
      "loss": 2.1904,
      "step": 9797
    },
    {
      "epoch": 0.06657100792857783,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009914260906759066,
      "loss": 2.156,
      "step": 9798
    },
    {
      "epoch": 0.06657780227517189,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991424106730865,
      "loss": 2.2713,
      "step": 9799
    },
    {
      "epoch": 0.06658459662176594,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009914221225582997,
      "loss": 2.4609,
      "step": 9800
    },
    {
      "epoch": 0.06659139096836,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009914201381582117,
      "loss": 2.2097,
      "step": 9801
    },
    {
      "epoch": 0.06659818531495405,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009914181535306016,
      "loss": 2.1594,
      "step": 9802
    },
    {
      "epoch": 0.06660497966154812,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000991416168675471,
      "loss": 2.3173,
      "step": 9803
    },
    {
      "epoch": 0.06661177400814218,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099141418359282,
      "loss": 2.2391,
      "step": 9804
    },
    {
      "epoch": 0.06661856835473623,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009914121982826497,
      "loss": 2.1915,
      "step": 9805
    },
    {
      "epoch": 0.06662536270133029,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009914102127449613,
      "loss": 2.088,
      "step": 9806
    },
    {
      "epoch": 0.06663215704792434,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009914082269797558,
      "loss": 2.3771,
      "step": 9807
    },
    {
      "epoch": 0.0666389513945184,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009914062409870337,
      "loss": 2.1901,
      "step": 9808
    },
    {
      "epoch": 0.06664574574111247,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009914042547667964,
      "loss": 2.2439,
      "step": 9809
    },
    {
      "epoch": 0.06665254008770652,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009914022683190445,
      "loss": 2.1546,
      "step": 9810
    },
    {
      "epoch": 0.06665933443430058,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009914002816437787,
      "loss": 2.2934,
      "step": 9811
    },
    {
      "epoch": 0.06666612878089463,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009913982947410006,
      "loss": 2.2066,
      "step": 9812
    },
    {
      "epoch": 0.0666729231274887,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009913963076107104,
      "loss": 2.3703,
      "step": 9813
    },
    {
      "epoch": 0.06667971747408276,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009913943202529095,
      "loss": 2.3074,
      "step": 9814
    },
    {
      "epoch": 0.06668651182067681,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009913923326675986,
      "loss": 2.2694,
      "step": 9815
    },
    {
      "epoch": 0.06669330616727087,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009913903448547788,
      "loss": 2.2159,
      "step": 9816
    },
    {
      "epoch": 0.06670010051386492,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009913883568144508,
      "loss": 2.3927,
      "step": 9817
    },
    {
      "epoch": 0.06670689486045898,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009913863685466157,
      "loss": 2.2456,
      "step": 9818
    },
    {
      "epoch": 0.06671368920705305,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009913843800512742,
      "loss": 2.2794,
      "step": 9819
    },
    {
      "epoch": 0.0667204835536471,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009913823913284274,
      "loss": 2.2076,
      "step": 9820
    },
    {
      "epoch": 0.06672727790024116,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009913804023780764,
      "loss": 2.3159,
      "step": 9821
    },
    {
      "epoch": 0.06673407224683521,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009913784132002217,
      "loss": 2.0542,
      "step": 9822
    },
    {
      "epoch": 0.06674086659342927,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009913764237948644,
      "loss": 2.2601,
      "step": 9823
    },
    {
      "epoch": 0.06674766094002332,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009913744341620058,
      "loss": 2.2569,
      "step": 9824
    },
    {
      "epoch": 0.06675445528661739,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009913724443016462,
      "loss": 2.3904,
      "step": 9825
    },
    {
      "epoch": 0.06676124963321145,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009913704542137865,
      "loss": 2.3635,
      "step": 9826
    },
    {
      "epoch": 0.0667680439798055,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009913684638984284,
      "loss": 2.2946,
      "step": 9827
    },
    {
      "epoch": 0.06677483832639956,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000991366473355572,
      "loss": 2.2324,
      "step": 9828
    },
    {
      "epoch": 0.06678163267299361,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009913644825852188,
      "loss": 2.3741,
      "step": 9829
    },
    {
      "epoch": 0.06678842701958768,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009913624915873693,
      "loss": 2.3269,
      "step": 9830
    },
    {
      "epoch": 0.06679522136618174,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009913605003620248,
      "loss": 2.121,
      "step": 9831
    },
    {
      "epoch": 0.06680201571277579,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009913585089091858,
      "loss": 2.1701,
      "step": 9832
    },
    {
      "epoch": 0.06680881005936985,
      "grad_norm": 0.125,
      "learning_rate": 0.0009913565172288534,
      "loss": 2.0814,
      "step": 9833
    },
    {
      "epoch": 0.0668156044059639,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009913545253210287,
      "loss": 2.2701,
      "step": 9834
    },
    {
      "epoch": 0.06682239875255797,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009913525331857126,
      "loss": 2.2078,
      "step": 9835
    },
    {
      "epoch": 0.06682919309915203,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009913505408229058,
      "loss": 2.272,
      "step": 9836
    },
    {
      "epoch": 0.06683598744574608,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009913485482326095,
      "loss": 2.076,
      "step": 9837
    },
    {
      "epoch": 0.06684278179234014,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009913465554148243,
      "loss": 2.2081,
      "step": 9838
    },
    {
      "epoch": 0.06684957613893419,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009913445623695512,
      "loss": 2.4091,
      "step": 9839
    },
    {
      "epoch": 0.06685637048552825,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009913425690967913,
      "loss": 2.4965,
      "step": 9840
    },
    {
      "epoch": 0.06686316483212232,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009913405755965453,
      "loss": 2.1453,
      "step": 9841
    },
    {
      "epoch": 0.06686995917871637,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009913385818688142,
      "loss": 2.2041,
      "step": 9842
    },
    {
      "epoch": 0.06687675352531043,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009913365879135993,
      "loss": 2.2257,
      "step": 9843
    },
    {
      "epoch": 0.06688354787190448,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000991334593730901,
      "loss": 2.1988,
      "step": 9844
    },
    {
      "epoch": 0.06689034221849854,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009913325993207204,
      "loss": 2.2507,
      "step": 9845
    },
    {
      "epoch": 0.0668971365650926,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009913306046830583,
      "loss": 2.2672,
      "step": 9846
    },
    {
      "epoch": 0.06690393091168666,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000991328609817916,
      "loss": 2.0534,
      "step": 9847
    },
    {
      "epoch": 0.06691072525828072,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009913266147252942,
      "loss": 2.3139,
      "step": 9848
    },
    {
      "epoch": 0.06691751960487477,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009913246194051936,
      "loss": 2.2285,
      "step": 9849
    },
    {
      "epoch": 0.06692431395146883,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009913226238576155,
      "loss": 2.2799,
      "step": 9850
    },
    {
      "epoch": 0.06693110829806288,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009913206280825605,
      "loss": 2.314,
      "step": 9851
    },
    {
      "epoch": 0.06693790264465695,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00099131863208003,
      "loss": 1.9469,
      "step": 9852
    },
    {
      "epoch": 0.06694469699125101,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009913166358500242,
      "loss": 2.2219,
      "step": 9853
    },
    {
      "epoch": 0.06695149133784506,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009913146393925446,
      "loss": 2.0409,
      "step": 9854
    },
    {
      "epoch": 0.06695828568443912,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000991312642707592,
      "loss": 2.2371,
      "step": 9855
    },
    {
      "epoch": 0.06696508003103317,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009913106457951674,
      "loss": 2.2055,
      "step": 9856
    },
    {
      "epoch": 0.06697187437762724,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009913086486552714,
      "loss": 2.25,
      "step": 9857
    },
    {
      "epoch": 0.0669786687242213,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009913066512879053,
      "loss": 2.2939,
      "step": 9858
    },
    {
      "epoch": 0.06698546307081535,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009913046536930697,
      "loss": 2.3937,
      "step": 9859
    },
    {
      "epoch": 0.06699225741740941,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009913026558707656,
      "loss": 2.2007,
      "step": 9860
    },
    {
      "epoch": 0.06699905176400346,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009913006578209942,
      "loss": 2.1612,
      "step": 9861
    },
    {
      "epoch": 0.06700584611059753,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009912986595437562,
      "loss": 2.3701,
      "step": 9862
    },
    {
      "epoch": 0.06701264045719159,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009912966610390525,
      "loss": 2.3618,
      "step": 9863
    },
    {
      "epoch": 0.06701943480378564,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009912946623068842,
      "loss": 2.2765,
      "step": 9864
    },
    {
      "epoch": 0.0670262291503797,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009912926633472519,
      "loss": 2.2978,
      "step": 9865
    },
    {
      "epoch": 0.06703302349697375,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009912906641601567,
      "loss": 2.1759,
      "step": 9866
    },
    {
      "epoch": 0.06703981784356781,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009912886647455998,
      "loss": 2.229,
      "step": 9867
    },
    {
      "epoch": 0.06704661219016188,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009912866651035817,
      "loss": 2.1538,
      "step": 9868
    },
    {
      "epoch": 0.06705340653675593,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009912846652341037,
      "loss": 2.337,
      "step": 9869
    },
    {
      "epoch": 0.06706020088334999,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009912826651371663,
      "loss": 2.2873,
      "step": 9870
    },
    {
      "epoch": 0.06706699522994404,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009912806648127708,
      "loss": 2.1613,
      "step": 9871
    },
    {
      "epoch": 0.0670737895765381,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000991278664260918,
      "loss": 2.2015,
      "step": 9872
    },
    {
      "epoch": 0.06708058392313215,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009912766634816085,
      "loss": 2.1904,
      "step": 9873
    },
    {
      "epoch": 0.06708737826972622,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991274662474844,
      "loss": 2.0789,
      "step": 9874
    },
    {
      "epoch": 0.06709417261632028,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009912726612406246,
      "loss": 2.177,
      "step": 9875
    },
    {
      "epoch": 0.06710096696291433,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009912706597789518,
      "loss": 2.209,
      "step": 9876
    },
    {
      "epoch": 0.06710776130950839,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009912686580898263,
      "loss": 2.2141,
      "step": 9877
    },
    {
      "epoch": 0.06711455565610244,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009912666561732487,
      "loss": 2.4228,
      "step": 9878
    },
    {
      "epoch": 0.0671213500026965,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009912646540292207,
      "loss": 2.2488,
      "step": 9879
    },
    {
      "epoch": 0.06712814434929057,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009912626516577427,
      "loss": 2.2152,
      "step": 9880
    },
    {
      "epoch": 0.06713493869588462,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009912606490588157,
      "loss": 2.2411,
      "step": 9881
    },
    {
      "epoch": 0.06714173304247868,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009912586462324407,
      "loss": 2.3874,
      "step": 9882
    },
    {
      "epoch": 0.06714852738907273,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009912566431786185,
      "loss": 2.298,
      "step": 9883
    },
    {
      "epoch": 0.0671553217356668,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009912546398973502,
      "loss": 2.3391,
      "step": 9884
    },
    {
      "epoch": 0.06716211608226086,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009912526363886364,
      "loss": 2.2201,
      "step": 9885
    },
    {
      "epoch": 0.06716891042885491,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009912506326524785,
      "loss": 2.353,
      "step": 9886
    },
    {
      "epoch": 0.06717570477544897,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009912486286888771,
      "loss": 2.2123,
      "step": 9887
    },
    {
      "epoch": 0.06718249912204302,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009912466244978333,
      "loss": 2.2742,
      "step": 9888
    },
    {
      "epoch": 0.06718929346863708,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000991244620079348,
      "loss": 2.3642,
      "step": 9889
    },
    {
      "epoch": 0.06719608781523115,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009912426154334217,
      "loss": 2.4535,
      "step": 9890
    },
    {
      "epoch": 0.0672028821618252,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000991240610560056,
      "loss": 2.3688,
      "step": 9891
    },
    {
      "epoch": 0.06720967650841926,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009912386054592515,
      "loss": 2.2013,
      "step": 9892
    },
    {
      "epoch": 0.06721647085501331,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000991236600131009,
      "loss": 2.39,
      "step": 9893
    },
    {
      "epoch": 0.06722326520160737,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009912345945753299,
      "loss": 2.2649,
      "step": 9894
    },
    {
      "epoch": 0.06723005954820142,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009912325887922145,
      "loss": 2.2737,
      "step": 9895
    },
    {
      "epoch": 0.06723685389479549,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009912305827816644,
      "loss": 2.4041,
      "step": 9896
    },
    {
      "epoch": 0.06724364824138955,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009912285765436798,
      "loss": 2.2389,
      "step": 9897
    },
    {
      "epoch": 0.0672504425879836,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009912265700782622,
      "loss": 2.188,
      "step": 9898
    },
    {
      "epoch": 0.06725723693457766,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009912245633854123,
      "loss": 2.2012,
      "step": 9899
    },
    {
      "epoch": 0.06726403128117171,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009912225564651309,
      "loss": 2.1859,
      "step": 9900
    },
    {
      "epoch": 0.06727082562776578,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009912205493174194,
      "loss": 2.1709,
      "step": 9901
    },
    {
      "epoch": 0.06727761997435984,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000991218541942278,
      "loss": 2.2191,
      "step": 9902
    },
    {
      "epoch": 0.06728441432095389,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009912165343397084,
      "loss": 2.3351,
      "step": 9903
    },
    {
      "epoch": 0.06729120866754795,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000991214526509711,
      "loss": 2.3064,
      "step": 9904
    },
    {
      "epoch": 0.067298003014142,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009912125184522872,
      "loss": 2.2918,
      "step": 9905
    },
    {
      "epoch": 0.06730479736073607,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009912105101674373,
      "loss": 2.2564,
      "step": 9906
    },
    {
      "epoch": 0.06731159170733013,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009912085016551627,
      "loss": 2.0387,
      "step": 9907
    },
    {
      "epoch": 0.06731838605392418,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000991206492915464,
      "loss": 2.2487,
      "step": 9908
    },
    {
      "epoch": 0.06732518040051824,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009912044839483426,
      "loss": 2.2737,
      "step": 9909
    },
    {
      "epoch": 0.06733197474711229,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000991202474753799,
      "loss": 2.3947,
      "step": 9910
    },
    {
      "epoch": 0.06733876909370635,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009912004653318343,
      "loss": 2.2513,
      "step": 9911
    },
    {
      "epoch": 0.06734556344030042,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009911984556824497,
      "loss": 2.3102,
      "step": 9912
    },
    {
      "epoch": 0.06735235778689447,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009911964458056454,
      "loss": 2.176,
      "step": 9913
    },
    {
      "epoch": 0.06735915213348853,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000991194435701423,
      "loss": 2.1366,
      "step": 9914
    },
    {
      "epoch": 0.06736594648008258,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009911924253697831,
      "loss": 2.1058,
      "step": 9915
    },
    {
      "epoch": 0.06737274082667664,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009911904148107269,
      "loss": 2.2145,
      "step": 9916
    },
    {
      "epoch": 0.06737953517327071,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991188404024255,
      "loss": 2.0675,
      "step": 9917
    },
    {
      "epoch": 0.06738632951986476,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009911863930103686,
      "loss": 2.2773,
      "step": 9918
    },
    {
      "epoch": 0.06739312386645882,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009911843817690686,
      "loss": 2.2124,
      "step": 9919
    },
    {
      "epoch": 0.06739991821305287,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009911823703003557,
      "loss": 2.2986,
      "step": 9920
    },
    {
      "epoch": 0.06740671255964693,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000991180358604231,
      "loss": 2.3333,
      "step": 9921
    },
    {
      "epoch": 0.06741350690624098,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009911783466806955,
      "loss": 2.2068,
      "step": 9922
    },
    {
      "epoch": 0.06742030125283505,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009911763345297501,
      "loss": 2.2267,
      "step": 9923
    },
    {
      "epoch": 0.06742709559942911,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009911743221513959,
      "loss": 2.1553,
      "step": 9924
    },
    {
      "epoch": 0.06743388994602316,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009911723095456333,
      "loss": 2.2883,
      "step": 9925
    },
    {
      "epoch": 0.06744068429261722,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009911702967124636,
      "loss": 2.2413,
      "step": 9926
    },
    {
      "epoch": 0.06744747863921127,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009911682836518875,
      "loss": 2.1871,
      "step": 9927
    },
    {
      "epoch": 0.06745427298580534,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009911662703639063,
      "loss": 2.2874,
      "step": 9928
    },
    {
      "epoch": 0.0674610673323994,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009911642568485209,
      "loss": 2.1741,
      "step": 9929
    },
    {
      "epoch": 0.06746786167899345,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009911622431057318,
      "loss": 2.2738,
      "step": 9930
    },
    {
      "epoch": 0.06747465602558751,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009911602291355404,
      "loss": 2.1328,
      "step": 9931
    },
    {
      "epoch": 0.06748145037218156,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009911582149379474,
      "loss": 2.1522,
      "step": 9932
    },
    {
      "epoch": 0.06748824471877563,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009911562005129539,
      "loss": 2.3685,
      "step": 9933
    },
    {
      "epoch": 0.06749503906536969,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009911541858605606,
      "loss": 2.2391,
      "step": 9934
    },
    {
      "epoch": 0.06750183341196374,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009911521709807685,
      "loss": 2.3338,
      "step": 9935
    },
    {
      "epoch": 0.0675086277585578,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009911501558735787,
      "loss": 2.2877,
      "step": 9936
    },
    {
      "epoch": 0.06751542210515185,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009911481405389918,
      "loss": 2.3195,
      "step": 9937
    },
    {
      "epoch": 0.06752221645174591,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991146124977009,
      "loss": 2.297,
      "step": 9938
    },
    {
      "epoch": 0.06752901079833998,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009911441091876314,
      "loss": 2.2766,
      "step": 9939
    },
    {
      "epoch": 0.06753580514493403,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009911420931708595,
      "loss": 2.111,
      "step": 9940
    },
    {
      "epoch": 0.06754259949152809,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009911400769266946,
      "loss": 2.3233,
      "step": 9941
    },
    {
      "epoch": 0.06754939383812214,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009911380604551371,
      "loss": 2.1387,
      "step": 9942
    },
    {
      "epoch": 0.0675561881847162,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009911360437561887,
      "loss": 2.0346,
      "step": 9943
    },
    {
      "epoch": 0.06756298253131025,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009911340268298498,
      "loss": 2.2838,
      "step": 9944
    },
    {
      "epoch": 0.06756977687790432,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009911320096761215,
      "loss": 2.1334,
      "step": 9945
    },
    {
      "epoch": 0.06757657122449838,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009911299922950045,
      "loss": 2.2132,
      "step": 9946
    },
    {
      "epoch": 0.06758336557109243,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009911279746865002,
      "loss": 2.3501,
      "step": 9947
    },
    {
      "epoch": 0.0675901599176865,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009911259568506092,
      "loss": 2.3433,
      "step": 9948
    },
    {
      "epoch": 0.06759695426428054,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009911239387873326,
      "loss": 2.3802,
      "step": 9949
    },
    {
      "epoch": 0.0676037486108746,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009911219204966713,
      "loss": 2.3985,
      "step": 9950
    },
    {
      "epoch": 0.06761054295746867,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009911199019786258,
      "loss": 2.3215,
      "step": 9951
    },
    {
      "epoch": 0.06761733730406272,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009911178832331976,
      "loss": 2.304,
      "step": 9952
    },
    {
      "epoch": 0.06762413165065678,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009911158642603877,
      "loss": 2.258,
      "step": 9953
    },
    {
      "epoch": 0.06763092599725083,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009911138450601966,
      "loss": 2.3289,
      "step": 9954
    },
    {
      "epoch": 0.0676377203438449,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009911118256326255,
      "loss": 2.3127,
      "step": 9955
    },
    {
      "epoch": 0.06764451469043896,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991109805977675,
      "loss": 2.1469,
      "step": 9956
    },
    {
      "epoch": 0.06765130903703301,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009911077860953466,
      "loss": 2.2242,
      "step": 9957
    },
    {
      "epoch": 0.06765810338362707,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009911057659856407,
      "loss": 2.2061,
      "step": 9958
    },
    {
      "epoch": 0.06766489773022112,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009911037456485589,
      "loss": 2.2019,
      "step": 9959
    },
    {
      "epoch": 0.06767169207681518,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009911017250841011,
      "loss": 2.5227,
      "step": 9960
    },
    {
      "epoch": 0.06767848642340925,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.000991099704292269,
      "loss": 2.2101,
      "step": 9961
    },
    {
      "epoch": 0.0676852807700033,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009910976832730636,
      "loss": 2.2114,
      "step": 9962
    },
    {
      "epoch": 0.06769207511659736,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009910956620264855,
      "loss": 2.1402,
      "step": 9963
    },
    {
      "epoch": 0.06769886946319141,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009910936405525357,
      "loss": 2.3295,
      "step": 9964
    },
    {
      "epoch": 0.06770566380978547,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009910916188512152,
      "loss": 2.2213,
      "step": 9965
    },
    {
      "epoch": 0.06771245815637952,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009910895969225248,
      "loss": 2.2307,
      "step": 9966
    },
    {
      "epoch": 0.06771925250297359,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009910875747664656,
      "loss": 2.389,
      "step": 9967
    },
    {
      "epoch": 0.06772604684956765,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009910855523830385,
      "loss": 2.3202,
      "step": 9968
    },
    {
      "epoch": 0.0677328411961617,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009910835297722445,
      "loss": 2.2339,
      "step": 9969
    },
    {
      "epoch": 0.06773963554275576,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009910815069340843,
      "loss": 2.2363,
      "step": 9970
    },
    {
      "epoch": 0.06774642988934981,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009910794838685591,
      "loss": 2.4211,
      "step": 9971
    },
    {
      "epoch": 0.06775322423594388,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009910774605756698,
      "loss": 2.2445,
      "step": 9972
    },
    {
      "epoch": 0.06776001858253794,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009910754370554171,
      "loss": 2.2186,
      "step": 9973
    },
    {
      "epoch": 0.06776681292913199,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009910734133078021,
      "loss": 2.3474,
      "step": 9974
    },
    {
      "epoch": 0.06777360727572605,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009910713893328258,
      "loss": 2.3139,
      "step": 9975
    },
    {
      "epoch": 0.0677804016223201,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009910693651304892,
      "loss": 2.3209,
      "step": 9976
    },
    {
      "epoch": 0.06778719596891417,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009910673407007929,
      "loss": 2.3652,
      "step": 9977
    },
    {
      "epoch": 0.06779399031550823,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009910653160437382,
      "loss": 2.2172,
      "step": 9978
    },
    {
      "epoch": 0.06780078466210228,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009910632911593257,
      "loss": 2.1327,
      "step": 9979
    },
    {
      "epoch": 0.06780757900869634,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000991061266047557,
      "loss": 2.3967,
      "step": 9980
    },
    {
      "epoch": 0.06781437335529039,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009910592407084322,
      "loss": 2.2009,
      "step": 9981
    },
    {
      "epoch": 0.06782116770188445,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009910572151419525,
      "loss": 2.293,
      "step": 9982
    },
    {
      "epoch": 0.06782796204847852,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000991055189348119,
      "loss": 2.3998,
      "step": 9983
    },
    {
      "epoch": 0.06783475639507257,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009910531633269327,
      "loss": 2.3161,
      "step": 9984
    },
    {
      "epoch": 0.06784155074166663,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009910511370783944,
      "loss": 2.3464,
      "step": 9985
    },
    {
      "epoch": 0.06784834508826068,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000991049110602505,
      "loss": 2.3466,
      "step": 9986
    },
    {
      "epoch": 0.06785513943485474,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009910470838992655,
      "loss": 2.2124,
      "step": 9987
    },
    {
      "epoch": 0.06786193378144881,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009910450569686768,
      "loss": 2.2145,
      "step": 9988
    },
    {
      "epoch": 0.06786872812804286,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00099104302981074,
      "loss": 2.2693,
      "step": 9989
    },
    {
      "epoch": 0.06787552247463692,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000991041002425456,
      "loss": 2.202,
      "step": 9990
    },
    {
      "epoch": 0.06788231682123097,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009910389748128252,
      "loss": 2.2517,
      "step": 9991
    },
    {
      "epoch": 0.06788911116782503,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009910369469728494,
      "loss": 2.2245,
      "step": 9992
    },
    {
      "epoch": 0.06789590551441908,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000991034918905529,
      "loss": 2.2108,
      "step": 9993
    },
    {
      "epoch": 0.06790269986101315,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000991032890610865,
      "loss": 2.2979,
      "step": 9994
    },
    {
      "epoch": 0.06790949420760721,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009910308620888584,
      "loss": 2.2337,
      "step": 9995
    },
    {
      "epoch": 0.06791628855420126,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009910288333395102,
      "loss": 2.1835,
      "step": 9996
    },
    {
      "epoch": 0.06792308290079532,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009910268043628212,
      "loss": 2.3186,
      "step": 9997
    },
    {
      "epoch": 0.06792987724738937,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009910247751587925,
      "loss": 2.3191,
      "step": 9998
    },
    {
      "epoch": 0.06793667159398344,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000991022745727425,
      "loss": 2.2475,
      "step": 9999
    },
    {
      "epoch": 0.0679434659405775,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009910207160687195,
      "loss": 2.2427,
      "step": 10000
    },
    {
      "epoch": 0.06795026028717155,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000991018686182677,
      "loss": 2.3303,
      "step": 10001
    },
    {
      "epoch": 0.06795705463376561,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009910166560692987,
      "loss": 2.1967,
      "step": 10002
    },
    {
      "epoch": 0.06796384898035966,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000991014625728585,
      "loss": 2.2062,
      "step": 10003
    },
    {
      "epoch": 0.06797064332695373,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009910125951605373,
      "loss": 2.213,
      "step": 10004
    },
    {
      "epoch": 0.06797743767354779,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009910105643651565,
      "loss": 2.2148,
      "step": 10005
    },
    {
      "epoch": 0.06798423202014184,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009910085333424432,
      "loss": 2.2664,
      "step": 10006
    },
    {
      "epoch": 0.0679910263667359,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000991006502092399,
      "loss": 2.3285,
      "step": 10007
    },
    {
      "epoch": 0.06799782071332995,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000991004470615024,
      "loss": 2.3691,
      "step": 10008
    },
    {
      "epoch": 0.06800461505992401,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009910024389103197,
      "loss": 2.2183,
      "step": 10009
    },
    {
      "epoch": 0.06801140940651808,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000991000406978287,
      "loss": 2.1949,
      "step": 10010
    },
    {
      "epoch": 0.06801820375311213,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009909983748189264,
      "loss": 2.2672,
      "step": 10011
    },
    {
      "epoch": 0.06802499809970619,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009909963424322395,
      "loss": 2.1753,
      "step": 10012
    },
    {
      "epoch": 0.06803179244630024,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009909943098182269,
      "loss": 2.2527,
      "step": 10013
    },
    {
      "epoch": 0.0680385867928943,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009909922769768896,
      "loss": 2.2556,
      "step": 10014
    },
    {
      "epoch": 0.06804538113948835,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009909902439082284,
      "loss": 2.4312,
      "step": 10015
    },
    {
      "epoch": 0.06805217548608242,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009909882106122444,
      "loss": 2.2713,
      "step": 10016
    },
    {
      "epoch": 0.06805896983267648,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009909861770889383,
      "loss": 2.3291,
      "step": 10017
    },
    {
      "epoch": 0.06806576417927053,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009909841433383114,
      "loss": 2.1991,
      "step": 10018
    },
    {
      "epoch": 0.0680725585258646,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009909821093603646,
      "loss": 2.1436,
      "step": 10019
    },
    {
      "epoch": 0.06807935287245864,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009909800751550985,
      "loss": 2.2804,
      "step": 10020
    },
    {
      "epoch": 0.0680861472190527,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009909780407225143,
      "loss": 2.3201,
      "step": 10021
    },
    {
      "epoch": 0.06809294156564677,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000990976006062613,
      "loss": 2.2916,
      "step": 10022
    },
    {
      "epoch": 0.06809973591224082,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009909739711753953,
      "loss": 2.3544,
      "step": 10023
    },
    {
      "epoch": 0.06810653025883488,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009909719360608623,
      "loss": 2.2851,
      "step": 10024
    },
    {
      "epoch": 0.06811332460542893,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000990969900719015,
      "loss": 2.1585,
      "step": 10025
    },
    {
      "epoch": 0.068120118952023,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009909678651498543,
      "loss": 2.2416,
      "step": 10026
    },
    {
      "epoch": 0.06812691329861706,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990965829353381,
      "loss": 2.2859,
      "step": 10027
    },
    {
      "epoch": 0.06813370764521111,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009909637933295961,
      "loss": 2.3211,
      "step": 10028
    },
    {
      "epoch": 0.06814050199180517,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009909617570785006,
      "loss": 2.354,
      "step": 10029
    },
    {
      "epoch": 0.06814729633839922,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009909597206000955,
      "loss": 2.2258,
      "step": 10030
    },
    {
      "epoch": 0.06815409068499328,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009909576838943817,
      "loss": 2.1899,
      "step": 10031
    },
    {
      "epoch": 0.06816088503158735,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099095564696136,
      "loss": 2.2691,
      "step": 10032
    },
    {
      "epoch": 0.0681676793781814,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009909536098010316,
      "loss": 2.3207,
      "step": 10033
    },
    {
      "epoch": 0.06817447372477546,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009909515724133974,
      "loss": 2.2911,
      "step": 10034
    },
    {
      "epoch": 0.06818126807136951,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000990949534798458,
      "loss": 2.3483,
      "step": 10035
    },
    {
      "epoch": 0.06818806241796357,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000990947496956215,
      "loss": 2.2291,
      "step": 10036
    },
    {
      "epoch": 0.06819485676455764,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009909454588866685,
      "loss": 2.1903,
      "step": 10037
    },
    {
      "epoch": 0.06820165111115169,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00099094342058982,
      "loss": 2.1219,
      "step": 10038
    },
    {
      "epoch": 0.06820844545774575,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009909413820656705,
      "loss": 2.1383,
      "step": 10039
    },
    {
      "epoch": 0.0682152398043398,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009909393433142205,
      "loss": 2.3255,
      "step": 10040
    },
    {
      "epoch": 0.06822203415093386,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009909373043354714,
      "loss": 2.187,
      "step": 10041
    },
    {
      "epoch": 0.06822882849752791,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000990935265129424,
      "loss": 2.1789,
      "step": 10042
    },
    {
      "epoch": 0.06823562284412198,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000990933225696079,
      "loss": 2.2731,
      "step": 10043
    },
    {
      "epoch": 0.06824241719071604,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009909311860354377,
      "loss": 2.2235,
      "step": 10044
    },
    {
      "epoch": 0.06824921153731009,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009909291461475008,
      "loss": 2.2552,
      "step": 10045
    },
    {
      "epoch": 0.06825600588390415,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009909271060322693,
      "loss": 2.2477,
      "step": 10046
    },
    {
      "epoch": 0.0682628002304982,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009909250656897444,
      "loss": 2.2107,
      "step": 10047
    },
    {
      "epoch": 0.06826959457709227,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009909230251199266,
      "loss": 2.2263,
      "step": 10048
    },
    {
      "epoch": 0.06827638892368633,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000990920984322817,
      "loss": 2.1868,
      "step": 10049
    },
    {
      "epoch": 0.06828318327028038,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000990918943298417,
      "loss": 2.2341,
      "step": 10050
    },
    {
      "epoch": 0.06828997761687444,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009909169020467267,
      "loss": 2.1999,
      "step": 10051
    },
    {
      "epoch": 0.06829677196346849,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009909148605677478,
      "loss": 2.3071,
      "step": 10052
    },
    {
      "epoch": 0.06830356631006256,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009909128188614808,
      "loss": 2.2148,
      "step": 10053
    },
    {
      "epoch": 0.06831036065665662,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990910776927927,
      "loss": 2.3047,
      "step": 10054
    },
    {
      "epoch": 0.06831715500325067,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009909087347670868,
      "loss": 2.2433,
      "step": 10055
    },
    {
      "epoch": 0.06832394934984473,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009909066923789618,
      "loss": 2.2189,
      "step": 10056
    },
    {
      "epoch": 0.06833074369643878,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009909046497635523,
      "loss": 2.3009,
      "step": 10057
    },
    {
      "epoch": 0.06833753804303284,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00099090260692086,
      "loss": 2.2028,
      "step": 10058
    },
    {
      "epoch": 0.06834433238962691,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000990900563850885,
      "loss": 2.282,
      "step": 10059
    },
    {
      "epoch": 0.06835112673622096,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009908985205536289,
      "loss": 2.2136,
      "step": 10060
    },
    {
      "epoch": 0.06835792108281502,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009908964770290925,
      "loss": 2.2522,
      "step": 10061
    },
    {
      "epoch": 0.06836471542940907,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009908944332772764,
      "loss": 2.1633,
      "step": 10062
    },
    {
      "epoch": 0.06837150977600313,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000990892389298182,
      "loss": 2.3959,
      "step": 10063
    },
    {
      "epoch": 0.06837830412259718,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00099089034509181,
      "loss": 2.2356,
      "step": 10064
    },
    {
      "epoch": 0.06838509846919125,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009908883006581614,
      "loss": 2.3602,
      "step": 10065
    },
    {
      "epoch": 0.06839189281578531,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009908862559972371,
      "loss": 2.2761,
      "step": 10066
    },
    {
      "epoch": 0.06839868716237936,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000990884211109038,
      "loss": 2.0482,
      "step": 10067
    },
    {
      "epoch": 0.06840548150897342,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009908821659935655,
      "loss": 2.2444,
      "step": 10068
    },
    {
      "epoch": 0.06841227585556747,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009908801206508198,
      "loss": 2.3365,
      "step": 10069
    },
    {
      "epoch": 0.06841907020216154,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009908780750808026,
      "loss": 2.1948,
      "step": 10070
    },
    {
      "epoch": 0.0684258645487556,
      "grad_norm": 0.15625,
      "learning_rate": 0.000990876029283514,
      "loss": 2.2923,
      "step": 10071
    },
    {
      "epoch": 0.06843265889534965,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009908739832589558,
      "loss": 2.1666,
      "step": 10072
    },
    {
      "epoch": 0.06843945324194371,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009908719370071285,
      "loss": 2.1661,
      "step": 10073
    },
    {
      "epoch": 0.06844624758853776,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009908698905280332,
      "loss": 2.2465,
      "step": 10074
    },
    {
      "epoch": 0.06845304193513183,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009908678438216704,
      "loss": 2.275,
      "step": 10075
    },
    {
      "epoch": 0.06845983628172589,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009908657968880417,
      "loss": 2.2787,
      "step": 10076
    },
    {
      "epoch": 0.06846663062831994,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009908637497271478,
      "loss": 2.1052,
      "step": 10077
    },
    {
      "epoch": 0.068473424974914,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009908617023389897,
      "loss": 2.3061,
      "step": 10078
    },
    {
      "epoch": 0.06848021932150805,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000990859654723568,
      "loss": 2.3065,
      "step": 10079
    },
    {
      "epoch": 0.06848701366810211,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000990857606880884,
      "loss": 2.2347,
      "step": 10080
    },
    {
      "epoch": 0.06849380801469618,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009908555588109384,
      "loss": 2.182,
      "step": 10081
    },
    {
      "epoch": 0.06850060236129023,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009908535105137327,
      "loss": 2.4238,
      "step": 10082
    },
    {
      "epoch": 0.06850739670788429,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009908514619892672,
      "loss": 2.146,
      "step": 10083
    },
    {
      "epoch": 0.06851419105447834,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009908494132375432,
      "loss": 2.2367,
      "step": 10084
    },
    {
      "epoch": 0.0685209854010724,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009908473642585615,
      "loss": 2.2348,
      "step": 10085
    },
    {
      "epoch": 0.06852777974766645,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000990845315052323,
      "loss": 2.1913,
      "step": 10086
    },
    {
      "epoch": 0.06853457409426052,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009908432656188287,
      "loss": 2.2537,
      "step": 10087
    },
    {
      "epoch": 0.06854136844085458,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009908412159580797,
      "loss": 2.2424,
      "step": 10088
    },
    {
      "epoch": 0.06854816278744863,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009908391660700768,
      "loss": 2.1083,
      "step": 10089
    },
    {
      "epoch": 0.0685549571340427,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000990837115954821,
      "loss": 2.2075,
      "step": 10090
    },
    {
      "epoch": 0.06856175148063674,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009908350656123133,
      "loss": 2.171,
      "step": 10091
    },
    {
      "epoch": 0.0685685458272308,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009908330150425547,
      "loss": 2.2155,
      "step": 10092
    },
    {
      "epoch": 0.06857534017382487,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009908309642455459,
      "loss": 2.1844,
      "step": 10093
    },
    {
      "epoch": 0.06858213452041892,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009908289132212879,
      "loss": 2.3307,
      "step": 10094
    },
    {
      "epoch": 0.06858892886701298,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000990826861969782,
      "loss": 2.2598,
      "step": 10095
    },
    {
      "epoch": 0.06859572321360703,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009908248104910286,
      "loss": 2.2411,
      "step": 10096
    },
    {
      "epoch": 0.0686025175602011,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000990822758785029,
      "loss": 2.2654,
      "step": 10097
    },
    {
      "epoch": 0.06860931190679516,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009908207068517845,
      "loss": 2.1981,
      "step": 10098
    },
    {
      "epoch": 0.06861610625338921,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009908186546912952,
      "loss": 2.3885,
      "step": 10099
    },
    {
      "epoch": 0.06862290059998327,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009908166023035625,
      "loss": 2.1399,
      "step": 10100
    },
    {
      "epoch": 0.06862969494657732,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009908145496885876,
      "loss": 2.0871,
      "step": 10101
    },
    {
      "epoch": 0.06863648929317138,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009908124968463709,
      "loss": 2.217,
      "step": 10102
    },
    {
      "epoch": 0.06864328363976545,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009908104437769138,
      "loss": 2.1777,
      "step": 10103
    },
    {
      "epoch": 0.0686500779863595,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000990808390480217,
      "loss": 2.3366,
      "step": 10104
    },
    {
      "epoch": 0.06865687233295356,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009908063369562816,
      "loss": 2.3955,
      "step": 10105
    },
    {
      "epoch": 0.06866366667954761,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009908042832051085,
      "loss": 2.1261,
      "step": 10106
    },
    {
      "epoch": 0.06867046102614167,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009908022292266986,
      "loss": 2.2032,
      "step": 10107
    },
    {
      "epoch": 0.06867725537273574,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009908001750210532,
      "loss": 2.2936,
      "step": 10108
    },
    {
      "epoch": 0.06868404971932979,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009907981205881725,
      "loss": 2.1717,
      "step": 10109
    },
    {
      "epoch": 0.06869084406592385,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000990796065928058,
      "loss": 2.4428,
      "step": 10110
    },
    {
      "epoch": 0.0686976384125179,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009907940110407108,
      "loss": 2.1173,
      "step": 10111
    },
    {
      "epoch": 0.06870443275911196,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009907919559261314,
      "loss": 2.2471,
      "step": 10112
    },
    {
      "epoch": 0.06871122710570601,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000990789900584321,
      "loss": 2.2119,
      "step": 10113
    },
    {
      "epoch": 0.06871802145230008,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009907878450152805,
      "loss": 2.1567,
      "step": 10114
    },
    {
      "epoch": 0.06872481579889414,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009907857892190108,
      "loss": 2.2986,
      "step": 10115
    },
    {
      "epoch": 0.06873161014548819,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000990783733195513,
      "loss": 2.3841,
      "step": 10116
    },
    {
      "epoch": 0.06873840449208225,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009907816769447878,
      "loss": 2.2531,
      "step": 10117
    },
    {
      "epoch": 0.0687451988386763,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009907796204668366,
      "loss": 2.2347,
      "step": 10118
    },
    {
      "epoch": 0.06875199318527037,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009907775637616597,
      "loss": 2.2406,
      "step": 10119
    },
    {
      "epoch": 0.06875878753186443,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009907755068292586,
      "loss": 2.2321,
      "step": 10120
    },
    {
      "epoch": 0.06876558187845848,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009907734496696341,
      "loss": 2.2116,
      "step": 10121
    },
    {
      "epoch": 0.06877237622505254,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000990771392282787,
      "loss": 2.3669,
      "step": 10122
    },
    {
      "epoch": 0.06877917057164659,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009907693346687184,
      "loss": 2.3505,
      "step": 10123
    },
    {
      "epoch": 0.06878596491824066,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009907672768274295,
      "loss": 2.3148,
      "step": 10124
    },
    {
      "epoch": 0.06879275926483472,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009907652187589205,
      "loss": 2.3185,
      "step": 10125
    },
    {
      "epoch": 0.06879955361142877,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000990763160463193,
      "loss": 2.3168,
      "step": 10126
    },
    {
      "epoch": 0.06880634795802283,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000990761101940248,
      "loss": 2.3535,
      "step": 10127
    },
    {
      "epoch": 0.06881314230461688,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009907590431900862,
      "loss": 2.4206,
      "step": 10128
    },
    {
      "epoch": 0.06881993665121094,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009907569842127085,
      "loss": 2.3418,
      "step": 10129
    },
    {
      "epoch": 0.06882673099780501,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009907549250081158,
      "loss": 2.2164,
      "step": 10130
    },
    {
      "epoch": 0.06883352534439906,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009907528655763093,
      "loss": 2.2752,
      "step": 10131
    },
    {
      "epoch": 0.06884031969099312,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00099075080591729,
      "loss": 2.2306,
      "step": 10132
    },
    {
      "epoch": 0.06884711403758717,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009907487460310584,
      "loss": 2.2283,
      "step": 10133
    },
    {
      "epoch": 0.06885390838418123,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000990746685917616,
      "loss": 2.1671,
      "step": 10134
    },
    {
      "epoch": 0.06886070273077528,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009907446255769635,
      "loss": 2.2634,
      "step": 10135
    },
    {
      "epoch": 0.06886749707736935,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009907425650091018,
      "loss": 2.2176,
      "step": 10136
    },
    {
      "epoch": 0.06887429142396341,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009907405042140318,
      "loss": 2.2971,
      "step": 10137
    },
    {
      "epoch": 0.06888108577055746,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009907384431917547,
      "loss": 2.047,
      "step": 10138
    },
    {
      "epoch": 0.06888788011715152,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009907363819422715,
      "loss": 2.3357,
      "step": 10139
    },
    {
      "epoch": 0.06889467446374557,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009907343204655827,
      "loss": 2.2982,
      "step": 10140
    },
    {
      "epoch": 0.06890146881033964,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009907322587616896,
      "loss": 2.1876,
      "step": 10141
    },
    {
      "epoch": 0.0689082631569337,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009907301968305933,
      "loss": 2.1976,
      "step": 10142
    },
    {
      "epoch": 0.06891505750352775,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009907281346722943,
      "loss": 2.29,
      "step": 10143
    },
    {
      "epoch": 0.06892185185012181,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009907260722867939,
      "loss": 2.1945,
      "step": 10144
    },
    {
      "epoch": 0.06892864619671586,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000990724009674093,
      "loss": 2.2688,
      "step": 10145
    },
    {
      "epoch": 0.06893544054330993,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009907219468341924,
      "loss": 2.1975,
      "step": 10146
    },
    {
      "epoch": 0.06894223488990399,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009907198837670933,
      "loss": 2.2699,
      "step": 10147
    },
    {
      "epoch": 0.06894902923649804,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009907178204727962,
      "loss": 2.1558,
      "step": 10148
    },
    {
      "epoch": 0.0689558235830921,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009907157569513027,
      "loss": 2.1896,
      "step": 10149
    },
    {
      "epoch": 0.06896261792968615,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009907136932026132,
      "loss": 2.0034,
      "step": 10150
    },
    {
      "epoch": 0.06896941227628021,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009907116292267291,
      "loss": 2.2247,
      "step": 10151
    },
    {
      "epoch": 0.06897620662287428,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009907095650236511,
      "loss": 2.1215,
      "step": 10152
    },
    {
      "epoch": 0.06898300096946833,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00099070750059338,
      "loss": 2.2717,
      "step": 10153
    },
    {
      "epoch": 0.06898979531606239,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009907054359359173,
      "loss": 2.1318,
      "step": 10154
    },
    {
      "epoch": 0.06899658966265644,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009907033710512634,
      "loss": 2.2063,
      "step": 10155
    },
    {
      "epoch": 0.0690033840092505,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009907013059394195,
      "loss": 2.169,
      "step": 10156
    },
    {
      "epoch": 0.06901017835584457,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009906992406003865,
      "loss": 2.2089,
      "step": 10157
    },
    {
      "epoch": 0.06901697270243862,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009906971750341653,
      "loss": 2.3456,
      "step": 10158
    },
    {
      "epoch": 0.06902376704903268,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000990695109240757,
      "loss": 2.1434,
      "step": 10159
    },
    {
      "epoch": 0.06903056139562673,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009906930432201626,
      "loss": 2.1598,
      "step": 10160
    },
    {
      "epoch": 0.0690373557422208,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009906909769723828,
      "loss": 2.4645,
      "step": 10161
    },
    {
      "epoch": 0.06904415008881484,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009906889104974188,
      "loss": 2.3305,
      "step": 10162
    },
    {
      "epoch": 0.0690509444354089,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009906868437952715,
      "loss": 2.3694,
      "step": 10163
    },
    {
      "epoch": 0.06905773878200297,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009906847768659415,
      "loss": 2.45,
      "step": 10164
    },
    {
      "epoch": 0.06906453312859702,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009906827097094304,
      "loss": 2.1446,
      "step": 10165
    },
    {
      "epoch": 0.06907132747519108,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009906806423257387,
      "loss": 2.3773,
      "step": 10166
    },
    {
      "epoch": 0.06907812182178513,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009906785747148675,
      "loss": 2.2853,
      "step": 10167
    },
    {
      "epoch": 0.0690849161683792,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009906765068768177,
      "loss": 2.3434,
      "step": 10168
    },
    {
      "epoch": 0.06909171051497326,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009906744388115906,
      "loss": 2.4204,
      "step": 10169
    },
    {
      "epoch": 0.06909850486156731,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009906723705191865,
      "loss": 2.2511,
      "step": 10170
    },
    {
      "epoch": 0.06910529920816137,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009906703019996069,
      "loss": 2.1773,
      "step": 10171
    },
    {
      "epoch": 0.06911209355475542,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009906682332528525,
      "loss": 2.3444,
      "step": 10172
    },
    {
      "epoch": 0.06911888790134949,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009906661642789243,
      "loss": 2.2314,
      "step": 10173
    },
    {
      "epoch": 0.06912568224794355,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009906640950778235,
      "loss": 2.3349,
      "step": 10174
    },
    {
      "epoch": 0.0691324765945376,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009906620256495506,
      "loss": 2.2867,
      "step": 10175
    },
    {
      "epoch": 0.06913927094113166,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009906599559941069,
      "loss": 2.3609,
      "step": 10176
    },
    {
      "epoch": 0.06914606528772571,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009906578861114932,
      "loss": 2.2474,
      "step": 10177
    },
    {
      "epoch": 0.06915285963431977,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009906558160017107,
      "loss": 2.4051,
      "step": 10178
    },
    {
      "epoch": 0.06915965398091384,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009906537456647602,
      "loss": 2.3102,
      "step": 10179
    },
    {
      "epoch": 0.06916644832750789,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009906516751006424,
      "loss": 2.3025,
      "step": 10180
    },
    {
      "epoch": 0.06917324267410195,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009906496043093585,
      "loss": 2.278,
      "step": 10181
    },
    {
      "epoch": 0.069180037020696,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009906475332909098,
      "loss": 2.2947,
      "step": 10182
    },
    {
      "epoch": 0.06918683136729006,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009906454620452964,
      "loss": 2.1859,
      "step": 10183
    },
    {
      "epoch": 0.06919362571388411,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009906433905725203,
      "loss": 2.2266,
      "step": 10184
    },
    {
      "epoch": 0.06920042006047818,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009906413188725816,
      "loss": 2.3025,
      "step": 10185
    },
    {
      "epoch": 0.06920721440707224,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009906392469454816,
      "loss": 2.296,
      "step": 10186
    },
    {
      "epoch": 0.06921400875366629,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009906371747912216,
      "loss": 2.2262,
      "step": 10187
    },
    {
      "epoch": 0.06922080310026035,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000990635102409802,
      "loss": 2.039,
      "step": 10188
    },
    {
      "epoch": 0.0692275974468544,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009906330298012239,
      "loss": 2.3081,
      "step": 10189
    },
    {
      "epoch": 0.06923439179344847,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009906309569654883,
      "loss": 2.4128,
      "step": 10190
    },
    {
      "epoch": 0.06924118614004253,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009906288839025962,
      "loss": 2.189,
      "step": 10191
    },
    {
      "epoch": 0.06924798048663658,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009906268106125487,
      "loss": 2.1842,
      "step": 10192
    },
    {
      "epoch": 0.06925477483323064,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009906247370953464,
      "loss": 2.2686,
      "step": 10193
    },
    {
      "epoch": 0.06926156917982469,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009906226633509907,
      "loss": 2.3875,
      "step": 10194
    },
    {
      "epoch": 0.06926836352641876,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009906205893794823,
      "loss": 2.1532,
      "step": 10195
    },
    {
      "epoch": 0.06927515787301282,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000990618515180822,
      "loss": 2.1541,
      "step": 10196
    },
    {
      "epoch": 0.06928195221960687,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009906164407550112,
      "loss": 2.1053,
      "step": 10197
    },
    {
      "epoch": 0.06928874656620093,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009906143661020503,
      "loss": 2.1712,
      "step": 10198
    },
    {
      "epoch": 0.06929554091279498,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0009906122912219407,
      "loss": 1.9762,
      "step": 10199
    },
    {
      "epoch": 0.06930233525938904,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009906102161146833,
      "loss": 2.2817,
      "step": 10200
    },
    {
      "epoch": 0.06930912960598311,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009906081407802792,
      "loss": 2.2129,
      "step": 10201
    },
    {
      "epoch": 0.06931592395257716,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009906060652187289,
      "loss": 2.3494,
      "step": 10202
    },
    {
      "epoch": 0.06932271829917122,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009906039894300336,
      "loss": 2.2706,
      "step": 10203
    },
    {
      "epoch": 0.06932951264576527,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009906019134141943,
      "loss": 2.1969,
      "step": 10204
    },
    {
      "epoch": 0.06933630699235933,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000990599837171212,
      "loss": 2.2104,
      "step": 10205
    },
    {
      "epoch": 0.06934310133895338,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009905977607010876,
      "loss": 2.2726,
      "step": 10206
    },
    {
      "epoch": 0.06934989568554745,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000990595684003822,
      "loss": 2.1823,
      "step": 10207
    },
    {
      "epoch": 0.06935669003214151,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009905936070794164,
      "loss": 2.1481,
      "step": 10208
    },
    {
      "epoch": 0.06936348437873556,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009905915299278714,
      "loss": 2.2227,
      "step": 10209
    },
    {
      "epoch": 0.06937027872532962,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000990589452549188,
      "loss": 2.2664,
      "step": 10210
    },
    {
      "epoch": 0.06937707307192367,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009905873749433676,
      "loss": 2.3293,
      "step": 10211
    },
    {
      "epoch": 0.06938386741851774,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009905852971104107,
      "loss": 2.2747,
      "step": 10212
    },
    {
      "epoch": 0.0693906617651118,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009905832190503186,
      "loss": 2.2246,
      "step": 10213
    },
    {
      "epoch": 0.06939745611170585,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009905811407630919,
      "loss": 2.291,
      "step": 10214
    },
    {
      "epoch": 0.06940425045829991,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009905790622487319,
      "loss": 2.202,
      "step": 10215
    },
    {
      "epoch": 0.06941104480489396,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009905769835072394,
      "loss": 2.1378,
      "step": 10216
    },
    {
      "epoch": 0.06941783915148803,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009905749045386154,
      "loss": 2.2253,
      "step": 10217
    },
    {
      "epoch": 0.06942463349808209,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009905728253428608,
      "loss": 2.3042,
      "step": 10218
    },
    {
      "epoch": 0.06943142784467614,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009905707459199765,
      "loss": 2.2683,
      "step": 10219
    },
    {
      "epoch": 0.0694382221912702,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009905686662699637,
      "loss": 2.2685,
      "step": 10220
    },
    {
      "epoch": 0.06944501653786425,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009905665863928232,
      "loss": 2.2233,
      "step": 10221
    },
    {
      "epoch": 0.06945181088445831,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990564506288556,
      "loss": 2.1722,
      "step": 10222
    },
    {
      "epoch": 0.06945860523105238,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009905624259571632,
      "loss": 2.1661,
      "step": 10223
    },
    {
      "epoch": 0.06946539957764643,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009905603453986452,
      "loss": 2.3135,
      "step": 10224
    },
    {
      "epoch": 0.06947219392424049,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009905582646130037,
      "loss": 2.2678,
      "step": 10225
    },
    {
      "epoch": 0.06947898827083454,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009905561836002393,
      "loss": 2.1817,
      "step": 10226
    },
    {
      "epoch": 0.0694857826174286,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000990554102360353,
      "loss": 2.1014,
      "step": 10227
    },
    {
      "epoch": 0.06949257696402267,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009905520208933459,
      "loss": 2.3281,
      "step": 10228
    },
    {
      "epoch": 0.06949937131061672,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009905499391992187,
      "loss": 2.3394,
      "step": 10229
    },
    {
      "epoch": 0.06950616565721078,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009905478572779724,
      "loss": 2.2276,
      "step": 10230
    },
    {
      "epoch": 0.06951296000380483,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009905457751296083,
      "loss": 2.3509,
      "step": 10231
    },
    {
      "epoch": 0.0695197543503989,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009905436927541268,
      "loss": 2.3253,
      "step": 10232
    },
    {
      "epoch": 0.06952654869699294,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009905416101515297,
      "loss": 2.2704,
      "step": 10233
    },
    {
      "epoch": 0.069533343043587,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000990539527321817,
      "loss": 2.278,
      "step": 10234
    },
    {
      "epoch": 0.06954013739018107,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009905374442649903,
      "loss": 2.1536,
      "step": 10235
    },
    {
      "epoch": 0.06954693173677512,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009905353609810505,
      "loss": 2.2779,
      "step": 10236
    },
    {
      "epoch": 0.06955372608336918,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009905332774699982,
      "loss": 2.0895,
      "step": 10237
    },
    {
      "epoch": 0.06956052042996323,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009905311937318348,
      "loss": 2.1968,
      "step": 10238
    },
    {
      "epoch": 0.0695673147765573,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000990529109766561,
      "loss": 2.3067,
      "step": 10239
    },
    {
      "epoch": 0.06957410912315136,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000990527025574178,
      "loss": 2.4099,
      "step": 10240
    },
    {
      "epoch": 0.06958090346974541,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009905249411546863,
      "loss": 2.1942,
      "step": 10241
    },
    {
      "epoch": 0.06958769781633947,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009905228565080875,
      "loss": 2.1864,
      "step": 10242
    },
    {
      "epoch": 0.06959449216293352,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000990520771634382,
      "loss": 2.2527,
      "step": 10243
    },
    {
      "epoch": 0.06960128650952759,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009905186865335712,
      "loss": 2.2946,
      "step": 10244
    },
    {
      "epoch": 0.06960808085612165,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009905166012056559,
      "loss": 2.2181,
      "step": 10245
    },
    {
      "epoch": 0.0696148752027157,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009905145156506368,
      "loss": 2.155,
      "step": 10246
    },
    {
      "epoch": 0.06962166954930976,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009905124298685154,
      "loss": 2.5458,
      "step": 10247
    },
    {
      "epoch": 0.06962846389590381,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009905103438592922,
      "loss": 2.1818,
      "step": 10248
    },
    {
      "epoch": 0.06963525824249787,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009905082576229683,
      "loss": 2.1929,
      "step": 10249
    },
    {
      "epoch": 0.06964205258909194,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009905061711595446,
      "loss": 2.2134,
      "step": 10250
    },
    {
      "epoch": 0.06964884693568599,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009905040844690224,
      "loss": 2.3853,
      "step": 10251
    },
    {
      "epoch": 0.06965564128228005,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009905019975514023,
      "loss": 2.2934,
      "step": 10252
    },
    {
      "epoch": 0.0696624356288741,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009904999104066853,
      "loss": 2.3425,
      "step": 10253
    },
    {
      "epoch": 0.06966922997546816,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009904978230348726,
      "loss": 2.2468,
      "step": 10254
    },
    {
      "epoch": 0.06967602432206221,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000990495735435965,
      "loss": 2.1833,
      "step": 10255
    },
    {
      "epoch": 0.06968281866865628,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009904936476099635,
      "loss": 2.2717,
      "step": 10256
    },
    {
      "epoch": 0.06968961301525034,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009904915595568692,
      "loss": 2.2704,
      "step": 10257
    },
    {
      "epoch": 0.06969640736184439,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009904894712766828,
      "loss": 2.2572,
      "step": 10258
    },
    {
      "epoch": 0.06970320170843845,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009904873827694053,
      "loss": 2.1439,
      "step": 10259
    },
    {
      "epoch": 0.0697099960550325,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990485294035038,
      "loss": 2.1898,
      "step": 10260
    },
    {
      "epoch": 0.06971679040162657,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009904832050735815,
      "loss": 2.2499,
      "step": 10261
    },
    {
      "epoch": 0.06972358474822063,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000990481115885037,
      "loss": 2.3103,
      "step": 10262
    },
    {
      "epoch": 0.06973037909481468,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000990479026469405,
      "loss": 2.2355,
      "step": 10263
    },
    {
      "epoch": 0.06973717344140874,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009904769368266873,
      "loss": 2.1222,
      "step": 10264
    },
    {
      "epoch": 0.06974396778800279,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009904748469568843,
      "loss": 2.2657,
      "step": 10265
    },
    {
      "epoch": 0.06975076213459686,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009904727568599968,
      "loss": 2.3233,
      "step": 10266
    },
    {
      "epoch": 0.06975755648119092,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009904706665360263,
      "loss": 2.2601,
      "step": 10267
    },
    {
      "epoch": 0.06976435082778497,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009904685759849735,
      "loss": 2.2916,
      "step": 10268
    },
    {
      "epoch": 0.06977114517437903,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009904664852068393,
      "loss": 2.1468,
      "step": 10269
    },
    {
      "epoch": 0.06977793952097308,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009904643942016246,
      "loss": 2.2997,
      "step": 10270
    },
    {
      "epoch": 0.06978473386756714,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009904623029693307,
      "loss": 2.1799,
      "step": 10271
    },
    {
      "epoch": 0.06979152821416121,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009904602115099583,
      "loss": 2.299,
      "step": 10272
    },
    {
      "epoch": 0.06979832256075526,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009904581198235084,
      "loss": 2.2984,
      "step": 10273
    },
    {
      "epoch": 0.06980511690734932,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000990456027909982,
      "loss": 2.154,
      "step": 10274
    },
    {
      "epoch": 0.06981191125394337,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009904539357693802,
      "loss": 2.3224,
      "step": 10275
    },
    {
      "epoch": 0.06981870560053743,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009904518434017038,
      "loss": 2.2471,
      "step": 10276
    },
    {
      "epoch": 0.0698254999471315,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000990449750806954,
      "loss": 2.2779,
      "step": 10277
    },
    {
      "epoch": 0.06983229429372555,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009904476579851314,
      "loss": 2.1822,
      "step": 10278
    },
    {
      "epoch": 0.06983908864031961,
      "grad_norm": 0.140625,
      "learning_rate": 0.000990445564936237,
      "loss": 2.3154,
      "step": 10279
    },
    {
      "epoch": 0.06984588298691366,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009904434716602722,
      "loss": 2.2145,
      "step": 10280
    },
    {
      "epoch": 0.06985267733350772,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009904413781572376,
      "loss": 2.2214,
      "step": 10281
    },
    {
      "epoch": 0.06985947168010177,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009904392844271343,
      "loss": 2.1454,
      "step": 10282
    },
    {
      "epoch": 0.06986626602669584,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009904371904699632,
      "loss": 2.281,
      "step": 10283
    },
    {
      "epoch": 0.0698730603732899,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009904350962857253,
      "loss": 2.296,
      "step": 10284
    },
    {
      "epoch": 0.06987985471988395,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009904330018744214,
      "loss": 2.1254,
      "step": 10285
    },
    {
      "epoch": 0.06988664906647801,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009904309072360528,
      "loss": 2.1936,
      "step": 10286
    },
    {
      "epoch": 0.06989344341307206,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009904288123706204,
      "loss": 2.1531,
      "step": 10287
    },
    {
      "epoch": 0.06990023775966613,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009904267172781251,
      "loss": 2.3951,
      "step": 10288
    },
    {
      "epoch": 0.06990703210626019,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009904246219585677,
      "loss": 2.2718,
      "step": 10289
    },
    {
      "epoch": 0.06991382645285424,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009904225264119493,
      "loss": 2.2033,
      "step": 10290
    },
    {
      "epoch": 0.0699206207994483,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009904204306382712,
      "loss": 2.1748,
      "step": 10291
    },
    {
      "epoch": 0.06992741514604235,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009904183346375337,
      "loss": 2.3494,
      "step": 10292
    },
    {
      "epoch": 0.06993420949263641,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009904162384097382,
      "loss": 2.1851,
      "step": 10293
    },
    {
      "epoch": 0.06994100383923048,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009904141419548858,
      "loss": 2.3142,
      "step": 10294
    },
    {
      "epoch": 0.06994779818582453,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000990412045272977,
      "loss": 2.258,
      "step": 10295
    },
    {
      "epoch": 0.06995459253241859,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009904099483640132,
      "loss": 2.1187,
      "step": 10296
    },
    {
      "epoch": 0.06996138687901264,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009904078512279952,
      "loss": 2.1923,
      "step": 10297
    },
    {
      "epoch": 0.0699681812256067,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000990405753864924,
      "loss": 2.0977,
      "step": 10298
    },
    {
      "epoch": 0.06997497557220077,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009904036562748007,
      "loss": 2.3906,
      "step": 10299
    },
    {
      "epoch": 0.06998176991879482,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000990401558457626,
      "loss": 2.0821,
      "step": 10300
    },
    {
      "epoch": 0.06998856426538888,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000990399460413401,
      "loss": 2.279,
      "step": 10301
    },
    {
      "epoch": 0.06999535861198293,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009903973621421267,
      "loss": 2.2498,
      "step": 10302
    },
    {
      "epoch": 0.070002152958577,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009903952636438039,
      "loss": 2.3126,
      "step": 10303
    },
    {
      "epoch": 0.07000894730517104,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009903931649184337,
      "loss": 2.2598,
      "step": 10304
    },
    {
      "epoch": 0.0700157416517651,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009903910659660173,
      "loss": 2.14,
      "step": 10305
    },
    {
      "epoch": 0.07002253599835917,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009903889667865554,
      "loss": 2.1642,
      "step": 10306
    },
    {
      "epoch": 0.07002933034495322,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009903868673800489,
      "loss": 2.0696,
      "step": 10307
    },
    {
      "epoch": 0.07003612469154728,
      "grad_norm": 0.140625,
      "learning_rate": 0.000990384767746499,
      "loss": 2.2935,
      "step": 10308
    },
    {
      "epoch": 0.07004291903814133,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009903826678859067,
      "loss": 2.2278,
      "step": 10309
    },
    {
      "epoch": 0.0700497133847354,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009903805677982727,
      "loss": 2.398,
      "step": 10310
    },
    {
      "epoch": 0.07005650773132946,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009903784674835981,
      "loss": 2.1645,
      "step": 10311
    },
    {
      "epoch": 0.07006330207792351,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009903763669418838,
      "loss": 2.236,
      "step": 10312
    },
    {
      "epoch": 0.07007009642451757,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000990374266173131,
      "loss": 2.1903,
      "step": 10313
    },
    {
      "epoch": 0.07007689077111162,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009903721651773406,
      "loss": 2.1837,
      "step": 10314
    },
    {
      "epoch": 0.07008368511770569,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009903700639545135,
      "loss": 2.2311,
      "step": 10315
    },
    {
      "epoch": 0.07009047946429975,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009903679625046507,
      "loss": 2.0906,
      "step": 10316
    },
    {
      "epoch": 0.0700972738108938,
      "grad_norm": 0.125,
      "learning_rate": 0.000990365860827753,
      "loss": 2.1015,
      "step": 10317
    },
    {
      "epoch": 0.07010406815748786,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009903637589238215,
      "loss": 2.4211,
      "step": 10318
    },
    {
      "epoch": 0.07011086250408191,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009903616567928574,
      "loss": 2.1286,
      "step": 10319
    },
    {
      "epoch": 0.07011765685067597,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009903595544348614,
      "loss": 2.3243,
      "step": 10320
    },
    {
      "epoch": 0.07012445119727004,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009903574518498345,
      "loss": 2.4087,
      "step": 10321
    },
    {
      "epoch": 0.07013124554386409,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009903553490377777,
      "loss": 2.1825,
      "step": 10322
    },
    {
      "epoch": 0.07013803989045815,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009903532459986921,
      "loss": 2.2909,
      "step": 10323
    },
    {
      "epoch": 0.0701448342370522,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009903511427325784,
      "loss": 2.3272,
      "step": 10324
    },
    {
      "epoch": 0.07015162858364626,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009903490392394378,
      "loss": 2.3274,
      "step": 10325
    },
    {
      "epoch": 0.07015842293024031,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009903469355192714,
      "loss": 2.2206,
      "step": 10326
    },
    {
      "epoch": 0.07016521727683438,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009903448315720798,
      "loss": 2.2162,
      "step": 10327
    },
    {
      "epoch": 0.07017201162342844,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009903427273978643,
      "loss": 2.3319,
      "step": 10328
    },
    {
      "epoch": 0.07017880597002249,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009903406229966256,
      "loss": 2.3731,
      "step": 10329
    },
    {
      "epoch": 0.07018560031661655,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009903385183683648,
      "loss": 2.2911,
      "step": 10330
    },
    {
      "epoch": 0.0701923946632106,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009903364135130832,
      "loss": 2.4622,
      "step": 10331
    },
    {
      "epoch": 0.07019918900980467,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009903343084307812,
      "loss": 2.2841,
      "step": 10332
    },
    {
      "epoch": 0.07020598335639873,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.00099033220312146,
      "loss": 2.1629,
      "step": 10333
    },
    {
      "epoch": 0.07021277770299278,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009903300975851205,
      "loss": 2.3595,
      "step": 10334
    },
    {
      "epoch": 0.07021957204958684,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000990327991821764,
      "loss": 2.1929,
      "step": 10335
    },
    {
      "epoch": 0.07022636639618089,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009903258858313913,
      "loss": 2.3683,
      "step": 10336
    },
    {
      "epoch": 0.07023316074277496,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009903237796140033,
      "loss": 2.293,
      "step": 10337
    },
    {
      "epoch": 0.07023995508936902,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000990321673169601,
      "loss": 2.1716,
      "step": 10338
    },
    {
      "epoch": 0.07024674943596307,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009903195664981853,
      "loss": 2.1537,
      "step": 10339
    },
    {
      "epoch": 0.07025354378255713,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009903174595997574,
      "loss": 2.1224,
      "step": 10340
    },
    {
      "epoch": 0.07026033812915118,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000990315352474318,
      "loss": 2.3289,
      "step": 10341
    },
    {
      "epoch": 0.07026713247574524,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009903132451218682,
      "loss": 2.3384,
      "step": 10342
    },
    {
      "epoch": 0.07027392682233931,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009903111375424091,
      "loss": 2.2961,
      "step": 10343
    },
    {
      "epoch": 0.07028072116893336,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009903090297359416,
      "loss": 2.2205,
      "step": 10344
    },
    {
      "epoch": 0.07028751551552742,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009903069217024666,
      "loss": 2.0726,
      "step": 10345
    },
    {
      "epoch": 0.07029430986212147,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000990304813441985,
      "loss": 2.2326,
      "step": 10346
    },
    {
      "epoch": 0.07030110420871553,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000990302704954498,
      "loss": 2.3244,
      "step": 10347
    },
    {
      "epoch": 0.0703078985553096,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009903005962400065,
      "loss": 2.2434,
      "step": 10348
    },
    {
      "epoch": 0.07031469290190365,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009902984872985111,
      "loss": 2.2227,
      "step": 10349
    },
    {
      "epoch": 0.07032148724849771,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009902963781300135,
      "loss": 2.4711,
      "step": 10350
    },
    {
      "epoch": 0.07032828159509176,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009902942687345143,
      "loss": 2.285,
      "step": 10351
    },
    {
      "epoch": 0.07033507594168582,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009902921591120144,
      "loss": 2.2831,
      "step": 10352
    },
    {
      "epoch": 0.07034187028827987,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009902900492625149,
      "loss": 2.1331,
      "step": 10353
    },
    {
      "epoch": 0.07034866463487394,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009902879391860166,
      "loss": 2.2542,
      "step": 10354
    },
    {
      "epoch": 0.070355458981468,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009902858288825208,
      "loss": 2.1393,
      "step": 10355
    },
    {
      "epoch": 0.07036225332806205,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990283718352028,
      "loss": 2.3246,
      "step": 10356
    },
    {
      "epoch": 0.07036904767465611,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009902816075945395,
      "loss": 2.1935,
      "step": 10357
    },
    {
      "epoch": 0.07037584202125016,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009902794966100564,
      "loss": 2.2687,
      "step": 10358
    },
    {
      "epoch": 0.07038263636784423,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009902773853985795,
      "loss": 2.2624,
      "step": 10359
    },
    {
      "epoch": 0.07038943071443829,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009902752739601097,
      "loss": 2.2496,
      "step": 10360
    },
    {
      "epoch": 0.07039622506103234,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000990273162294648,
      "loss": 2.2227,
      "step": 10361
    },
    {
      "epoch": 0.0704030194076264,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009902710504021957,
      "loss": 2.0622,
      "step": 10362
    },
    {
      "epoch": 0.07040981375422045,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009902689382827535,
      "loss": 2.4451,
      "step": 10363
    },
    {
      "epoch": 0.07041660810081452,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009902668259363222,
      "loss": 2.1393,
      "step": 10364
    },
    {
      "epoch": 0.07042340244740858,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000990264713362903,
      "loss": 2.2666,
      "step": 10365
    },
    {
      "epoch": 0.07043019679400263,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000990262600562497,
      "loss": 2.2462,
      "step": 10366
    },
    {
      "epoch": 0.07043699114059669,
      "grad_norm": 0.140625,
      "learning_rate": 0.000990260487535105,
      "loss": 2.2704,
      "step": 10367
    },
    {
      "epoch": 0.07044378548719074,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009902583742807279,
      "loss": 2.3073,
      "step": 10368
    },
    {
      "epoch": 0.0704505798337848,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000990256260799367,
      "loss": 2.2718,
      "step": 10369
    },
    {
      "epoch": 0.07045737418037887,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009902541470910228,
      "loss": 2.2422,
      "step": 10370
    },
    {
      "epoch": 0.07046416852697292,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009902520331556968,
      "loss": 2.2951,
      "step": 10371
    },
    {
      "epoch": 0.07047096287356698,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009902499189933895,
      "loss": 2.1056,
      "step": 10372
    },
    {
      "epoch": 0.07047775722016103,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009902478046041024,
      "loss": 2.2296,
      "step": 10373
    },
    {
      "epoch": 0.0704845515667551,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000990245689987836,
      "loss": 2.2883,
      "step": 10374
    },
    {
      "epoch": 0.07049134591334914,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009902435751445915,
      "loss": 2.3155,
      "step": 10375
    },
    {
      "epoch": 0.0704981402599432,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009902414600743699,
      "loss": 2.2871,
      "step": 10376
    },
    {
      "epoch": 0.07050493460653727,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009902393447771718,
      "loss": 2.5563,
      "step": 10377
    },
    {
      "epoch": 0.07051172895313132,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000990237229252999,
      "loss": 2.2941,
      "step": 10378
    },
    {
      "epoch": 0.07051852329972538,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009902351135018517,
      "loss": 2.2246,
      "step": 10379
    },
    {
      "epoch": 0.07052531764631943,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009902329975237311,
      "loss": 2.2335,
      "step": 10380
    },
    {
      "epoch": 0.0705321119929135,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009902308813186384,
      "loss": 2.2297,
      "step": 10381
    },
    {
      "epoch": 0.07053890633950756,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009902287648865743,
      "loss": 2.205,
      "step": 10382
    },
    {
      "epoch": 0.07054570068610161,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009902266482275399,
      "loss": 2.3277,
      "step": 10383
    },
    {
      "epoch": 0.07055249503269567,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009902245313415364,
      "loss": 2.1674,
      "step": 10384
    },
    {
      "epoch": 0.07055928937928972,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009902224142285642,
      "loss": 2.2802,
      "step": 10385
    },
    {
      "epoch": 0.07056608372588379,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000990220296888625,
      "loss": 2.2466,
      "step": 10386
    },
    {
      "epoch": 0.07057287807247785,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009902181793217193,
      "loss": 2.2574,
      "step": 10387
    },
    {
      "epoch": 0.0705796724190719,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000990216061527848,
      "loss": 2.2544,
      "step": 10388
    },
    {
      "epoch": 0.07058646676566596,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009902139435070127,
      "loss": 2.127,
      "step": 10389
    },
    {
      "epoch": 0.07059326111226001,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009902118252592138,
      "loss": 2.3224,
      "step": 10390
    },
    {
      "epoch": 0.07060005545885407,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009902097067844523,
      "loss": 2.3871,
      "step": 10391
    },
    {
      "epoch": 0.07060684980544814,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009902075880827293,
      "loss": 2.1963,
      "step": 10392
    },
    {
      "epoch": 0.07061364415204219,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000990205469154046,
      "loss": 2.1732,
      "step": 10393
    },
    {
      "epoch": 0.07062043849863625,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009902033499984031,
      "loss": 2.4099,
      "step": 10394
    },
    {
      "epoch": 0.0706272328452303,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009902012306158016,
      "loss": 2.1715,
      "step": 10395
    },
    {
      "epoch": 0.07063402719182436,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009901991110062428,
      "loss": 2.2721,
      "step": 10396
    },
    {
      "epoch": 0.07064082153841843,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990196991169727,
      "loss": 2.1478,
      "step": 10397
    },
    {
      "epoch": 0.07064761588501248,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000990194871106256,
      "loss": 2.256,
      "step": 10398
    },
    {
      "epoch": 0.07065441023160654,
      "grad_norm": 0.140625,
      "learning_rate": 0.00099019275081583,
      "loss": 2.243,
      "step": 10399
    },
    {
      "epoch": 0.07066120457820059,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000990190630298451,
      "loss": 2.301,
      "step": 10400
    },
    {
      "epoch": 0.07066799892479465,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009901885095541189,
      "loss": 2.2753,
      "step": 10401
    },
    {
      "epoch": 0.0706747932713887,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000990186388582835,
      "loss": 2.2105,
      "step": 10402
    },
    {
      "epoch": 0.07068158761798277,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009901842673846009,
      "loss": 2.2092,
      "step": 10403
    },
    {
      "epoch": 0.07068838196457683,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009901821459594167,
      "loss": 1.9543,
      "step": 10404
    },
    {
      "epoch": 0.07069517631117088,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000990180024307284,
      "loss": 2.1675,
      "step": 10405
    },
    {
      "epoch": 0.07070197065776494,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009901779024282034,
      "loss": 2.3421,
      "step": 10406
    },
    {
      "epoch": 0.07070876500435899,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009901757803221763,
      "loss": 2.2239,
      "step": 10407
    },
    {
      "epoch": 0.07071555935095306,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009901736579892033,
      "loss": 2.2663,
      "step": 10408
    },
    {
      "epoch": 0.07072235369754712,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009901715354292855,
      "loss": 2.314,
      "step": 10409
    },
    {
      "epoch": 0.07072914804414117,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009901694126424239,
      "loss": 2.3667,
      "step": 10410
    },
    {
      "epoch": 0.07073594239073523,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009901672896286194,
      "loss": 2.3223,
      "step": 10411
    },
    {
      "epoch": 0.07074273673732928,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009901651663878732,
      "loss": 2.3455,
      "step": 10412
    },
    {
      "epoch": 0.07074953108392334,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009901630429201861,
      "loss": 2.2191,
      "step": 10413
    },
    {
      "epoch": 0.07075632543051741,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000990160919225559,
      "loss": 2.23,
      "step": 10414
    },
    {
      "epoch": 0.07076311977711146,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009901587953039931,
      "loss": 2.141,
      "step": 10415
    },
    {
      "epoch": 0.07076991412370552,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009901566711554895,
      "loss": 2.3366,
      "step": 10416
    },
    {
      "epoch": 0.07077670847029957,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009901545467800488,
      "loss": 2.3073,
      "step": 10417
    },
    {
      "epoch": 0.07078350281689363,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009901524221776724,
      "loss": 2.2379,
      "step": 10418
    },
    {
      "epoch": 0.0707902971634877,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009901502973483607,
      "loss": 2.4067,
      "step": 10419
    },
    {
      "epoch": 0.07079709151008175,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000990148172292115,
      "loss": 2.3396,
      "step": 10420
    },
    {
      "epoch": 0.07080388585667581,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009901460470089366,
      "loss": 2.1598,
      "step": 10421
    },
    {
      "epoch": 0.07081068020326986,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009901439214988262,
      "loss": 2.2405,
      "step": 10422
    },
    {
      "epoch": 0.07081747454986392,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009901417957617847,
      "loss": 2.3281,
      "step": 10423
    },
    {
      "epoch": 0.07082426889645797,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990139669797813,
      "loss": 2.1313,
      "step": 10424
    },
    {
      "epoch": 0.07083106324305204,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009901375436069124,
      "loss": 2.3867,
      "step": 10425
    },
    {
      "epoch": 0.0708378575896461,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009901354171890837,
      "loss": 2.3014,
      "step": 10426
    },
    {
      "epoch": 0.07084465193624015,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009901332905443278,
      "loss": 2.2027,
      "step": 10427
    },
    {
      "epoch": 0.07085144628283421,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009901311636726462,
      "loss": 2.3409,
      "step": 10428
    },
    {
      "epoch": 0.07085824062942826,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009901290365740391,
      "loss": 2.1426,
      "step": 10429
    },
    {
      "epoch": 0.07086503497602233,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009901269092485081,
      "loss": 2.2314,
      "step": 10430
    },
    {
      "epoch": 0.07087182932261639,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000990124781696054,
      "loss": 2.3132,
      "step": 10431
    },
    {
      "epoch": 0.07087862366921044,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009901226539166775,
      "loss": 2.0124,
      "step": 10432
    },
    {
      "epoch": 0.0708854180158045,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009901205259103798,
      "loss": 2.2376,
      "step": 10433
    },
    {
      "epoch": 0.07089221236239855,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009901183976771622,
      "loss": 2.2462,
      "step": 10434
    },
    {
      "epoch": 0.07089900670899262,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009901162692170253,
      "loss": 2.0465,
      "step": 10435
    },
    {
      "epoch": 0.07090580105558668,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00099011414052997,
      "loss": 2.3604,
      "step": 10436
    },
    {
      "epoch": 0.07091259540218073,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009901120116159977,
      "loss": 2.3429,
      "step": 10437
    },
    {
      "epoch": 0.07091938974877479,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009901098824751091,
      "loss": 2.2812,
      "step": 10438
    },
    {
      "epoch": 0.07092618409536884,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009901077531073053,
      "loss": 2.2447,
      "step": 10439
    },
    {
      "epoch": 0.0709329784419629,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000990105623512587,
      "loss": 2.2906,
      "step": 10440
    },
    {
      "epoch": 0.07093977278855697,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009901034936909557,
      "loss": 2.2107,
      "step": 10441
    },
    {
      "epoch": 0.07094656713515102,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000990101363642412,
      "loss": 2.0391,
      "step": 10442
    },
    {
      "epoch": 0.07095336148174508,
      "grad_norm": 0.125,
      "learning_rate": 0.0009900992333669568,
      "loss": 2.2159,
      "step": 10443
    },
    {
      "epoch": 0.07096015582833913,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009900971028645915,
      "loss": 2.2195,
      "step": 10444
    },
    {
      "epoch": 0.0709669501749332,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009900949721353166,
      "loss": 2.3505,
      "step": 10445
    },
    {
      "epoch": 0.07097374452152724,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009900928411791337,
      "loss": 2.2543,
      "step": 10446
    },
    {
      "epoch": 0.0709805388681213,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009900907099960432,
      "loss": 2.2074,
      "step": 10447
    },
    {
      "epoch": 0.07098733321471537,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009900885785860466,
      "loss": 2.2991,
      "step": 10448
    },
    {
      "epoch": 0.07099412756130942,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009900864469491442,
      "loss": 2.3181,
      "step": 10449
    },
    {
      "epoch": 0.07100092190790348,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009900843150853376,
      "loss": 2.2351,
      "step": 10450
    },
    {
      "epoch": 0.07100771625449753,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009900821829946277,
      "loss": 2.1493,
      "step": 10451
    },
    {
      "epoch": 0.0710145106010916,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000990080050677015,
      "loss": 2.3171,
      "step": 10452
    },
    {
      "epoch": 0.07102130494768566,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009900779181325012,
      "loss": 2.2186,
      "step": 10453
    },
    {
      "epoch": 0.07102809929427971,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000990075785361087,
      "loss": 2.3281,
      "step": 10454
    },
    {
      "epoch": 0.07103489364087377,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009900736523627731,
      "loss": 2.2111,
      "step": 10455
    },
    {
      "epoch": 0.07104168798746782,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009900715191375608,
      "loss": 2.2847,
      "step": 10456
    },
    {
      "epoch": 0.07104848233406189,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990069385685451,
      "loss": 2.2154,
      "step": 10457
    },
    {
      "epoch": 0.07105527668065595,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009900672520064446,
      "loss": 2.2574,
      "step": 10458
    },
    {
      "epoch": 0.07106207102725,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009900651181005428,
      "loss": 2.2387,
      "step": 10459
    },
    {
      "epoch": 0.07106886537384406,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009900629839677464,
      "loss": 2.1753,
      "step": 10460
    },
    {
      "epoch": 0.07107565972043811,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009900608496080565,
      "loss": 2.3868,
      "step": 10461
    },
    {
      "epoch": 0.07108245406703217,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009900587150214741,
      "loss": 2.2792,
      "step": 10462
    },
    {
      "epoch": 0.07108924841362624,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009900565802080001,
      "loss": 2.3435,
      "step": 10463
    },
    {
      "epoch": 0.07109604276022029,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009900544451676354,
      "loss": 2.2703,
      "step": 10464
    },
    {
      "epoch": 0.07110283710681435,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009900523099003811,
      "loss": 2.2582,
      "step": 10465
    },
    {
      "epoch": 0.0711096314534084,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009900501744062385,
      "loss": 2.2768,
      "step": 10466
    },
    {
      "epoch": 0.07111642580000246,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000990048038685208,
      "loss": 2.0895,
      "step": 10467
    },
    {
      "epoch": 0.07112322014659653,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000990045902737291,
      "loss": 2.3541,
      "step": 10468
    },
    {
      "epoch": 0.07113001449319058,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009900437665624883,
      "loss": 2.2755,
      "step": 10469
    },
    {
      "epoch": 0.07113680883978464,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000990041630160801,
      "loss": 2.2163,
      "step": 10470
    },
    {
      "epoch": 0.07114360318637869,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00099003949353223,
      "loss": 2.2923,
      "step": 10471
    },
    {
      "epoch": 0.07115039753297275,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009900373566767765,
      "loss": 2.2037,
      "step": 10472
    },
    {
      "epoch": 0.0711571918795668,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009900352195944412,
      "loss": 2.2863,
      "step": 10473
    },
    {
      "epoch": 0.07116398622616087,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000990033082285225,
      "loss": 2.2935,
      "step": 10474
    },
    {
      "epoch": 0.07117078057275493,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009900309447491294,
      "loss": 2.1684,
      "step": 10475
    },
    {
      "epoch": 0.07117757491934898,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000990028806986155,
      "loss": 2.2946,
      "step": 10476
    },
    {
      "epoch": 0.07118436926594304,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009900266689963028,
      "loss": 2.4269,
      "step": 10477
    },
    {
      "epoch": 0.07119116361253709,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009900245307795739,
      "loss": 2.1769,
      "step": 10478
    },
    {
      "epoch": 0.07119795795913116,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009900223923359692,
      "loss": 2.3354,
      "step": 10479
    },
    {
      "epoch": 0.07120475230572522,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00099002025366549,
      "loss": 2.233,
      "step": 10480
    },
    {
      "epoch": 0.07121154665231927,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000990018114768137,
      "loss": 2.1886,
      "step": 10481
    },
    {
      "epoch": 0.07121834099891333,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009900159756439108,
      "loss": 2.3124,
      "step": 10482
    },
    {
      "epoch": 0.07122513534550738,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009900138362928133,
      "loss": 2.1951,
      "step": 10483
    },
    {
      "epoch": 0.07123192969210144,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009900116967148447,
      "loss": 2.257,
      "step": 10484
    },
    {
      "epoch": 0.07123872403869551,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009900095569100064,
      "loss": 2.2445,
      "step": 10485
    },
    {
      "epoch": 0.07124551838528956,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009900074168782995,
      "loss": 2.2774,
      "step": 10486
    },
    {
      "epoch": 0.07125231273188362,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009900052766197245,
      "loss": 2.2195,
      "step": 10487
    },
    {
      "epoch": 0.07125910707847767,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000990003136134283,
      "loss": 2.2398,
      "step": 10488
    },
    {
      "epoch": 0.07126590142507173,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009900009954219753,
      "loss": 2.389,
      "step": 10489
    },
    {
      "epoch": 0.0712726957716658,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009899988544828028,
      "loss": 2.2334,
      "step": 10490
    },
    {
      "epoch": 0.07127949011825985,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009899967133167667,
      "loss": 2.0984,
      "step": 10491
    },
    {
      "epoch": 0.07128628446485391,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009899945719238678,
      "loss": 2.2579,
      "step": 10492
    },
    {
      "epoch": 0.07129307881144796,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009899924303041067,
      "loss": 2.1002,
      "step": 10493
    },
    {
      "epoch": 0.07129987315804202,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000989990288457485,
      "loss": 2.2979,
      "step": 10494
    },
    {
      "epoch": 0.07130666750463607,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009899881463840032,
      "loss": 2.2421,
      "step": 10495
    },
    {
      "epoch": 0.07131346185123014,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009899860040836627,
      "loss": 2.2341,
      "step": 10496
    },
    {
      "epoch": 0.0713202561978242,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000989983861556464,
      "loss": 2.2514,
      "step": 10497
    },
    {
      "epoch": 0.07132705054441825,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009899817188024085,
      "loss": 2.1554,
      "step": 10498
    },
    {
      "epoch": 0.07133384489101231,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000989979575821497,
      "loss": 2.3265,
      "step": 10499
    },
    {
      "epoch": 0.07134063923760636,
      "grad_norm": 0.140625,
      "learning_rate": 0.000989977432613731,
      "loss": 2.4677,
      "step": 10500
    },
    {
      "epoch": 0.07134743358420043,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009899752891791107,
      "loss": 2.3892,
      "step": 10501
    },
    {
      "epoch": 0.07135422793079449,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009899731455176375,
      "loss": 2.2163,
      "step": 10502
    },
    {
      "epoch": 0.07136102227738854,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009899710016293124,
      "loss": 2.3667,
      "step": 10503
    },
    {
      "epoch": 0.0713678166239826,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009899688575141364,
      "loss": 2.2155,
      "step": 10504
    },
    {
      "epoch": 0.07137461097057665,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009899667131721105,
      "loss": 2.1171,
      "step": 10505
    },
    {
      "epoch": 0.07138140531717072,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009899645686032353,
      "loss": 2.1828,
      "step": 10506
    },
    {
      "epoch": 0.07138819966376478,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009899624238075124,
      "loss": 2.1805,
      "step": 10507
    },
    {
      "epoch": 0.07139499401035883,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009899602787849424,
      "loss": 2.1382,
      "step": 10508
    },
    {
      "epoch": 0.07140178835695289,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009899581335355266,
      "loss": 2.1981,
      "step": 10509
    },
    {
      "epoch": 0.07140858270354694,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009899559880592655,
      "loss": 2.3999,
      "step": 10510
    },
    {
      "epoch": 0.071415377050141,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009899538423561605,
      "loss": 2.1798,
      "step": 10511
    },
    {
      "epoch": 0.07142217139673507,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009899516964262127,
      "loss": 2.1879,
      "step": 10512
    },
    {
      "epoch": 0.07142896574332912,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009899495502694226,
      "loss": 2.0917,
      "step": 10513
    },
    {
      "epoch": 0.07143576008992318,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009899474038857917,
      "loss": 2.2338,
      "step": 10514
    },
    {
      "epoch": 0.07144255443651723,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009899452572753205,
      "loss": 2.0821,
      "step": 10515
    },
    {
      "epoch": 0.0714493487831113,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009899431104380106,
      "loss": 2.095,
      "step": 10516
    },
    {
      "epoch": 0.07145614312970536,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009899409633738624,
      "loss": 2.3136,
      "step": 10517
    },
    {
      "epoch": 0.0714629374762994,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009899388160828774,
      "loss": 2.3331,
      "step": 10518
    },
    {
      "epoch": 0.07146973182289347,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000989936668565056,
      "loss": 2.3018,
      "step": 10519
    },
    {
      "epoch": 0.07147652616948752,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009899345208203999,
      "loss": 2.4165,
      "step": 10520
    },
    {
      "epoch": 0.07148332051608158,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009899323728489093,
      "loss": 2.2152,
      "step": 10521
    },
    {
      "epoch": 0.07149011486267563,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009899302246505859,
      "loss": 2.2875,
      "step": 10522
    },
    {
      "epoch": 0.0714969092092697,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009899280762254304,
      "loss": 2.1581,
      "step": 10523
    },
    {
      "epoch": 0.07150370355586376,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009899259275734438,
      "loss": 2.1194,
      "step": 10524
    },
    {
      "epoch": 0.07151049790245781,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009899237786946272,
      "loss": 2.2611,
      "step": 10525
    },
    {
      "epoch": 0.07151729224905187,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009899216295889814,
      "loss": 2.2604,
      "step": 10526
    },
    {
      "epoch": 0.07152408659564592,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009899194802565077,
      "loss": 2.2711,
      "step": 10527
    },
    {
      "epoch": 0.07153088094223999,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009899173306972067,
      "loss": 2.2598,
      "step": 10528
    },
    {
      "epoch": 0.07153767528883405,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009899151809110796,
      "loss": 2.3467,
      "step": 10529
    },
    {
      "epoch": 0.0715444696354281,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009899130308981274,
      "loss": 2.0412,
      "step": 10530
    },
    {
      "epoch": 0.07155126398202216,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000989910880658351,
      "loss": 2.2463,
      "step": 10531
    },
    {
      "epoch": 0.07155805832861621,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009899087301917517,
      "loss": 2.2137,
      "step": 10532
    },
    {
      "epoch": 0.07156485267521027,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009899065794983302,
      "loss": 2.1479,
      "step": 10533
    },
    {
      "epoch": 0.07157164702180434,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009899044285780875,
      "loss": 2.3823,
      "step": 10534
    },
    {
      "epoch": 0.07157844136839839,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009899022774310247,
      "loss": 2.1795,
      "step": 10535
    },
    {
      "epoch": 0.07158523571499245,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009899001260571427,
      "loss": 2.281,
      "step": 10536
    },
    {
      "epoch": 0.0715920300615865,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009898979744564426,
      "loss": 2.2802,
      "step": 10537
    },
    {
      "epoch": 0.07159882440818056,
      "grad_norm": 0.1845703125,
      "learning_rate": 0.0009898958226289254,
      "loss": 2.1867,
      "step": 10538
    },
    {
      "epoch": 0.07160561875477463,
      "grad_norm": 0.140625,
      "learning_rate": 0.000989893670574592,
      "loss": 2.2896,
      "step": 10539
    },
    {
      "epoch": 0.07161241310136868,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009898915182934434,
      "loss": 2.2945,
      "step": 10540
    },
    {
      "epoch": 0.07161920744796274,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009898893657854808,
      "loss": 2.0534,
      "step": 10541
    },
    {
      "epoch": 0.07162600179455679,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009898872130507047,
      "loss": 2.3643,
      "step": 10542
    },
    {
      "epoch": 0.07163279614115085,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009898850600891169,
      "loss": 2.1948,
      "step": 10543
    },
    {
      "epoch": 0.0716395904877449,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009898829069007176,
      "loss": 2.2985,
      "step": 10544
    },
    {
      "epoch": 0.07164638483433897,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009898807534855084,
      "loss": 2.285,
      "step": 10545
    },
    {
      "epoch": 0.07165317918093303,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009898785998434898,
      "loss": 2.2639,
      "step": 10546
    },
    {
      "epoch": 0.07165997352752708,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009898764459746632,
      "loss": 2.3196,
      "step": 10547
    },
    {
      "epoch": 0.07166676787412114,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009898742918790294,
      "loss": 2.244,
      "step": 10548
    },
    {
      "epoch": 0.07167356222071519,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009898721375565893,
      "loss": 2.2597,
      "step": 10549
    },
    {
      "epoch": 0.07168035656730926,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000989869983007344,
      "loss": 2.2411,
      "step": 10550
    },
    {
      "epoch": 0.07168715091390332,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009898678282312945,
      "loss": 2.1842,
      "step": 10551
    },
    {
      "epoch": 0.07169394526049737,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009898656732284419,
      "loss": 2.3425,
      "step": 10552
    },
    {
      "epoch": 0.07170073960709143,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009898635179987872,
      "loss": 2.3576,
      "step": 10553
    },
    {
      "epoch": 0.07170753395368548,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000989861362542331,
      "loss": 2.3777,
      "step": 10554
    },
    {
      "epoch": 0.07171432830027955,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009898592068590748,
      "loss": 2.2082,
      "step": 10555
    },
    {
      "epoch": 0.07172112264687361,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009898570509490195,
      "loss": 2.2088,
      "step": 10556
    },
    {
      "epoch": 0.07172791699346766,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009898548948121659,
      "loss": 2.4163,
      "step": 10557
    },
    {
      "epoch": 0.07173471134006172,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009898527384485153,
      "loss": 2.2386,
      "step": 10558
    },
    {
      "epoch": 0.07174150568665577,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000989850581858068,
      "loss": 2.4041,
      "step": 10559
    },
    {
      "epoch": 0.07174830003324983,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009898484250408259,
      "loss": 2.2677,
      "step": 10560
    },
    {
      "epoch": 0.0717550943798439,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009898462679967894,
      "loss": 2.2125,
      "step": 10561
    },
    {
      "epoch": 0.07176188872643795,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00098984411072596,
      "loss": 2.2403,
      "step": 10562
    },
    {
      "epoch": 0.07176868307303201,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000989841953228338,
      "loss": 2.2387,
      "step": 10563
    },
    {
      "epoch": 0.07177547741962606,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000989839795503925,
      "loss": 2.2247,
      "step": 10564
    },
    {
      "epoch": 0.07178227176622012,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000989837637552722,
      "loss": 2.2991,
      "step": 10565
    },
    {
      "epoch": 0.07178906611281417,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009898354793747295,
      "loss": 2.3235,
      "step": 10566
    },
    {
      "epoch": 0.07179586045940824,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009898333209699488,
      "loss": 2.2478,
      "step": 10567
    },
    {
      "epoch": 0.0718026548060023,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009898311623383808,
      "loss": 2.2253,
      "step": 10568
    },
    {
      "epoch": 0.07180944915259635,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009898290034800268,
      "loss": 2.1398,
      "step": 10569
    },
    {
      "epoch": 0.07181624349919041,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009898268443948874,
      "loss": 2.2912,
      "step": 10570
    },
    {
      "epoch": 0.07182303784578446,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000989824685082964,
      "loss": 2.2523,
      "step": 10571
    },
    {
      "epoch": 0.07182983219237853,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009898225255442572,
      "loss": 2.3709,
      "step": 10572
    },
    {
      "epoch": 0.07183662653897259,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009898203657787684,
      "loss": 2.2184,
      "step": 10573
    },
    {
      "epoch": 0.07184342088556664,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009898182057864982,
      "loss": 2.2119,
      "step": 10574
    },
    {
      "epoch": 0.0718502152321607,
      "grad_norm": 0.16015625,
      "learning_rate": 0.000989816045567448,
      "loss": 2.1755,
      "step": 10575
    },
    {
      "epoch": 0.07185700957875475,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009898138851216183,
      "loss": 2.139,
      "step": 10576
    },
    {
      "epoch": 0.07186380392534882,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009898117244490105,
      "loss": 2.219,
      "step": 10577
    },
    {
      "epoch": 0.07187059827194288,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009898095635496255,
      "loss": 1.9971,
      "step": 10578
    },
    {
      "epoch": 0.07187739261853693,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009898074024234642,
      "loss": 2.298,
      "step": 10579
    },
    {
      "epoch": 0.07188418696513099,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009898052410705278,
      "loss": 2.2997,
      "step": 10580
    },
    {
      "epoch": 0.07189098131172504,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009898030794908172,
      "loss": 2.1266,
      "step": 10581
    },
    {
      "epoch": 0.0718977756583191,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009898009176843334,
      "loss": 2.0594,
      "step": 10582
    },
    {
      "epoch": 0.07190457000491317,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009897987556510772,
      "loss": 1.9862,
      "step": 10583
    },
    {
      "epoch": 0.07191136435150722,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00098979659339105,
      "loss": 2.1454,
      "step": 10584
    },
    {
      "epoch": 0.07191815869810128,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009897944309042526,
      "loss": 2.1803,
      "step": 10585
    },
    {
      "epoch": 0.07192495304469533,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000989792268190686,
      "loss": 2.3994,
      "step": 10586
    },
    {
      "epoch": 0.0719317473912894,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000989790105250351,
      "loss": 2.2313,
      "step": 10587
    },
    {
      "epoch": 0.07193854173788346,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009897879420832489,
      "loss": 2.283,
      "step": 10588
    },
    {
      "epoch": 0.07194533608447751,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009897857786893807,
      "loss": 2.2593,
      "step": 10589
    },
    {
      "epoch": 0.07195213043107157,
      "grad_norm": 0.125,
      "learning_rate": 0.0009897836150687472,
      "loss": 2.0284,
      "step": 10590
    },
    {
      "epoch": 0.07195892477766562,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009897814512213495,
      "loss": 2.2828,
      "step": 10591
    },
    {
      "epoch": 0.07196571912425968,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009897792871471885,
      "loss": 2.2563,
      "step": 10592
    },
    {
      "epoch": 0.07197251347085373,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009897771228462656,
      "loss": 2.2727,
      "step": 10593
    },
    {
      "epoch": 0.0719793078174478,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000989774958318581,
      "loss": 2.1941,
      "step": 10594
    },
    {
      "epoch": 0.07198610216404186,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009897727935641368,
      "loss": 2.0967,
      "step": 10595
    },
    {
      "epoch": 0.07199289651063591,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009897706285829332,
      "loss": 2.2572,
      "step": 10596
    },
    {
      "epoch": 0.07199969085722997,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009897684633749712,
      "loss": 2.1178,
      "step": 10597
    },
    {
      "epoch": 0.07200648520382402,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009897662979402521,
      "loss": 2.1837,
      "step": 10598
    },
    {
      "epoch": 0.07201327955041809,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009897641322787769,
      "loss": 2.1903,
      "step": 10599
    },
    {
      "epoch": 0.07202007389701215,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009897619663905466,
      "loss": 2.2128,
      "step": 10600
    },
    {
      "epoch": 0.0720268682436062,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009897598002755622,
      "loss": 2.2429,
      "step": 10601
    },
    {
      "epoch": 0.07203366259020026,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009897576339338242,
      "loss": 2.3448,
      "step": 10602
    },
    {
      "epoch": 0.07204045693679431,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009897554673653343,
      "loss": 2.2421,
      "step": 10603
    },
    {
      "epoch": 0.07204725128338837,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009897533005700935,
      "loss": 2.4673,
      "step": 10604
    },
    {
      "epoch": 0.07205404562998244,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009897511335481021,
      "loss": 2.2865,
      "step": 10605
    },
    {
      "epoch": 0.07206083997657649,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009897489662993616,
      "loss": 2.2386,
      "step": 10606
    },
    {
      "epoch": 0.07206763432317055,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009897467988238733,
      "loss": 2.2333,
      "step": 10607
    },
    {
      "epoch": 0.0720744286697646,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009897446311216377,
      "loss": 2.2438,
      "step": 10608
    },
    {
      "epoch": 0.07208122301635866,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009897424631926557,
      "loss": 2.3533,
      "step": 10609
    },
    {
      "epoch": 0.07208801736295273,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009897402950369287,
      "loss": 2.2157,
      "step": 10610
    },
    {
      "epoch": 0.07209481170954678,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009897381266544575,
      "loss": 2.3284,
      "step": 10611
    },
    {
      "epoch": 0.07210160605614084,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009897359580452432,
      "loss": 2.1433,
      "step": 10612
    },
    {
      "epoch": 0.07210840040273489,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009897337892092867,
      "loss": 2.3798,
      "step": 10613
    },
    {
      "epoch": 0.07211519474932895,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009897316201465895,
      "loss": 2.2378,
      "step": 10614
    },
    {
      "epoch": 0.072121989095923,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009897294508571517,
      "loss": 2.2728,
      "step": 10615
    },
    {
      "epoch": 0.07212878344251707,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009897272813409747,
      "loss": 2.1852,
      "step": 10616
    },
    {
      "epoch": 0.07213557778911113,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00098972511159806,
      "loss": 2.2063,
      "step": 10617
    },
    {
      "epoch": 0.07214237213570518,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000989722941628408,
      "loss": 2.1533,
      "step": 10618
    },
    {
      "epoch": 0.07214916648229924,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009897207714320197,
      "loss": 2.3084,
      "step": 10619
    },
    {
      "epoch": 0.07215596082889329,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009897186010088966,
      "loss": 2.1936,
      "step": 10620
    },
    {
      "epoch": 0.07216275517548736,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009897164303590393,
      "loss": 2.2302,
      "step": 10621
    },
    {
      "epoch": 0.07216954952208142,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009897142594824488,
      "loss": 2.1784,
      "step": 10622
    },
    {
      "epoch": 0.07217634386867547,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009897120883791265,
      "loss": 2.1046,
      "step": 10623
    },
    {
      "epoch": 0.07218313821526953,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009897099170490728,
      "loss": 2.159,
      "step": 10624
    },
    {
      "epoch": 0.07218993256186358,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009897077454922893,
      "loss": 2.2504,
      "step": 10625
    },
    {
      "epoch": 0.07219672690845765,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009897055737087762,
      "loss": 2.1041,
      "step": 10626
    },
    {
      "epoch": 0.07220352125505171,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009897034016985355,
      "loss": 2.1915,
      "step": 10627
    },
    {
      "epoch": 0.07221031560164576,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009897012294615678,
      "loss": 2.1894,
      "step": 10628
    },
    {
      "epoch": 0.07221710994823982,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009896990569978738,
      "loss": 2.2047,
      "step": 10629
    },
    {
      "epoch": 0.07222390429483387,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009896968843074547,
      "loss": 2.3648,
      "step": 10630
    },
    {
      "epoch": 0.07223069864142793,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009896947113903118,
      "loss": 2.2575,
      "step": 10631
    },
    {
      "epoch": 0.072237492988022,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009896925382464458,
      "loss": 2.1906,
      "step": 10632
    },
    {
      "epoch": 0.07224428733461605,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009896903648758577,
      "loss": 2.1187,
      "step": 10633
    },
    {
      "epoch": 0.07225108168121011,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009896881912785486,
      "loss": 2.3323,
      "step": 10634
    },
    {
      "epoch": 0.07225787602780416,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009896860174545194,
      "loss": 2.1449,
      "step": 10635
    },
    {
      "epoch": 0.07226467037439822,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009896838434037714,
      "loss": 2.3541,
      "step": 10636
    },
    {
      "epoch": 0.07227146472099229,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009896816691263052,
      "loss": 2.2749,
      "step": 10637
    },
    {
      "epoch": 0.07227825906758634,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000989679494622122,
      "loss": 2.2507,
      "step": 10638
    },
    {
      "epoch": 0.0722850534141804,
      "grad_norm": 0.185546875,
      "learning_rate": 0.0009896773198912229,
      "loss": 2.2213,
      "step": 10639
    },
    {
      "epoch": 0.07229184776077445,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009896751449336087,
      "loss": 2.4147,
      "step": 10640
    },
    {
      "epoch": 0.07229864210736851,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009896729697492807,
      "loss": 2.1975,
      "step": 10641
    },
    {
      "epoch": 0.07230543645396256,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009896707943382397,
      "loss": 2.3079,
      "step": 10642
    },
    {
      "epoch": 0.07231223080055663,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009896686187004866,
      "loss": 2.3377,
      "step": 10643
    },
    {
      "epoch": 0.07231902514715069,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009896664428360228,
      "loss": 2.1491,
      "step": 10644
    },
    {
      "epoch": 0.07232581949374474,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009896642667448488,
      "loss": 2.1641,
      "step": 10645
    },
    {
      "epoch": 0.0723326138403388,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009896620904269657,
      "loss": 2.2073,
      "step": 10646
    },
    {
      "epoch": 0.07233940818693285,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009896599138823749,
      "loss": 2.1712,
      "step": 10647
    },
    {
      "epoch": 0.07234620253352692,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009896577371110774,
      "loss": 2.1611,
      "step": 10648
    },
    {
      "epoch": 0.07235299688012098,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009896555601130736,
      "loss": 2.222,
      "step": 10649
    },
    {
      "epoch": 0.07235979122671503,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000989653382888365,
      "loss": 2.3306,
      "step": 10650
    },
    {
      "epoch": 0.07236658557330909,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009896512054369526,
      "loss": 2.4066,
      "step": 10651
    },
    {
      "epoch": 0.07237337991990314,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009896490277588373,
      "loss": 2.296,
      "step": 10652
    },
    {
      "epoch": 0.0723801742664972,
      "grad_norm": 0.140625,
      "learning_rate": 0.00098964684985402,
      "loss": 2.2482,
      "step": 10653
    },
    {
      "epoch": 0.07238696861309127,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009896446717225019,
      "loss": 2.1867,
      "step": 10654
    },
    {
      "epoch": 0.07239376295968532,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009896424933642839,
      "loss": 2.1648,
      "step": 10655
    },
    {
      "epoch": 0.07240055730627938,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000989640314779367,
      "loss": 2.4764,
      "step": 10656
    },
    {
      "epoch": 0.07240735165287343,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009896381359677522,
      "loss": 2.0485,
      "step": 10657
    },
    {
      "epoch": 0.0724141459994675,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009896359569294406,
      "loss": 2.2458,
      "step": 10658
    },
    {
      "epoch": 0.07242094034606156,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009896337776644332,
      "loss": 2.2246,
      "step": 10659
    },
    {
      "epoch": 0.07242773469265561,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009896315981727311,
      "loss": 2.2821,
      "step": 10660
    },
    {
      "epoch": 0.07243452903924967,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000989629418454335,
      "loss": 2.1659,
      "step": 10661
    },
    {
      "epoch": 0.07244132338584372,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009896272385092463,
      "loss": 2.1214,
      "step": 10662
    },
    {
      "epoch": 0.07244811773243778,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009896250583374656,
      "loss": 2.1593,
      "step": 10663
    },
    {
      "epoch": 0.07245491207903183,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009896228779389942,
      "loss": 2.2571,
      "step": 10664
    },
    {
      "epoch": 0.0724617064256259,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000989620697313833,
      "loss": 2.4103,
      "step": 10665
    },
    {
      "epoch": 0.07246850077221996,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000989618516461983,
      "loss": 2.2588,
      "step": 10666
    },
    {
      "epoch": 0.07247529511881401,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009896163353834452,
      "loss": 2.2981,
      "step": 10667
    },
    {
      "epoch": 0.07248208946540807,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009896141540782209,
      "loss": 2.3061,
      "step": 10668
    },
    {
      "epoch": 0.07248888381200212,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009896119725463107,
      "loss": 2.1936,
      "step": 10669
    },
    {
      "epoch": 0.07249567815859619,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009896097907877157,
      "loss": 2.2959,
      "step": 10670
    },
    {
      "epoch": 0.07250247250519025,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009896076088024368,
      "loss": 2.2993,
      "step": 10671
    },
    {
      "epoch": 0.0725092668517843,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009896054265904756,
      "loss": 2.2685,
      "step": 10672
    },
    {
      "epoch": 0.07251606119837836,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009896032441518324,
      "loss": 2.3587,
      "step": 10673
    },
    {
      "epoch": 0.07252285554497241,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009896010614865087,
      "loss": 2.1719,
      "step": 10674
    },
    {
      "epoch": 0.07252964989156648,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009895988785945053,
      "loss": 2.244,
      "step": 10675
    },
    {
      "epoch": 0.07253644423816054,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009895966954758231,
      "loss": 2.1265,
      "step": 10676
    },
    {
      "epoch": 0.07254323858475459,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009895945121304634,
      "loss": 2.1672,
      "step": 10677
    },
    {
      "epoch": 0.07255003293134865,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009895923285584269,
      "loss": 2.1728,
      "step": 10678
    },
    {
      "epoch": 0.0725568272779427,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000989590144759715,
      "loss": 2.1886,
      "step": 10679
    },
    {
      "epoch": 0.07256362162453676,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009895879607343281,
      "loss": 2.3404,
      "step": 10680
    },
    {
      "epoch": 0.07257041597113083,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009895857764822679,
      "loss": 2.2562,
      "step": 10681
    },
    {
      "epoch": 0.07257721031772488,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000989583592003535,
      "loss": 2.0469,
      "step": 10682
    },
    {
      "epoch": 0.07258400466431894,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009895814072981306,
      "loss": 2.1026,
      "step": 10683
    },
    {
      "epoch": 0.07259079901091299,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009895792223660555,
      "loss": 2.2401,
      "step": 10684
    },
    {
      "epoch": 0.07259759335750705,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009895770372073107,
      "loss": 2.2921,
      "step": 10685
    },
    {
      "epoch": 0.0726043877041011,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009895748518218977,
      "loss": 2.3247,
      "step": 10686
    },
    {
      "epoch": 0.07261118205069517,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009895726662098169,
      "loss": 2.2884,
      "step": 10687
    },
    {
      "epoch": 0.07261797639728923,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009895704803710697,
      "loss": 2.5238,
      "step": 10688
    },
    {
      "epoch": 0.07262477074388328,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009895682943056568,
      "loss": 2.1856,
      "step": 10689
    },
    {
      "epoch": 0.07263156509047734,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009895661080135795,
      "loss": 2.3696,
      "step": 10690
    },
    {
      "epoch": 0.07263835943707139,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000989563921494839,
      "loss": 2.2363,
      "step": 10691
    },
    {
      "epoch": 0.07264515378366546,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009895617347494355,
      "loss": 2.1264,
      "step": 10692
    },
    {
      "epoch": 0.07265194813025952,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009895595477773709,
      "loss": 2.1355,
      "step": 10693
    },
    {
      "epoch": 0.07265874247685357,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009895573605786456,
      "loss": 2.3336,
      "step": 10694
    },
    {
      "epoch": 0.07266553682344763,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009895551731532609,
      "loss": 2.1365,
      "step": 10695
    },
    {
      "epoch": 0.07267233117004168,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009895529855012178,
      "loss": 2.2245,
      "step": 10696
    },
    {
      "epoch": 0.07267912551663575,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009895507976225174,
      "loss": 2.1491,
      "step": 10697
    },
    {
      "epoch": 0.07268591986322981,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009895486095171605,
      "loss": 2.3131,
      "step": 10698
    },
    {
      "epoch": 0.07269271420982386,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009895464211851482,
      "loss": 2.4007,
      "step": 10699
    },
    {
      "epoch": 0.07269950855641792,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009895442326264814,
      "loss": 2.1883,
      "step": 10700
    },
    {
      "epoch": 0.07270630290301197,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009895420438411615,
      "loss": 2.1985,
      "step": 10701
    },
    {
      "epoch": 0.07271309724960603,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000989539854829189,
      "loss": 2.1703,
      "step": 10702
    },
    {
      "epoch": 0.0727198915962001,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009895376655905653,
      "loss": 2.3437,
      "step": 10703
    },
    {
      "epoch": 0.07272668594279415,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009895354761252913,
      "loss": 2.2916,
      "step": 10704
    },
    {
      "epoch": 0.07273348028938821,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000989533286433368,
      "loss": 2.1083,
      "step": 10705
    },
    {
      "epoch": 0.07274027463598226,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009895310965147962,
      "loss": 2.2289,
      "step": 10706
    },
    {
      "epoch": 0.07274706898257632,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009895289063695772,
      "loss": 2.3447,
      "step": 10707
    },
    {
      "epoch": 0.07275386332917039,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000989526715997712,
      "loss": 2.2818,
      "step": 10708
    },
    {
      "epoch": 0.07276065767576444,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009895245253992017,
      "loss": 2.1159,
      "step": 10709
    },
    {
      "epoch": 0.0727674520223585,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009895223345740467,
      "loss": 2.19,
      "step": 10710
    },
    {
      "epoch": 0.07277424636895255,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009895201435222488,
      "loss": 2.3551,
      "step": 10711
    },
    {
      "epoch": 0.07278104071554661,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009895179522438086,
      "loss": 2.3968,
      "step": 10712
    },
    {
      "epoch": 0.07278783506214066,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009895157607387273,
      "loss": 2.1266,
      "step": 10713
    },
    {
      "epoch": 0.07279462940873473,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009895135690070056,
      "loss": 2.2168,
      "step": 10714
    },
    {
      "epoch": 0.07280142375532879,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009895113770486451,
      "loss": 2.293,
      "step": 10715
    },
    {
      "epoch": 0.07280821810192284,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000989509184863646,
      "loss": 2.2486,
      "step": 10716
    },
    {
      "epoch": 0.0728150124485169,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009895069924520101,
      "loss": 2.3108,
      "step": 10717
    },
    {
      "epoch": 0.07282180679511095,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009895047998137378,
      "loss": 2.287,
      "step": 10718
    },
    {
      "epoch": 0.07282860114170502,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009895026069488306,
      "loss": 2.0787,
      "step": 10719
    },
    {
      "epoch": 0.07283539548829908,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009895004138572893,
      "loss": 2.2492,
      "step": 10720
    },
    {
      "epoch": 0.07284218983489313,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009894982205391149,
      "loss": 2.1926,
      "step": 10721
    },
    {
      "epoch": 0.07284898418148719,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009894960269943083,
      "loss": 2.2386,
      "step": 10722
    },
    {
      "epoch": 0.07285577852808124,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009894938332228708,
      "loss": 2.2373,
      "step": 10723
    },
    {
      "epoch": 0.0728625728746753,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009894916392248032,
      "loss": 2.2022,
      "step": 10724
    },
    {
      "epoch": 0.07286936722126937,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009894894450001067,
      "loss": 2.2129,
      "step": 10725
    },
    {
      "epoch": 0.07287616156786342,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000989487250548782,
      "loss": 2.1321,
      "step": 10726
    },
    {
      "epoch": 0.07288295591445748,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009894850558708303,
      "loss": 2.2111,
      "step": 10727
    },
    {
      "epoch": 0.07288975026105153,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009894828609662528,
      "loss": 2.2896,
      "step": 10728
    },
    {
      "epoch": 0.0728965446076456,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009894806658350503,
      "loss": 2.3183,
      "step": 10729
    },
    {
      "epoch": 0.07290333895423966,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000989478470477224,
      "loss": 2.1578,
      "step": 10730
    },
    {
      "epoch": 0.07291013330083371,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009894762748927745,
      "loss": 2.2381,
      "step": 10731
    },
    {
      "epoch": 0.07291692764742777,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009894740790817032,
      "loss": 2.2647,
      "step": 10732
    },
    {
      "epoch": 0.07292372199402182,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000989471883044011,
      "loss": 2.213,
      "step": 10733
    },
    {
      "epoch": 0.07293051634061588,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000989469686779699,
      "loss": 2.2487,
      "step": 10734
    },
    {
      "epoch": 0.07293731068720993,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000989467490288768,
      "loss": 2.397,
      "step": 10735
    },
    {
      "epoch": 0.072944105033804,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009894652935712193,
      "loss": 2.1505,
      "step": 10736
    },
    {
      "epoch": 0.07295089938039806,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000989463096627054,
      "loss": 2.2943,
      "step": 10737
    },
    {
      "epoch": 0.07295769372699211,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009894608994562726,
      "loss": 2.2818,
      "step": 10738
    },
    {
      "epoch": 0.07296448807358617,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009894587020588764,
      "loss": 2.164,
      "step": 10739
    },
    {
      "epoch": 0.07297128242018022,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009894565044348666,
      "loss": 2.4188,
      "step": 10740
    },
    {
      "epoch": 0.07297807676677429,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009894543065842437,
      "loss": 2.281,
      "step": 10741
    },
    {
      "epoch": 0.07298487111336835,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009894521085070094,
      "loss": 2.3801,
      "step": 10742
    },
    {
      "epoch": 0.0729916654599624,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009894499102031645,
      "loss": 2.3312,
      "step": 10743
    },
    {
      "epoch": 0.07299845980655646,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009894477116727095,
      "loss": 2.3587,
      "step": 10744
    },
    {
      "epoch": 0.07300525415315051,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009894455129156462,
      "loss": 2.2655,
      "step": 10745
    },
    {
      "epoch": 0.07301204849974458,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000989443313931975,
      "loss": 2.2322,
      "step": 10746
    },
    {
      "epoch": 0.07301884284633864,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009894411147216974,
      "loss": 2.1646,
      "step": 10747
    },
    {
      "epoch": 0.07302563719293269,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000989438915284814,
      "loss": 2.1745,
      "step": 10748
    },
    {
      "epoch": 0.07303243153952675,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009894367156213259,
      "loss": 2.3068,
      "step": 10749
    },
    {
      "epoch": 0.0730392258861208,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009894345157312344,
      "loss": 2.2003,
      "step": 10750
    },
    {
      "epoch": 0.07304602023271486,
      "grad_norm": 0.150390625,
      "learning_rate": 0.00098943231561454,
      "loss": 2.2723,
      "step": 10751
    },
    {
      "epoch": 0.07305281457930893,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009894301152712443,
      "loss": 2.171,
      "step": 10752
    },
    {
      "epoch": 0.07305960892590298,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000989427914701348,
      "loss": 2.219,
      "step": 10753
    },
    {
      "epoch": 0.07306640327249704,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009894257139048523,
      "loss": 2.139,
      "step": 10754
    },
    {
      "epoch": 0.07307319761909109,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009894235128817582,
      "loss": 2.2301,
      "step": 10755
    },
    {
      "epoch": 0.07307999196568515,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009894213116320664,
      "loss": 2.1781,
      "step": 10756
    },
    {
      "epoch": 0.07308678631227922,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009894191101557782,
      "loss": 2.092,
      "step": 10757
    },
    {
      "epoch": 0.07309358065887327,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009894169084528946,
      "loss": 2.3099,
      "step": 10758
    },
    {
      "epoch": 0.07310037500546733,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009894147065234167,
      "loss": 2.3114,
      "step": 10759
    },
    {
      "epoch": 0.07310716935206138,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009894125043673453,
      "loss": 2.1947,
      "step": 10760
    },
    {
      "epoch": 0.07311396369865544,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009894103019846812,
      "loss": 2.183,
      "step": 10761
    },
    {
      "epoch": 0.07312075804524949,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009894080993754263,
      "loss": 2.4206,
      "step": 10762
    },
    {
      "epoch": 0.07312755239184356,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009894058965395807,
      "loss": 2.2938,
      "step": 10763
    },
    {
      "epoch": 0.07313434673843762,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009894036934771457,
      "loss": 2.2807,
      "step": 10764
    },
    {
      "epoch": 0.07314114108503167,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009894014901881227,
      "loss": 2.2487,
      "step": 10765
    },
    {
      "epoch": 0.07314793543162573,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009893992866725123,
      "loss": 2.2296,
      "step": 10766
    },
    {
      "epoch": 0.07315472977821978,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009893970829303155,
      "loss": 2.2545,
      "step": 10767
    },
    {
      "epoch": 0.07316152412481385,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009893948789615338,
      "loss": 2.2583,
      "step": 10768
    },
    {
      "epoch": 0.07316831847140791,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009893926747661674,
      "loss": 2.1931,
      "step": 10769
    },
    {
      "epoch": 0.07317511281800196,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009893904703442181,
      "loss": 2.1585,
      "step": 10770
    },
    {
      "epoch": 0.07318190716459602,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009893882656956867,
      "loss": 2.2083,
      "step": 10771
    },
    {
      "epoch": 0.07318870151119007,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000989386060820574,
      "loss": 2.1774,
      "step": 10772
    },
    {
      "epoch": 0.07319549585778413,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009893838557188813,
      "loss": 2.2386,
      "step": 10773
    },
    {
      "epoch": 0.0732022902043782,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009893816503906093,
      "loss": 2.2035,
      "step": 10774
    },
    {
      "epoch": 0.07320908455097225,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009893794448357595,
      "loss": 2.3295,
      "step": 10775
    },
    {
      "epoch": 0.07321587889756631,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009893772390543324,
      "loss": 2.3863,
      "step": 10776
    },
    {
      "epoch": 0.07322267324416036,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009893750330463291,
      "loss": 2.3066,
      "step": 10777
    },
    {
      "epoch": 0.07322946759075442,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000989372826811751,
      "loss": 2.1959,
      "step": 10778
    },
    {
      "epoch": 0.07323626193734849,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009893706203505988,
      "loss": 2.1877,
      "step": 10779
    },
    {
      "epoch": 0.07324305628394254,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009893684136628737,
      "loss": 2.2587,
      "step": 10780
    },
    {
      "epoch": 0.0732498506305366,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009893662067485764,
      "loss": 2.0905,
      "step": 10781
    },
    {
      "epoch": 0.07325664497713065,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009893639996077085,
      "loss": 2.2959,
      "step": 10782
    },
    {
      "epoch": 0.07326343932372471,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009893617922402704,
      "loss": 2.069,
      "step": 10783
    },
    {
      "epoch": 0.07327023367031876,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009893595846462637,
      "loss": 2.2103,
      "step": 10784
    },
    {
      "epoch": 0.07327702801691283,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009893573768256889,
      "loss": 2.3299,
      "step": 10785
    },
    {
      "epoch": 0.07328382236350689,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009893551687785473,
      "loss": 2.3659,
      "step": 10786
    },
    {
      "epoch": 0.07329061671010094,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009893529605048397,
      "loss": 2.1334,
      "step": 10787
    },
    {
      "epoch": 0.073297411056695,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009893507520045675,
      "loss": 2.1023,
      "step": 10788
    },
    {
      "epoch": 0.07330420540328905,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009893485432777315,
      "loss": 2.3537,
      "step": 10789
    },
    {
      "epoch": 0.07331099974988312,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009893463343243326,
      "loss": 2.2329,
      "step": 10790
    },
    {
      "epoch": 0.07331779409647718,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009893441251443722,
      "loss": 2.2025,
      "step": 10791
    },
    {
      "epoch": 0.07332458844307123,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000989341915737851,
      "loss": 2.1662,
      "step": 10792
    },
    {
      "epoch": 0.07333138278966529,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009893397061047698,
      "loss": 2.2017,
      "step": 10793
    },
    {
      "epoch": 0.07333817713625934,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009893374962451303,
      "loss": 2.2327,
      "step": 10794
    },
    {
      "epoch": 0.0733449714828534,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000989335286158933,
      "loss": 2.2428,
      "step": 10795
    },
    {
      "epoch": 0.07335176582944747,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000989333075846179,
      "loss": 2.1062,
      "step": 10796
    },
    {
      "epoch": 0.07335856017604152,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009893308653068696,
      "loss": 2.2148,
      "step": 10797
    },
    {
      "epoch": 0.07336535452263558,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009893286545410054,
      "loss": 2.0702,
      "step": 10798
    },
    {
      "epoch": 0.07337214886922963,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009893264435485876,
      "loss": 2.2537,
      "step": 10799
    },
    {
      "epoch": 0.0733789432158237,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009893242323296175,
      "loss": 2.2552,
      "step": 10800
    },
    {
      "epoch": 0.07338573756241776,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009893220208840958,
      "loss": 2.2098,
      "step": 10801
    },
    {
      "epoch": 0.07339253190901181,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009893198092120238,
      "loss": 2.1797,
      "step": 10802
    },
    {
      "epoch": 0.07339932625560587,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000989317597313402,
      "loss": 2.26,
      "step": 10803
    },
    {
      "epoch": 0.07340612060219992,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000989315385188232,
      "loss": 2.1123,
      "step": 10804
    },
    {
      "epoch": 0.07341291494879398,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009893131728365143,
      "loss": 2.2786,
      "step": 10805
    },
    {
      "epoch": 0.07341970929538803,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009893109602582505,
      "loss": 2.3092,
      "step": 10806
    },
    {
      "epoch": 0.0734265036419821,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009893087474534413,
      "loss": 2.3056,
      "step": 10807
    },
    {
      "epoch": 0.07343329798857616,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009893065344220877,
      "loss": 2.3329,
      "step": 10808
    },
    {
      "epoch": 0.07344009233517021,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009893043211641906,
      "loss": 2.0296,
      "step": 10809
    },
    {
      "epoch": 0.07344688668176427,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009893021076797515,
      "loss": 2.1951,
      "step": 10810
    },
    {
      "epoch": 0.07345368102835832,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009892998939687709,
      "loss": 2.1595,
      "step": 10811
    },
    {
      "epoch": 0.07346047537495239,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009892976800312504,
      "loss": 2.2616,
      "step": 10812
    },
    {
      "epoch": 0.07346726972154645,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009892954658671904,
      "loss": 2.076,
      "step": 10813
    },
    {
      "epoch": 0.0734740640681405,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009892932514765922,
      "loss": 2.3228,
      "step": 10814
    },
    {
      "epoch": 0.07348085841473456,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009892910368594569,
      "loss": 2.2079,
      "step": 10815
    },
    {
      "epoch": 0.07348765276132861,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009892888220157856,
      "loss": 2.3013,
      "step": 10816
    },
    {
      "epoch": 0.07349444710792268,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009892866069455791,
      "loss": 2.1631,
      "step": 10817
    },
    {
      "epoch": 0.07350124145451674,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009892843916488384,
      "loss": 2.1385,
      "step": 10818
    },
    {
      "epoch": 0.07350803580111079,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009892821761255646,
      "loss": 2.037,
      "step": 10819
    },
    {
      "epoch": 0.07351483014770485,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000989279960375759,
      "loss": 2.2319,
      "step": 10820
    },
    {
      "epoch": 0.0735216244942989,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000989277744399422,
      "loss": 2.2915,
      "step": 10821
    },
    {
      "epoch": 0.07352841884089296,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009892755281965555,
      "loss": 2.2534,
      "step": 10822
    },
    {
      "epoch": 0.07353521318748703,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009892733117671598,
      "loss": 2.3958,
      "step": 10823
    },
    {
      "epoch": 0.07354200753408108,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000989271095111236,
      "loss": 2.2078,
      "step": 10824
    },
    {
      "epoch": 0.07354880188067514,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009892688782287854,
      "loss": 2.3807,
      "step": 10825
    },
    {
      "epoch": 0.07355559622726919,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009892666611198092,
      "loss": 1.9566,
      "step": 10826
    },
    {
      "epoch": 0.07356239057386325,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009892644437843078,
      "loss": 2.2812,
      "step": 10827
    },
    {
      "epoch": 0.07356918492045732,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009892622262222827,
      "loss": 2.4237,
      "step": 10828
    },
    {
      "epoch": 0.07357597926705137,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009892600084337349,
      "loss": 2.2628,
      "step": 10829
    },
    {
      "epoch": 0.07358277361364543,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000989257790418665,
      "loss": 2.1293,
      "step": 10830
    },
    {
      "epoch": 0.07358956796023948,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009892555721770747,
      "loss": 2.3851,
      "step": 10831
    },
    {
      "epoch": 0.07359636230683354,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009892533537089645,
      "loss": 2.2362,
      "step": 10832
    },
    {
      "epoch": 0.07360315665342759,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009892511350143357,
      "loss": 2.3901,
      "step": 10833
    },
    {
      "epoch": 0.07360995100002166,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009892489160931892,
      "loss": 2.1704,
      "step": 10834
    },
    {
      "epoch": 0.07361674534661572,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000989246696945526,
      "loss": 2.2766,
      "step": 10835
    },
    {
      "epoch": 0.07362353969320977,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009892444775713473,
      "loss": 2.2386,
      "step": 10836
    },
    {
      "epoch": 0.07363033403980383,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000989242257970654,
      "loss": 2.1732,
      "step": 10837
    },
    {
      "epoch": 0.07363712838639788,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009892400381434471,
      "loss": 2.1304,
      "step": 10838
    },
    {
      "epoch": 0.07364392273299195,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009892378180897277,
      "loss": 2.1495,
      "step": 10839
    },
    {
      "epoch": 0.07365071707958601,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009892355978094968,
      "loss": 2.3967,
      "step": 10840
    },
    {
      "epoch": 0.07365751142618006,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009892333773027554,
      "loss": 2.3194,
      "step": 10841
    },
    {
      "epoch": 0.07366430577277412,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009892311565695045,
      "loss": 2.2644,
      "step": 10842
    },
    {
      "epoch": 0.07367110011936817,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009892289356097453,
      "loss": 2.1701,
      "step": 10843
    },
    {
      "epoch": 0.07367789446596223,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009892267144234787,
      "loss": 2.3063,
      "step": 10844
    },
    {
      "epoch": 0.0736846888125563,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009892244930107055,
      "loss": 2.2921,
      "step": 10845
    },
    {
      "epoch": 0.07369148315915035,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009892222713714272,
      "loss": 2.2504,
      "step": 10846
    },
    {
      "epoch": 0.07369827750574441,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009892200495056444,
      "loss": 2.4104,
      "step": 10847
    },
    {
      "epoch": 0.07370507185233846,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009892178274133587,
      "loss": 2.1956,
      "step": 10848
    },
    {
      "epoch": 0.07371186619893252,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009892156050945705,
      "loss": 2.3539,
      "step": 10849
    },
    {
      "epoch": 0.07371866054552659,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000989213382549281,
      "loss": 2.3117,
      "step": 10850
    },
    {
      "epoch": 0.07372545489212064,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009892111597774916,
      "loss": 2.3048,
      "step": 10851
    },
    {
      "epoch": 0.0737322492387147,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009892089367792027,
      "loss": 2.4134,
      "step": 10852
    },
    {
      "epoch": 0.07373904358530875,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000989206713554416,
      "loss": 2.3071,
      "step": 10853
    },
    {
      "epoch": 0.07374583793190281,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000989204490103132,
      "loss": 2.2782,
      "step": 10854
    },
    {
      "epoch": 0.07375263227849686,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009892022664253518,
      "loss": 2.29,
      "step": 10855
    },
    {
      "epoch": 0.07375942662509093,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000989200042521077,
      "loss": 2.347,
      "step": 10856
    },
    {
      "epoch": 0.07376622097168499,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009891978183903078,
      "loss": 2.2698,
      "step": 10857
    },
    {
      "epoch": 0.07377301531827904,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009891955940330456,
      "loss": 2.3739,
      "step": 10858
    },
    {
      "epoch": 0.0737798096648731,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009891933694492917,
      "loss": 2.3536,
      "step": 10859
    },
    {
      "epoch": 0.07378660401146715,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009891911446390468,
      "loss": 2.2197,
      "step": 10860
    },
    {
      "epoch": 0.07379339835806122,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009891889196023118,
      "loss": 1.9653,
      "step": 10861
    },
    {
      "epoch": 0.07380019270465528,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009891866943390881,
      "loss": 2.2449,
      "step": 10862
    },
    {
      "epoch": 0.07380698705124933,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009891844688493768,
      "loss": 2.3002,
      "step": 10863
    },
    {
      "epoch": 0.07381378139784339,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009891822431331784,
      "loss": 2.2817,
      "step": 10864
    },
    {
      "epoch": 0.07382057574443744,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009891800171904943,
      "loss": 2.2212,
      "step": 10865
    },
    {
      "epoch": 0.0738273700910315,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009891777910213256,
      "loss": 2.1533,
      "step": 10866
    },
    {
      "epoch": 0.07383416443762557,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000989175564625673,
      "loss": 2.278,
      "step": 10867
    },
    {
      "epoch": 0.07384095878421962,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009891733380035378,
      "loss": 2.279,
      "step": 10868
    },
    {
      "epoch": 0.07384775313081368,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009891711111549208,
      "loss": 2.2426,
      "step": 10869
    },
    {
      "epoch": 0.07385454747740773,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009891688840798234,
      "loss": 2.2513,
      "step": 10870
    },
    {
      "epoch": 0.0738613418240018,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009891666567782465,
      "loss": 2.3241,
      "step": 10871
    },
    {
      "epoch": 0.07386813617059586,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009891644292501906,
      "loss": 2.1506,
      "step": 10872
    },
    {
      "epoch": 0.07387493051718991,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009891622014956576,
      "loss": 2.1698,
      "step": 10873
    },
    {
      "epoch": 0.07388172486378397,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009891599735146479,
      "loss": 2.2729,
      "step": 10874
    },
    {
      "epoch": 0.07388851921037802,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009891577453071628,
      "loss": 2.2272,
      "step": 10875
    },
    {
      "epoch": 0.07389531355697208,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009891555168732033,
      "loss": 2.2519,
      "step": 10876
    },
    {
      "epoch": 0.07390210790356615,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009891532882127705,
      "loss": 2.1215,
      "step": 10877
    },
    {
      "epoch": 0.0739089022501602,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009891510593258651,
      "loss": 2.4061,
      "step": 10878
    },
    {
      "epoch": 0.07391569659675426,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009891488302124884,
      "loss": 2.1328,
      "step": 10879
    },
    {
      "epoch": 0.07392249094334831,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009891466008726416,
      "loss": 2.2969,
      "step": 10880
    },
    {
      "epoch": 0.07392928528994237,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009891443713063255,
      "loss": 2.227,
      "step": 10881
    },
    {
      "epoch": 0.07393607963653642,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000989142141513541,
      "loss": 2.1934,
      "step": 10882
    },
    {
      "epoch": 0.07394287398313049,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009891399114942893,
      "loss": 2.2414,
      "step": 10883
    },
    {
      "epoch": 0.07394966832972455,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009891376812485714,
      "loss": 2.3639,
      "step": 10884
    },
    {
      "epoch": 0.0739564626763186,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009891354507763886,
      "loss": 2.3498,
      "step": 10885
    },
    {
      "epoch": 0.07396325702291266,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009891332200777417,
      "loss": 2.2628,
      "step": 10886
    },
    {
      "epoch": 0.07397005136950671,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009891309891526315,
      "loss": 2.1669,
      "step": 10887
    },
    {
      "epoch": 0.07397684571610078,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009891287580010592,
      "loss": 2.174,
      "step": 10888
    },
    {
      "epoch": 0.07398364006269484,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009891265266230261,
      "loss": 2.1391,
      "step": 10889
    },
    {
      "epoch": 0.07399043440928889,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009891242950185331,
      "loss": 2.1264,
      "step": 10890
    },
    {
      "epoch": 0.07399722875588295,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000989122063187581,
      "loss": 2.3207,
      "step": 10891
    },
    {
      "epoch": 0.074004023102477,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000989119831130171,
      "loss": 2.1703,
      "step": 10892
    },
    {
      "epoch": 0.07401081744907106,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009891175988463042,
      "loss": 2.1796,
      "step": 10893
    },
    {
      "epoch": 0.07401761179566513,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009891153663359813,
      "loss": 2.3326,
      "step": 10894
    },
    {
      "epoch": 0.07402440614225918,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009891131335992038,
      "loss": 2.2452,
      "step": 10895
    },
    {
      "epoch": 0.07403120048885324,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009891109006359726,
      "loss": 2.2589,
      "step": 10896
    },
    {
      "epoch": 0.07403799483544729,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009891086674462886,
      "loss": 2.1152,
      "step": 10897
    },
    {
      "epoch": 0.07404478918204135,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000989106434030153,
      "loss": 2.3244,
      "step": 10898
    },
    {
      "epoch": 0.07405158352863542,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009891042003875665,
      "loss": 2.2671,
      "step": 10899
    },
    {
      "epoch": 0.07405837787522947,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009891019665185304,
      "loss": 2.2565,
      "step": 10900
    },
    {
      "epoch": 0.07406517222182353,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009890997324230458,
      "loss": 2.1895,
      "step": 10901
    },
    {
      "epoch": 0.07407196656841758,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009890974981011135,
      "loss": 2.2925,
      "step": 10902
    },
    {
      "epoch": 0.07407876091501164,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000989095263552735,
      "loss": 2.3303,
      "step": 10903
    },
    {
      "epoch": 0.07408555526160569,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009890930287779105,
      "loss": 2.0997,
      "step": 10904
    },
    {
      "epoch": 0.07409234960819976,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009890907937766418,
      "loss": 2.1815,
      "step": 10905
    },
    {
      "epoch": 0.07409914395479382,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009890885585489297,
      "loss": 2.1296,
      "step": 10906
    },
    {
      "epoch": 0.07410593830138787,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009890863230947752,
      "loss": 2.13,
      "step": 10907
    },
    {
      "epoch": 0.07411273264798193,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009890840874141793,
      "loss": 2.1644,
      "step": 10908
    },
    {
      "epoch": 0.07411952699457598,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009890818515071432,
      "loss": 2.0198,
      "step": 10909
    },
    {
      "epoch": 0.07412632134117005,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009890796153736677,
      "loss": 2.2715,
      "step": 10910
    },
    {
      "epoch": 0.07413311568776411,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009890773790137539,
      "loss": 2.2099,
      "step": 10911
    },
    {
      "epoch": 0.07413991003435816,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009890751424274028,
      "loss": 2.2166,
      "step": 10912
    },
    {
      "epoch": 0.07414670438095222,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009890729056146158,
      "loss": 2.2064,
      "step": 10913
    },
    {
      "epoch": 0.07415349872754627,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009890706685753935,
      "loss": 2.3154,
      "step": 10914
    },
    {
      "epoch": 0.07416029307414033,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000989068431309737,
      "loss": 2.2374,
      "step": 10915
    },
    {
      "epoch": 0.0741670874207344,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009890661938176475,
      "loss": 2.2784,
      "step": 10916
    },
    {
      "epoch": 0.07417388176732845,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000989063956099126,
      "loss": 2.3851,
      "step": 10917
    },
    {
      "epoch": 0.07418067611392251,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009890617181541733,
      "loss": 2.112,
      "step": 10918
    },
    {
      "epoch": 0.07418747046051656,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009890594799827907,
      "loss": 2.1981,
      "step": 10919
    },
    {
      "epoch": 0.07419426480711062,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009890572415849795,
      "loss": 2.2123,
      "step": 10920
    },
    {
      "epoch": 0.07420105915370469,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009890550029607401,
      "loss": 2.2053,
      "step": 10921
    },
    {
      "epoch": 0.07420785350029874,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009890527641100739,
      "loss": 2.1655,
      "step": 10922
    },
    {
      "epoch": 0.0742146478468928,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009890505250329817,
      "loss": 2.2731,
      "step": 10923
    },
    {
      "epoch": 0.07422144219348685,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000989048285729465,
      "loss": 2.2346,
      "step": 10924
    },
    {
      "epoch": 0.07422823654008091,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009890460461995244,
      "loss": 2.2896,
      "step": 10925
    },
    {
      "epoch": 0.07423503088667496,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009890438064431612,
      "loss": 2.1986,
      "step": 10926
    },
    {
      "epoch": 0.07424182523326903,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009890415664603762,
      "loss": 2.1588,
      "step": 10927
    },
    {
      "epoch": 0.07424861957986309,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009890393262511705,
      "loss": 2.2724,
      "step": 10928
    },
    {
      "epoch": 0.07425541392645714,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009890370858155453,
      "loss": 2.333,
      "step": 10929
    },
    {
      "epoch": 0.0742622082730512,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009890348451535016,
      "loss": 2.2989,
      "step": 10930
    },
    {
      "epoch": 0.07426900261964525,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009890326042650403,
      "loss": 2.3675,
      "step": 10931
    },
    {
      "epoch": 0.07427579696623932,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009890303631501626,
      "loss": 2.1507,
      "step": 10932
    },
    {
      "epoch": 0.07428259131283338,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009890281218088693,
      "loss": 2.1435,
      "step": 10933
    },
    {
      "epoch": 0.07428938565942743,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009890258802411618,
      "loss": 2.263,
      "step": 10934
    },
    {
      "epoch": 0.07429618000602149,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009890236384470408,
      "loss": 2.3875,
      "step": 10935
    },
    {
      "epoch": 0.07430297435261554,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009890213964265074,
      "loss": 2.2784,
      "step": 10936
    },
    {
      "epoch": 0.0743097686992096,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009890191541795628,
      "loss": 2.097,
      "step": 10937
    },
    {
      "epoch": 0.07431656304580367,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009890169117062078,
      "loss": 2.2596,
      "step": 10938
    },
    {
      "epoch": 0.07432335739239772,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009890146690064436,
      "loss": 2.1611,
      "step": 10939
    },
    {
      "epoch": 0.07433015173899178,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009890124260802714,
      "loss": 2.2023,
      "step": 10940
    },
    {
      "epoch": 0.07433694608558583,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000989010182927692,
      "loss": 2.2461,
      "step": 10941
    },
    {
      "epoch": 0.0743437404321799,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009890079395487063,
      "loss": 2.301,
      "step": 10942
    },
    {
      "epoch": 0.07435053477877396,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009890056959433156,
      "loss": 2.2933,
      "step": 10943
    },
    {
      "epoch": 0.07435732912536801,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000989003452111521,
      "loss": 2.3363,
      "step": 10944
    },
    {
      "epoch": 0.07436412347196207,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009890012080533233,
      "loss": 2.3173,
      "step": 10945
    },
    {
      "epoch": 0.07437091781855612,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009889989637687235,
      "loss": 2.2859,
      "step": 10946
    },
    {
      "epoch": 0.07437771216515018,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000988996719257723,
      "loss": 2.3232,
      "step": 10947
    },
    {
      "epoch": 0.07438450651174425,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009889944745203228,
      "loss": 2.121,
      "step": 10948
    },
    {
      "epoch": 0.0743913008583383,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009889922295565235,
      "loss": 2.2958,
      "step": 10949
    },
    {
      "epoch": 0.07439809520493236,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009889899843663263,
      "loss": 2.3082,
      "step": 10950
    },
    {
      "epoch": 0.07440488955152641,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009889877389497325,
      "loss": 2.3911,
      "step": 10951
    },
    {
      "epoch": 0.07441168389812047,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000988985493306743,
      "loss": 2.3841,
      "step": 10952
    },
    {
      "epoch": 0.07441847824471452,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009889832474373587,
      "loss": 2.3649,
      "step": 10953
    },
    {
      "epoch": 0.07442527259130859,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000988981001341581,
      "loss": 2.3703,
      "step": 10954
    },
    {
      "epoch": 0.07443206693790265,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009889787550194107,
      "loss": 2.1813,
      "step": 10955
    },
    {
      "epoch": 0.0744388612844967,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009889765084708487,
      "loss": 2.37,
      "step": 10956
    },
    {
      "epoch": 0.07444565563109076,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009889742616958962,
      "loss": 2.2658,
      "step": 10957
    },
    {
      "epoch": 0.07445244997768481,
      "grad_norm": 0.140625,
      "learning_rate": 0.000988972014694554,
      "loss": 2.1083,
      "step": 10958
    },
    {
      "epoch": 0.07445924432427888,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009889697674668237,
      "loss": 2.2581,
      "step": 10959
    },
    {
      "epoch": 0.07446603867087294,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009889675200127057,
      "loss": 2.2315,
      "step": 10960
    },
    {
      "epoch": 0.07447283301746699,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009889652723322018,
      "loss": 2.2579,
      "step": 10961
    },
    {
      "epoch": 0.07447962736406105,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000988963024425312,
      "loss": 2.3007,
      "step": 10962
    },
    {
      "epoch": 0.0744864217106551,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009889607762920383,
      "loss": 2.2409,
      "step": 10963
    },
    {
      "epoch": 0.07449321605724916,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009889585279323815,
      "loss": 2.165,
      "step": 10964
    },
    {
      "epoch": 0.07450001040384323,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009889562793463423,
      "loss": 2.2588,
      "step": 10965
    },
    {
      "epoch": 0.07450680475043728,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009889540305339218,
      "loss": 2.2289,
      "step": 10966
    },
    {
      "epoch": 0.07451359909703134,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009889517814951215,
      "loss": 2.2832,
      "step": 10967
    },
    {
      "epoch": 0.07452039344362539,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009889495322299418,
      "loss": 2.2213,
      "step": 10968
    },
    {
      "epoch": 0.07452718779021945,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009889472827383844,
      "loss": 2.1587,
      "step": 10969
    },
    {
      "epoch": 0.07453398213681352,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009889450330204498,
      "loss": 2.2123,
      "step": 10970
    },
    {
      "epoch": 0.07454077648340757,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009889427830761393,
      "loss": 2.2195,
      "step": 10971
    },
    {
      "epoch": 0.07454757083000163,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000988940532905454,
      "loss": 2.2339,
      "step": 10972
    },
    {
      "epoch": 0.07455436517659568,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009889382825083947,
      "loss": 2.2213,
      "step": 10973
    },
    {
      "epoch": 0.07456115952318974,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009889360318849627,
      "loss": 2.3496,
      "step": 10974
    },
    {
      "epoch": 0.07456795386978379,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009889337810351589,
      "loss": 2.2886,
      "step": 10975
    },
    {
      "epoch": 0.07457474821637786,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009889315299589842,
      "loss": 2.1972,
      "step": 10976
    },
    {
      "epoch": 0.07458154256297192,
      "grad_norm": 0.125,
      "learning_rate": 0.0009889292786564399,
      "loss": 2.0325,
      "step": 10977
    },
    {
      "epoch": 0.07458833690956597,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009889270271275271,
      "loss": 2.3036,
      "step": 10978
    },
    {
      "epoch": 0.07459513125616003,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009889247753722464,
      "loss": 2.1217,
      "step": 10979
    },
    {
      "epoch": 0.07460192560275408,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009889225233905995,
      "loss": 2.1879,
      "step": 10980
    },
    {
      "epoch": 0.07460871994934815,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009889202711825867,
      "loss": 2.2251,
      "step": 10981
    },
    {
      "epoch": 0.07461551429594221,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009889180187482097,
      "loss": 2.1267,
      "step": 10982
    },
    {
      "epoch": 0.07462230864253626,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009889157660874692,
      "loss": 2.3643,
      "step": 10983
    },
    {
      "epoch": 0.07462910298913032,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009889135132003662,
      "loss": 2.1551,
      "step": 10984
    },
    {
      "epoch": 0.07463589733572437,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000988911260086902,
      "loss": 2.1906,
      "step": 10985
    },
    {
      "epoch": 0.07464269168231843,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009889090067470773,
      "loss": 2.2831,
      "step": 10986
    },
    {
      "epoch": 0.0746494860289125,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009889067531808934,
      "loss": 2.3347,
      "step": 10987
    },
    {
      "epoch": 0.07465628037550655,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009889044993883513,
      "loss": 2.2561,
      "step": 10988
    },
    {
      "epoch": 0.07466307472210061,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000988902245369452,
      "loss": 2.1918,
      "step": 10989
    },
    {
      "epoch": 0.07466986906869466,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009888999911241967,
      "loss": 2.2339,
      "step": 10990
    },
    {
      "epoch": 0.07467666341528872,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009888977366525862,
      "loss": 2.1607,
      "step": 10991
    },
    {
      "epoch": 0.07468345776188279,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009888954819546217,
      "loss": 2.223,
      "step": 10992
    },
    {
      "epoch": 0.07469025210847684,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000988893227030304,
      "loss": 2.3409,
      "step": 10993
    },
    {
      "epoch": 0.0746970464550709,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009888909718796346,
      "loss": 2.4734,
      "step": 10994
    },
    {
      "epoch": 0.07470384080166495,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009888887165026145,
      "loss": 2.3479,
      "step": 10995
    },
    {
      "epoch": 0.07471063514825901,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009888864608992441,
      "loss": 2.1066,
      "step": 10996
    },
    {
      "epoch": 0.07471742949485308,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000988884205069525,
      "loss": 2.2371,
      "step": 10997
    },
    {
      "epoch": 0.07472422384144713,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000988881949013458,
      "loss": 2.2164,
      "step": 10998
    },
    {
      "epoch": 0.07473101818804119,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009888796927310444,
      "loss": 2.101,
      "step": 10999
    },
    {
      "epoch": 0.07473781253463524,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000988877436222285,
      "loss": 2.1938,
      "step": 11000
    },
    {
      "epoch": 0.0747446068812293,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009888751794871811,
      "loss": 2.4517,
      "step": 11001
    },
    {
      "epoch": 0.07475140122782335,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009888729225257336,
      "loss": 2.3258,
      "step": 11002
    },
    {
      "epoch": 0.07475819557441742,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009888706653379434,
      "loss": 2.3411,
      "step": 11003
    },
    {
      "epoch": 0.07476498992101148,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009888684079238117,
      "loss": 2.0681,
      "step": 11004
    },
    {
      "epoch": 0.07477178426760553,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009888661502833395,
      "loss": 2.1805,
      "step": 11005
    },
    {
      "epoch": 0.07477857861419959,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000988863892416528,
      "loss": 2.2786,
      "step": 11006
    },
    {
      "epoch": 0.07478537296079364,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009888616343233783,
      "loss": 2.2532,
      "step": 11007
    },
    {
      "epoch": 0.0747921673073877,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000988859376003891,
      "loss": 2.2663,
      "step": 11008
    },
    {
      "epoch": 0.07479896165398177,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009888571174580675,
      "loss": 2.3136,
      "step": 11009
    },
    {
      "epoch": 0.07480575600057582,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009888548586859088,
      "loss": 2.3576,
      "step": 11010
    },
    {
      "epoch": 0.07481255034716988,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009888525996874157,
      "loss": 2.1928,
      "step": 11011
    },
    {
      "epoch": 0.07481934469376393,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009888503404625896,
      "loss": 2.3765,
      "step": 11012
    },
    {
      "epoch": 0.074826139040358,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009888480810114314,
      "loss": 2.211,
      "step": 11013
    },
    {
      "epoch": 0.07483293338695206,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009888458213339422,
      "loss": 2.3478,
      "step": 11014
    },
    {
      "epoch": 0.07483972773354611,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009888435614301228,
      "loss": 2.3308,
      "step": 11015
    },
    {
      "epoch": 0.07484652208014017,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009888413012999746,
      "loss": 2.2129,
      "step": 11016
    },
    {
      "epoch": 0.07485331642673422,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009888390409434984,
      "loss": 2.3931,
      "step": 11017
    },
    {
      "epoch": 0.07486011077332828,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009888367803606954,
      "loss": 2.156,
      "step": 11018
    },
    {
      "epoch": 0.07486690511992235,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009888345195515665,
      "loss": 2.2846,
      "step": 11019
    },
    {
      "epoch": 0.0748736994665164,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009888322585161128,
      "loss": 2.2032,
      "step": 11020
    },
    {
      "epoch": 0.07488049381311046,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009888299972543356,
      "loss": 2.3366,
      "step": 11021
    },
    {
      "epoch": 0.07488728815970451,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009888277357662353,
      "loss": 2.2611,
      "step": 11022
    },
    {
      "epoch": 0.07489408250629857,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009888254740518137,
      "loss": 2.1774,
      "step": 11023
    },
    {
      "epoch": 0.07490087685289262,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009888232121110713,
      "loss": 2.2292,
      "step": 11024
    },
    {
      "epoch": 0.07490767119948669,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009888209499440096,
      "loss": 2.2583,
      "step": 11025
    },
    {
      "epoch": 0.07491446554608075,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009888186875506293,
      "loss": 2.1828,
      "step": 11026
    },
    {
      "epoch": 0.0749212598926748,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009888164249309316,
      "loss": 2.2056,
      "step": 11027
    },
    {
      "epoch": 0.07492805423926886,
      "grad_norm": 0.125,
      "learning_rate": 0.0009888141620849173,
      "loss": 2.2374,
      "step": 11028
    },
    {
      "epoch": 0.07493484858586291,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000988811899012588,
      "loss": 2.1403,
      "step": 11029
    },
    {
      "epoch": 0.07494164293245698,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000988809635713944,
      "loss": 2.19,
      "step": 11030
    },
    {
      "epoch": 0.07494843727905104,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009888073721889868,
      "loss": 2.3479,
      "step": 11031
    },
    {
      "epoch": 0.07495523162564509,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009888051084377177,
      "loss": 2.3808,
      "step": 11032
    },
    {
      "epoch": 0.07496202597223915,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009888028444601372,
      "loss": 2.2124,
      "step": 11033
    },
    {
      "epoch": 0.0749688203188332,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009888005802562464,
      "loss": 2.319,
      "step": 11034
    },
    {
      "epoch": 0.07497561466542726,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009887983158260469,
      "loss": 2.0987,
      "step": 11035
    },
    {
      "epoch": 0.07498240901202133,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009887960511695392,
      "loss": 2.1982,
      "step": 11036
    },
    {
      "epoch": 0.07498920335861538,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009887937862867247,
      "loss": 2.076,
      "step": 11037
    },
    {
      "epoch": 0.07499599770520944,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000988791521177604,
      "loss": 2.3154,
      "step": 11038
    },
    {
      "epoch": 0.07500279205180349,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009887892558421787,
      "loss": 2.2206,
      "step": 11039
    },
    {
      "epoch": 0.07500958639839755,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009887869902804494,
      "loss": 2.3802,
      "step": 11040
    },
    {
      "epoch": 0.07501638074499162,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009887847244924176,
      "loss": 2.3615,
      "step": 11041
    },
    {
      "epoch": 0.07502317509158567,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009887824584780837,
      "loss": 2.2235,
      "step": 11042
    },
    {
      "epoch": 0.07502996943817973,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009887801922374494,
      "loss": 2.3033,
      "step": 11043
    },
    {
      "epoch": 0.07503676378477378,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009887779257705153,
      "loss": 2.1877,
      "step": 11044
    },
    {
      "epoch": 0.07504355813136784,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009887756590772827,
      "loss": 2.3597,
      "step": 11045
    },
    {
      "epoch": 0.07505035247796189,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009887733921577525,
      "loss": 2.1369,
      "step": 11046
    },
    {
      "epoch": 0.07505714682455596,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000988771125011926,
      "loss": 2.1546,
      "step": 11047
    },
    {
      "epoch": 0.07506394117115002,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000988768857639804,
      "loss": 2.1969,
      "step": 11048
    },
    {
      "epoch": 0.07507073551774407,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009887665900413876,
      "loss": 2.1433,
      "step": 11049
    },
    {
      "epoch": 0.07507752986433813,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009887643222166779,
      "loss": 2.1066,
      "step": 11050
    },
    {
      "epoch": 0.07508432421093218,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009887620541656757,
      "loss": 2.1525,
      "step": 11051
    },
    {
      "epoch": 0.07509111855752625,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009887597858883824,
      "loss": 2.1141,
      "step": 11052
    },
    {
      "epoch": 0.07509791290412031,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009887575173847991,
      "loss": 2.0591,
      "step": 11053
    },
    {
      "epoch": 0.07510470725071436,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009887552486549265,
      "loss": 2.2261,
      "step": 11054
    },
    {
      "epoch": 0.07511150159730842,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009887529796987658,
      "loss": 2.2068,
      "step": 11055
    },
    {
      "epoch": 0.07511829594390247,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009887507105163184,
      "loss": 2.1481,
      "step": 11056
    },
    {
      "epoch": 0.07512509029049654,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009887484411075845,
      "loss": 2.2431,
      "step": 11057
    },
    {
      "epoch": 0.0751318846370906,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009887461714725662,
      "loss": 2.3106,
      "step": 11058
    },
    {
      "epoch": 0.07513867898368465,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009887439016112636,
      "loss": 2.0913,
      "step": 11059
    },
    {
      "epoch": 0.07514547333027871,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009887416315236784,
      "loss": 2.1368,
      "step": 11060
    },
    {
      "epoch": 0.07515226767687276,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009887393612098112,
      "loss": 2.1325,
      "step": 11061
    },
    {
      "epoch": 0.07515906202346682,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009887370906696636,
      "loss": 2.2827,
      "step": 11062
    },
    {
      "epoch": 0.07516585637006089,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000988734819903236,
      "loss": 2.1965,
      "step": 11063
    },
    {
      "epoch": 0.07517265071665494,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00098873254891053,
      "loss": 2.2807,
      "step": 11064
    },
    {
      "epoch": 0.075179445063249,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009887302776915466,
      "loss": 2.3297,
      "step": 11065
    },
    {
      "epoch": 0.07518623940984305,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009887280062462864,
      "loss": 2.2108,
      "step": 11066
    },
    {
      "epoch": 0.07519303375643711,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009887257345747508,
      "loss": 2.2212,
      "step": 11067
    },
    {
      "epoch": 0.07519982810303118,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009887234626769408,
      "loss": 2.3357,
      "step": 11068
    },
    {
      "epoch": 0.07520662244962523,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009887211905528575,
      "loss": 2.3607,
      "step": 11069
    },
    {
      "epoch": 0.07521341679621929,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009887189182025019,
      "loss": 2.2508,
      "step": 11070
    },
    {
      "epoch": 0.07522021114281334,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009887166456258748,
      "loss": 2.1686,
      "step": 11071
    },
    {
      "epoch": 0.0752270054894074,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009887143728229777,
      "loss": 2.2177,
      "step": 11072
    },
    {
      "epoch": 0.07523379983600145,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009887120997938115,
      "loss": 2.1807,
      "step": 11073
    },
    {
      "epoch": 0.07524059418259552,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009887098265383772,
      "loss": 2.3536,
      "step": 11074
    },
    {
      "epoch": 0.07524738852918958,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009887075530566757,
      "loss": 2.3088,
      "step": 11075
    },
    {
      "epoch": 0.07525418287578363,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009887052793487083,
      "loss": 2.1418,
      "step": 11076
    },
    {
      "epoch": 0.07526097722237769,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009887030054144758,
      "loss": 2.2583,
      "step": 11077
    },
    {
      "epoch": 0.07526777156897174,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009887007312539796,
      "loss": 2.2222,
      "step": 11078
    },
    {
      "epoch": 0.0752745659155658,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009886984568672206,
      "loss": 2.3111,
      "step": 11079
    },
    {
      "epoch": 0.07528136026215987,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009886961822541996,
      "loss": 2.2624,
      "step": 11080
    },
    {
      "epoch": 0.07528815460875392,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000988693907414918,
      "loss": 2.2206,
      "step": 11081
    },
    {
      "epoch": 0.07529494895534798,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000988691632349377,
      "loss": 2.235,
      "step": 11082
    },
    {
      "epoch": 0.07530174330194203,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000988689357057577,
      "loss": 2.2315,
      "step": 11083
    },
    {
      "epoch": 0.0753085376485361,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009886870815395195,
      "loss": 2.1218,
      "step": 11084
    },
    {
      "epoch": 0.07531533199513016,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009886848057952054,
      "loss": 2.1699,
      "step": 11085
    },
    {
      "epoch": 0.07532212634172421,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000988682529824636,
      "loss": 2.1617,
      "step": 11086
    },
    {
      "epoch": 0.07532892068831827,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009886802536278121,
      "loss": 2.3185,
      "step": 11087
    },
    {
      "epoch": 0.07533571503491232,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000988677977204735,
      "loss": 2.1831,
      "step": 11088
    },
    {
      "epoch": 0.07534250938150638,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009886757005554054,
      "loss": 2.4715,
      "step": 11089
    },
    {
      "epoch": 0.07534930372810045,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009886734236798248,
      "loss": 2.3547,
      "step": 11090
    },
    {
      "epoch": 0.0753560980746945,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009886711465779939,
      "loss": 2.1795,
      "step": 11091
    },
    {
      "epoch": 0.07536289242128856,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009886688692499136,
      "loss": 2.2451,
      "step": 11092
    },
    {
      "epoch": 0.07536968676788261,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009886665916955856,
      "loss": 2.1864,
      "step": 11093
    },
    {
      "epoch": 0.07537648111447667,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009886643139150103,
      "loss": 2.2693,
      "step": 11094
    },
    {
      "epoch": 0.07538327546107072,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009886620359081891,
      "loss": 2.1675,
      "step": 11095
    },
    {
      "epoch": 0.07539006980766479,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000988659757675123,
      "loss": 2.1884,
      "step": 11096
    },
    {
      "epoch": 0.07539686415425885,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000988657479215813,
      "loss": 2.2933,
      "step": 11097
    },
    {
      "epoch": 0.0754036585008529,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009886552005302603,
      "loss": 2.1726,
      "step": 11098
    },
    {
      "epoch": 0.07541045284744696,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009886529216184658,
      "loss": 2.2654,
      "step": 11099
    },
    {
      "epoch": 0.07541724719404101,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009886506424804306,
      "loss": 2.2359,
      "step": 11100
    },
    {
      "epoch": 0.07542404154063508,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009886483631161556,
      "loss": 2.183,
      "step": 11101
    },
    {
      "epoch": 0.07543083588722914,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009886460835256423,
      "loss": 2.3165,
      "step": 11102
    },
    {
      "epoch": 0.07543763023382319,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000988643803708891,
      "loss": 2.2783,
      "step": 11103
    },
    {
      "epoch": 0.07544442458041725,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009886415236659038,
      "loss": 2.2432,
      "step": 11104
    },
    {
      "epoch": 0.0754512189270113,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009886392433966808,
      "loss": 2.2099,
      "step": 11105
    },
    {
      "epoch": 0.07545801327360536,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009886369629012236,
      "loss": 2.2916,
      "step": 11106
    },
    {
      "epoch": 0.07546480762019943,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000988634682179533,
      "loss": 2.262,
      "step": 11107
    },
    {
      "epoch": 0.07547160196679348,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009886324012316101,
      "loss": 2.1853,
      "step": 11108
    },
    {
      "epoch": 0.07547839631338754,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000988630120057456,
      "loss": 2.2605,
      "step": 11109
    },
    {
      "epoch": 0.07548519065998159,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000988627838657072,
      "loss": 2.377,
      "step": 11110
    },
    {
      "epoch": 0.07549198500657565,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009886255570304588,
      "loss": 2.3489,
      "step": 11111
    },
    {
      "epoch": 0.07549877935316972,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009886232751776173,
      "loss": 2.2093,
      "step": 11112
    },
    {
      "epoch": 0.07550557369976377,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000988620993098549,
      "loss": 2.2542,
      "step": 11113
    },
    {
      "epoch": 0.07551236804635783,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000988618710793255,
      "loss": 2.2101,
      "step": 11114
    },
    {
      "epoch": 0.07551916239295188,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009886164282617359,
      "loss": 2.2539,
      "step": 11115
    },
    {
      "epoch": 0.07552595673954594,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000988614145503993,
      "loss": 2.0856,
      "step": 11116
    },
    {
      "epoch": 0.07553275108614001,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009886118625200275,
      "loss": 2.2146,
      "step": 11117
    },
    {
      "epoch": 0.07553954543273406,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009886095793098402,
      "loss": 2.1268,
      "step": 11118
    },
    {
      "epoch": 0.07554633977932812,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009886072958734322,
      "loss": 2.1428,
      "step": 11119
    },
    {
      "epoch": 0.07555313412592217,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009886050122108047,
      "loss": 2.1748,
      "step": 11120
    },
    {
      "epoch": 0.07555992847251623,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009886027283219588,
      "loss": 2.2602,
      "step": 11121
    },
    {
      "epoch": 0.07556672281911028,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009886004442068953,
      "loss": 2.2607,
      "step": 11122
    },
    {
      "epoch": 0.07557351716570435,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009885981598656154,
      "loss": 2.2525,
      "step": 11123
    },
    {
      "epoch": 0.07558031151229841,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00098859587529812,
      "loss": 2.3146,
      "step": 11124
    },
    {
      "epoch": 0.07558710585889246,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009885935905044106,
      "loss": 2.3191,
      "step": 11125
    },
    {
      "epoch": 0.07559390020548652,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009885913054844878,
      "loss": 2.3336,
      "step": 11126
    },
    {
      "epoch": 0.07560069455208057,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009885890202383529,
      "loss": 2.2376,
      "step": 11127
    },
    {
      "epoch": 0.07560748889867464,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009885867347660068,
      "loss": 2.0757,
      "step": 11128
    },
    {
      "epoch": 0.0756142832452687,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009885844490674508,
      "loss": 2.2592,
      "step": 11129
    },
    {
      "epoch": 0.07562107759186275,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009885821631426858,
      "loss": 2.2988,
      "step": 11130
    },
    {
      "epoch": 0.07562787193845681,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009885798769917126,
      "loss": 2.1046,
      "step": 11131
    },
    {
      "epoch": 0.07563466628505086,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009885775906145326,
      "loss": 2.2321,
      "step": 11132
    },
    {
      "epoch": 0.07564146063164492,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000988575304011147,
      "loss": 2.2791,
      "step": 11133
    },
    {
      "epoch": 0.07564825497823899,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009885730171815563,
      "loss": 2.2951,
      "step": 11134
    },
    {
      "epoch": 0.07565504932483304,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009885707301257622,
      "loss": 2.1323,
      "step": 11135
    },
    {
      "epoch": 0.0756618436714271,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009885684428437653,
      "loss": 2.2333,
      "step": 11136
    },
    {
      "epoch": 0.07566863801802115,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009885661553355669,
      "loss": 2.3537,
      "step": 11137
    },
    {
      "epoch": 0.07567543236461521,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009885638676011678,
      "loss": 2.2333,
      "step": 11138
    },
    {
      "epoch": 0.07568222671120928,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009885615796405695,
      "loss": 2.2119,
      "step": 11139
    },
    {
      "epoch": 0.07568902105780333,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009885592914537728,
      "loss": 2.1025,
      "step": 11140
    },
    {
      "epoch": 0.07569581540439739,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009885570030407785,
      "loss": 2.1296,
      "step": 11141
    },
    {
      "epoch": 0.07570260975099144,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009885547144015878,
      "loss": 2.2731,
      "step": 11142
    },
    {
      "epoch": 0.0757094040975855,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009885524255362022,
      "loss": 2.2263,
      "step": 11143
    },
    {
      "epoch": 0.07571619844417955,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009885501364446222,
      "loss": 2.2439,
      "step": 11144
    },
    {
      "epoch": 0.07572299279077362,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009885478471268491,
      "loss": 2.0481,
      "step": 11145
    },
    {
      "epoch": 0.07572978713736768,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009885455575828841,
      "loss": 2.3929,
      "step": 11146
    },
    {
      "epoch": 0.07573658148396173,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000988543267812728,
      "loss": 2.2441,
      "step": 11147
    },
    {
      "epoch": 0.07574337583055579,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009885409778163821,
      "loss": 2.1793,
      "step": 11148
    },
    {
      "epoch": 0.07575017017714984,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009885386875938473,
      "loss": 2.2786,
      "step": 11149
    },
    {
      "epoch": 0.0757569645237439,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009885363971451247,
      "loss": 2.2505,
      "step": 11150
    },
    {
      "epoch": 0.07576375887033797,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009885341064702151,
      "loss": 2.3191,
      "step": 11151
    },
    {
      "epoch": 0.07577055321693202,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00098853181556912,
      "loss": 2.2419,
      "step": 11152
    },
    {
      "epoch": 0.07577734756352608,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009885295244418404,
      "loss": 2.3916,
      "step": 11153
    },
    {
      "epoch": 0.07578414191012013,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000988527233088377,
      "loss": 2.0532,
      "step": 11154
    },
    {
      "epoch": 0.0757909362567142,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009885249415087312,
      "loss": 2.2184,
      "step": 11155
    },
    {
      "epoch": 0.07579773060330826,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009885226497029038,
      "loss": 2.4285,
      "step": 11156
    },
    {
      "epoch": 0.07580452494990231,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009885203576708963,
      "loss": 2.2311,
      "step": 11157
    },
    {
      "epoch": 0.07581131929649637,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009885180654127092,
      "loss": 2.2616,
      "step": 11158
    },
    {
      "epoch": 0.07581811364309042,
      "grad_norm": 0.140625,
      "learning_rate": 0.000988515772928344,
      "loss": 2.2226,
      "step": 11159
    },
    {
      "epoch": 0.07582490798968448,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009885134802178016,
      "loss": 2.3033,
      "step": 11160
    },
    {
      "epoch": 0.07583170233627855,
      "grad_norm": 0.140625,
      "learning_rate": 0.000988511187281083,
      "loss": 2.3668,
      "step": 11161
    },
    {
      "epoch": 0.0758384966828726,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009885088941181891,
      "loss": 2.1259,
      "step": 11162
    },
    {
      "epoch": 0.07584529102946666,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009885066007291217,
      "loss": 2.0882,
      "step": 11163
    },
    {
      "epoch": 0.07585208537606071,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000988504307113881,
      "loss": 2.1196,
      "step": 11164
    },
    {
      "epoch": 0.07585887972265477,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009885020132724685,
      "loss": 2.3794,
      "step": 11165
    },
    {
      "epoch": 0.07586567406924882,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000988499719204885,
      "loss": 2.0667,
      "step": 11166
    },
    {
      "epoch": 0.07587246841584289,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000988497424911132,
      "loss": 2.3042,
      "step": 11167
    },
    {
      "epoch": 0.07587926276243695,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00098849513039121,
      "loss": 2.1024,
      "step": 11168
    },
    {
      "epoch": 0.075886057109031,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009884928356451207,
      "loss": 2.04,
      "step": 11169
    },
    {
      "epoch": 0.07589285145562506,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009884905406728646,
      "loss": 2.3838,
      "step": 11170
    },
    {
      "epoch": 0.07589964580221911,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000988488245474443,
      "loss": 2.1662,
      "step": 11171
    },
    {
      "epoch": 0.07590644014881318,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009884859500498568,
      "loss": 2.1253,
      "step": 11172
    },
    {
      "epoch": 0.07591323449540724,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009884836543991075,
      "loss": 2.2731,
      "step": 11173
    },
    {
      "epoch": 0.07592002884200129,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009884813585221957,
      "loss": 2.3499,
      "step": 11174
    },
    {
      "epoch": 0.07592682318859535,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009884790624191226,
      "loss": 2.2778,
      "step": 11175
    },
    {
      "epoch": 0.0759336175351894,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009884767660898895,
      "loss": 2.2302,
      "step": 11176
    },
    {
      "epoch": 0.07594041188178347,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000988474469534497,
      "loss": 2.4015,
      "step": 11177
    },
    {
      "epoch": 0.07594720622837753,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009884721727529465,
      "loss": 2.2843,
      "step": 11178
    },
    {
      "epoch": 0.07595400057497158,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000988469875745239,
      "loss": 1.9425,
      "step": 11179
    },
    {
      "epoch": 0.07596079492156564,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009884675785113756,
      "loss": 2.31,
      "step": 11180
    },
    {
      "epoch": 0.07596758926815969,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009884652810513573,
      "loss": 2.2846,
      "step": 11181
    },
    {
      "epoch": 0.07597438361475375,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009884629833651851,
      "loss": 2.1201,
      "step": 11182
    },
    {
      "epoch": 0.07598117796134782,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00098846068545286,
      "loss": 2.2375,
      "step": 11183
    },
    {
      "epoch": 0.07598797230794187,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009884583873143835,
      "loss": 2.1373,
      "step": 11184
    },
    {
      "epoch": 0.07599476665453593,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009884560889497564,
      "loss": 2.2691,
      "step": 11185
    },
    {
      "epoch": 0.07600156100112998,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009884537903589795,
      "loss": 2.2117,
      "step": 11186
    },
    {
      "epoch": 0.07600835534772404,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009884514915420542,
      "loss": 2.5272,
      "step": 11187
    },
    {
      "epoch": 0.07601514969431811,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009884491924989816,
      "loss": 2.1896,
      "step": 11188
    },
    {
      "epoch": 0.07602194404091216,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009884468932297623,
      "loss": 2.2018,
      "step": 11189
    },
    {
      "epoch": 0.07602873838750622,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009884445937343978,
      "loss": 2.2406,
      "step": 11190
    },
    {
      "epoch": 0.07603553273410027,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009884422940128893,
      "loss": 2.2495,
      "step": 11191
    },
    {
      "epoch": 0.07604232708069433,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009884399940652373,
      "loss": 2.2056,
      "step": 11192
    },
    {
      "epoch": 0.07604912142728838,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009884376938914434,
      "loss": 2.276,
      "step": 11193
    },
    {
      "epoch": 0.07605591577388245,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009884353934915082,
      "loss": 2.3869,
      "step": 11194
    },
    {
      "epoch": 0.07606271012047651,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009884330928654332,
      "loss": 2.076,
      "step": 11195
    },
    {
      "epoch": 0.07606950446707056,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009884307920132193,
      "loss": 2.2214,
      "step": 11196
    },
    {
      "epoch": 0.07607629881366462,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009884284909348676,
      "loss": 2.2145,
      "step": 11197
    },
    {
      "epoch": 0.07608309316025867,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009884261896303788,
      "loss": 2.2539,
      "step": 11198
    },
    {
      "epoch": 0.07608988750685274,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009884238880997545,
      "loss": 2.2859,
      "step": 11199
    },
    {
      "epoch": 0.0760966818534468,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009884215863429956,
      "loss": 2.2926,
      "step": 11200
    },
    {
      "epoch": 0.07610347620004085,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009884192843601029,
      "loss": 2.174,
      "step": 11201
    },
    {
      "epoch": 0.07611027054663491,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009884169821510777,
      "loss": 2.2045,
      "step": 11202
    },
    {
      "epoch": 0.07611706489322896,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009884146797159213,
      "loss": 2.2988,
      "step": 11203
    },
    {
      "epoch": 0.07612385923982302,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009884123770546342,
      "loss": 2.2169,
      "step": 11204
    },
    {
      "epoch": 0.07613065358641709,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000988410074167218,
      "loss": 2.1666,
      "step": 11205
    },
    {
      "epoch": 0.07613744793301114,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009884077710536733,
      "loss": 2.1865,
      "step": 11206
    },
    {
      "epoch": 0.0761442422796052,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009884054677140016,
      "loss": 2.1484,
      "step": 11207
    },
    {
      "epoch": 0.07615103662619925,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009884031641482036,
      "loss": 2.26,
      "step": 11208
    },
    {
      "epoch": 0.07615783097279331,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009884008603562807,
      "loss": 2.2615,
      "step": 11209
    },
    {
      "epoch": 0.07616462531938738,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009883985563382339,
      "loss": 2.078,
      "step": 11210
    },
    {
      "epoch": 0.07617141966598143,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009883962520940638,
      "loss": 2.231,
      "step": 11211
    },
    {
      "epoch": 0.07617821401257549,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009883939476237722,
      "loss": 2.2908,
      "step": 11212
    },
    {
      "epoch": 0.07618500835916954,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009883916429273595,
      "loss": 2.1304,
      "step": 11213
    },
    {
      "epoch": 0.0761918027057636,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009883893380048271,
      "loss": 2.2835,
      "step": 11214
    },
    {
      "epoch": 0.07619859705235765,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009883870328561762,
      "loss": 2.2924,
      "step": 11215
    },
    {
      "epoch": 0.07620539139895172,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009883847274814077,
      "loss": 2.3831,
      "step": 11216
    },
    {
      "epoch": 0.07621218574554578,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009883824218805226,
      "loss": 2.2931,
      "step": 11217
    },
    {
      "epoch": 0.07621898009213983,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009883801160535221,
      "loss": 2.1243,
      "step": 11218
    },
    {
      "epoch": 0.07622577443873389,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000988377810000407,
      "loss": 2.0853,
      "step": 11219
    },
    {
      "epoch": 0.07623256878532794,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009883755037211788,
      "loss": 2.1774,
      "step": 11220
    },
    {
      "epoch": 0.076239363131922,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009883731972158384,
      "loss": 2.3292,
      "step": 11221
    },
    {
      "epoch": 0.07624615747851607,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009883708904843867,
      "loss": 2.1379,
      "step": 11222
    },
    {
      "epoch": 0.07625295182511012,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009883685835268248,
      "loss": 2.2743,
      "step": 11223
    },
    {
      "epoch": 0.07625974617170418,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009883662763431537,
      "loss": 2.1709,
      "step": 11224
    },
    {
      "epoch": 0.07626654051829823,
      "grad_norm": 0.140625,
      "learning_rate": 0.000988363968933375,
      "loss": 2.2475,
      "step": 11225
    },
    {
      "epoch": 0.0762733348648923,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000988361661297489,
      "loss": 2.1963,
      "step": 11226
    },
    {
      "epoch": 0.07628012921148636,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009883593534354975,
      "loss": 2.3536,
      "step": 11227
    },
    {
      "epoch": 0.07628692355808041,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009883570453474012,
      "loss": 2.238,
      "step": 11228
    },
    {
      "epoch": 0.07629371790467447,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009883547370332008,
      "loss": 2.2184,
      "step": 11229
    },
    {
      "epoch": 0.07630051225126852,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009883524284928981,
      "loss": 2.1812,
      "step": 11230
    },
    {
      "epoch": 0.07630730659786258,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009883501197264937,
      "loss": 2.2353,
      "step": 11231
    },
    {
      "epoch": 0.07631410094445665,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009883478107339887,
      "loss": 2.2956,
      "step": 11232
    },
    {
      "epoch": 0.0763208952910507,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009883455015153844,
      "loss": 2.2857,
      "step": 11233
    },
    {
      "epoch": 0.07632768963764476,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009883431920706819,
      "loss": 2.2237,
      "step": 11234
    },
    {
      "epoch": 0.07633448398423881,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009883408823998818,
      "loss": 2.194,
      "step": 11235
    },
    {
      "epoch": 0.07634127833083287,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009883385725029856,
      "loss": 2.2015,
      "step": 11236
    },
    {
      "epoch": 0.07634807267742692,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009883362623799943,
      "loss": 2.397,
      "step": 11237
    },
    {
      "epoch": 0.07635486702402099,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009883339520309086,
      "loss": 2.2014,
      "step": 11238
    },
    {
      "epoch": 0.07636166137061505,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009883316414557304,
      "loss": 2.1766,
      "step": 11239
    },
    {
      "epoch": 0.0763684557172091,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009883293306544598,
      "loss": 2.2459,
      "step": 11240
    },
    {
      "epoch": 0.07637525006380316,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009883270196270986,
      "loss": 2.1888,
      "step": 11241
    },
    {
      "epoch": 0.07638204441039721,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009883247083736473,
      "loss": 2.1376,
      "step": 11242
    },
    {
      "epoch": 0.07638883875699128,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009883223968941075,
      "loss": 2.2621,
      "step": 11243
    },
    {
      "epoch": 0.07639563310358534,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00098832008518848,
      "loss": 2.2986,
      "step": 11244
    },
    {
      "epoch": 0.07640242745017939,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000988317773256766,
      "loss": 2.1582,
      "step": 11245
    },
    {
      "epoch": 0.07640922179677345,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009883154610989661,
      "loss": 2.2007,
      "step": 11246
    },
    {
      "epoch": 0.0764160161433675,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009883131487150821,
      "loss": 2.246,
      "step": 11247
    },
    {
      "epoch": 0.07642281048996157,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009883108361051145,
      "loss": 2.2829,
      "step": 11248
    },
    {
      "epoch": 0.07642960483655563,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009883085232690646,
      "loss": 2.2121,
      "step": 11249
    },
    {
      "epoch": 0.07643639918314968,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009883062102069337,
      "loss": 2.2452,
      "step": 11250
    },
    {
      "epoch": 0.07644319352974374,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009883038969187223,
      "loss": 2.3777,
      "step": 11251
    },
    {
      "epoch": 0.07644998787633779,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009883015834044318,
      "loss": 2.2809,
      "step": 11252
    },
    {
      "epoch": 0.07645678222293185,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009882992696640634,
      "loss": 2.2185,
      "step": 11253
    },
    {
      "epoch": 0.07646357656952592,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000988296955697618,
      "loss": 2.406,
      "step": 11254
    },
    {
      "epoch": 0.07647037091611997,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000988294641505097,
      "loss": 2.2066,
      "step": 11255
    },
    {
      "epoch": 0.07647716526271403,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009882923270865007,
      "loss": 2.2506,
      "step": 11256
    },
    {
      "epoch": 0.07648395960930808,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000988290012441831,
      "loss": 2.2522,
      "step": 11257
    },
    {
      "epoch": 0.07649075395590214,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009882876975710884,
      "loss": 2.3357,
      "step": 11258
    },
    {
      "epoch": 0.07649754830249621,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009882853824742742,
      "loss": 2.2343,
      "step": 11259
    },
    {
      "epoch": 0.07650434264909026,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009882830671513897,
      "loss": 2.0836,
      "step": 11260
    },
    {
      "epoch": 0.07651113699568432,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009882807516024355,
      "loss": 2.3006,
      "step": 11261
    },
    {
      "epoch": 0.07651793134227837,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000988278435827413,
      "loss": 2.2525,
      "step": 11262
    },
    {
      "epoch": 0.07652472568887243,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000988276119826323,
      "loss": 2.336,
      "step": 11263
    },
    {
      "epoch": 0.07653152003546648,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000988273803599167,
      "loss": 2.3963,
      "step": 11264
    },
    {
      "epoch": 0.07653831438206055,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009882714871459457,
      "loss": 2.1908,
      "step": 11265
    },
    {
      "epoch": 0.07654510872865461,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009882691704666605,
      "loss": 2.2036,
      "step": 11266
    },
    {
      "epoch": 0.07655190307524866,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009882668535613121,
      "loss": 2.2886,
      "step": 11267
    },
    {
      "epoch": 0.07655869742184272,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009882645364299018,
      "loss": 2.2883,
      "step": 11268
    },
    {
      "epoch": 0.07656549176843677,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009882622190724307,
      "loss": 2.1962,
      "step": 11269
    },
    {
      "epoch": 0.07657228611503084,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009882599014888996,
      "loss": 2.2226,
      "step": 11270
    },
    {
      "epoch": 0.0765790804616249,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009882575836793097,
      "loss": 2.1582,
      "step": 11271
    },
    {
      "epoch": 0.07658587480821895,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009882552656436624,
      "loss": 2.1702,
      "step": 11272
    },
    {
      "epoch": 0.07659266915481301,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009882529473819583,
      "loss": 2.2066,
      "step": 11273
    },
    {
      "epoch": 0.07659946350140706,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009882506288941988,
      "loss": 2.249,
      "step": 11274
    },
    {
      "epoch": 0.07660625784800112,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009882483101803848,
      "loss": 2.2281,
      "step": 11275
    },
    {
      "epoch": 0.07661305219459519,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009882459912405174,
      "loss": 2.1635,
      "step": 11276
    },
    {
      "epoch": 0.07661984654118924,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009882436720745978,
      "loss": 2.2017,
      "step": 11277
    },
    {
      "epoch": 0.0766266408877833,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000988241352682627,
      "loss": 2.2668,
      "step": 11278
    },
    {
      "epoch": 0.07663343523437735,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000988239033064606,
      "loss": 2.2419,
      "step": 11279
    },
    {
      "epoch": 0.07664022958097141,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009882367132205358,
      "loss": 2.1525,
      "step": 11280
    },
    {
      "epoch": 0.07664702392756548,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009882343931504178,
      "loss": 2.1169,
      "step": 11281
    },
    {
      "epoch": 0.07665381827415953,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009882320728542527,
      "loss": 2.3749,
      "step": 11282
    },
    {
      "epoch": 0.07666061262075359,
      "grad_norm": 0.140625,
      "learning_rate": 0.000988229752332042,
      "loss": 2.2031,
      "step": 11283
    },
    {
      "epoch": 0.07666740696734764,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009882274315837862,
      "loss": 2.3689,
      "step": 11284
    },
    {
      "epoch": 0.0766742013139417,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009882251106094868,
      "loss": 2.1928,
      "step": 11285
    },
    {
      "epoch": 0.07668099566053575,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009882227894091448,
      "loss": 2.0349,
      "step": 11286
    },
    {
      "epoch": 0.07668779000712982,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009882204679827614,
      "loss": 2.2216,
      "step": 11287
    },
    {
      "epoch": 0.07669458435372388,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009882181463303375,
      "loss": 2.2782,
      "step": 11288
    },
    {
      "epoch": 0.07670137870031793,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000988215824451874,
      "loss": 2.407,
      "step": 11289
    },
    {
      "epoch": 0.07670817304691199,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009882135023473722,
      "loss": 2.1482,
      "step": 11290
    },
    {
      "epoch": 0.07671496739350604,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009882111800168332,
      "loss": 2.1597,
      "step": 11291
    },
    {
      "epoch": 0.0767217617401001,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009882088574602581,
      "loss": 2.3294,
      "step": 11292
    },
    {
      "epoch": 0.07672855608669417,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009882065346776478,
      "loss": 2.3222,
      "step": 11293
    },
    {
      "epoch": 0.07673535043328822,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009882042116690034,
      "loss": 2.2107,
      "step": 11294
    },
    {
      "epoch": 0.07674214477988228,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009882018884343263,
      "loss": 2.1703,
      "step": 11295
    },
    {
      "epoch": 0.07674893912647633,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000988199564973617,
      "loss": 2.0477,
      "step": 11296
    },
    {
      "epoch": 0.0767557334730704,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000988197241286877,
      "loss": 2.2088,
      "step": 11297
    },
    {
      "epoch": 0.07676252781966446,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009881949173741074,
      "loss": 2.238,
      "step": 11298
    },
    {
      "epoch": 0.07676932216625851,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009881925932353091,
      "loss": 2.2622,
      "step": 11299
    },
    {
      "epoch": 0.07677611651285257,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000988190268870483,
      "loss": 2.1962,
      "step": 11300
    },
    {
      "epoch": 0.07678291085944662,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009881879442796308,
      "loss": 2.4141,
      "step": 11301
    },
    {
      "epoch": 0.07678970520604068,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000988185619462753,
      "loss": 2.2513,
      "step": 11302
    },
    {
      "epoch": 0.07679649955263475,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.0009881832944198508,
      "loss": 2.2795,
      "step": 11303
    },
    {
      "epoch": 0.0768032938992288,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009881809691509253,
      "loss": 2.3003,
      "step": 11304
    },
    {
      "epoch": 0.07681008824582286,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009881786436559778,
      "loss": 2.2506,
      "step": 11305
    },
    {
      "epoch": 0.07681688259241691,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009881763179350092,
      "loss": 2.2657,
      "step": 11306
    },
    {
      "epoch": 0.07682367693901097,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009881739919880203,
      "loss": 2.4564,
      "step": 11307
    },
    {
      "epoch": 0.07683047128560504,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009881716658150126,
      "loss": 2.2693,
      "step": 11308
    },
    {
      "epoch": 0.07683726563219909,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000988169339415987,
      "loss": 2.2045,
      "step": 11309
    },
    {
      "epoch": 0.07684405997879315,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009881670127909446,
      "loss": 2.2662,
      "step": 11310
    },
    {
      "epoch": 0.0768508543253872,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009881646859398862,
      "loss": 2.2708,
      "step": 11311
    },
    {
      "epoch": 0.07685764867198126,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009881623588628135,
      "loss": 2.33,
      "step": 11312
    },
    {
      "epoch": 0.07686444301857531,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009881600315597271,
      "loss": 2.1942,
      "step": 11313
    },
    {
      "epoch": 0.07687123736516938,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009881577040306282,
      "loss": 2.2753,
      "step": 11314
    },
    {
      "epoch": 0.07687803171176344,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000988155376275518,
      "loss": 2.1866,
      "step": 11315
    },
    {
      "epoch": 0.07688482605835749,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009881530482943974,
      "loss": 2.2903,
      "step": 11316
    },
    {
      "epoch": 0.07689162040495155,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009881507200872675,
      "loss": 2.2093,
      "step": 11317
    },
    {
      "epoch": 0.0768984147515456,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009881483916541294,
      "loss": 2.2701,
      "step": 11318
    },
    {
      "epoch": 0.07690520909813967,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000988146062994984,
      "loss": 2.2216,
      "step": 11319
    },
    {
      "epoch": 0.07691200344473373,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000988143734109833,
      "loss": 2.2839,
      "step": 11320
    },
    {
      "epoch": 0.07691879779132778,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009881414049986768,
      "loss": 2.3812,
      "step": 11321
    },
    {
      "epoch": 0.07692559213792184,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009881390756615166,
      "loss": 2.2176,
      "step": 11322
    },
    {
      "epoch": 0.07693238648451589,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009881367460983536,
      "loss": 2.3039,
      "step": 11323
    },
    {
      "epoch": 0.07693918083110995,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009881344163091893,
      "loss": 2.2487,
      "step": 11324
    },
    {
      "epoch": 0.07694597517770402,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0009881320862940239,
      "loss": 2.246,
      "step": 11325
    },
    {
      "epoch": 0.07695276952429807,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009881297560528593,
      "loss": 2.3712,
      "step": 11326
    },
    {
      "epoch": 0.07695956387089213,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000988127425585696,
      "loss": 2.3252,
      "step": 11327
    },
    {
      "epoch": 0.07696635821748618,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009881250948925353,
      "loss": 2.111,
      "step": 11328
    },
    {
      "epoch": 0.07697315256408024,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009881227639733784,
      "loss": 2.0933,
      "step": 11329
    },
    {
      "epoch": 0.07697994691067431,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000988120432828226,
      "loss": 2.4365,
      "step": 11330
    },
    {
      "epoch": 0.07698674125726836,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009881181014570798,
      "loss": 2.1466,
      "step": 11331
    },
    {
      "epoch": 0.07699353560386242,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009881157698599403,
      "loss": 2.2757,
      "step": 11332
    },
    {
      "epoch": 0.07700032995045647,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009881134380368087,
      "loss": 2.3057,
      "step": 11333
    },
    {
      "epoch": 0.07700712429705053,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009881111059876864,
      "loss": 2.1957,
      "step": 11334
    },
    {
      "epoch": 0.07701391864364458,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000988108773712574,
      "loss": 2.1965,
      "step": 11335
    },
    {
      "epoch": 0.07702071299023865,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000988106441211473,
      "loss": 2.1305,
      "step": 11336
    },
    {
      "epoch": 0.07702750733683271,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009881041084843843,
      "loss": 2.1849,
      "step": 11337
    },
    {
      "epoch": 0.07703430168342676,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000988101775531309,
      "loss": 2.1864,
      "step": 11338
    },
    {
      "epoch": 0.07704109603002082,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009880994423522482,
      "loss": 2.3415,
      "step": 11339
    },
    {
      "epoch": 0.07704789037661487,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009880971089472029,
      "loss": 2.2309,
      "step": 11340
    },
    {
      "epoch": 0.07705468472320894,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009880947753161742,
      "loss": 2.2619,
      "step": 11341
    },
    {
      "epoch": 0.077061479069803,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009880924414591634,
      "loss": 2.188,
      "step": 11342
    },
    {
      "epoch": 0.07706827341639705,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009880901073761712,
      "loss": 2.3379,
      "step": 11343
    },
    {
      "epoch": 0.07707506776299111,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000988087773067199,
      "loss": 2.2963,
      "step": 11344
    },
    {
      "epoch": 0.07708186210958516,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009880854385322477,
      "loss": 2.3908,
      "step": 11345
    },
    {
      "epoch": 0.07708865645617922,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009880831037713184,
      "loss": 2.4095,
      "step": 11346
    },
    {
      "epoch": 0.07709545080277329,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009880807687844121,
      "loss": 2.3025,
      "step": 11347
    },
    {
      "epoch": 0.07710224514936734,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009880784335715303,
      "loss": 2.2804,
      "step": 11348
    },
    {
      "epoch": 0.0771090394959614,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009880760981326735,
      "loss": 2.3766,
      "step": 11349
    },
    {
      "epoch": 0.07711583384255545,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009880737624678433,
      "loss": 2.2061,
      "step": 11350
    },
    {
      "epoch": 0.07712262818914951,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009880714265770405,
      "loss": 2.2274,
      "step": 11351
    },
    {
      "epoch": 0.07712942253574358,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000988069090460266,
      "loss": 2.2364,
      "step": 11352
    },
    {
      "epoch": 0.07713621688233763,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009880667541175214,
      "loss": 2.1061,
      "step": 11353
    },
    {
      "epoch": 0.07714301122893169,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009880644175488075,
      "loss": 2.1893,
      "step": 11354
    },
    {
      "epoch": 0.07714980557552574,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000988062080754125,
      "loss": 2.3529,
      "step": 11355
    },
    {
      "epoch": 0.0771565999221198,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009880597437334755,
      "loss": 2.1642,
      "step": 11356
    },
    {
      "epoch": 0.07716339426871385,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00098805740648686,
      "loss": 2.3066,
      "step": 11357
    },
    {
      "epoch": 0.07717018861530792,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009880550690142796,
      "loss": 2.271,
      "step": 11358
    },
    {
      "epoch": 0.07717698296190198,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009880527313157353,
      "loss": 2.1675,
      "step": 11359
    },
    {
      "epoch": 0.07718377730849603,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000988050393391228,
      "loss": 2.2196,
      "step": 11360
    },
    {
      "epoch": 0.07719057165509009,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.000988048055240759,
      "loss": 2.2179,
      "step": 11361
    },
    {
      "epoch": 0.07719736600168414,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009880457168643295,
      "loss": 2.2871,
      "step": 11362
    },
    {
      "epoch": 0.0772041603482782,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009880433782619405,
      "loss": 2.2226,
      "step": 11363
    },
    {
      "epoch": 0.07721095469487227,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009880410394335927,
      "loss": 2.2396,
      "step": 11364
    },
    {
      "epoch": 0.07721774904146632,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009880387003792877,
      "loss": 2.224,
      "step": 11365
    },
    {
      "epoch": 0.07722454338806038,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009880363610990264,
      "loss": 2.1326,
      "step": 11366
    },
    {
      "epoch": 0.07723133773465443,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009880340215928098,
      "loss": 2.3236,
      "step": 11367
    },
    {
      "epoch": 0.0772381320812485,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000988031681860639,
      "loss": 2.2706,
      "step": 11368
    },
    {
      "epoch": 0.07724492642784256,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000988029341902515,
      "loss": 2.4291,
      "step": 11369
    },
    {
      "epoch": 0.07725172077443661,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009880270017184392,
      "loss": 2.1033,
      "step": 11370
    },
    {
      "epoch": 0.07725851512103067,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009880246613084124,
      "loss": 2.2742,
      "step": 11371
    },
    {
      "epoch": 0.07726530946762472,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000988022320672436,
      "loss": 2.3057,
      "step": 11372
    },
    {
      "epoch": 0.07727210381421878,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009880199798105106,
      "loss": 2.1901,
      "step": 11373
    },
    {
      "epoch": 0.07727889816081285,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009880176387226378,
      "loss": 2.2023,
      "step": 11374
    },
    {
      "epoch": 0.0772856925074069,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009880152974088181,
      "loss": 2.191,
      "step": 11375
    },
    {
      "epoch": 0.07729248685400096,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000988012955869053,
      "loss": 2.2889,
      "step": 11376
    },
    {
      "epoch": 0.07729928120059501,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009880106141033438,
      "loss": 2.2894,
      "step": 11377
    },
    {
      "epoch": 0.07730607554718907,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009880082721116909,
      "loss": 2.2153,
      "step": 11378
    },
    {
      "epoch": 0.07731286989378314,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009880059298940959,
      "loss": 2.1696,
      "step": 11379
    },
    {
      "epoch": 0.07731966424037719,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009880035874505599,
      "loss": 2.2353,
      "step": 11380
    },
    {
      "epoch": 0.07732645858697125,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009880012447810837,
      "loss": 2.0881,
      "step": 11381
    },
    {
      "epoch": 0.0773332529335653,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009879989018856685,
      "loss": 2.2272,
      "step": 11382
    },
    {
      "epoch": 0.07734004728015936,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009879965587643153,
      "loss": 2.1825,
      "step": 11383
    },
    {
      "epoch": 0.07734684162675341,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009879942154170256,
      "loss": 2.1278,
      "step": 11384
    },
    {
      "epoch": 0.07735363597334748,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009879918718438,
      "loss": 2.1789,
      "step": 11385
    },
    {
      "epoch": 0.07736043031994154,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009879895280446398,
      "loss": 2.1224,
      "step": 11386
    },
    {
      "epoch": 0.07736722466653559,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000987987184019546,
      "loss": 2.2866,
      "step": 11387
    },
    {
      "epoch": 0.07737401901312965,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009879848397685196,
      "loss": 2.151,
      "step": 11388
    },
    {
      "epoch": 0.0773808133597237,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000987982495291562,
      "loss": 2.303,
      "step": 11389
    },
    {
      "epoch": 0.07738760770631777,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000987980150588674,
      "loss": 2.3021,
      "step": 11390
    },
    {
      "epoch": 0.07739440205291183,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009879778056598567,
      "loss": 2.2656,
      "step": 11391
    },
    {
      "epoch": 0.07740119639950588,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009879754605051114,
      "loss": 2.2879,
      "step": 11392
    },
    {
      "epoch": 0.07740799074609994,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009879731151244391,
      "loss": 2.2376,
      "step": 11393
    },
    {
      "epoch": 0.07741478509269399,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009879707695178408,
      "loss": 2.2393,
      "step": 11394
    },
    {
      "epoch": 0.07742157943928805,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009879684236853176,
      "loss": 2.3221,
      "step": 11395
    },
    {
      "epoch": 0.07742837378588212,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009879660776268707,
      "loss": 2.2125,
      "step": 11396
    },
    {
      "epoch": 0.07743516813247617,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000987963731342501,
      "loss": 2.2277,
      "step": 11397
    },
    {
      "epoch": 0.07744196247907023,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009879613848322096,
      "loss": 2.2141,
      "step": 11398
    },
    {
      "epoch": 0.07744875682566428,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009879590380959978,
      "loss": 2.2061,
      "step": 11399
    },
    {
      "epoch": 0.07745555117225834,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009879566911338666,
      "loss": 2.3052,
      "step": 11400
    },
    {
      "epoch": 0.07746234551885241,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000987954343945817,
      "loss": 2.2644,
      "step": 11401
    },
    {
      "epoch": 0.07746913986544646,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00098795199653185,
      "loss": 2.2819,
      "step": 11402
    },
    {
      "epoch": 0.07747593421204052,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000987949648891967,
      "loss": 2.0724,
      "step": 11403
    },
    {
      "epoch": 0.07748272855863457,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009879473010261687,
      "loss": 2.4112,
      "step": 11404
    },
    {
      "epoch": 0.07748952290522863,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009879449529344568,
      "loss": 2.1249,
      "step": 11405
    },
    {
      "epoch": 0.07749631725182268,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009879426046168315,
      "loss": 2.2949,
      "step": 11406
    },
    {
      "epoch": 0.07750311159841675,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009879402560732948,
      "loss": 2.1249,
      "step": 11407
    },
    {
      "epoch": 0.07750990594501081,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000987937907303847,
      "loss": 2.175,
      "step": 11408
    },
    {
      "epoch": 0.07751670029160486,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009879355583084898,
      "loss": 2.172,
      "step": 11409
    },
    {
      "epoch": 0.07752349463819892,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000987933209087224,
      "loss": 2.2432,
      "step": 11410
    },
    {
      "epoch": 0.07753028898479297,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009879308596400505,
      "loss": 2.1418,
      "step": 11411
    },
    {
      "epoch": 0.07753708333138704,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000987928509966971,
      "loss": 2.2839,
      "step": 11412
    },
    {
      "epoch": 0.0775438776779811,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009879261600679856,
      "loss": 2.2095,
      "step": 11413
    },
    {
      "epoch": 0.07755067202457515,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009879238099430965,
      "loss": 2.1444,
      "step": 11414
    },
    {
      "epoch": 0.07755746637116921,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.000987921459592304,
      "loss": 2.4573,
      "step": 11415
    },
    {
      "epoch": 0.07756426071776326,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009879191090156098,
      "loss": 2.3209,
      "step": 11416
    },
    {
      "epoch": 0.07757105506435732,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009879167582130142,
      "loss": 2.3221,
      "step": 11417
    },
    {
      "epoch": 0.07757784941095139,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000987914407184519,
      "loss": 2.2877,
      "step": 11418
    },
    {
      "epoch": 0.07758464375754544,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000987912055930125,
      "loss": 2.2482,
      "step": 11419
    },
    {
      "epoch": 0.0775914381041395,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000987909704449833,
      "loss": 2.186,
      "step": 11420
    },
    {
      "epoch": 0.07759823245073355,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000987907352743645,
      "loss": 2.1298,
      "step": 11421
    },
    {
      "epoch": 0.07760502679732761,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000987905000811561,
      "loss": 2.2558,
      "step": 11422
    },
    {
      "epoch": 0.07761182114392168,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009879026486535828,
      "loss": 2.2392,
      "step": 11423
    },
    {
      "epoch": 0.07761861549051573,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000987900296269711,
      "loss": 2.169,
      "step": 11424
    },
    {
      "epoch": 0.07762540983710979,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009878979436599472,
      "loss": 2.2396,
      "step": 11425
    },
    {
      "epoch": 0.07763220418370384,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009878955908242922,
      "loss": 2.2699,
      "step": 11426
    },
    {
      "epoch": 0.0776389985302979,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000987893237762747,
      "loss": 2.1991,
      "step": 11427
    },
    {
      "epoch": 0.07764579287689197,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009878908844753133,
      "loss": 2.1308,
      "step": 11428
    },
    {
      "epoch": 0.07765258722348602,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000987888530961991,
      "loss": 2.1285,
      "step": 11429
    },
    {
      "epoch": 0.07765938157008008,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009878861772227826,
      "loss": 2.146,
      "step": 11430
    },
    {
      "epoch": 0.07766617591667413,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000987883823257688,
      "loss": 2.1977,
      "step": 11431
    },
    {
      "epoch": 0.07767297026326819,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000987881469066709,
      "loss": 2.3361,
      "step": 11432
    },
    {
      "epoch": 0.07767976460986224,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009878791146498466,
      "loss": 2.2891,
      "step": 11433
    },
    {
      "epoch": 0.0776865589564563,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009878767600071015,
      "loss": 2.1494,
      "step": 11434
    },
    {
      "epoch": 0.07769335330305037,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009878744051384752,
      "loss": 2.1871,
      "step": 11435
    },
    {
      "epoch": 0.07770014764964442,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009878720500439684,
      "loss": 2.2675,
      "step": 11436
    },
    {
      "epoch": 0.07770694199623848,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009878696947235826,
      "loss": 2.2274,
      "step": 11437
    },
    {
      "epoch": 0.07771373634283253,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000987867339177319,
      "loss": 2.1701,
      "step": 11438
    },
    {
      "epoch": 0.0777205306894266,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000987864983405178,
      "loss": 2.2041,
      "step": 11439
    },
    {
      "epoch": 0.07772732503602066,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009878626274071613,
      "loss": 2.3019,
      "step": 11440
    },
    {
      "epoch": 0.07773411938261471,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009878602711832698,
      "loss": 2.3405,
      "step": 11441
    },
    {
      "epoch": 0.07774091372920877,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009878579147335044,
      "loss": 2.3717,
      "step": 11442
    },
    {
      "epoch": 0.07774770807580282,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009878555580578666,
      "loss": 2.245,
      "step": 11443
    },
    {
      "epoch": 0.07775450242239688,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009878532011563573,
      "loss": 2.3099,
      "step": 11444
    },
    {
      "epoch": 0.07776129676899095,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009878508440289773,
      "loss": 2.2145,
      "step": 11445
    },
    {
      "epoch": 0.077768091115585,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009878484866757283,
      "loss": 2.2519,
      "step": 11446
    },
    {
      "epoch": 0.07777488546217906,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009878461290966107,
      "loss": 2.2648,
      "step": 11447
    },
    {
      "epoch": 0.07778167980877311,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009878437712916262,
      "loss": 2.2952,
      "step": 11448
    },
    {
      "epoch": 0.07778847415536717,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009878414132607756,
      "loss": 2.2286,
      "step": 11449
    },
    {
      "epoch": 0.07779526850196124,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009878390550040598,
      "loss": 2.3964,
      "step": 11450
    },
    {
      "epoch": 0.07780206284855529,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009878366965214802,
      "loss": 2.2555,
      "step": 11451
    },
    {
      "epoch": 0.07780885719514935,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000987834337813038,
      "loss": 2.1914,
      "step": 11452
    },
    {
      "epoch": 0.0778156515417434,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009878319788787337,
      "loss": 2.0479,
      "step": 11453
    },
    {
      "epoch": 0.07782244588833746,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009878296197185692,
      "loss": 2.2808,
      "step": 11454
    },
    {
      "epoch": 0.07782924023493151,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000987827260332545,
      "loss": 2.0967,
      "step": 11455
    },
    {
      "epoch": 0.07783603458152558,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009878249007206623,
      "loss": 2.315,
      "step": 11456
    },
    {
      "epoch": 0.07784282892811964,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009878225408829224,
      "loss": 2.2177,
      "step": 11457
    },
    {
      "epoch": 0.07784962327471369,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000987820180819326,
      "loss": 2.2087,
      "step": 11458
    },
    {
      "epoch": 0.07785641762130775,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000987817820529875,
      "loss": 2.1243,
      "step": 11459
    },
    {
      "epoch": 0.0778632119679018,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009878154600145694,
      "loss": 2.1465,
      "step": 11460
    },
    {
      "epoch": 0.07787000631449587,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000987813099273411,
      "loss": 2.3033,
      "step": 11461
    },
    {
      "epoch": 0.07787680066108993,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009878107383064008,
      "loss": 2.2419,
      "step": 11462
    },
    {
      "epoch": 0.07788359500768398,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009878083771135398,
      "loss": 2.2973,
      "step": 11463
    },
    {
      "epoch": 0.07789038935427804,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000987806015694829,
      "loss": 2.4083,
      "step": 11464
    },
    {
      "epoch": 0.07789718370087209,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009878036540502697,
      "loss": 2.1648,
      "step": 11465
    },
    {
      "epoch": 0.07790397804746615,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000987801292179863,
      "loss": 2.2316,
      "step": 11466
    },
    {
      "epoch": 0.07791077239406022,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009877989300836098,
      "loss": 2.1549,
      "step": 11467
    },
    {
      "epoch": 0.07791756674065427,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009877965677615112,
      "loss": 2.3519,
      "step": 11468
    },
    {
      "epoch": 0.07792436108724833,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009877942052135685,
      "loss": 2.23,
      "step": 11469
    },
    {
      "epoch": 0.07793115543384238,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009877918424397827,
      "loss": 2.2482,
      "step": 11470
    },
    {
      "epoch": 0.07793794978043644,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009877894794401547,
      "loss": 2.2503,
      "step": 11471
    },
    {
      "epoch": 0.07794474412703051,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009877871162146858,
      "loss": 2.2341,
      "step": 11472
    },
    {
      "epoch": 0.07795153847362456,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009877847527633773,
      "loss": 2.2424,
      "step": 11473
    },
    {
      "epoch": 0.07795833282021862,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009877823890862298,
      "loss": 2.1632,
      "step": 11474
    },
    {
      "epoch": 0.07796512716681267,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009877800251832448,
      "loss": 2.1593,
      "step": 11475
    },
    {
      "epoch": 0.07797192151340673,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000987777661054423,
      "loss": 2.2692,
      "step": 11476
    },
    {
      "epoch": 0.07797871586000078,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009877752966997659,
      "loss": 2.2253,
      "step": 11477
    },
    {
      "epoch": 0.07798551020659485,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009877729321192746,
      "loss": 2.1408,
      "step": 11478
    },
    {
      "epoch": 0.07799230455318891,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009877705673129499,
      "loss": 2.2636,
      "step": 11479
    },
    {
      "epoch": 0.07799909889978296,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009877682022807928,
      "loss": 2.1489,
      "step": 11480
    },
    {
      "epoch": 0.07800589324637702,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009877658370228048,
      "loss": 2.3583,
      "step": 11481
    },
    {
      "epoch": 0.07801268759297107,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000987763471538987,
      "loss": 2.2485,
      "step": 11482
    },
    {
      "epoch": 0.07801948193956514,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009877611058293398,
      "loss": 2.2603,
      "step": 11483
    },
    {
      "epoch": 0.0780262762861592,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009877587398938652,
      "loss": 2.31,
      "step": 11484
    },
    {
      "epoch": 0.07803307063275325,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009877563737325638,
      "loss": 2.3061,
      "step": 11485
    },
    {
      "epoch": 0.07803986497934731,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009877540073454367,
      "loss": 2.1548,
      "step": 11486
    },
    {
      "epoch": 0.07804665932594136,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009877516407324853,
      "loss": 2.3867,
      "step": 11487
    },
    {
      "epoch": 0.07805345367253542,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009877492738937105,
      "loss": 2.2357,
      "step": 11488
    },
    {
      "epoch": 0.07806024801912949,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000987746906829113,
      "loss": 2.1477,
      "step": 11489
    },
    {
      "epoch": 0.07806704236572354,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009877445395386946,
      "loss": 2.4351,
      "step": 11490
    },
    {
      "epoch": 0.0780738367123176,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000987742172022456,
      "loss": 2.2141,
      "step": 11491
    },
    {
      "epoch": 0.07808063105891165,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009877398042803982,
      "loss": 2.3708,
      "step": 11492
    },
    {
      "epoch": 0.07808742540550571,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009877374363125226,
      "loss": 2.1973,
      "step": 11493
    },
    {
      "epoch": 0.07809421975209978,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009877350681188302,
      "loss": 2.2431,
      "step": 11494
    },
    {
      "epoch": 0.07810101409869383,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009877326996993222,
      "loss": 2.3177,
      "step": 11495
    },
    {
      "epoch": 0.07810780844528789,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009877303310539991,
      "loss": 2.3043,
      "step": 11496
    },
    {
      "epoch": 0.07811460279188194,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009877279621828628,
      "loss": 2.1334,
      "step": 11497
    },
    {
      "epoch": 0.078121397138476,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000987725593085914,
      "loss": 2.1042,
      "step": 11498
    },
    {
      "epoch": 0.07812819148507007,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009877232237631539,
      "loss": 2.3152,
      "step": 11499
    },
    {
      "epoch": 0.07813498583166412,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009877208542145833,
      "loss": 2.1304,
      "step": 11500
    },
    {
      "epoch": 0.07814178017825818,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009877184844402037,
      "loss": 2.3036,
      "step": 11501
    },
    {
      "epoch": 0.07814857452485223,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000987716114440016,
      "loss": 2.0697,
      "step": 11502
    },
    {
      "epoch": 0.07815536887144629,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009877137442140214,
      "loss": 2.2981,
      "step": 11503
    },
    {
      "epoch": 0.07816216321804034,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009877113737622209,
      "loss": 2.1206,
      "step": 11504
    },
    {
      "epoch": 0.0781689575646344,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009877090030846157,
      "loss": 2.0842,
      "step": 11505
    },
    {
      "epoch": 0.07817575191122847,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009877066321812066,
      "loss": 2.2569,
      "step": 11506
    },
    {
      "epoch": 0.07818254625782252,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000987704261051995,
      "loss": 2.3644,
      "step": 11507
    },
    {
      "epoch": 0.07818934060441658,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000987701889696982,
      "loss": 2.1786,
      "step": 11508
    },
    {
      "epoch": 0.07819613495101063,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009876995181161685,
      "loss": 2.3319,
      "step": 11509
    },
    {
      "epoch": 0.0782029292976047,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009876971463095559,
      "loss": 2.1373,
      "step": 11510
    },
    {
      "epoch": 0.07820972364419876,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000987694774277145,
      "loss": 2.3636,
      "step": 11511
    },
    {
      "epoch": 0.07821651799079281,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009876924020189369,
      "loss": 2.1021,
      "step": 11512
    },
    {
      "epoch": 0.07822331233738687,
      "grad_norm": 0.15625,
      "learning_rate": 0.000987690029534933,
      "loss": 2.2468,
      "step": 11513
    },
    {
      "epoch": 0.07823010668398092,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009876876568251341,
      "loss": 2.2556,
      "step": 11514
    },
    {
      "epoch": 0.07823690103057498,
      "grad_norm": 0.2333984375,
      "learning_rate": 0.0009876852838895413,
      "loss": 2.2373,
      "step": 11515
    },
    {
      "epoch": 0.07824369537716905,
      "grad_norm": 0.1640625,
      "learning_rate": 0.000987682910728156,
      "loss": 2.314,
      "step": 11516
    },
    {
      "epoch": 0.0782504897237631,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009876805373409791,
      "loss": 2.4241,
      "step": 11517
    },
    {
      "epoch": 0.07825728407035716,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009876781637280117,
      "loss": 2.288,
      "step": 11518
    },
    {
      "epoch": 0.07826407841695121,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009876757898892548,
      "loss": 2.2197,
      "step": 11519
    },
    {
      "epoch": 0.07827087276354527,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009876734158247096,
      "loss": 2.2167,
      "step": 11520
    },
    {
      "epoch": 0.07827766711013934,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009876710415343774,
      "loss": 2.2096,
      "step": 11521
    },
    {
      "epoch": 0.07828446145673339,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000987668667018259,
      "loss": 2.1824,
      "step": 11522
    },
    {
      "epoch": 0.07829125580332745,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009876662922763554,
      "loss": 2.3784,
      "step": 11523
    },
    {
      "epoch": 0.0782980501499215,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009876639173086682,
      "loss": 2.1635,
      "step": 11524
    },
    {
      "epoch": 0.07830484449651556,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000987661542115198,
      "loss": 2.153,
      "step": 11525
    },
    {
      "epoch": 0.07831163884310961,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000987659166695946,
      "loss": 2.1099,
      "step": 11526
    },
    {
      "epoch": 0.07831843318970368,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009876567910509135,
      "loss": 2.2288,
      "step": 11527
    },
    {
      "epoch": 0.07832522753629774,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009876544151801016,
      "loss": 2.3008,
      "step": 11528
    },
    {
      "epoch": 0.07833202188289179,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009876520390835113,
      "loss": 2.1968,
      "step": 11529
    },
    {
      "epoch": 0.07833881622948585,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009876496627611434,
      "loss": 2.3186,
      "step": 11530
    },
    {
      "epoch": 0.0783456105760799,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009876472862129996,
      "loss": 2.3195,
      "step": 11531
    },
    {
      "epoch": 0.07835240492267397,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009876449094390807,
      "loss": 2.0873,
      "step": 11532
    },
    {
      "epoch": 0.07835919926926803,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009876425324393877,
      "loss": 2.2217,
      "step": 11533
    },
    {
      "epoch": 0.07836599361586208,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009876401552139217,
      "loss": 2.214,
      "step": 11534
    },
    {
      "epoch": 0.07837278796245614,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.000987637777762684,
      "loss": 2.3434,
      "step": 11535
    },
    {
      "epoch": 0.07837958230905019,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009876354000856757,
      "loss": 2.1102,
      "step": 11536
    },
    {
      "epoch": 0.07838637665564425,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009876330221828978,
      "loss": 2.2027,
      "step": 11537
    },
    {
      "epoch": 0.07839317100223832,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009876306440543513,
      "loss": 2.1283,
      "step": 11538
    },
    {
      "epoch": 0.07839996534883237,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009876282657000372,
      "loss": 2.2971,
      "step": 11539
    },
    {
      "epoch": 0.07840675969542643,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009876258871199571,
      "loss": 2.1954,
      "step": 11540
    },
    {
      "epoch": 0.07841355404202048,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009876235083141117,
      "loss": 2.3555,
      "step": 11541
    },
    {
      "epoch": 0.07842034838861454,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009876211292825024,
      "loss": 2.4226,
      "step": 11542
    },
    {
      "epoch": 0.07842714273520861,
      "grad_norm": 0.140625,
      "learning_rate": 0.00098761875002513,
      "loss": 2.2422,
      "step": 11543
    },
    {
      "epoch": 0.07843393708180266,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009876163705419953,
      "loss": 2.0614,
      "step": 11544
    },
    {
      "epoch": 0.07844073142839672,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009876139908331003,
      "loss": 2.1551,
      "step": 11545
    },
    {
      "epoch": 0.07844752577499077,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009876116108984453,
      "loss": 2.179,
      "step": 11546
    },
    {
      "epoch": 0.07845432012158483,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000987609230738032,
      "loss": 2.2154,
      "step": 11547
    },
    {
      "epoch": 0.0784611144681789,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009876068503518612,
      "loss": 2.1851,
      "step": 11548
    },
    {
      "epoch": 0.07846790881477295,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009876044697399338,
      "loss": 2.1206,
      "step": 11549
    },
    {
      "epoch": 0.07847470316136701,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009876020889022512,
      "loss": 2.3084,
      "step": 11550
    },
    {
      "epoch": 0.07848149750796106,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009875997078388145,
      "loss": 2.1807,
      "step": 11551
    },
    {
      "epoch": 0.07848829185455512,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009875973265496248,
      "loss": 2.206,
      "step": 11552
    },
    {
      "epoch": 0.07849508620114917,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000987594945034683,
      "loss": 2.1328,
      "step": 11553
    },
    {
      "epoch": 0.07850188054774324,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009875925632939903,
      "loss": 2.2814,
      "step": 11554
    },
    {
      "epoch": 0.0785086748943373,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009875901813275478,
      "loss": 2.4612,
      "step": 11555
    },
    {
      "epoch": 0.07851546924093135,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009875877991353567,
      "loss": 2.1287,
      "step": 11556
    },
    {
      "epoch": 0.07852226358752541,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000987585416717418,
      "loss": 2.2944,
      "step": 11557
    },
    {
      "epoch": 0.07852905793411946,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000987583034073733,
      "loss": 2.312,
      "step": 11558
    },
    {
      "epoch": 0.07853585228071353,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009875806512043026,
      "loss": 2.2004,
      "step": 11559
    },
    {
      "epoch": 0.07854264662730759,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009875782681091278,
      "loss": 2.2154,
      "step": 11560
    },
    {
      "epoch": 0.07854944097390164,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00098757588478821,
      "loss": 2.2092,
      "step": 11561
    },
    {
      "epoch": 0.0785562353204957,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00098757350124155,
      "loss": 2.1973,
      "step": 11562
    },
    {
      "epoch": 0.07856302966708975,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009875711174691492,
      "loss": 2.2576,
      "step": 11563
    },
    {
      "epoch": 0.07856982401368381,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009875687334710086,
      "loss": 2.3138,
      "step": 11564
    },
    {
      "epoch": 0.07857661836027788,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000987566349247129,
      "loss": 2.322,
      "step": 11565
    },
    {
      "epoch": 0.07858341270687193,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009875639647975122,
      "loss": 2.2919,
      "step": 11566
    },
    {
      "epoch": 0.07859020705346599,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009875615801221585,
      "loss": 2.2415,
      "step": 11567
    },
    {
      "epoch": 0.07859700140006004,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009875591952210696,
      "loss": 2.2981,
      "step": 11568
    },
    {
      "epoch": 0.0786037957466541,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009875568100942464,
      "loss": 2.2306,
      "step": 11569
    },
    {
      "epoch": 0.07861059009324817,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009875544247416896,
      "loss": 2.1753,
      "step": 11570
    },
    {
      "epoch": 0.07861738443984222,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009875520391634009,
      "loss": 2.249,
      "step": 11571
    },
    {
      "epoch": 0.07862417878643628,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009875496533593812,
      "loss": 2.3586,
      "step": 11572
    },
    {
      "epoch": 0.07863097313303033,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009875472673296318,
      "loss": 2.3247,
      "step": 11573
    },
    {
      "epoch": 0.0786377674796244,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009875448810741534,
      "loss": 2.1972,
      "step": 11574
    },
    {
      "epoch": 0.07864456182621844,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009875424945929474,
      "loss": 2.2877,
      "step": 11575
    },
    {
      "epoch": 0.0786513561728125,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009875401078860149,
      "loss": 2.3438,
      "step": 11576
    },
    {
      "epoch": 0.07865815051940657,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009875377209533566,
      "loss": 2.1835,
      "step": 11577
    },
    {
      "epoch": 0.07866494486600062,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009875353337949742,
      "loss": 2.3283,
      "step": 11578
    },
    {
      "epoch": 0.07867173921259468,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009875329464108685,
      "loss": 2.2573,
      "step": 11579
    },
    {
      "epoch": 0.07867853355918873,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009875305588010405,
      "loss": 2.1629,
      "step": 11580
    },
    {
      "epoch": 0.0786853279057828,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009875281709654917,
      "loss": 2.2501,
      "step": 11581
    },
    {
      "epoch": 0.07869212225237686,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009875257829042227,
      "loss": 2.1715,
      "step": 11582
    },
    {
      "epoch": 0.07869891659897091,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000987523394617235,
      "loss": 2.1077,
      "step": 11583
    },
    {
      "epoch": 0.07870571094556497,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009875210061045295,
      "loss": 2.1637,
      "step": 11584
    },
    {
      "epoch": 0.07871250529215902,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009875186173661072,
      "loss": 2.3653,
      "step": 11585
    },
    {
      "epoch": 0.07871929963875308,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009875162284019696,
      "loss": 2.1791,
      "step": 11586
    },
    {
      "epoch": 0.07872609398534715,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009875138392121175,
      "loss": 2.264,
      "step": 11587
    },
    {
      "epoch": 0.0787328883319412,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000987511449796552,
      "loss": 2.3035,
      "step": 11588
    },
    {
      "epoch": 0.07873968267853526,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0009875090601552746,
      "loss": 2.3222,
      "step": 11589
    },
    {
      "epoch": 0.07874647702512931,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009875066702882858,
      "loss": 2.4781,
      "step": 11590
    },
    {
      "epoch": 0.07875327137172337,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000987504280195587,
      "loss": 2.2595,
      "step": 11591
    },
    {
      "epoch": 0.07876006571831744,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009875018898771793,
      "loss": 2.2853,
      "step": 11592
    },
    {
      "epoch": 0.07876686006491149,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000987499499333064,
      "loss": 2.1612,
      "step": 11593
    },
    {
      "epoch": 0.07877365441150555,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009874971085632419,
      "loss": 2.2215,
      "step": 11594
    },
    {
      "epoch": 0.0787804487580996,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0009874947175677142,
      "loss": 2.1956,
      "step": 11595
    },
    {
      "epoch": 0.07878724310469366,
      "grad_norm": 0.2578125,
      "learning_rate": 0.0009874923263464821,
      "loss": 2.3337,
      "step": 11596
    },
    {
      "epoch": 0.07879403745128771,
      "grad_norm": 0.361328125,
      "learning_rate": 0.0009874899348995465,
      "loss": 2.2299,
      "step": 11597
    },
    {
      "epoch": 0.07880083179788178,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009874875432269088,
      "loss": 2.355,
      "step": 11598
    },
    {
      "epoch": 0.07880762614447584,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009874851513285699,
      "loss": 2.312,
      "step": 11599
    },
    {
      "epoch": 0.07881442049106989,
      "grad_norm": 0.169921875,
      "learning_rate": 0.000987482759204531,
      "loss": 2.1352,
      "step": 11600
    },
    {
      "epoch": 0.07882121483766395,
      "grad_norm": 0.166015625,
      "learning_rate": 0.000987480366854793,
      "loss": 2.1949,
      "step": 11601
    },
    {
      "epoch": 0.078828009184258,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009874779742793576,
      "loss": 2.3281,
      "step": 11602
    },
    {
      "epoch": 0.07883480353085207,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009874755814782252,
      "loss": 2.2513,
      "step": 11603
    },
    {
      "epoch": 0.07884159787744613,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009874731884513973,
      "loss": 2.2608,
      "step": 11604
    },
    {
      "epoch": 0.07884839222404018,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009874707951988747,
      "loss": 2.2959,
      "step": 11605
    },
    {
      "epoch": 0.07885518657063424,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009874684017206587,
      "loss": 2.0392,
      "step": 11606
    },
    {
      "epoch": 0.07886198091722829,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009874660080167507,
      "loss": 2.3478,
      "step": 11607
    },
    {
      "epoch": 0.07886877526382235,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009874636140871514,
      "loss": 2.1598,
      "step": 11608
    },
    {
      "epoch": 0.07887556961041642,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009874612199318618,
      "loss": 2.2402,
      "step": 11609
    },
    {
      "epoch": 0.07888236395701047,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009874588255508836,
      "loss": 2.3635,
      "step": 11610
    },
    {
      "epoch": 0.07888915830360453,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009874564309442174,
      "loss": 2.2625,
      "step": 11611
    },
    {
      "epoch": 0.07889595265019858,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009874540361118644,
      "loss": 2.2128,
      "step": 11612
    },
    {
      "epoch": 0.07890274699679264,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009874516410538258,
      "loss": 2.2652,
      "step": 11613
    },
    {
      "epoch": 0.07890954134338671,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009874492457701027,
      "loss": 2.2455,
      "step": 11614
    },
    {
      "epoch": 0.07891633568998076,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009874468502606962,
      "loss": 2.2471,
      "step": 11615
    },
    {
      "epoch": 0.07892313003657482,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009874444545256073,
      "loss": 2.2677,
      "step": 11616
    },
    {
      "epoch": 0.07892992438316887,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009874420585648373,
      "loss": 2.286,
      "step": 11617
    },
    {
      "epoch": 0.07893671872976293,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009874396623783873,
      "loss": 2.2755,
      "step": 11618
    },
    {
      "epoch": 0.078943513076357,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009874372659662582,
      "loss": 2.2091,
      "step": 11619
    },
    {
      "epoch": 0.07895030742295105,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009874348693284514,
      "loss": 2.2454,
      "step": 11620
    },
    {
      "epoch": 0.07895710176954511,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009874324724649676,
      "loss": 2.3013,
      "step": 11621
    },
    {
      "epoch": 0.07896389611613916,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009874300753758083,
      "loss": 2.3584,
      "step": 11622
    },
    {
      "epoch": 0.07897069046273322,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009874276780609745,
      "loss": 2.2964,
      "step": 11623
    },
    {
      "epoch": 0.07897748480932727,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000987425280520467,
      "loss": 2.1229,
      "step": 11624
    },
    {
      "epoch": 0.07898427915592134,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009874228827542876,
      "loss": 2.1072,
      "step": 11625
    },
    {
      "epoch": 0.0789910735025154,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009874204847624367,
      "loss": 2.1085,
      "step": 11626
    },
    {
      "epoch": 0.07899786784910945,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009874180865449159,
      "loss": 2.2801,
      "step": 11627
    },
    {
      "epoch": 0.07900466219570351,
      "grad_norm": 0.140625,
      "learning_rate": 0.000987415688101726,
      "loss": 2.1357,
      "step": 11628
    },
    {
      "epoch": 0.07901145654229756,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009874132894328682,
      "loss": 2.2323,
      "step": 11629
    },
    {
      "epoch": 0.07901825088889163,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009874108905383436,
      "loss": 2.164,
      "step": 11630
    },
    {
      "epoch": 0.07902504523548569,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009874084914181533,
      "loss": 2.2601,
      "step": 11631
    },
    {
      "epoch": 0.07903183958207974,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009874060920722987,
      "loss": 2.1219,
      "step": 11632
    },
    {
      "epoch": 0.0790386339286738,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009874036925007807,
      "loss": 2.3083,
      "step": 11633
    },
    {
      "epoch": 0.07904542827526785,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009874012927036,
      "loss": 2.2227,
      "step": 11634
    },
    {
      "epoch": 0.07905222262186191,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009873988926807583,
      "loss": 2.4188,
      "step": 11635
    },
    {
      "epoch": 0.07905901696845598,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009873964924322564,
      "loss": 2.4189,
      "step": 11636
    },
    {
      "epoch": 0.07906581131505003,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009873940919580958,
      "loss": 2.2914,
      "step": 11637
    },
    {
      "epoch": 0.07907260566164409,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000987391691258277,
      "loss": 2.3321,
      "step": 11638
    },
    {
      "epoch": 0.07907940000823814,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009873892903328016,
      "loss": 2.1569,
      "step": 11639
    },
    {
      "epoch": 0.0790861943548322,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009873868891816704,
      "loss": 2.3323,
      "step": 11640
    },
    {
      "epoch": 0.07909298870142627,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009873844878048848,
      "loss": 2.2011,
      "step": 11641
    },
    {
      "epoch": 0.07909978304802032,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009873820862024458,
      "loss": 2.057,
      "step": 11642
    },
    {
      "epoch": 0.07910657739461438,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009873796843743543,
      "loss": 2.1059,
      "step": 11643
    },
    {
      "epoch": 0.07911337174120843,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009873772823206118,
      "loss": 2.3557,
      "step": 11644
    },
    {
      "epoch": 0.0791201660878025,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009873748800412192,
      "loss": 2.1048,
      "step": 11645
    },
    {
      "epoch": 0.07912696043439654,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009873724775361775,
      "loss": 2.2139,
      "step": 11646
    },
    {
      "epoch": 0.0791337547809906,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000987370074805488,
      "loss": 2.2209,
      "step": 11647
    },
    {
      "epoch": 0.07914054912758467,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009873676718491517,
      "loss": 2.1844,
      "step": 11648
    },
    {
      "epoch": 0.07914734347417872,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009873652686671697,
      "loss": 2.2221,
      "step": 11649
    },
    {
      "epoch": 0.07915413782077278,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009873628652595431,
      "loss": 2.1296,
      "step": 11650
    },
    {
      "epoch": 0.07916093216736683,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009873604616262733,
      "loss": 2.1037,
      "step": 11651
    },
    {
      "epoch": 0.0791677265139609,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009873580577673612,
      "loss": 2.1573,
      "step": 11652
    },
    {
      "epoch": 0.07917452086055496,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009873556536828078,
      "loss": 2.364,
      "step": 11653
    },
    {
      "epoch": 0.07918131520714901,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009873532493726142,
      "loss": 2.2265,
      "step": 11654
    },
    {
      "epoch": 0.07918810955374307,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009873508448367818,
      "loss": 2.2608,
      "step": 11655
    },
    {
      "epoch": 0.07919490390033712,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009873484400753116,
      "loss": 2.2268,
      "step": 11656
    },
    {
      "epoch": 0.07920169824693118,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009873460350882045,
      "loss": 2.2967,
      "step": 11657
    },
    {
      "epoch": 0.07920849259352525,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009873436298754618,
      "loss": 2.1634,
      "step": 11658
    },
    {
      "epoch": 0.0792152869401193,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009873412244370847,
      "loss": 2.1951,
      "step": 11659
    },
    {
      "epoch": 0.07922208128671336,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009873388187730742,
      "loss": 2.2879,
      "step": 11660
    },
    {
      "epoch": 0.07922887563330741,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009873364128834313,
      "loss": 2.1945,
      "step": 11661
    },
    {
      "epoch": 0.07923566997990147,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009873340067681573,
      "loss": 2.211,
      "step": 11662
    },
    {
      "epoch": 0.07924246432649554,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009873316004272532,
      "loss": 2.149,
      "step": 11663
    },
    {
      "epoch": 0.07924925867308959,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009873291938607203,
      "loss": 2.1375,
      "step": 11664
    },
    {
      "epoch": 0.07925605301968365,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009873267870685594,
      "loss": 2.3096,
      "step": 11665
    },
    {
      "epoch": 0.0792628473662777,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009873243800507718,
      "loss": 2.2467,
      "step": 11666
    },
    {
      "epoch": 0.07926964171287176,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009873219728073585,
      "loss": 2.2531,
      "step": 11667
    },
    {
      "epoch": 0.07927643605946583,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009873195653383208,
      "loss": 2.2078,
      "step": 11668
    },
    {
      "epoch": 0.07928323040605988,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.00098731715764366,
      "loss": 2.1568,
      "step": 11669
    },
    {
      "epoch": 0.07929002475265394,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009873147497233767,
      "loss": 2.3223,
      "step": 11670
    },
    {
      "epoch": 0.07929681909924799,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009873123415774722,
      "loss": 2.338,
      "step": 11671
    },
    {
      "epoch": 0.07930361344584205,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009873099332059476,
      "loss": 2.0372,
      "step": 11672
    },
    {
      "epoch": 0.0793104077924361,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009873075246088043,
      "loss": 2.3229,
      "step": 11673
    },
    {
      "epoch": 0.07931720213903017,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009873051157860433,
      "loss": 2.4006,
      "step": 11674
    },
    {
      "epoch": 0.07932399648562423,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009873027067376654,
      "loss": 2.1336,
      "step": 11675
    },
    {
      "epoch": 0.07933079083221828,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000987300297463672,
      "loss": 2.3578,
      "step": 11676
    },
    {
      "epoch": 0.07933758517881234,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000987297887964064,
      "loss": 2.2483,
      "step": 11677
    },
    {
      "epoch": 0.07934437952540639,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009872954782388429,
      "loss": 2.3255,
      "step": 11678
    },
    {
      "epoch": 0.07935117387200046,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009872930682880096,
      "loss": 2.0958,
      "step": 11679
    },
    {
      "epoch": 0.07935796821859452,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009872906581115649,
      "loss": 2.1872,
      "step": 11680
    },
    {
      "epoch": 0.07936476256518857,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009872882477095105,
      "loss": 2.2304,
      "step": 11681
    },
    {
      "epoch": 0.07937155691178263,
      "grad_norm": 0.140625,
      "learning_rate": 0.000987285837081847,
      "loss": 2.2252,
      "step": 11682
    },
    {
      "epoch": 0.07937835125837668,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000987283426228576,
      "loss": 2.3547,
      "step": 11683
    },
    {
      "epoch": 0.07938514560497074,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009872810151496981,
      "loss": 2.2007,
      "step": 11684
    },
    {
      "epoch": 0.07939193995156481,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009872786038452148,
      "loss": 2.13,
      "step": 11685
    },
    {
      "epoch": 0.07939873429815886,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009872761923151272,
      "loss": 2.3758,
      "step": 11686
    },
    {
      "epoch": 0.07940552864475292,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000987273780559436,
      "loss": 2.2439,
      "step": 11687
    },
    {
      "epoch": 0.07941232299134697,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000987271368578143,
      "loss": 2.2876,
      "step": 11688
    },
    {
      "epoch": 0.07941911733794103,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009872689563712486,
      "loss": 2.1759,
      "step": 11689
    },
    {
      "epoch": 0.0794259116845351,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009872665439387544,
      "loss": 2.31,
      "step": 11690
    },
    {
      "epoch": 0.07943270603112915,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009872641312806615,
      "loss": 2.2461,
      "step": 11691
    },
    {
      "epoch": 0.07943950037772321,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009872617183969707,
      "loss": 2.2032,
      "step": 11692
    },
    {
      "epoch": 0.07944629472431726,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009872593052876835,
      "loss": 2.3566,
      "step": 11693
    },
    {
      "epoch": 0.07945308907091132,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009872568919528006,
      "loss": 2.2693,
      "step": 11694
    },
    {
      "epoch": 0.07945988341750537,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009872544783923237,
      "loss": 2.2796,
      "step": 11695
    },
    {
      "epoch": 0.07946667776409944,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009872520646062532,
      "loss": 2.2877,
      "step": 11696
    },
    {
      "epoch": 0.0794734721106935,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009872496505945907,
      "loss": 2.2585,
      "step": 11697
    },
    {
      "epoch": 0.07948026645728755,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009872472363573373,
      "loss": 2.0393,
      "step": 11698
    },
    {
      "epoch": 0.07948706080388161,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009872448218944938,
      "loss": 2.0762,
      "step": 11699
    },
    {
      "epoch": 0.07949385515047566,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009872424072060618,
      "loss": 2.1897,
      "step": 11700
    },
    {
      "epoch": 0.07950064949706973,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000987239992292042,
      "loss": 2.1567,
      "step": 11701
    },
    {
      "epoch": 0.07950744384366379,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009872375771524356,
      "loss": 2.3319,
      "step": 11702
    },
    {
      "epoch": 0.07951423819025784,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000987235161787244,
      "loss": 2.2701,
      "step": 11703
    },
    {
      "epoch": 0.0795210325368519,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000987232746196468,
      "loss": 2.2464,
      "step": 11704
    },
    {
      "epoch": 0.07952782688344595,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009872303303801087,
      "loss": 2.0902,
      "step": 11705
    },
    {
      "epoch": 0.07953462123004001,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009872279143381675,
      "loss": 2.3571,
      "step": 11706
    },
    {
      "epoch": 0.07954141557663408,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009872254980706453,
      "loss": 2.175,
      "step": 11707
    },
    {
      "epoch": 0.07954820992322813,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009872230815775433,
      "loss": 2.102,
      "step": 11708
    },
    {
      "epoch": 0.07955500426982219,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009872206648588627,
      "loss": 2.2578,
      "step": 11709
    },
    {
      "epoch": 0.07956179861641624,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009872182479146044,
      "loss": 2.2928,
      "step": 11710
    },
    {
      "epoch": 0.0795685929630103,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009872158307447697,
      "loss": 2.2254,
      "step": 11711
    },
    {
      "epoch": 0.07957538730960437,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009872134133493595,
      "loss": 2.2717,
      "step": 11712
    },
    {
      "epoch": 0.07958218165619842,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000987210995728375,
      "loss": 2.3238,
      "step": 11713
    },
    {
      "epoch": 0.07958897600279248,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009872085778818175,
      "loss": 2.2209,
      "step": 11714
    },
    {
      "epoch": 0.07959577034938653,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009872061598096883,
      "loss": 2.2061,
      "step": 11715
    },
    {
      "epoch": 0.0796025646959806,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000987203741511988,
      "loss": 2.3145,
      "step": 11716
    },
    {
      "epoch": 0.07960935904257464,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000987201322988718,
      "loss": 2.2065,
      "step": 11717
    },
    {
      "epoch": 0.0796161533891687,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009871989042398793,
      "loss": 2.2389,
      "step": 11718
    },
    {
      "epoch": 0.07962294773576277,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000987196485265473,
      "loss": 2.2346,
      "step": 11719
    },
    {
      "epoch": 0.07962974208235682,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009871940660655006,
      "loss": 2.1566,
      "step": 11720
    },
    {
      "epoch": 0.07963653642895088,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009871916466399628,
      "loss": 2.161,
      "step": 11721
    },
    {
      "epoch": 0.07964333077554493,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009871892269888607,
      "loss": 2.1851,
      "step": 11722
    },
    {
      "epoch": 0.079650125122139,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009871868071121958,
      "loss": 2.1405,
      "step": 11723
    },
    {
      "epoch": 0.07965691946873306,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009871843870099689,
      "loss": 2.2074,
      "step": 11724
    },
    {
      "epoch": 0.07966371381532711,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009871819666821813,
      "loss": 2.2149,
      "step": 11725
    },
    {
      "epoch": 0.07967050816192117,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000987179546128834,
      "loss": 2.2604,
      "step": 11726
    },
    {
      "epoch": 0.07967730250851522,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000987177125349928,
      "loss": 2.2195,
      "step": 11727
    },
    {
      "epoch": 0.07968409685510928,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009871747043454647,
      "loss": 2.1337,
      "step": 11728
    },
    {
      "epoch": 0.07969089120170335,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000987172283115445,
      "loss": 2.0457,
      "step": 11729
    },
    {
      "epoch": 0.0796976855482974,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009871698616598704,
      "loss": 2.0943,
      "step": 11730
    },
    {
      "epoch": 0.07970447989489146,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009871674399787413,
      "loss": 2.1497,
      "step": 11731
    },
    {
      "epoch": 0.07971127424148551,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009871650180720596,
      "loss": 2.1595,
      "step": 11732
    },
    {
      "epoch": 0.07971806858807957,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009871625959398261,
      "loss": 2.2533,
      "step": 11733
    },
    {
      "epoch": 0.07972486293467364,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009871601735820417,
      "loss": 2.1418,
      "step": 11734
    },
    {
      "epoch": 0.07973165728126769,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009871577509987077,
      "loss": 2.1504,
      "step": 11735
    },
    {
      "epoch": 0.07973845162786175,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009871553281898253,
      "loss": 2.1727,
      "step": 11736
    },
    {
      "epoch": 0.0797452459744558,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009871529051553958,
      "loss": 2.1749,
      "step": 11737
    },
    {
      "epoch": 0.07975204032104986,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009871504818954198,
      "loss": 2.305,
      "step": 11738
    },
    {
      "epoch": 0.07975883466764393,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009871480584098989,
      "loss": 2.4084,
      "step": 11739
    },
    {
      "epoch": 0.07976562901423798,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009871456346988339,
      "loss": 2.2987,
      "step": 11740
    },
    {
      "epoch": 0.07977242336083204,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000987143210762226,
      "loss": 2.1525,
      "step": 11741
    },
    {
      "epoch": 0.07977921770742609,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009871407866000766,
      "loss": 2.1409,
      "step": 11742
    },
    {
      "epoch": 0.07978601205402015,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009871383622123864,
      "loss": 2.3067,
      "step": 11743
    },
    {
      "epoch": 0.0797928064006142,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009871359375991568,
      "loss": 2.1468,
      "step": 11744
    },
    {
      "epoch": 0.07979960074720827,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000987133512760389,
      "loss": 2.3603,
      "step": 11745
    },
    {
      "epoch": 0.07980639509380233,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000987131087696084,
      "loss": 2.1906,
      "step": 11746
    },
    {
      "epoch": 0.07981318944039638,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009871286624062426,
      "loss": 2.1055,
      "step": 11747
    },
    {
      "epoch": 0.07981998378699044,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009871262368908661,
      "loss": 2.1857,
      "step": 11748
    },
    {
      "epoch": 0.07982677813358449,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009871238111499561,
      "loss": 2.1824,
      "step": 11749
    },
    {
      "epoch": 0.07983357248017856,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009871213851835132,
      "loss": 2.1817,
      "step": 11750
    },
    {
      "epoch": 0.07984036682677262,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000987118958991539,
      "loss": 2.2891,
      "step": 11751
    },
    {
      "epoch": 0.07984716117336667,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009871165325740338,
      "loss": 2.0448,
      "step": 11752
    },
    {
      "epoch": 0.07985395551996073,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009871141059309995,
      "loss": 2.3319,
      "step": 11753
    },
    {
      "epoch": 0.07986074986655478,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009871116790624369,
      "loss": 2.3157,
      "step": 11754
    },
    {
      "epoch": 0.07986754421314884,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009871092519683471,
      "loss": 2.2278,
      "step": 11755
    },
    {
      "epoch": 0.07987433855974291,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009871068246487314,
      "loss": 2.2736,
      "step": 11756
    },
    {
      "epoch": 0.07988113290633696,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009871043971035908,
      "loss": 2.1394,
      "step": 11757
    },
    {
      "epoch": 0.07988792725293102,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009871019693329264,
      "loss": 2.3211,
      "step": 11758
    },
    {
      "epoch": 0.07989472159952507,
      "grad_norm": 0.11669921875,
      "learning_rate": 0.0009870995413367397,
      "loss": 2.1718,
      "step": 11759
    },
    {
      "epoch": 0.07990151594611913,
      "grad_norm": 0.12109375,
      "learning_rate": 0.000987097113115031,
      "loss": 2.2121,
      "step": 11760
    },
    {
      "epoch": 0.0799083102927132,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009870946846678023,
      "loss": 2.2423,
      "step": 11761
    },
    {
      "epoch": 0.07991510463930725,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009870922559950541,
      "loss": 2.1177,
      "step": 11762
    },
    {
      "epoch": 0.07992189898590131,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009870898270967878,
      "loss": 2.3263,
      "step": 11763
    },
    {
      "epoch": 0.07992869333249536,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009870873979730046,
      "loss": 2.1155,
      "step": 11764
    },
    {
      "epoch": 0.07993548767908942,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009870849686237056,
      "loss": 2.196,
      "step": 11765
    },
    {
      "epoch": 0.07994228202568347,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009870825390488915,
      "loss": 2.3374,
      "step": 11766
    },
    {
      "epoch": 0.07994907637227754,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009870801092485642,
      "loss": 2.2296,
      "step": 11767
    },
    {
      "epoch": 0.0799558707188716,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009870776792227241,
      "loss": 2.2629,
      "step": 11768
    },
    {
      "epoch": 0.07996266506546565,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009870752489713726,
      "loss": 2.3554,
      "step": 11769
    },
    {
      "epoch": 0.07996945941205971,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000987072818494511,
      "loss": 2.2158,
      "step": 11770
    },
    {
      "epoch": 0.07997625375865376,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009870703877921402,
      "loss": 2.1414,
      "step": 11771
    },
    {
      "epoch": 0.07998304810524783,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009870679568642613,
      "loss": 2.316,
      "step": 11772
    },
    {
      "epoch": 0.07998984245184189,
      "grad_norm": 0.125,
      "learning_rate": 0.0009870655257108757,
      "loss": 2.1034,
      "step": 11773
    },
    {
      "epoch": 0.07999663679843594,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009870630943319843,
      "loss": 2.1748,
      "step": 11774
    },
    {
      "epoch": 0.08000343114503,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009870606627275882,
      "loss": 2.2163,
      "step": 11775
    },
    {
      "epoch": 0.08001022549162405,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009870582308976888,
      "loss": 2.3183,
      "step": 11776
    },
    {
      "epoch": 0.08001022549162405,
      "eval_loss": 2.29274845123291,
      "eval_runtime": 2364.3381,
      "eval_samples_per_second": 15.968,
      "eval_steps_per_second": 3.992,
      "step": 11776
    },
    {
      "epoch": 0.08001701983821811,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009870557988422867,
      "loss": 2.2005,
      "step": 11777
    },
    {
      "epoch": 0.08002381418481218,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009870533665613836,
      "loss": 2.3048,
      "step": 11778
    },
    {
      "epoch": 0.08003060853140623,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009870509340549802,
      "loss": 2.1665,
      "step": 11779
    },
    {
      "epoch": 0.08003740287800029,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000987048501323078,
      "loss": 2.246,
      "step": 11780
    },
    {
      "epoch": 0.08004419722459434,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009870460683656778,
      "loss": 2.1736,
      "step": 11781
    },
    {
      "epoch": 0.0800509915711884,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009870436351827809,
      "loss": 2.3496,
      "step": 11782
    },
    {
      "epoch": 0.08005778591778247,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009870412017743883,
      "loss": 2.2227,
      "step": 11783
    },
    {
      "epoch": 0.08006458026437652,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009870387681405013,
      "loss": 2.2078,
      "step": 11784
    },
    {
      "epoch": 0.08007137461097058,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009870363342811207,
      "loss": 2.1967,
      "step": 11785
    },
    {
      "epoch": 0.08007816895756463,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009870339001962483,
      "loss": 2.1996,
      "step": 11786
    },
    {
      "epoch": 0.0800849633041587,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009870314658858845,
      "loss": 2.3173,
      "step": 11787
    },
    {
      "epoch": 0.08009175765075276,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009870290313500308,
      "loss": 2.1677,
      "step": 11788
    },
    {
      "epoch": 0.0800985519973468,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009870265965886882,
      "loss": 2.1994,
      "step": 11789
    },
    {
      "epoch": 0.08010534634394087,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009870241616018577,
      "loss": 2.2053,
      "step": 11790
    },
    {
      "epoch": 0.08011214069053492,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000987021726389541,
      "loss": 2.1199,
      "step": 11791
    },
    {
      "epoch": 0.08011893503712898,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009870192909517384,
      "loss": 2.2323,
      "step": 11792
    },
    {
      "epoch": 0.08012572938372303,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009870168552884517,
      "loss": 2.2777,
      "step": 11793
    },
    {
      "epoch": 0.0801325237303171,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009870144193996817,
      "loss": 2.3011,
      "step": 11794
    },
    {
      "epoch": 0.08013931807691116,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009870119832854295,
      "loss": 2.1736,
      "step": 11795
    },
    {
      "epoch": 0.08014611242350521,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009870095469456965,
      "loss": 2.3154,
      "step": 11796
    },
    {
      "epoch": 0.08015290677009927,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009870071103804836,
      "loss": 2.1303,
      "step": 11797
    },
    {
      "epoch": 0.08015970111669332,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009870046735897921,
      "loss": 2.3445,
      "step": 11798
    },
    {
      "epoch": 0.08016649546328738,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000987002236573623,
      "loss": 2.3563,
      "step": 11799
    },
    {
      "epoch": 0.08017328980988145,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009869997993319773,
      "loss": 2.2405,
      "step": 11800
    },
    {
      "epoch": 0.0801800841564755,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009869973618648565,
      "loss": 2.0799,
      "step": 11801
    },
    {
      "epoch": 0.08018687850306956,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009869949241722614,
      "loss": 2.2213,
      "step": 11802
    },
    {
      "epoch": 0.08019367284966361,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000986992486254193,
      "loss": 2.2853,
      "step": 11803
    },
    {
      "epoch": 0.08020046719625767,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000986990048110653,
      "loss": 2.1119,
      "step": 11804
    },
    {
      "epoch": 0.08020726154285174,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009869876097416419,
      "loss": 2.2561,
      "step": 11805
    },
    {
      "epoch": 0.08021405588944579,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009869851711471614,
      "loss": 2.276,
      "step": 11806
    },
    {
      "epoch": 0.08022085023603985,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000986982732327212,
      "loss": 2.178,
      "step": 11807
    },
    {
      "epoch": 0.0802276445826339,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009869802932817955,
      "loss": 2.2564,
      "step": 11808
    },
    {
      "epoch": 0.08023443892922796,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009869778540109128,
      "loss": 2.1929,
      "step": 11809
    },
    {
      "epoch": 0.08024123327582203,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009869754145145646,
      "loss": 2.2818,
      "step": 11810
    },
    {
      "epoch": 0.08024802762241608,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009869729747927524,
      "loss": 2.1805,
      "step": 11811
    },
    {
      "epoch": 0.08025482196901014,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009869705348454776,
      "loss": 2.175,
      "step": 11812
    },
    {
      "epoch": 0.08026161631560419,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009869680946727407,
      "loss": 2.3728,
      "step": 11813
    },
    {
      "epoch": 0.08026841066219825,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009869656542745434,
      "loss": 2.2543,
      "step": 11814
    },
    {
      "epoch": 0.0802752050087923,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009869632136508865,
      "loss": 2.3365,
      "step": 11815
    },
    {
      "epoch": 0.08028199935538637,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009869607728017712,
      "loss": 2.2827,
      "step": 11816
    },
    {
      "epoch": 0.08028879370198043,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009869583317271985,
      "loss": 2.1263,
      "step": 11817
    },
    {
      "epoch": 0.08029558804857448,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009869558904271698,
      "loss": 2.1648,
      "step": 11818
    },
    {
      "epoch": 0.08030238239516854,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000986953448901686,
      "loss": 2.2809,
      "step": 11819
    },
    {
      "epoch": 0.08030917674176259,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009869510071507485,
      "loss": 2.1193,
      "step": 11820
    },
    {
      "epoch": 0.08031597108835666,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009869485651743583,
      "loss": 2.1966,
      "step": 11821
    },
    {
      "epoch": 0.08032276543495072,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009869461229725163,
      "loss": 2.0816,
      "step": 11822
    },
    {
      "epoch": 0.08032955978154477,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000986943680545224,
      "loss": 2.1609,
      "step": 11823
    },
    {
      "epoch": 0.08033635412813883,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009869412378924823,
      "loss": 2.2765,
      "step": 11824
    },
    {
      "epoch": 0.08034314847473288,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009869387950142924,
      "loss": 2.2696,
      "step": 11825
    },
    {
      "epoch": 0.08034994282132694,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009869363519106552,
      "loss": 2.1241,
      "step": 11826
    },
    {
      "epoch": 0.08035673716792101,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009869339085815723,
      "loss": 2.3373,
      "step": 11827
    },
    {
      "epoch": 0.08036353151451506,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009869314650270446,
      "loss": 2.2189,
      "step": 11828
    },
    {
      "epoch": 0.08037032586110912,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009869290212470732,
      "loss": 2.1374,
      "step": 11829
    },
    {
      "epoch": 0.08037712020770317,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009869265772416592,
      "loss": 2.0888,
      "step": 11830
    },
    {
      "epoch": 0.08038391455429723,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009869241330108037,
      "loss": 2.3695,
      "step": 11831
    },
    {
      "epoch": 0.0803907089008913,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000986921688554508,
      "loss": 2.2005,
      "step": 11832
    },
    {
      "epoch": 0.08039750324748535,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009869192438727733,
      "loss": 2.3844,
      "step": 11833
    },
    {
      "epoch": 0.08040429759407941,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009869167989656003,
      "loss": 2.2054,
      "step": 11834
    },
    {
      "epoch": 0.08041109194067346,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009869143538329904,
      "loss": 2.2201,
      "step": 11835
    },
    {
      "epoch": 0.08041788628726752,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000986911908474945,
      "loss": 2.0829,
      "step": 11836
    },
    {
      "epoch": 0.08042468063386157,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009869094628914647,
      "loss": 2.2851,
      "step": 11837
    },
    {
      "epoch": 0.08043147498045564,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009869070170825508,
      "loss": 2.326,
      "step": 11838
    },
    {
      "epoch": 0.0804382693270497,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000986904571048205,
      "loss": 2.2657,
      "step": 11839
    },
    {
      "epoch": 0.08044506367364375,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009869021247884276,
      "loss": 2.2015,
      "step": 11840
    },
    {
      "epoch": 0.08045185802023781,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009868996783032203,
      "loss": 2.1887,
      "step": 11841
    },
    {
      "epoch": 0.08045865236683186,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000986897231592584,
      "loss": 2.3686,
      "step": 11842
    },
    {
      "epoch": 0.08046544671342593,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009868947846565198,
      "loss": 2.1051,
      "step": 11843
    },
    {
      "epoch": 0.08047224106001999,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009868923374950287,
      "loss": 2.1662,
      "step": 11844
    },
    {
      "epoch": 0.08047903540661404,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009868898901081123,
      "loss": 2.3959,
      "step": 11845
    },
    {
      "epoch": 0.0804858297532081,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009868874424957715,
      "loss": 2.3566,
      "step": 11846
    },
    {
      "epoch": 0.08049262409980215,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009868849946580072,
      "loss": 2.2104,
      "step": 11847
    },
    {
      "epoch": 0.08049941844639621,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009868825465948207,
      "loss": 2.3103,
      "step": 11848
    },
    {
      "epoch": 0.08050621279299028,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009868800983062133,
      "loss": 2.2613,
      "step": 11849
    },
    {
      "epoch": 0.08051300713958433,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009868776497921861,
      "loss": 2.2173,
      "step": 11850
    },
    {
      "epoch": 0.08051980148617839,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.00098687520105274,
      "loss": 2.3462,
      "step": 11851
    },
    {
      "epoch": 0.08052659583277244,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000986872752087876,
      "loss": 2.3682,
      "step": 11852
    },
    {
      "epoch": 0.0805333901793665,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000986870302897596,
      "loss": 2.2435,
      "step": 11853
    },
    {
      "epoch": 0.08054018452596057,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009868678534819004,
      "loss": 2.1535,
      "step": 11854
    },
    {
      "epoch": 0.08054697887255462,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009868654038407904,
      "loss": 2.1891,
      "step": 11855
    },
    {
      "epoch": 0.08055377321914868,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009868629539742674,
      "loss": 2.2965,
      "step": 11856
    },
    {
      "epoch": 0.08056056756574273,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009868605038823326,
      "loss": 2.2643,
      "step": 11857
    },
    {
      "epoch": 0.0805673619123368,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009868580535649868,
      "loss": 2.3586,
      "step": 11858
    },
    {
      "epoch": 0.08057415625893086,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009868556030222316,
      "loss": 2.2898,
      "step": 11859
    },
    {
      "epoch": 0.0805809506055249,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009868531522540674,
      "loss": 2.2574,
      "step": 11860
    },
    {
      "epoch": 0.08058774495211897,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009868507012604961,
      "loss": 2.1285,
      "step": 11861
    },
    {
      "epoch": 0.08059453929871302,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009868482500415183,
      "loss": 2.1738,
      "step": 11862
    },
    {
      "epoch": 0.08060133364530708,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009868457985971356,
      "loss": 2.3466,
      "step": 11863
    },
    {
      "epoch": 0.08060812799190113,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009868433469273487,
      "loss": 2.3001,
      "step": 11864
    },
    {
      "epoch": 0.0806149223384952,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009868408950321587,
      "loss": 2.0464,
      "step": 11865
    },
    {
      "epoch": 0.08062171668508926,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009868384429115673,
      "loss": 2.3806,
      "step": 11866
    },
    {
      "epoch": 0.08062851103168331,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000986835990565575,
      "loss": 2.1771,
      "step": 11867
    },
    {
      "epoch": 0.08063530537827737,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009868335379941834,
      "loss": 2.1311,
      "step": 11868
    },
    {
      "epoch": 0.08064209972487142,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009868310851973935,
      "loss": 2.2489,
      "step": 11869
    },
    {
      "epoch": 0.08064889407146549,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009868286321752062,
      "loss": 2.3409,
      "step": 11870
    },
    {
      "epoch": 0.08065568841805955,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000986826178927623,
      "loss": 2.2411,
      "step": 11871
    },
    {
      "epoch": 0.0806624827646536,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009868237254546447,
      "loss": 2.143,
      "step": 11872
    },
    {
      "epoch": 0.08066927711124766,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009868212717562728,
      "loss": 2.3586,
      "step": 11873
    },
    {
      "epoch": 0.08067607145784171,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000986818817832508,
      "loss": 2.2284,
      "step": 11874
    },
    {
      "epoch": 0.08068286580443577,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009868163636833518,
      "loss": 2.2711,
      "step": 11875
    },
    {
      "epoch": 0.08068966015102984,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009868139093088051,
      "loss": 2.22,
      "step": 11876
    },
    {
      "epoch": 0.08069645449762389,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009868114547088693,
      "loss": 2.1079,
      "step": 11877
    },
    {
      "epoch": 0.08070324884421795,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009868089998835454,
      "loss": 2.0922,
      "step": 11878
    },
    {
      "epoch": 0.080710043190812,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009868065448328342,
      "loss": 2.1004,
      "step": 11879
    },
    {
      "epoch": 0.08071683753740606,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009868040895567374,
      "loss": 2.0988,
      "step": 11880
    },
    {
      "epoch": 0.08072363188400013,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009868016340552558,
      "loss": 2.022,
      "step": 11881
    },
    {
      "epoch": 0.08073042623059418,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000986799178328391,
      "loss": 2.1677,
      "step": 11882
    },
    {
      "epoch": 0.08073722057718824,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009867967223761433,
      "loss": 2.2426,
      "step": 11883
    },
    {
      "epoch": 0.08074401492378229,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009867942661985144,
      "loss": 2.1749,
      "step": 11884
    },
    {
      "epoch": 0.08075080927037635,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009867918097955055,
      "loss": 2.1929,
      "step": 11885
    },
    {
      "epoch": 0.0807576036169704,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009867893531671174,
      "loss": 2.2701,
      "step": 11886
    },
    {
      "epoch": 0.08076439796356447,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009867868963133516,
      "loss": 2.1758,
      "step": 11887
    },
    {
      "epoch": 0.08077119231015853,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009867844392342088,
      "loss": 2.243,
      "step": 11888
    },
    {
      "epoch": 0.08077798665675258,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009867819819296903,
      "loss": 2.1864,
      "step": 11889
    },
    {
      "epoch": 0.08078478100334664,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009867795243997976,
      "loss": 2.2833,
      "step": 11890
    },
    {
      "epoch": 0.08079157534994069,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009867770666445316,
      "loss": 2.2282,
      "step": 11891
    },
    {
      "epoch": 0.08079836969653476,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000986774608663893,
      "loss": 2.1521,
      "step": 11892
    },
    {
      "epoch": 0.08080516404312882,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000986772150457884,
      "loss": 2.2602,
      "step": 11893
    },
    {
      "epoch": 0.08081195838972287,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009867696920265046,
      "loss": 2.1375,
      "step": 11894
    },
    {
      "epoch": 0.08081875273631693,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009867672333697564,
      "loss": 2.2642,
      "step": 11895
    },
    {
      "epoch": 0.08082554708291098,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009867647744876407,
      "loss": 2.192,
      "step": 11896
    },
    {
      "epoch": 0.08083234142950504,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009867623153801584,
      "loss": 2.2888,
      "step": 11897
    },
    {
      "epoch": 0.08083913577609911,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009867598560473108,
      "loss": 2.1473,
      "step": 11898
    },
    {
      "epoch": 0.08084593012269316,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000986757396489099,
      "loss": 2.2594,
      "step": 11899
    },
    {
      "epoch": 0.08085272446928722,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000986754936705524,
      "loss": 2.2678,
      "step": 11900
    },
    {
      "epoch": 0.08085951881588127,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000986752476696587,
      "loss": 2.2928,
      "step": 11901
    },
    {
      "epoch": 0.08086631316247533,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009867500164622894,
      "loss": 2.1231,
      "step": 11902
    },
    {
      "epoch": 0.0808731075090694,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009867475560026318,
      "loss": 2.2978,
      "step": 11903
    },
    {
      "epoch": 0.08087990185566345,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000986745095317616,
      "loss": 2.2824,
      "step": 11904
    },
    {
      "epoch": 0.08088669620225751,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009867426344072425,
      "loss": 2.1731,
      "step": 11905
    },
    {
      "epoch": 0.08089349054885156,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000986740173271513,
      "loss": 2.2001,
      "step": 11906
    },
    {
      "epoch": 0.08090028489544562,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009867377119104283,
      "loss": 2.1624,
      "step": 11907
    },
    {
      "epoch": 0.08090707924203969,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009867352503239895,
      "loss": 2.3376,
      "step": 11908
    },
    {
      "epoch": 0.08091387358863374,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000986732788512198,
      "loss": 2.2032,
      "step": 11909
    },
    {
      "epoch": 0.0809206679352278,
      "grad_norm": 0.2109375,
      "learning_rate": 0.0009867303264750548,
      "loss": 2.2233,
      "step": 11910
    },
    {
      "epoch": 0.08092746228182185,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000986727864212561,
      "loss": 2.2099,
      "step": 11911
    },
    {
      "epoch": 0.08093425662841591,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009867254017247179,
      "loss": 2.2471,
      "step": 11912
    },
    {
      "epoch": 0.08094105097500996,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009867229390115263,
      "loss": 2.1317,
      "step": 11913
    },
    {
      "epoch": 0.08094784532160403,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009867204760729876,
      "loss": 2.3552,
      "step": 11914
    },
    {
      "epoch": 0.08095463966819809,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000986718012909103,
      "loss": 2.1128,
      "step": 11915
    },
    {
      "epoch": 0.08096143401479214,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009867155495198734,
      "loss": 2.2528,
      "step": 11916
    },
    {
      "epoch": 0.0809682283613862,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009867130859053002,
      "loss": 2.2911,
      "step": 11917
    },
    {
      "epoch": 0.08097502270798025,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009867106220653845,
      "loss": 2.1393,
      "step": 11918
    },
    {
      "epoch": 0.08098181705457431,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000986708158000127,
      "loss": 2.2477,
      "step": 11919
    },
    {
      "epoch": 0.08098861140116838,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009867056937095295,
      "loss": 2.1498,
      "step": 11920
    },
    {
      "epoch": 0.08099540574776243,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000986703229193593,
      "loss": 2.2105,
      "step": 11921
    },
    {
      "epoch": 0.08100220009435649,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009867007644523184,
      "loss": 2.1643,
      "step": 11922
    },
    {
      "epoch": 0.08100899444095054,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009866982994857067,
      "loss": 2.2943,
      "step": 11923
    },
    {
      "epoch": 0.0810157887875446,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009866958342937594,
      "loss": 2.2625,
      "step": 11924
    },
    {
      "epoch": 0.08102258313413867,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009866933688764774,
      "loss": 2.2589,
      "step": 11925
    },
    {
      "epoch": 0.08102937748073272,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000986690903233862,
      "loss": 2.2704,
      "step": 11926
    },
    {
      "epoch": 0.08103617182732678,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009866884373659144,
      "loss": 2.3162,
      "step": 11927
    },
    {
      "epoch": 0.08104296617392083,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009866859712726354,
      "loss": 2.0423,
      "step": 11928
    },
    {
      "epoch": 0.0810497605205149,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009866835049540266,
      "loss": 2.2262,
      "step": 11929
    },
    {
      "epoch": 0.08105655486710896,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000986681038410089,
      "loss": 2.1826,
      "step": 11930
    },
    {
      "epoch": 0.081063349213703,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009866785716408234,
      "loss": 2.2718,
      "step": 11931
    },
    {
      "epoch": 0.08107014356029707,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009866761046462313,
      "loss": 2.1841,
      "step": 11932
    },
    {
      "epoch": 0.08107693790689112,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000986673637426314,
      "loss": 2.2136,
      "step": 11933
    },
    {
      "epoch": 0.08108373225348518,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000986671169981072,
      "loss": 2.3563,
      "step": 11934
    },
    {
      "epoch": 0.08109052660007923,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000986668702310507,
      "loss": 2.2908,
      "step": 11935
    },
    {
      "epoch": 0.0810973209466733,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009866662344146199,
      "loss": 2.1392,
      "step": 11936
    },
    {
      "epoch": 0.08110411529326736,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000986663766293412,
      "loss": 2.3219,
      "step": 11937
    },
    {
      "epoch": 0.08111090963986141,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009866612979468844,
      "loss": 2.3633,
      "step": 11938
    },
    {
      "epoch": 0.08111770398645547,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000986658829375038,
      "loss": 2.3557,
      "step": 11939
    },
    {
      "epoch": 0.08112449833304952,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009866563605778742,
      "loss": 2.3944,
      "step": 11940
    },
    {
      "epoch": 0.08113129267964359,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000986653891555394,
      "loss": 2.2329,
      "step": 11941
    },
    {
      "epoch": 0.08113808702623765,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000986651422307599,
      "loss": 2.2019,
      "step": 11942
    },
    {
      "epoch": 0.0811448813728317,
      "grad_norm": 0.20703125,
      "learning_rate": 0.0009866489528344896,
      "loss": 2.3859,
      "step": 11943
    },
    {
      "epoch": 0.08115167571942576,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009866464831360675,
      "loss": 2.2635,
      "step": 11944
    },
    {
      "epoch": 0.08115847006601981,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009866440132123336,
      "loss": 2.2,
      "step": 11945
    },
    {
      "epoch": 0.08116526441261387,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009866415430632891,
      "loss": 2.155,
      "step": 11946
    },
    {
      "epoch": 0.08117205875920794,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009866390726889353,
      "loss": 2.2294,
      "step": 11947
    },
    {
      "epoch": 0.08117885310580199,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000986636602089273,
      "loss": 2.3377,
      "step": 11948
    },
    {
      "epoch": 0.08118564745239605,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009866341312643034,
      "loss": 2.2029,
      "step": 11949
    },
    {
      "epoch": 0.0811924417989901,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000986631660214028,
      "loss": 2.2124,
      "step": 11950
    },
    {
      "epoch": 0.08119923614558416,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009866291889384475,
      "loss": 2.0675,
      "step": 11951
    },
    {
      "epoch": 0.08120603049217823,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009866267174375636,
      "loss": 2.2156,
      "step": 11952
    },
    {
      "epoch": 0.08121282483877228,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009866242457113768,
      "loss": 2.2837,
      "step": 11953
    },
    {
      "epoch": 0.08121961918536634,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009866217737598889,
      "loss": 2.1757,
      "step": 11954
    },
    {
      "epoch": 0.08122641353196039,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009866193015831005,
      "loss": 2.2738,
      "step": 11955
    },
    {
      "epoch": 0.08123320787855445,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009866168291810128,
      "loss": 2.2028,
      "step": 11956
    },
    {
      "epoch": 0.0812400022251485,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009866143565536271,
      "loss": 2.209,
      "step": 11957
    },
    {
      "epoch": 0.08124679657174257,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009866118837009448,
      "loss": 2.1785,
      "step": 11958
    },
    {
      "epoch": 0.08125359091833663,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009866094106229667,
      "loss": 2.2617,
      "step": 11959
    },
    {
      "epoch": 0.08126038526493068,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009866069373196938,
      "loss": 2.3174,
      "step": 11960
    },
    {
      "epoch": 0.08126717961152474,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009866044637911278,
      "loss": 2.1028,
      "step": 11961
    },
    {
      "epoch": 0.08127397395811879,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000986601990037269,
      "loss": 2.3385,
      "step": 11962
    },
    {
      "epoch": 0.08128076830471286,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009865995160581195,
      "loss": 2.1578,
      "step": 11963
    },
    {
      "epoch": 0.08128756265130692,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00098659704185368,
      "loss": 2.1335,
      "step": 11964
    },
    {
      "epoch": 0.08129435699790097,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009865945674239515,
      "loss": 2.1096,
      "step": 11965
    },
    {
      "epoch": 0.08130115134449503,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009865920927689354,
      "loss": 2.2066,
      "step": 11966
    },
    {
      "epoch": 0.08130794569108908,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009865896178886326,
      "loss": 2.3126,
      "step": 11967
    },
    {
      "epoch": 0.08131474003768314,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009865871427830444,
      "loss": 2.1987,
      "step": 11968
    },
    {
      "epoch": 0.08132153438427721,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000986584667452172,
      "loss": 2.2584,
      "step": 11969
    },
    {
      "epoch": 0.08132832873087126,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009865821918960166,
      "loss": 2.1602,
      "step": 11970
    },
    {
      "epoch": 0.08133512307746532,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000986579716114579,
      "loss": 2.1476,
      "step": 11971
    },
    {
      "epoch": 0.08134191742405937,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009865772401078607,
      "loss": 2.2365,
      "step": 11972
    },
    {
      "epoch": 0.08134871177065343,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009865747638758626,
      "loss": 2.1784,
      "step": 11973
    },
    {
      "epoch": 0.0813555061172475,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009865722874185861,
      "loss": 2.1358,
      "step": 11974
    },
    {
      "epoch": 0.08136230046384155,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009865698107360322,
      "loss": 2.2024,
      "step": 11975
    },
    {
      "epoch": 0.08136909481043561,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009865673338282019,
      "loss": 2.1605,
      "step": 11976
    },
    {
      "epoch": 0.08137588915702966,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009865648566950967,
      "loss": 2.3596,
      "step": 11977
    },
    {
      "epoch": 0.08138268350362372,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009865623793367173,
      "loss": 2.2753,
      "step": 11978
    },
    {
      "epoch": 0.08138947785021779,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009865599017530654,
      "loss": 2.1074,
      "step": 11979
    },
    {
      "epoch": 0.08139627219681184,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009865574239441414,
      "loss": 2.2189,
      "step": 11980
    },
    {
      "epoch": 0.0814030665434059,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009865549459099475,
      "loss": 2.2266,
      "step": 11981
    },
    {
      "epoch": 0.08140986088999995,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009865524676504838,
      "loss": 2.2172,
      "step": 11982
    },
    {
      "epoch": 0.08141665523659401,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000986549989165752,
      "loss": 2.2503,
      "step": 11983
    },
    {
      "epoch": 0.08142344958318806,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009865475104557531,
      "loss": 2.3091,
      "step": 11984
    },
    {
      "epoch": 0.08143024392978213,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009865450315204884,
      "loss": 2.1949,
      "step": 11985
    },
    {
      "epoch": 0.08143703827637619,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009865425523599587,
      "loss": 2.0568,
      "step": 11986
    },
    {
      "epoch": 0.08144383262297024,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009865400729741655,
      "loss": 2.2665,
      "step": 11987
    },
    {
      "epoch": 0.0814506269695643,
      "grad_norm": 0.140625,
      "learning_rate": 0.00098653759336311,
      "loss": 2.2221,
      "step": 11988
    },
    {
      "epoch": 0.08145742131615835,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000986535113526793,
      "loss": 2.1895,
      "step": 11989
    },
    {
      "epoch": 0.08146421566275241,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009865326334652159,
      "loss": 2.3171,
      "step": 11990
    },
    {
      "epoch": 0.08147101000934648,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009865301531783797,
      "loss": 2.2738,
      "step": 11991
    },
    {
      "epoch": 0.08147780435594053,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009865276726662856,
      "loss": 2.088,
      "step": 11992
    },
    {
      "epoch": 0.08148459870253459,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009865251919289348,
      "loss": 2.2996,
      "step": 11993
    },
    {
      "epoch": 0.08149139304912864,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009865227109663284,
      "loss": 2.2651,
      "step": 11994
    },
    {
      "epoch": 0.0814981873957227,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009865202297784676,
      "loss": 2.3057,
      "step": 11995
    },
    {
      "epoch": 0.08150498174231677,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009865177483653533,
      "loss": 2.1187,
      "step": 11996
    },
    {
      "epoch": 0.08151177608891082,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009865152667269872,
      "loss": 2.2173,
      "step": 11997
    },
    {
      "epoch": 0.08151857043550488,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00098651278486337,
      "loss": 2.1498,
      "step": 11998
    },
    {
      "epoch": 0.08152536478209893,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009865103027745029,
      "loss": 2.4483,
      "step": 11999
    },
    {
      "epoch": 0.081532159128693,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000986507820460387,
      "loss": 2.2663,
      "step": 12000
    },
    {
      "epoch": 0.08153895347528706,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009865053379210238,
      "loss": 2.3901,
      "step": 12001
    },
    {
      "epoch": 0.0815457478218811,
      "grad_norm": 0.140625,
      "learning_rate": 0.000986502855156414,
      "loss": 2.2094,
      "step": 12002
    },
    {
      "epoch": 0.08155254216847517,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000986500372166559,
      "loss": 2.3277,
      "step": 12003
    },
    {
      "epoch": 0.08155933651506922,
      "grad_norm": 0.15625,
      "learning_rate": 0.00098649788895146,
      "loss": 2.2636,
      "step": 12004
    },
    {
      "epoch": 0.08156613086166328,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000986495405511118,
      "loss": 2.2389,
      "step": 12005
    },
    {
      "epoch": 0.08157292520825733,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009864929218455342,
      "loss": 2.2871,
      "step": 12006
    },
    {
      "epoch": 0.0815797195548514,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009864904379547096,
      "loss": 2.1907,
      "step": 12007
    },
    {
      "epoch": 0.08158651390144546,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009864879538386457,
      "loss": 2.2504,
      "step": 12008
    },
    {
      "epoch": 0.08159330824803951,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009864854694973436,
      "loss": 2.2171,
      "step": 12009
    },
    {
      "epoch": 0.08160010259463357,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000986482984930804,
      "loss": 2.1767,
      "step": 12010
    },
    {
      "epoch": 0.08160689694122762,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009864805001390284,
      "loss": 2.2612,
      "step": 12011
    },
    {
      "epoch": 0.08161369128782169,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000986478015122018,
      "loss": 2.2125,
      "step": 12012
    },
    {
      "epoch": 0.08162048563441575,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009864755298797739,
      "loss": 2.2305,
      "step": 12013
    },
    {
      "epoch": 0.0816272799810098,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000986473044412297,
      "loss": 2.2531,
      "step": 12014
    },
    {
      "epoch": 0.08163407432760386,
      "grad_norm": 0.140625,
      "learning_rate": 0.000986470558719589,
      "loss": 2.1456,
      "step": 12015
    },
    {
      "epoch": 0.08164086867419791,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009864680728016505,
      "loss": 2.1679,
      "step": 12016
    },
    {
      "epoch": 0.08164766302079197,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009864655866584828,
      "loss": 2.268,
      "step": 12017
    },
    {
      "epoch": 0.08165445736738604,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009864631002900871,
      "loss": 2.2113,
      "step": 12018
    },
    {
      "epoch": 0.08166125171398009,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009864606136964647,
      "loss": 2.1862,
      "step": 12019
    },
    {
      "epoch": 0.08166804606057415,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009864581268776167,
      "loss": 2.2188,
      "step": 12020
    },
    {
      "epoch": 0.0816748404071682,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000986455639833544,
      "loss": 2.1103,
      "step": 12021
    },
    {
      "epoch": 0.08168163475376226,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000986453152564248,
      "loss": 2.2589,
      "step": 12022
    },
    {
      "epoch": 0.08168842910035633,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009864506650697296,
      "loss": 2.1605,
      "step": 12023
    },
    {
      "epoch": 0.08169522344695038,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009864481773499902,
      "loss": 2.1614,
      "step": 12024
    },
    {
      "epoch": 0.08170201779354444,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000986445689405031,
      "loss": 2.1158,
      "step": 12025
    },
    {
      "epoch": 0.08170881214013849,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009864432012348528,
      "loss": 2.3679,
      "step": 12026
    },
    {
      "epoch": 0.08171560648673255,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000986440712839457,
      "loss": 2.2814,
      "step": 12027
    },
    {
      "epoch": 0.08172240083332662,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009864382242188447,
      "loss": 2.1455,
      "step": 12028
    },
    {
      "epoch": 0.08172919517992067,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009864357353730174,
      "loss": 2.1571,
      "step": 12029
    },
    {
      "epoch": 0.08173598952651473,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009864332463019757,
      "loss": 2.3215,
      "step": 12030
    },
    {
      "epoch": 0.08174278387310878,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009864307570057209,
      "loss": 2.4005,
      "step": 12031
    },
    {
      "epoch": 0.08174957821970284,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009864282674842545,
      "loss": 2.1666,
      "step": 12032
    },
    {
      "epoch": 0.08175637256629689,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009864257777375772,
      "loss": 2.2591,
      "step": 12033
    },
    {
      "epoch": 0.08176316691289096,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009864232877656903,
      "loss": 2.3307,
      "step": 12034
    },
    {
      "epoch": 0.08176996125948502,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000986420797568595,
      "loss": 2.2046,
      "step": 12035
    },
    {
      "epoch": 0.08177675560607907,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009864183071462927,
      "loss": 2.2708,
      "step": 12036
    },
    {
      "epoch": 0.08178354995267313,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000986415816498784,
      "loss": 2.228,
      "step": 12037
    },
    {
      "epoch": 0.08179034429926718,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009864133256260706,
      "loss": 2.3654,
      "step": 12038
    },
    {
      "epoch": 0.08179713864586124,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009864108345281532,
      "loss": 2.2657,
      "step": 12039
    },
    {
      "epoch": 0.08180393299245531,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009864083432050332,
      "loss": 2.2104,
      "step": 12040
    },
    {
      "epoch": 0.08181072733904936,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009864058516567118,
      "loss": 2.1148,
      "step": 12041
    },
    {
      "epoch": 0.08181752168564342,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00098640335988319,
      "loss": 2.1719,
      "step": 12042
    },
    {
      "epoch": 0.08182431603223747,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009864008678844688,
      "loss": 2.27,
      "step": 12043
    },
    {
      "epoch": 0.08183111037883153,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00098639837566055,
      "loss": 2.2543,
      "step": 12044
    },
    {
      "epoch": 0.0818379047254256,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000986395883211434,
      "loss": 2.2307,
      "step": 12045
    },
    {
      "epoch": 0.08184469907201965,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009863933905371225,
      "loss": 2.2657,
      "step": 12046
    },
    {
      "epoch": 0.08185149341861371,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009863908976376163,
      "loss": 2.2555,
      "step": 12047
    },
    {
      "epoch": 0.08185828776520776,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009863884045129168,
      "loss": 2.1961,
      "step": 12048
    },
    {
      "epoch": 0.08186508211180182,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000986385911163025,
      "loss": 2.3125,
      "step": 12049
    },
    {
      "epoch": 0.08187187645839589,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000986383417587942,
      "loss": 2.1008,
      "step": 12050
    },
    {
      "epoch": 0.08187867080498994,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009863809237876692,
      "loss": 2.1803,
      "step": 12051
    },
    {
      "epoch": 0.081885465151584,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009863784297622076,
      "loss": 2.3974,
      "step": 12052
    },
    {
      "epoch": 0.08189225949817805,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009863759355115585,
      "loss": 2.1662,
      "step": 12053
    },
    {
      "epoch": 0.08189905384477211,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009863734410357225,
      "loss": 2.3048,
      "step": 12054
    },
    {
      "epoch": 0.08190584819136616,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009863709463347014,
      "loss": 2.0873,
      "step": 12055
    },
    {
      "epoch": 0.08191264253796023,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009863684514084964,
      "loss": 2.2306,
      "step": 12056
    },
    {
      "epoch": 0.08191943688455429,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009863659562571081,
      "loss": 2.1743,
      "step": 12057
    },
    {
      "epoch": 0.08192623123114834,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009863634608805378,
      "loss": 2.1894,
      "step": 12058
    },
    {
      "epoch": 0.0819330255777424,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000986360965278787,
      "loss": 2.2235,
      "step": 12059
    },
    {
      "epoch": 0.08193981992433645,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009863584694518566,
      "loss": 2.1149,
      "step": 12060
    },
    {
      "epoch": 0.08194661427093052,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000986355973399748,
      "loss": 2.1542,
      "step": 12061
    },
    {
      "epoch": 0.08195340861752458,
      "grad_norm": 0.228515625,
      "learning_rate": 0.000986353477122462,
      "loss": 2.1627,
      "step": 12062
    },
    {
      "epoch": 0.08196020296411863,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009863509806199997,
      "loss": 2.2907,
      "step": 12063
    },
    {
      "epoch": 0.08196699731071269,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009863484838923626,
      "loss": 2.3157,
      "step": 12064
    },
    {
      "epoch": 0.08197379165730674,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009863459869395518,
      "loss": 2.2963,
      "step": 12065
    },
    {
      "epoch": 0.0819805860039008,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009863434897615683,
      "loss": 2.1644,
      "step": 12066
    },
    {
      "epoch": 0.08198738035049487,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009863409923584134,
      "loss": 2.1018,
      "step": 12067
    },
    {
      "epoch": 0.08199417469708892,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009863384947300883,
      "loss": 2.2401,
      "step": 12068
    },
    {
      "epoch": 0.08200096904368298,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000986335996876594,
      "loss": 2.2499,
      "step": 12069
    },
    {
      "epoch": 0.08200776339027703,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009863334987979316,
      "loss": 2.1907,
      "step": 12070
    },
    {
      "epoch": 0.0820145577368711,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009863310004941024,
      "loss": 2.1719,
      "step": 12071
    },
    {
      "epoch": 0.08202135208346516,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009863285019651076,
      "loss": 2.2897,
      "step": 12072
    },
    {
      "epoch": 0.0820281464300592,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009863260032109483,
      "loss": 2.1523,
      "step": 12073
    },
    {
      "epoch": 0.08203494077665327,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009863235042316255,
      "loss": 2.1665,
      "step": 12074
    },
    {
      "epoch": 0.08204173512324732,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009863210050271405,
      "loss": 2.1901,
      "step": 12075
    },
    {
      "epoch": 0.08204852946984138,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009863185055974946,
      "loss": 2.2377,
      "step": 12076
    },
    {
      "epoch": 0.08205532381643543,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009863160059426887,
      "loss": 2.0342,
      "step": 12077
    },
    {
      "epoch": 0.0820621181630295,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000986313506062724,
      "loss": 2.1724,
      "step": 12078
    },
    {
      "epoch": 0.08206891250962356,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009863110059576017,
      "loss": 2.2114,
      "step": 12079
    },
    {
      "epoch": 0.08207570685621761,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009863085056273232,
      "loss": 2.2474,
      "step": 12080
    },
    {
      "epoch": 0.08208250120281167,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009863060050718893,
      "loss": 2.2214,
      "step": 12081
    },
    {
      "epoch": 0.08208929554940572,
      "grad_norm": 0.125,
      "learning_rate": 0.0009863035042913014,
      "loss": 2.1778,
      "step": 12082
    },
    {
      "epoch": 0.08209608989599979,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009863010032855604,
      "loss": 2.3868,
      "step": 12083
    },
    {
      "epoch": 0.08210288424259385,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009862985020546678,
      "loss": 2.1641,
      "step": 12084
    },
    {
      "epoch": 0.0821096785891879,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009862960005986242,
      "loss": 2.0389,
      "step": 12085
    },
    {
      "epoch": 0.08211647293578196,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009862934989174315,
      "loss": 2.1927,
      "step": 12086
    },
    {
      "epoch": 0.08212326728237601,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009862909970110904,
      "loss": 2.4038,
      "step": 12087
    },
    {
      "epoch": 0.08213006162897007,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000986288494879602,
      "loss": 2.2642,
      "step": 12088
    },
    {
      "epoch": 0.08213685597556414,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009862859925229678,
      "loss": 2.1966,
      "step": 12089
    },
    {
      "epoch": 0.08214365032215819,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009862834899411887,
      "loss": 2.3139,
      "step": 12090
    },
    {
      "epoch": 0.08215044466875225,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009862809871342658,
      "loss": 2.205,
      "step": 12091
    },
    {
      "epoch": 0.0821572390153463,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009862784841022005,
      "loss": 2.1593,
      "step": 12092
    },
    {
      "epoch": 0.08216403336194036,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009862759808449938,
      "loss": 2.0974,
      "step": 12093
    },
    {
      "epoch": 0.08217082770853443,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000986273477362647,
      "loss": 2.2742,
      "step": 12094
    },
    {
      "epoch": 0.08217762205512848,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009862709736551609,
      "loss": 2.231,
      "step": 12095
    },
    {
      "epoch": 0.08218441640172254,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009862684697225371,
      "loss": 2.2014,
      "step": 12096
    },
    {
      "epoch": 0.08219121074831659,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009862659655647766,
      "loss": 2.2697,
      "step": 12097
    },
    {
      "epoch": 0.08219800509491065,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009862634611818805,
      "loss": 2.2485,
      "step": 12098
    },
    {
      "epoch": 0.08220479944150472,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00098626095657385,
      "loss": 2.2006,
      "step": 12099
    },
    {
      "epoch": 0.08221159378809877,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009862584517406861,
      "loss": 2.1885,
      "step": 12100
    },
    {
      "epoch": 0.08221838813469283,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009862559466823901,
      "loss": 2.2712,
      "step": 12101
    },
    {
      "epoch": 0.08222518248128688,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009862534413989635,
      "loss": 2.3288,
      "step": 12102
    },
    {
      "epoch": 0.08223197682788094,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009862509358904068,
      "loss": 2.1499,
      "step": 12103
    },
    {
      "epoch": 0.08223877117447499,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009862484301567216,
      "loss": 2.3674,
      "step": 12104
    },
    {
      "epoch": 0.08224556552106906,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009862459241979091,
      "loss": 2.1538,
      "step": 12105
    },
    {
      "epoch": 0.08225235986766312,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009862434180139702,
      "loss": 2.2486,
      "step": 12106
    },
    {
      "epoch": 0.08225915421425717,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009862409116049061,
      "loss": 2.2964,
      "step": 12107
    },
    {
      "epoch": 0.08226594856085123,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000986238404970718,
      "loss": 2.3059,
      "step": 12108
    },
    {
      "epoch": 0.08227274290744528,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009862358981114073,
      "loss": 2.189,
      "step": 12109
    },
    {
      "epoch": 0.08227953725403934,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009862333910269747,
      "loss": 2.2453,
      "step": 12110
    },
    {
      "epoch": 0.08228633160063341,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009862308837174218,
      "loss": 2.1257,
      "step": 12111
    },
    {
      "epoch": 0.08229312594722746,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009862283761827495,
      "loss": 2.335,
      "step": 12112
    },
    {
      "epoch": 0.08229992029382152,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000986225868422959,
      "loss": 2.3301,
      "step": 12113
    },
    {
      "epoch": 0.08230671464041557,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009862233604380516,
      "loss": 2.2149,
      "step": 12114
    },
    {
      "epoch": 0.08231350898700963,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009862208522280283,
      "loss": 2.2362,
      "step": 12115
    },
    {
      "epoch": 0.0823203033336037,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009862183437928905,
      "loss": 2.1964,
      "step": 12116
    },
    {
      "epoch": 0.08232709768019775,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000986215835132639,
      "loss": 2.1621,
      "step": 12117
    },
    {
      "epoch": 0.08233389202679181,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000986213326247275,
      "loss": 2.1776,
      "step": 12118
    },
    {
      "epoch": 0.08234068637338586,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009862108171368,
      "loss": 2.342,
      "step": 12119
    },
    {
      "epoch": 0.08234748071997992,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000986208307801215,
      "loss": 2.3222,
      "step": 12120
    },
    {
      "epoch": 0.08235427506657399,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000986205798240521,
      "loss": 2.1387,
      "step": 12121
    },
    {
      "epoch": 0.08236106941316804,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009862032884547195,
      "loss": 2.35,
      "step": 12122
    },
    {
      "epoch": 0.0823678637597621,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009862007784438112,
      "loss": 2.2728,
      "step": 12123
    },
    {
      "epoch": 0.08237465810635615,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009861982682077975,
      "loss": 2.2462,
      "step": 12124
    },
    {
      "epoch": 0.08238145245295021,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00098619575774668,
      "loss": 2.1637,
      "step": 12125
    },
    {
      "epoch": 0.08238824679954426,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000986193247060459,
      "loss": 2.4306,
      "step": 12126
    },
    {
      "epoch": 0.08239504114613833,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000986190736149136,
      "loss": 2.1489,
      "step": 12127
    },
    {
      "epoch": 0.08240183549273239,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009861882250127125,
      "loss": 2.2543,
      "step": 12128
    },
    {
      "epoch": 0.08240862983932644,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009861857136511895,
      "loss": 2.2081,
      "step": 12129
    },
    {
      "epoch": 0.0824154241859205,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000986183202064568,
      "loss": 2.1715,
      "step": 12130
    },
    {
      "epoch": 0.08242221853251455,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009861806902528491,
      "loss": 2.1433,
      "step": 12131
    },
    {
      "epoch": 0.08242901287910862,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009861781782160345,
      "loss": 2.0415,
      "step": 12132
    },
    {
      "epoch": 0.08243580722570268,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009861756659541247,
      "loss": 2.0832,
      "step": 12133
    },
    {
      "epoch": 0.08244260157229673,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009861731534671213,
      "loss": 2.2052,
      "step": 12134
    },
    {
      "epoch": 0.08244939591889079,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000986170640755025,
      "loss": 2.2598,
      "step": 12135
    },
    {
      "epoch": 0.08245619026548484,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009861681278178374,
      "loss": 2.1595,
      "step": 12136
    },
    {
      "epoch": 0.0824629846120789,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009861656146555597,
      "loss": 2.1061,
      "step": 12137
    },
    {
      "epoch": 0.08246977895867297,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009861631012681928,
      "loss": 2.1449,
      "step": 12138
    },
    {
      "epoch": 0.08247657330526702,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000986160587655738,
      "loss": 2.3883,
      "step": 12139
    },
    {
      "epoch": 0.08248336765186108,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009861580738181963,
      "loss": 2.1741,
      "step": 12140
    },
    {
      "epoch": 0.08249016199845513,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000986155559755569,
      "loss": 2.3769,
      "step": 12141
    },
    {
      "epoch": 0.0824969563450492,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009861530454678574,
      "loss": 2.218,
      "step": 12142
    },
    {
      "epoch": 0.08250375069164326,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009861505309550623,
      "loss": 2.2376,
      "step": 12143
    },
    {
      "epoch": 0.0825105450382373,
      "grad_norm": 0.125,
      "learning_rate": 0.0009861480162171852,
      "loss": 2.2423,
      "step": 12144
    },
    {
      "epoch": 0.08251733938483137,
      "grad_norm": 0.140625,
      "learning_rate": 0.000986145501254227,
      "loss": 2.1677,
      "step": 12145
    },
    {
      "epoch": 0.08252413373142542,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009861429860661892,
      "loss": 2.3018,
      "step": 12146
    },
    {
      "epoch": 0.08253092807801948,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009861404706530729,
      "loss": 2.2479,
      "step": 12147
    },
    {
      "epoch": 0.08253772242461355,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009861379550148788,
      "loss": 2.1728,
      "step": 12148
    },
    {
      "epoch": 0.0825445167712076,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009861354391516085,
      "loss": 2.2294,
      "step": 12149
    },
    {
      "epoch": 0.08255131111780166,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009861329230632632,
      "loss": 2.2758,
      "step": 12150
    },
    {
      "epoch": 0.08255810546439571,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009861304067498439,
      "loss": 2.3289,
      "step": 12151
    },
    {
      "epoch": 0.08256489981098977,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009861278902113517,
      "loss": 2.3569,
      "step": 12152
    },
    {
      "epoch": 0.08257169415758382,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009861253734477877,
      "loss": 2.2426,
      "step": 12153
    },
    {
      "epoch": 0.08257848850417789,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009861228564591535,
      "loss": 2.2679,
      "step": 12154
    },
    {
      "epoch": 0.08258528285077195,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00098612033924545,
      "loss": 2.3287,
      "step": 12155
    },
    {
      "epoch": 0.082592077197366,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009861178218066782,
      "loss": 2.0934,
      "step": 12156
    },
    {
      "epoch": 0.08259887154396006,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009861153041428395,
      "loss": 2.2403,
      "step": 12157
    },
    {
      "epoch": 0.08260566589055411,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000986112786253935,
      "loss": 2.188,
      "step": 12158
    },
    {
      "epoch": 0.08261246023714817,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009861102681399656,
      "loss": 2.1776,
      "step": 12159
    },
    {
      "epoch": 0.08261925458374224,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000986107749800933,
      "loss": 2.2128,
      "step": 12160
    },
    {
      "epoch": 0.08262604893033629,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009861052312368382,
      "loss": 2.2057,
      "step": 12161
    },
    {
      "epoch": 0.08263284327693035,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000986102712447682,
      "loss": 2.1578,
      "step": 12162
    },
    {
      "epoch": 0.0826396376235244,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009861001934334658,
      "loss": 2.1196,
      "step": 12163
    },
    {
      "epoch": 0.08264643197011846,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000986097674194191,
      "loss": 2.2662,
      "step": 12164
    },
    {
      "epoch": 0.08265322631671253,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009860951547298585,
      "loss": 2.2656,
      "step": 12165
    },
    {
      "epoch": 0.08266002066330658,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009860926350404693,
      "loss": 2.2349,
      "step": 12166
    },
    {
      "epoch": 0.08266681500990064,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000986090115126025,
      "loss": 2.2157,
      "step": 12167
    },
    {
      "epoch": 0.08267360935649469,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009860875949865265,
      "loss": 2.2569,
      "step": 12168
    },
    {
      "epoch": 0.08268040370308875,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009860850746219749,
      "loss": 2.2266,
      "step": 12169
    },
    {
      "epoch": 0.08268719804968282,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009860825540323717,
      "loss": 2.1076,
      "step": 12170
    },
    {
      "epoch": 0.08269399239627687,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009860800332177177,
      "loss": 2.3013,
      "step": 12171
    },
    {
      "epoch": 0.08270078674287093,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009860775121780143,
      "loss": 2.2712,
      "step": 12172
    },
    {
      "epoch": 0.08270758108946498,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009860749909132624,
      "loss": 2.2203,
      "step": 12173
    },
    {
      "epoch": 0.08271437543605904,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009860724694234636,
      "loss": 2.1424,
      "step": 12174
    },
    {
      "epoch": 0.08272116978265309,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009860699477086186,
      "loss": 2.1434,
      "step": 12175
    },
    {
      "epoch": 0.08272796412924716,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009860674257687288,
      "loss": 2.2912,
      "step": 12176
    },
    {
      "epoch": 0.08273475847584122,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009860649036037956,
      "loss": 2.147,
      "step": 12177
    },
    {
      "epoch": 0.08274155282243527,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009860623812138196,
      "loss": 2.2151,
      "step": 12178
    },
    {
      "epoch": 0.08274834716902933,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009860598585988026,
      "loss": 2.1396,
      "step": 12179
    },
    {
      "epoch": 0.08275514151562338,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009860573357587453,
      "loss": 2.3139,
      "step": 12180
    },
    {
      "epoch": 0.08276193586221745,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000986054812693649,
      "loss": 2.234,
      "step": 12181
    },
    {
      "epoch": 0.08276873020881151,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009860522894035149,
      "loss": 2.4241,
      "step": 12182
    },
    {
      "epoch": 0.08277552455540556,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009860497658883442,
      "loss": 2.2031,
      "step": 12183
    },
    {
      "epoch": 0.08278231890199962,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009860472421481382,
      "loss": 2.2588,
      "step": 12184
    },
    {
      "epoch": 0.08278911324859367,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009860447181828976,
      "loss": 2.216,
      "step": 12185
    },
    {
      "epoch": 0.08279590759518773,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000986042193992624,
      "loss": 2.2871,
      "step": 12186
    },
    {
      "epoch": 0.0828027019417818,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009860396695773185,
      "loss": 2.2605,
      "step": 12187
    },
    {
      "epoch": 0.08280949628837585,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000986037144936982,
      "loss": 2.1401,
      "step": 12188
    },
    {
      "epoch": 0.08281629063496991,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000986034620071616,
      "loss": 2.1555,
      "step": 12189
    },
    {
      "epoch": 0.08282308498156396,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009860320949812216,
      "loss": 2.1743,
      "step": 12190
    },
    {
      "epoch": 0.08282987932815802,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009860295696657998,
      "loss": 2.2048,
      "step": 12191
    },
    {
      "epoch": 0.08283667367475209,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000986027044125352,
      "loss": 2.0748,
      "step": 12192
    },
    {
      "epoch": 0.08284346802134614,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009860245183598792,
      "loss": 2.1188,
      "step": 12193
    },
    {
      "epoch": 0.0828502623679402,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009860219923693825,
      "loss": 2.3483,
      "step": 12194
    },
    {
      "epoch": 0.08285705671453425,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009860194661538634,
      "loss": 2.1866,
      "step": 12195
    },
    {
      "epoch": 0.08286385106112831,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009860169397133227,
      "loss": 2.1975,
      "step": 12196
    },
    {
      "epoch": 0.08287064540772236,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000986014413047762,
      "loss": 2.1273,
      "step": 12197
    },
    {
      "epoch": 0.08287743975431643,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009860118861571819,
      "loss": 2.2483,
      "step": 12198
    },
    {
      "epoch": 0.08288423410091049,
      "grad_norm": 0.16796875,
      "learning_rate": 0.000986009359041584,
      "loss": 2.4349,
      "step": 12199
    },
    {
      "epoch": 0.08289102844750454,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009860068317009692,
      "loss": 2.2675,
      "step": 12200
    },
    {
      "epoch": 0.0828978227940986,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009860043041353388,
      "loss": 2.2703,
      "step": 12201
    },
    {
      "epoch": 0.08290461714069265,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009860017763446943,
      "loss": 2.1813,
      "step": 12202
    },
    {
      "epoch": 0.08291141148728672,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009859992483290361,
      "loss": 2.1481,
      "step": 12203
    },
    {
      "epoch": 0.08291820583388078,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009859967200883662,
      "loss": 2.2833,
      "step": 12204
    },
    {
      "epoch": 0.08292500018047483,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009859941916226852,
      "loss": 2.3185,
      "step": 12205
    },
    {
      "epoch": 0.08293179452706889,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009859916629319945,
      "loss": 2.2455,
      "step": 12206
    },
    {
      "epoch": 0.08293858887366294,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009859891340162953,
      "loss": 2.2201,
      "step": 12207
    },
    {
      "epoch": 0.082945383220257,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009859866048755885,
      "loss": 2.2602,
      "step": 12208
    },
    {
      "epoch": 0.08295217756685107,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009859840755098758,
      "loss": 2.1795,
      "step": 12209
    },
    {
      "epoch": 0.08295897191344512,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000985981545919158,
      "loss": 2.2204,
      "step": 12210
    },
    {
      "epoch": 0.08296576626003918,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000985979016103436,
      "loss": 2.3628,
      "step": 12211
    },
    {
      "epoch": 0.08297256060663323,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009859764860627113,
      "loss": 2.3879,
      "step": 12212
    },
    {
      "epoch": 0.0829793549532273,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009859739557969854,
      "loss": 2.3006,
      "step": 12213
    },
    {
      "epoch": 0.08298614929982136,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009859714253062587,
      "loss": 2.296,
      "step": 12214
    },
    {
      "epoch": 0.0829929436464154,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009859688945905332,
      "loss": 2.1832,
      "step": 12215
    },
    {
      "epoch": 0.08299973799300947,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009859663636498095,
      "loss": 2.1165,
      "step": 12216
    },
    {
      "epoch": 0.08300653233960352,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000985963832484089,
      "loss": 2.2137,
      "step": 12217
    },
    {
      "epoch": 0.08301332668619758,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009859613010933727,
      "loss": 2.1869,
      "step": 12218
    },
    {
      "epoch": 0.08302012103279165,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000985958769477662,
      "loss": 2.1384,
      "step": 12219
    },
    {
      "epoch": 0.0830269153793857,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000985956237636958,
      "loss": 2.216,
      "step": 12220
    },
    {
      "epoch": 0.08303370972597976,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009859537055712617,
      "loss": 2.2062,
      "step": 12221
    },
    {
      "epoch": 0.08304050407257381,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009859511732805747,
      "loss": 2.1432,
      "step": 12222
    },
    {
      "epoch": 0.08304729841916787,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009859486407648975,
      "loss": 2.2218,
      "step": 12223
    },
    {
      "epoch": 0.08305409276576192,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009859461080242318,
      "loss": 2.3424,
      "step": 12224
    },
    {
      "epoch": 0.08306088711235599,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009859435750585786,
      "loss": 2.3252,
      "step": 12225
    },
    {
      "epoch": 0.08306768145895005,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009859410418679392,
      "loss": 2.1662,
      "step": 12226
    },
    {
      "epoch": 0.0830744758055441,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009859385084523145,
      "loss": 2.4682,
      "step": 12227
    },
    {
      "epoch": 0.08308127015213816,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.000985935974811706,
      "loss": 2.0863,
      "step": 12228
    },
    {
      "epoch": 0.08308806449873221,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009859334409461145,
      "loss": 2.2451,
      "step": 12229
    },
    {
      "epoch": 0.08309485884532627,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009859309068555417,
      "loss": 2.3342,
      "step": 12230
    },
    {
      "epoch": 0.08310165319192034,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009859283725399883,
      "loss": 2.261,
      "step": 12231
    },
    {
      "epoch": 0.08310844753851439,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009859258379994555,
      "loss": 2.2738,
      "step": 12232
    },
    {
      "epoch": 0.08311524188510845,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009859233032339448,
      "loss": 2.1911,
      "step": 12233
    },
    {
      "epoch": 0.0831220362317025,
      "grad_norm": 0.125,
      "learning_rate": 0.000985920768243457,
      "loss": 2.2005,
      "step": 12234
    },
    {
      "epoch": 0.08312883057829656,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009859182330279937,
      "loss": 2.2718,
      "step": 12235
    },
    {
      "epoch": 0.08313562492489063,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009859156975875556,
      "loss": 2.2599,
      "step": 12236
    },
    {
      "epoch": 0.08314241927148468,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009859131619221442,
      "loss": 2.1515,
      "step": 12237
    },
    {
      "epoch": 0.08314921361807874,
      "grad_norm": 0.125,
      "learning_rate": 0.0009859106260317607,
      "loss": 2.1585,
      "step": 12238
    },
    {
      "epoch": 0.08315600796467279,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000985908089916406,
      "loss": 2.2366,
      "step": 12239
    },
    {
      "epoch": 0.08316280231126685,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009859055535760815,
      "loss": 2.3112,
      "step": 12240
    },
    {
      "epoch": 0.08316959665786092,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009859030170107882,
      "loss": 2.2768,
      "step": 12241
    },
    {
      "epoch": 0.08317639100445497,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009859004802205275,
      "loss": 2.1483,
      "step": 12242
    },
    {
      "epoch": 0.08318318535104903,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009858979432053004,
      "loss": 2.1455,
      "step": 12243
    },
    {
      "epoch": 0.08318997969764308,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000985895405965108,
      "loss": 2.2874,
      "step": 12244
    },
    {
      "epoch": 0.08319677404423714,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009858928684999518,
      "loss": 2.2953,
      "step": 12245
    },
    {
      "epoch": 0.08320356839083119,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009858903308098326,
      "loss": 2.075,
      "step": 12246
    },
    {
      "epoch": 0.08321036273742526,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009858877928947518,
      "loss": 2.1226,
      "step": 12247
    },
    {
      "epoch": 0.08321715708401932,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009858852547547105,
      "loss": 2.3246,
      "step": 12248
    },
    {
      "epoch": 0.08322395143061337,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00098588271638971,
      "loss": 2.1847,
      "step": 12249
    },
    {
      "epoch": 0.08323074577720743,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009858801777997514,
      "loss": 2.1917,
      "step": 12250
    },
    {
      "epoch": 0.08323754012380148,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009858776389848356,
      "loss": 2.1509,
      "step": 12251
    },
    {
      "epoch": 0.08324433447039555,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009858750999449642,
      "loss": 2.1203,
      "step": 12252
    },
    {
      "epoch": 0.08325112881698961,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009858725606801382,
      "loss": 2.1964,
      "step": 12253
    },
    {
      "epoch": 0.08325792316358366,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000985870021190359,
      "loss": 2.2407,
      "step": 12254
    },
    {
      "epoch": 0.08326471751017772,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009858674814756272,
      "loss": 2.1213,
      "step": 12255
    },
    {
      "epoch": 0.08327151185677177,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009858649415359444,
      "loss": 2.2352,
      "step": 12256
    },
    {
      "epoch": 0.08327830620336583,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009858624013713117,
      "loss": 2.2294,
      "step": 12257
    },
    {
      "epoch": 0.0832851005499599,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009858598609817303,
      "loss": 2.3087,
      "step": 12258
    },
    {
      "epoch": 0.08329189489655395,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009858573203672014,
      "loss": 2.2825,
      "step": 12259
    },
    {
      "epoch": 0.08329868924314801,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009858547795277262,
      "loss": 2.2565,
      "step": 12260
    },
    {
      "epoch": 0.08330548358974206,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009858522384633057,
      "loss": 2.3221,
      "step": 12261
    },
    {
      "epoch": 0.08331227793633612,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000985849697173941,
      "loss": 2.2883,
      "step": 12262
    },
    {
      "epoch": 0.08331907228293019,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009858471556596338,
      "loss": 2.227,
      "step": 12263
    },
    {
      "epoch": 0.08332586662952424,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009858446139203846,
      "loss": 2.2693,
      "step": 12264
    },
    {
      "epoch": 0.0833326609761183,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009858420719561951,
      "loss": 2.1746,
      "step": 12265
    },
    {
      "epoch": 0.08333945532271235,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009858395297670664,
      "loss": 2.1602,
      "step": 12266
    },
    {
      "epoch": 0.08334624966930641,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009858369873529996,
      "loss": 2.3705,
      "step": 12267
    },
    {
      "epoch": 0.08335304401590048,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009858344447139955,
      "loss": 2.3022,
      "step": 12268
    },
    {
      "epoch": 0.08335983836249453,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000985831901850056,
      "loss": 2.1911,
      "step": 12269
    },
    {
      "epoch": 0.08336663270908859,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009858293587611817,
      "loss": 2.4208,
      "step": 12270
    },
    {
      "epoch": 0.08337342705568264,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009858268154473739,
      "loss": 2.2505,
      "step": 12271
    },
    {
      "epoch": 0.0833802214022767,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000985824271908634,
      "loss": 2.223,
      "step": 12272
    },
    {
      "epoch": 0.08338701574887075,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009858217281449632,
      "loss": 2.2654,
      "step": 12273
    },
    {
      "epoch": 0.08339381009546482,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000985819184156362,
      "loss": 2.3515,
      "step": 12274
    },
    {
      "epoch": 0.08340060444205888,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009858166399428325,
      "loss": 2.1762,
      "step": 12275
    },
    {
      "epoch": 0.08340739878865293,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009858140955043752,
      "loss": 2.171,
      "step": 12276
    },
    {
      "epoch": 0.08341419313524699,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009858115508409918,
      "loss": 2.2319,
      "step": 12277
    },
    {
      "epoch": 0.08342098748184104,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009858090059526833,
      "loss": 2.2528,
      "step": 12278
    },
    {
      "epoch": 0.0834277818284351,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009858064608394505,
      "loss": 2.2087,
      "step": 12279
    },
    {
      "epoch": 0.08343457617502917,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000985803915501295,
      "loss": 2.1633,
      "step": 12280
    },
    {
      "epoch": 0.08344137052162322,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009858013699382178,
      "loss": 2.1405,
      "step": 12281
    },
    {
      "epoch": 0.08344816486821728,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009857988241502203,
      "loss": 2.0974,
      "step": 12282
    },
    {
      "epoch": 0.08345495921481133,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009857962781373033,
      "loss": 2.3057,
      "step": 12283
    },
    {
      "epoch": 0.0834617535614054,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009857937318994683,
      "loss": 2.2357,
      "step": 12284
    },
    {
      "epoch": 0.08346854790799946,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009857911854367163,
      "loss": 2.0937,
      "step": 12285
    },
    {
      "epoch": 0.08347534225459351,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009857886387490487,
      "loss": 2.3869,
      "step": 12286
    },
    {
      "epoch": 0.08348213660118757,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009857860918364664,
      "loss": 2.3656,
      "step": 12287
    },
    {
      "epoch": 0.08348893094778162,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009857835446989707,
      "loss": 2.2311,
      "step": 12288
    },
    {
      "epoch": 0.08349572529437568,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009857809973365628,
      "loss": 2.1951,
      "step": 12289
    },
    {
      "epoch": 0.08350251964096975,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000985778449749244,
      "loss": 2.2125,
      "step": 12290
    },
    {
      "epoch": 0.0835093139875638,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009857759019370152,
      "loss": 2.2842,
      "step": 12291
    },
    {
      "epoch": 0.08351610833415786,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009857733538998778,
      "loss": 2.1377,
      "step": 12292
    },
    {
      "epoch": 0.08352290268075191,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009857708056378329,
      "loss": 2.1416,
      "step": 12293
    },
    {
      "epoch": 0.08352969702734597,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009857682571508817,
      "loss": 2.1757,
      "step": 12294
    },
    {
      "epoch": 0.08353649137394002,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009857657084390253,
      "loss": 2.125,
      "step": 12295
    },
    {
      "epoch": 0.08354328572053409,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000985763159502265,
      "loss": 2.2767,
      "step": 12296
    },
    {
      "epoch": 0.08355008006712815,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009857606103406019,
      "loss": 2.1887,
      "step": 12297
    },
    {
      "epoch": 0.0835568744137222,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009857580609540374,
      "loss": 2.2137,
      "step": 12298
    },
    {
      "epoch": 0.08356366876031626,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009857555113425723,
      "loss": 2.2123,
      "step": 12299
    },
    {
      "epoch": 0.08357046310691031,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009857529615062079,
      "loss": 2.2436,
      "step": 12300
    },
    {
      "epoch": 0.08357725745350437,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009857504114449455,
      "loss": 2.1953,
      "step": 12301
    },
    {
      "epoch": 0.08358405180009844,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009857478611587861,
      "loss": 2.168,
      "step": 12302
    },
    {
      "epoch": 0.08359084614669249,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009857453106477312,
      "loss": 2.0687,
      "step": 12303
    },
    {
      "epoch": 0.08359764049328655,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000985742759911782,
      "loss": 2.2346,
      "step": 12304
    },
    {
      "epoch": 0.0836044348398806,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009857402089509392,
      "loss": 2.2243,
      "step": 12305
    },
    {
      "epoch": 0.08361122918647466,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009857376577652043,
      "loss": 2.2652,
      "step": 12306
    },
    {
      "epoch": 0.08361802353306873,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009857351063545783,
      "loss": 2.187,
      "step": 12307
    },
    {
      "epoch": 0.08362481787966278,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009857325547190626,
      "loss": 2.2506,
      "step": 12308
    },
    {
      "epoch": 0.08363161222625684,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009857300028586582,
      "loss": 2.1638,
      "step": 12309
    },
    {
      "epoch": 0.08363840657285089,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009857274507733666,
      "loss": 2.0694,
      "step": 12310
    },
    {
      "epoch": 0.08364520091944495,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009857248984631885,
      "loss": 2.2652,
      "step": 12311
    },
    {
      "epoch": 0.08365199526603902,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009857223459281256,
      "loss": 2.3046,
      "step": 12312
    },
    {
      "epoch": 0.08365878961263307,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009857197931681786,
      "loss": 2.146,
      "step": 12313
    },
    {
      "epoch": 0.08366558395922713,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009857172401833492,
      "loss": 2.2031,
      "step": 12314
    },
    {
      "epoch": 0.08367237830582118,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000985714686973638,
      "loss": 2.237,
      "step": 12315
    },
    {
      "epoch": 0.08367917265241524,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009857121335390467,
      "loss": 2.2679,
      "step": 12316
    },
    {
      "epoch": 0.08368596699900929,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009857095798795762,
      "loss": 2.1738,
      "step": 12317
    },
    {
      "epoch": 0.08369276134560336,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009857070259952274,
      "loss": 2.359,
      "step": 12318
    },
    {
      "epoch": 0.08369955569219742,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009857044718860023,
      "loss": 2.294,
      "step": 12319
    },
    {
      "epoch": 0.08370635003879147,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009857019175519013,
      "loss": 2.0783,
      "step": 12320
    },
    {
      "epoch": 0.08371314438538553,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000985699362992926,
      "loss": 2.2282,
      "step": 12321
    },
    {
      "epoch": 0.08371993873197958,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009856968082090775,
      "loss": 2.1586,
      "step": 12322
    },
    {
      "epoch": 0.08372673307857365,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009856942532003566,
      "loss": 2.2665,
      "step": 12323
    },
    {
      "epoch": 0.08373352742516771,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009856916979667652,
      "loss": 2.4171,
      "step": 12324
    },
    {
      "epoch": 0.08374032177176176,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009856891425083039,
      "loss": 2.2448,
      "step": 12325
    },
    {
      "epoch": 0.08374711611835582,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009856865868249742,
      "loss": 2.2836,
      "step": 12326
    },
    {
      "epoch": 0.08375391046494987,
      "grad_norm": 0.11962890625,
      "learning_rate": 0.000985684030916777,
      "loss": 2.067,
      "step": 12327
    },
    {
      "epoch": 0.08376070481154393,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000985681474783714,
      "loss": 2.2698,
      "step": 12328
    },
    {
      "epoch": 0.083767499158138,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000985678918425786,
      "loss": 2.352,
      "step": 12329
    },
    {
      "epoch": 0.08377429350473205,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000985676361842994,
      "loss": 2.2526,
      "step": 12330
    },
    {
      "epoch": 0.08378108785132611,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009856738050353395,
      "loss": 2.3306,
      "step": 12331
    },
    {
      "epoch": 0.08378788219792016,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009856712480028236,
      "loss": 2.2656,
      "step": 12332
    },
    {
      "epoch": 0.08379467654451422,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009856686907454475,
      "loss": 2.1141,
      "step": 12333
    },
    {
      "epoch": 0.08380147089110829,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009856661332632124,
      "loss": 2.2869,
      "step": 12334
    },
    {
      "epoch": 0.08380826523770234,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009856635755561194,
      "loss": 2.2487,
      "step": 12335
    },
    {
      "epoch": 0.0838150595842964,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009856610176241698,
      "loss": 2.106,
      "step": 12336
    },
    {
      "epoch": 0.08382185393089045,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009856584594673647,
      "loss": 2.2348,
      "step": 12337
    },
    {
      "epoch": 0.08382864827748451,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009856559010857052,
      "loss": 2.1136,
      "step": 12338
    },
    {
      "epoch": 0.08383544262407858,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009856533424791928,
      "loss": 2.1638,
      "step": 12339
    },
    {
      "epoch": 0.08384223697067263,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000985650783647828,
      "loss": 2.1208,
      "step": 12340
    },
    {
      "epoch": 0.08384903131726669,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009856482245916129,
      "loss": 2.3035,
      "step": 12341
    },
    {
      "epoch": 0.08385582566386074,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000985645665310548,
      "loss": 2.2682,
      "step": 12342
    },
    {
      "epoch": 0.0838626200104548,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000985643105804635,
      "loss": 2.2599,
      "step": 12343
    },
    {
      "epoch": 0.08386941435704885,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009856405460738746,
      "loss": 2.2412,
      "step": 12344
    },
    {
      "epoch": 0.08387620870364292,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009856379861182684,
      "loss": 2.087,
      "step": 12345
    },
    {
      "epoch": 0.08388300305023698,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000985635425937817,
      "loss": 2.2161,
      "step": 12346
    },
    {
      "epoch": 0.08388979739683103,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009856328655325223,
      "loss": 2.322,
      "step": 12347
    },
    {
      "epoch": 0.08389659174342509,
      "grad_norm": 0.140625,
      "learning_rate": 0.000985630304902385,
      "loss": 2.3676,
      "step": 12348
    },
    {
      "epoch": 0.08390338609001914,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009856277440474066,
      "loss": 2.3722,
      "step": 12349
    },
    {
      "epoch": 0.0839101804366132,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009856251829675879,
      "loss": 2.0534,
      "step": 12350
    },
    {
      "epoch": 0.08391697478320727,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009856226216629306,
      "loss": 2.1283,
      "step": 12351
    },
    {
      "epoch": 0.08392376912980132,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009856200601334353,
      "loss": 2.1691,
      "step": 12352
    },
    {
      "epoch": 0.08393056347639538,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009856174983791034,
      "loss": 2.0608,
      "step": 12353
    },
    {
      "epoch": 0.08393735782298943,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009856149363999364,
      "loss": 2.2887,
      "step": 12354
    },
    {
      "epoch": 0.0839441521695835,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009856123741959354,
      "loss": 2.0686,
      "step": 12355
    },
    {
      "epoch": 0.08395094651617756,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009856098117671012,
      "loss": 2.105,
      "step": 12356
    },
    {
      "epoch": 0.08395774086277161,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000985607249113435,
      "loss": 2.1898,
      "step": 12357
    },
    {
      "epoch": 0.08396453520936567,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009856046862349386,
      "loss": 2.1633,
      "step": 12358
    },
    {
      "epoch": 0.08397132955595972,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009856021231316127,
      "loss": 2.2003,
      "step": 12359
    },
    {
      "epoch": 0.08397812390255378,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009855995598034586,
      "loss": 2.3305,
      "step": 12360
    },
    {
      "epoch": 0.08398491824914785,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009855969962504773,
      "loss": 2.2056,
      "step": 12361
    },
    {
      "epoch": 0.0839917125957419,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009855944324726703,
      "loss": 2.3269,
      "step": 12362
    },
    {
      "epoch": 0.08399850694233596,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009855918684700386,
      "loss": 2.0927,
      "step": 12363
    },
    {
      "epoch": 0.08400530128893001,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009855893042425834,
      "loss": 2.2403,
      "step": 12364
    },
    {
      "epoch": 0.08401209563552407,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000985586739790306,
      "loss": 2.2892,
      "step": 12365
    },
    {
      "epoch": 0.08401888998211812,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009855841751132074,
      "loss": 2.1043,
      "step": 12366
    },
    {
      "epoch": 0.08402568432871219,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009855816102112892,
      "loss": 2.2845,
      "step": 12367
    },
    {
      "epoch": 0.08403247867530625,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000985579045084552,
      "loss": 2.3543,
      "step": 12368
    },
    {
      "epoch": 0.0840392730219003,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009855764797329974,
      "loss": 2.2276,
      "step": 12369
    },
    {
      "epoch": 0.08404606736849436,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009855739141566263,
      "loss": 2.1742,
      "step": 12370
    },
    {
      "epoch": 0.08405286171508841,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00098557134835544,
      "loss": 2.2205,
      "step": 12371
    },
    {
      "epoch": 0.08405965606168248,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.00098556878232944,
      "loss": 2.3743,
      "step": 12372
    },
    {
      "epoch": 0.08406645040827654,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000985566216078627,
      "loss": 2.2246,
      "step": 12373
    },
    {
      "epoch": 0.08407324475487059,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009855636496030025,
      "loss": 2.3302,
      "step": 12374
    },
    {
      "epoch": 0.08408003910146465,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009855610829025677,
      "loss": 2.2743,
      "step": 12375
    },
    {
      "epoch": 0.0840868334480587,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009855585159773235,
      "loss": 2.4136,
      "step": 12376
    },
    {
      "epoch": 0.08409362779465276,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009855559488272715,
      "loss": 2.3321,
      "step": 12377
    },
    {
      "epoch": 0.08410042214124683,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009855533814524123,
      "loss": 2.228,
      "step": 12378
    },
    {
      "epoch": 0.08410721648784088,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009855508138527477,
      "loss": 2.1666,
      "step": 12379
    },
    {
      "epoch": 0.08411401083443494,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009855482460282786,
      "loss": 2.1703,
      "step": 12380
    },
    {
      "epoch": 0.08412080518102899,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009855456779790063,
      "loss": 2.2284,
      "step": 12381
    },
    {
      "epoch": 0.08412759952762305,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009855431097049317,
      "loss": 2.4244,
      "step": 12382
    },
    {
      "epoch": 0.08413439387421712,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009855405412060563,
      "loss": 2.1702,
      "step": 12383
    },
    {
      "epoch": 0.08414118822081117,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009855379724823814,
      "loss": 2.2707,
      "step": 12384
    },
    {
      "epoch": 0.08414798256740523,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009855354035339077,
      "loss": 2.1334,
      "step": 12385
    },
    {
      "epoch": 0.08415477691399928,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000985532834360637,
      "loss": 2.1647,
      "step": 12386
    },
    {
      "epoch": 0.08416157126059334,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00098553026496257,
      "loss": 2.1734,
      "step": 12387
    },
    {
      "epoch": 0.0841683656071874,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009855276953397078,
      "loss": 2.3666,
      "step": 12388
    },
    {
      "epoch": 0.08417515995378146,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009855251254920522,
      "loss": 2.3548,
      "step": 12389
    },
    {
      "epoch": 0.08418195430037552,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000985522555419604,
      "loss": 2.2271,
      "step": 12390
    },
    {
      "epoch": 0.08418874864696957,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009855199851223641,
      "loss": 2.3011,
      "step": 12391
    },
    {
      "epoch": 0.08419554299356363,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009855174146003343,
      "loss": 2.3536,
      "step": 12392
    },
    {
      "epoch": 0.08420233734015768,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009855148438535155,
      "loss": 2.092,
      "step": 12393
    },
    {
      "epoch": 0.08420913168675175,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009855122728819087,
      "loss": 2.1764,
      "step": 12394
    },
    {
      "epoch": 0.08421592603334581,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009855097016855155,
      "loss": 2.4414,
      "step": 12395
    },
    {
      "epoch": 0.08422272037993986,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000985507130264337,
      "loss": 2.2066,
      "step": 12396
    },
    {
      "epoch": 0.08422951472653392,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009855045586183739,
      "loss": 2.2039,
      "step": 12397
    },
    {
      "epoch": 0.08423630907312797,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009855019867476278,
      "loss": 2.1314,
      "step": 12398
    },
    {
      "epoch": 0.08424310341972203,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009854994146521001,
      "loss": 2.2525,
      "step": 12399
    },
    {
      "epoch": 0.0842498977663161,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009854968423317915,
      "loss": 2.3532,
      "step": 12400
    },
    {
      "epoch": 0.08425669211291015,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009854942697867036,
      "loss": 2.3927,
      "step": 12401
    },
    {
      "epoch": 0.08426348645950421,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009854916970168373,
      "loss": 2.1966,
      "step": 12402
    },
    {
      "epoch": 0.08427028080609826,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000985489124022194,
      "loss": 2.2871,
      "step": 12403
    },
    {
      "epoch": 0.08427707515269232,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009854865508027747,
      "loss": 2.0796,
      "step": 12404
    },
    {
      "epoch": 0.08428386949928639,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009854839773585807,
      "loss": 2.2479,
      "step": 12405
    },
    {
      "epoch": 0.08429066384588044,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009854814036896133,
      "loss": 2.2271,
      "step": 12406
    },
    {
      "epoch": 0.0842974581924745,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009854788297958734,
      "loss": 2.2172,
      "step": 12407
    },
    {
      "epoch": 0.08430425253906855,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009854762556773626,
      "loss": 2.3245,
      "step": 12408
    },
    {
      "epoch": 0.08431104688566261,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009854736813340818,
      "loss": 2.1576,
      "step": 12409
    },
    {
      "epoch": 0.08431784123225668,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009854711067660321,
      "loss": 2.2546,
      "step": 12410
    },
    {
      "epoch": 0.08432463557885073,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009854685319732148,
      "loss": 2.2577,
      "step": 12411
    },
    {
      "epoch": 0.08433142992544479,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009854659569556313,
      "loss": 2.1425,
      "step": 12412
    },
    {
      "epoch": 0.08433822427203884,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009854633817132827,
      "loss": 2.154,
      "step": 12413
    },
    {
      "epoch": 0.0843450186186329,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00098546080624617,
      "loss": 2.3492,
      "step": 12414
    },
    {
      "epoch": 0.08435181296522695,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009854582305542944,
      "loss": 2.1488,
      "step": 12415
    },
    {
      "epoch": 0.08435860731182102,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009854556546376573,
      "loss": 2.2289,
      "step": 12416
    },
    {
      "epoch": 0.08436540165841508,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00098545307849626,
      "loss": 2.2755,
      "step": 12417
    },
    {
      "epoch": 0.08437219600500913,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009854505021301033,
      "loss": 2.3862,
      "step": 12418
    },
    {
      "epoch": 0.08437899035160319,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009854479255391887,
      "loss": 2.1874,
      "step": 12419
    },
    {
      "epoch": 0.08438578469819724,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000985445348723517,
      "loss": 2.114,
      "step": 12420
    },
    {
      "epoch": 0.0843925790447913,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009854427716830898,
      "loss": 2.1605,
      "step": 12421
    },
    {
      "epoch": 0.08439937339138537,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009854401944179084,
      "loss": 2.2703,
      "step": 12422
    },
    {
      "epoch": 0.08440616773797942,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009854376169279736,
      "loss": 2.3229,
      "step": 12423
    },
    {
      "epoch": 0.08441296208457348,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009854350392132865,
      "loss": 2.247,
      "step": 12424
    },
    {
      "epoch": 0.08441975643116753,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009854324612738488,
      "loss": 2.1921,
      "step": 12425
    },
    {
      "epoch": 0.0844265507777616,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009854298831096615,
      "loss": 2.2082,
      "step": 12426
    },
    {
      "epoch": 0.08443334512435566,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009854273047207255,
      "loss": 2.0972,
      "step": 12427
    },
    {
      "epoch": 0.08444013947094971,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009854247261070423,
      "loss": 2.2948,
      "step": 12428
    },
    {
      "epoch": 0.08444693381754377,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009854221472686132,
      "loss": 2.2084,
      "step": 12429
    },
    {
      "epoch": 0.08445372816413782,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000985419568205439,
      "loss": 2.1651,
      "step": 12430
    },
    {
      "epoch": 0.08446052251073188,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000985416988917521,
      "loss": 2.176,
      "step": 12431
    },
    {
      "epoch": 0.08446731685732595,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000985414409404861,
      "loss": 2.2555,
      "step": 12432
    },
    {
      "epoch": 0.08447411120392,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009854118296674591,
      "loss": 2.0945,
      "step": 12433
    },
    {
      "epoch": 0.08448090555051406,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009854092497053174,
      "loss": 2.2916,
      "step": 12434
    },
    {
      "epoch": 0.08448769989710811,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009854066695184367,
      "loss": 2.2981,
      "step": 12435
    },
    {
      "epoch": 0.08449449424370217,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009854040891068183,
      "loss": 2.1846,
      "step": 12436
    },
    {
      "epoch": 0.08450128859029622,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009854015084704635,
      "loss": 2.1385,
      "step": 12437
    },
    {
      "epoch": 0.08450808293689029,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009853989276093732,
      "loss": 2.2781,
      "step": 12438
    },
    {
      "epoch": 0.08451487728348435,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009853963465235487,
      "loss": 2.2533,
      "step": 12439
    },
    {
      "epoch": 0.0845216716300784,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009853937652129915,
      "loss": 2.1172,
      "step": 12440
    },
    {
      "epoch": 0.08452846597667246,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009853911836777023,
      "loss": 2.121,
      "step": 12441
    },
    {
      "epoch": 0.08453526032326651,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009853886019176828,
      "loss": 2.3389,
      "step": 12442
    },
    {
      "epoch": 0.08454205466986058,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009853860199329338,
      "loss": 2.1375,
      "step": 12443
    },
    {
      "epoch": 0.08454884901645464,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009853834377234567,
      "loss": 2.2912,
      "step": 12444
    },
    {
      "epoch": 0.08455564336304869,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009853808552892525,
      "loss": 2.2551,
      "step": 12445
    },
    {
      "epoch": 0.08456243770964275,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009853782726303227,
      "loss": 2.256,
      "step": 12446
    },
    {
      "epoch": 0.0845692320562368,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009853756897466681,
      "loss": 2.0397,
      "step": 12447
    },
    {
      "epoch": 0.08457602640283086,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009853731066382903,
      "loss": 2.2183,
      "step": 12448
    },
    {
      "epoch": 0.08458282074942493,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009853705233051904,
      "loss": 2.4247,
      "step": 12449
    },
    {
      "epoch": 0.08458961509601898,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009853679397473693,
      "loss": 2.3146,
      "step": 12450
    },
    {
      "epoch": 0.08459640944261304,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009853653559648286,
      "loss": 2.0523,
      "step": 12451
    },
    {
      "epoch": 0.08460320378920709,
      "grad_norm": 0.140625,
      "learning_rate": 0.000985362771957569,
      "loss": 2.2957,
      "step": 12452
    },
    {
      "epoch": 0.08460999813580115,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009853601877255922,
      "loss": 2.2295,
      "step": 12453
    },
    {
      "epoch": 0.08461679248239522,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009853576032688993,
      "loss": 2.3047,
      "step": 12454
    },
    {
      "epoch": 0.08462358682898927,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009853550185874912,
      "loss": 2.3253,
      "step": 12455
    },
    {
      "epoch": 0.08463038117558333,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009853524336813693,
      "loss": 2.156,
      "step": 12456
    },
    {
      "epoch": 0.08463717552217738,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000985349848550535,
      "loss": 2.3601,
      "step": 12457
    },
    {
      "epoch": 0.08464396986877144,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009853472631949892,
      "loss": 2.126,
      "step": 12458
    },
    {
      "epoch": 0.0846507642153655,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000985344677614733,
      "loss": 2.1985,
      "step": 12459
    },
    {
      "epoch": 0.08465755856195956,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000985342091809768,
      "loss": 2.0959,
      "step": 12460
    },
    {
      "epoch": 0.08466435290855362,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009853395057800951,
      "loss": 2.3035,
      "step": 12461
    },
    {
      "epoch": 0.08467114725514767,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009853369195257154,
      "loss": 2.2216,
      "step": 12462
    },
    {
      "epoch": 0.08467794160174173,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009853343330466305,
      "loss": 2.1355,
      "step": 12463
    },
    {
      "epoch": 0.08468473594833578,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009853317463428414,
      "loss": 2.2106,
      "step": 12464
    },
    {
      "epoch": 0.08469153029492985,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009853291594143492,
      "loss": 2.1359,
      "step": 12465
    },
    {
      "epoch": 0.08469832464152391,
      "grad_norm": 0.140625,
      "learning_rate": 0.000985326572261155,
      "loss": 2.2396,
      "step": 12466
    },
    {
      "epoch": 0.08470511898811796,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009853239848832602,
      "loss": 2.2113,
      "step": 12467
    },
    {
      "epoch": 0.08471191333471202,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000985321397280666,
      "loss": 2.1964,
      "step": 12468
    },
    {
      "epoch": 0.08471870768130607,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009853188094533737,
      "loss": 2.1779,
      "step": 12469
    },
    {
      "epoch": 0.08472550202790013,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009853162214013842,
      "loss": 2.2089,
      "step": 12470
    },
    {
      "epoch": 0.0847322963744942,
      "grad_norm": 0.140625,
      "learning_rate": 0.000985313633124699,
      "loss": 2.3515,
      "step": 12471
    },
    {
      "epoch": 0.08473909072108825,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009853110446233191,
      "loss": 2.28,
      "step": 12472
    },
    {
      "epoch": 0.08474588506768231,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009853084558972456,
      "loss": 2.2169,
      "step": 12473
    },
    {
      "epoch": 0.08475267941427636,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00098530586694648,
      "loss": 2.3071,
      "step": 12474
    },
    {
      "epoch": 0.08475947376087042,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009853032777710233,
      "loss": 2.2413,
      "step": 12475
    },
    {
      "epoch": 0.08476626810746449,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009853006883708767,
      "loss": 2.4601,
      "step": 12476
    },
    {
      "epoch": 0.08477306245405854,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009852980987460416,
      "loss": 2.2785,
      "step": 12477
    },
    {
      "epoch": 0.0847798568006526,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000985295508896519,
      "loss": 2.1613,
      "step": 12478
    },
    {
      "epoch": 0.08478665114724665,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009852929188223101,
      "loss": 2.1136,
      "step": 12479
    },
    {
      "epoch": 0.08479344549384071,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009852903285234161,
      "loss": 2.3074,
      "step": 12480
    },
    {
      "epoch": 0.08480023984043478,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009852877379998382,
      "loss": 2.2096,
      "step": 12481
    },
    {
      "epoch": 0.08480703418702883,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009852851472515777,
      "loss": 2.1247,
      "step": 12482
    },
    {
      "epoch": 0.08481382853362289,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009852825562786359,
      "loss": 2.1372,
      "step": 12483
    },
    {
      "epoch": 0.08482062288021694,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009852799650810135,
      "loss": 2.0494,
      "step": 12484
    },
    {
      "epoch": 0.084827417226811,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009852773736587124,
      "loss": 2.3024,
      "step": 12485
    },
    {
      "epoch": 0.08483421157340505,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000985274782011733,
      "loss": 2.1123,
      "step": 12486
    },
    {
      "epoch": 0.08484100591999912,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009852721901400774,
      "loss": 2.3237,
      "step": 12487
    },
    {
      "epoch": 0.08484780026659318,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009852695980437462,
      "loss": 2.2962,
      "step": 12488
    },
    {
      "epoch": 0.08485459461318723,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009852670057227406,
      "loss": 2.2685,
      "step": 12489
    },
    {
      "epoch": 0.08486138895978129,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000985264413177062,
      "loss": 2.2134,
      "step": 12490
    },
    {
      "epoch": 0.08486818330637534,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009852618204067116,
      "loss": 2.2697,
      "step": 12491
    },
    {
      "epoch": 0.0848749776529694,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009852592274116904,
      "loss": 2.2119,
      "step": 12492
    },
    {
      "epoch": 0.08488177199956347,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000985256634192,
      "loss": 2.3282,
      "step": 12493
    },
    {
      "epoch": 0.08488856634615752,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009852540407476412,
      "loss": 2.1163,
      "step": 12494
    },
    {
      "epoch": 0.08489536069275158,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009852514470786154,
      "loss": 2.114,
      "step": 12495
    },
    {
      "epoch": 0.08490215503934563,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009852488531849235,
      "loss": 2.311,
      "step": 12496
    },
    {
      "epoch": 0.0849089493859397,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009852462590665672,
      "loss": 2.2271,
      "step": 12497
    },
    {
      "epoch": 0.08491574373253376,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009852436647235475,
      "loss": 2.1661,
      "step": 12498
    },
    {
      "epoch": 0.08492253807912781,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009852410701558653,
      "loss": 2.2149,
      "step": 12499
    },
    {
      "epoch": 0.08492933242572187,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009852384753635222,
      "loss": 2.2376,
      "step": 12500
    },
    {
      "epoch": 0.08493612677231592,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000985235880346519,
      "loss": 2.0629,
      "step": 12501
    },
    {
      "epoch": 0.08494292111890998,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009852332851048575,
      "loss": 2.1674,
      "step": 12502
    },
    {
      "epoch": 0.08494971546550405,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009852306896385385,
      "loss": 2.063,
      "step": 12503
    },
    {
      "epoch": 0.0849565098120981,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009852280939475632,
      "loss": 2.2899,
      "step": 12504
    },
    {
      "epoch": 0.08496330415869216,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009852254980319327,
      "loss": 2.3072,
      "step": 12505
    },
    {
      "epoch": 0.08497009850528621,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009852229018916486,
      "loss": 2.1735,
      "step": 12506
    },
    {
      "epoch": 0.08497689285188027,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009852203055267118,
      "loss": 2.258,
      "step": 12507
    },
    {
      "epoch": 0.08498368719847434,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009852177089371232,
      "loss": 2.2198,
      "step": 12508
    },
    {
      "epoch": 0.08499048154506839,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009852151121228848,
      "loss": 2.3669,
      "step": 12509
    },
    {
      "epoch": 0.08499727589166245,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009852125150839972,
      "loss": 2.2288,
      "step": 12510
    },
    {
      "epoch": 0.0850040702382565,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000985209917820462,
      "loss": 2.3702,
      "step": 12511
    },
    {
      "epoch": 0.08501086458485056,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009852073203322797,
      "loss": 2.1154,
      "step": 12512
    },
    {
      "epoch": 0.08501765893144461,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009852047226194523,
      "loss": 2.2035,
      "step": 12513
    },
    {
      "epoch": 0.08502445327803868,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009852021246819806,
      "loss": 2.3124,
      "step": 12514
    },
    {
      "epoch": 0.08503124762463274,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009851995265198659,
      "loss": 2.2696,
      "step": 12515
    },
    {
      "epoch": 0.08503804197122679,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009851969281331095,
      "loss": 2.4179,
      "step": 12516
    },
    {
      "epoch": 0.08504483631782085,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009851943295217122,
      "loss": 2.2718,
      "step": 12517
    },
    {
      "epoch": 0.0850516306644149,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009851917306856757,
      "loss": 2.256,
      "step": 12518
    },
    {
      "epoch": 0.08505842501100896,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000985189131625001,
      "loss": 2.2229,
      "step": 12519
    },
    {
      "epoch": 0.08506521935760303,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009851865323396891,
      "loss": 2.2839,
      "step": 12520
    },
    {
      "epoch": 0.08507201370419708,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009851839328297416,
      "loss": 2.124,
      "step": 12521
    },
    {
      "epoch": 0.08507880805079114,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009851813330951594,
      "loss": 2.1823,
      "step": 12522
    },
    {
      "epoch": 0.08508560239738519,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009851787331359438,
      "loss": 2.11,
      "step": 12523
    },
    {
      "epoch": 0.08509239674397925,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000985176132952096,
      "loss": 2.2572,
      "step": 12524
    },
    {
      "epoch": 0.08509919109057332,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009851735325436175,
      "loss": 2.1272,
      "step": 12525
    },
    {
      "epoch": 0.08510598543716737,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009851709319105089,
      "loss": 2.1623,
      "step": 12526
    },
    {
      "epoch": 0.08511277978376143,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009851683310527718,
      "loss": 2.3806,
      "step": 12527
    },
    {
      "epoch": 0.08511957413035548,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009851657299704073,
      "loss": 2.1989,
      "step": 12528
    },
    {
      "epoch": 0.08512636847694954,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009851631286634166,
      "loss": 2.1386,
      "step": 12529
    },
    {
      "epoch": 0.0851331628235436,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000985160527131801,
      "loss": 2.1648,
      "step": 12530
    },
    {
      "epoch": 0.08513995717013766,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009851579253755617,
      "loss": 2.1451,
      "step": 12531
    },
    {
      "epoch": 0.08514675151673172,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009851553233946996,
      "loss": 2.3553,
      "step": 12532
    },
    {
      "epoch": 0.08515354586332577,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009851527211892164,
      "loss": 2.2607,
      "step": 12533
    },
    {
      "epoch": 0.08516034020991983,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000985150118759113,
      "loss": 2.1721,
      "step": 12534
    },
    {
      "epoch": 0.08516713455651388,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009851475161043906,
      "loss": 2.1523,
      "step": 12535
    },
    {
      "epoch": 0.08517392890310795,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009851449132250506,
      "loss": 2.0625,
      "step": 12536
    },
    {
      "epoch": 0.08518072324970201,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000985142310121094,
      "loss": 2.1667,
      "step": 12537
    },
    {
      "epoch": 0.08518751759629606,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000985139706792522,
      "loss": 2.2904,
      "step": 12538
    },
    {
      "epoch": 0.08519431194289012,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000985137103239336,
      "loss": 2.2916,
      "step": 12539
    },
    {
      "epoch": 0.08520110628948417,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009851344994615367,
      "loss": 2.197,
      "step": 12540
    },
    {
      "epoch": 0.08520790063607823,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000985131895459126,
      "loss": 2.3368,
      "step": 12541
    },
    {
      "epoch": 0.0852146949826723,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009851292912321048,
      "loss": 2.1669,
      "step": 12542
    },
    {
      "epoch": 0.08522148932926635,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009851266867804743,
      "loss": 2.4419,
      "step": 12543
    },
    {
      "epoch": 0.08522828367586041,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009851240821042357,
      "loss": 2.0707,
      "step": 12544
    },
    {
      "epoch": 0.08523507802245446,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009851214772033901,
      "loss": 2.2434,
      "step": 12545
    },
    {
      "epoch": 0.08524187236904852,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009851188720779388,
      "loss": 2.245,
      "step": 12546
    },
    {
      "epoch": 0.08524866671564259,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009851162667278833,
      "loss": 2.1983,
      "step": 12547
    },
    {
      "epoch": 0.08525546106223664,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009851136611532243,
      "loss": 2.1429,
      "step": 12548
    },
    {
      "epoch": 0.0852622554088307,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009851110553539635,
      "loss": 2.2962,
      "step": 12549
    },
    {
      "epoch": 0.08526904975542475,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009851084493301015,
      "loss": 2.1872,
      "step": 12550
    },
    {
      "epoch": 0.08527584410201881,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009851058430816399,
      "loss": 2.1899,
      "step": 12551
    },
    {
      "epoch": 0.08528263844861288,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00098510323660858,
      "loss": 2.3441,
      "step": 12552
    },
    {
      "epoch": 0.08528943279520693,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000985100629910923,
      "loss": 2.2829,
      "step": 12553
    },
    {
      "epoch": 0.08529622714180099,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009850980229886698,
      "loss": 2.31,
      "step": 12554
    },
    {
      "epoch": 0.08530302148839504,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009850954158418215,
      "loss": 2.2516,
      "step": 12555
    },
    {
      "epoch": 0.0853098158349891,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009850928084703797,
      "loss": 2.2138,
      "step": 12556
    },
    {
      "epoch": 0.08531661018158315,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000985090200874346,
      "loss": 2.3123,
      "step": 12557
    },
    {
      "epoch": 0.08532340452817722,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009850875930537207,
      "loss": 2.1232,
      "step": 12558
    },
    {
      "epoch": 0.08533019887477128,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009850849850085053,
      "loss": 2.0676,
      "step": 12559
    },
    {
      "epoch": 0.08533699322136533,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009850823767387012,
      "loss": 2.4666,
      "step": 12560
    },
    {
      "epoch": 0.08534378756795939,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009850797682443096,
      "loss": 2.1762,
      "step": 12561
    },
    {
      "epoch": 0.08535058191455344,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009850771595253315,
      "loss": 2.2832,
      "step": 12562
    },
    {
      "epoch": 0.0853573762611475,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009850745505817684,
      "loss": 2.1685,
      "step": 12563
    },
    {
      "epoch": 0.08536417060774157,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009850719414136212,
      "loss": 2.2997,
      "step": 12564
    },
    {
      "epoch": 0.08537096495433562,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009850693320208913,
      "loss": 2.2018,
      "step": 12565
    },
    {
      "epoch": 0.08537775930092968,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00098506672240358,
      "loss": 2.1474,
      "step": 12566
    },
    {
      "epoch": 0.08538455364752373,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009850641125616882,
      "loss": 2.1946,
      "step": 12567
    },
    {
      "epoch": 0.0853913479941178,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009850615024952175,
      "loss": 2.2481,
      "step": 12568
    },
    {
      "epoch": 0.08539814234071186,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009850588922041686,
      "loss": 2.176,
      "step": 12569
    },
    {
      "epoch": 0.08540493668730591,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009850562816885431,
      "loss": 2.3684,
      "step": 12570
    },
    {
      "epoch": 0.08541173103389997,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009850536709483423,
      "loss": 2.0701,
      "step": 12571
    },
    {
      "epoch": 0.08541852538049402,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000985051059983567,
      "loss": 2.2678,
      "step": 12572
    },
    {
      "epoch": 0.08542531972708808,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009850484487942185,
      "loss": 2.1793,
      "step": 12573
    },
    {
      "epoch": 0.08543211407368215,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009850458373802983,
      "loss": 2.3422,
      "step": 12574
    },
    {
      "epoch": 0.0854389084202762,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009850432257418073,
      "loss": 2.3757,
      "step": 12575
    },
    {
      "epoch": 0.08544570276687026,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000985040613878747,
      "loss": 2.16,
      "step": 12576
    },
    {
      "epoch": 0.08545249711346431,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009850380017911184,
      "loss": 2.4043,
      "step": 12577
    },
    {
      "epoch": 0.08545929146005837,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009850353894789228,
      "loss": 2.1715,
      "step": 12578
    },
    {
      "epoch": 0.08546608580665244,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009850327769421615,
      "loss": 2.2534,
      "step": 12579
    },
    {
      "epoch": 0.08547288015324649,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009850301641808354,
      "loss": 2.2314,
      "step": 12580
    },
    {
      "epoch": 0.08547967449984055,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009850275511949459,
      "loss": 2.1042,
      "step": 12581
    },
    {
      "epoch": 0.0854864688464346,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009850249379844943,
      "loss": 2.2962,
      "step": 12582
    },
    {
      "epoch": 0.08549326319302866,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009850223245494816,
      "loss": 2.2668,
      "step": 12583
    },
    {
      "epoch": 0.08550005753962271,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009850197108899092,
      "loss": 2.3074,
      "step": 12584
    },
    {
      "epoch": 0.08550685188621678,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009850170970057781,
      "loss": 2.274,
      "step": 12585
    },
    {
      "epoch": 0.08551364623281084,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.00098501448289709,
      "loss": 2.3761,
      "step": 12586
    },
    {
      "epoch": 0.08552044057940489,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009850118685638455,
      "loss": 2.23,
      "step": 12587
    },
    {
      "epoch": 0.08552723492599895,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000985009254006046,
      "loss": 2.2998,
      "step": 12588
    },
    {
      "epoch": 0.085534029272593,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009850066392236927,
      "loss": 2.2863,
      "step": 12589
    },
    {
      "epoch": 0.08554082361918706,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000985004024216787,
      "loss": 2.3966,
      "step": 12590
    },
    {
      "epoch": 0.08554761796578113,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009850014089853303,
      "loss": 2.2449,
      "step": 12591
    },
    {
      "epoch": 0.08555441231237518,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009849987935293231,
      "loss": 2.2217,
      "step": 12592
    },
    {
      "epoch": 0.08556120665896924,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009849961778487673,
      "loss": 2.2147,
      "step": 12593
    },
    {
      "epoch": 0.08556800100556329,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009849935619436637,
      "loss": 2.1395,
      "step": 12594
    },
    {
      "epoch": 0.08557479535215735,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009849909458140135,
      "loss": 2.2,
      "step": 12595
    },
    {
      "epoch": 0.08558158969875142,
      "grad_norm": 0.125,
      "learning_rate": 0.0009849883294598182,
      "loss": 2.0572,
      "step": 12596
    },
    {
      "epoch": 0.08558838404534547,
      "grad_norm": 0.125,
      "learning_rate": 0.0009849857128810789,
      "loss": 2.0639,
      "step": 12597
    },
    {
      "epoch": 0.08559517839193953,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009849830960777966,
      "loss": 2.2047,
      "step": 12598
    },
    {
      "epoch": 0.08560197273853358,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000984980479049973,
      "loss": 2.2283,
      "step": 12599
    },
    {
      "epoch": 0.08560876708512764,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009849778617976087,
      "loss": 2.2944,
      "step": 12600
    },
    {
      "epoch": 0.0856155614317217,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009849752443207055,
      "loss": 2.2267,
      "step": 12601
    },
    {
      "epoch": 0.08562235577831576,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000984972626619264,
      "loss": 2.2532,
      "step": 12602
    },
    {
      "epoch": 0.08562915012490982,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009849700086932858,
      "loss": 2.0898,
      "step": 12603
    },
    {
      "epoch": 0.08563594447150387,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009849673905427722,
      "loss": 2.1378,
      "step": 12604
    },
    {
      "epoch": 0.08564273881809793,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009849647721677241,
      "loss": 2.3634,
      "step": 12605
    },
    {
      "epoch": 0.08564953316469198,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000984962153568143,
      "loss": 2.2736,
      "step": 12606
    },
    {
      "epoch": 0.08565632751128605,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00098495953474403,
      "loss": 2.2477,
      "step": 12607
    },
    {
      "epoch": 0.08566312185788011,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009849569156953862,
      "loss": 2.2265,
      "step": 12608
    },
    {
      "epoch": 0.08566991620447416,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009849542964222129,
      "loss": 2.222,
      "step": 12609
    },
    {
      "epoch": 0.08567671055106822,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009849516769245114,
      "loss": 2.0964,
      "step": 12610
    },
    {
      "epoch": 0.08568350489766227,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009849490572022828,
      "loss": 2.1896,
      "step": 12611
    },
    {
      "epoch": 0.08569029924425633,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009849464372555284,
      "loss": 2.3812,
      "step": 12612
    },
    {
      "epoch": 0.0856970935908504,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009849438170842493,
      "loss": 2.2968,
      "step": 12613
    },
    {
      "epoch": 0.08570388793744445,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009849411966884468,
      "loss": 2.3007,
      "step": 12614
    },
    {
      "epoch": 0.08571068228403851,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984938576068122,
      "loss": 2.1739,
      "step": 12615
    },
    {
      "epoch": 0.08571747663063256,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000984935955223276,
      "loss": 2.1279,
      "step": 12616
    },
    {
      "epoch": 0.08572427097722662,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009849333341539106,
      "loss": 2.2975,
      "step": 12617
    },
    {
      "epoch": 0.08573106532382069,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009849307128600266,
      "loss": 2.3038,
      "step": 12618
    },
    {
      "epoch": 0.08573785967041474,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009849280913416252,
      "loss": 2.3307,
      "step": 12619
    },
    {
      "epoch": 0.0857446540170088,
      "grad_norm": 0.125,
      "learning_rate": 0.0009849254695987076,
      "loss": 2.1256,
      "step": 12620
    },
    {
      "epoch": 0.08575144836360285,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000984922847631275,
      "loss": 1.9637,
      "step": 12621
    },
    {
      "epoch": 0.08575824271019691,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009849202254393287,
      "loss": 2.3392,
      "step": 12622
    },
    {
      "epoch": 0.08576503705679098,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00098491760302287,
      "loss": 2.3042,
      "step": 12623
    },
    {
      "epoch": 0.08577183140338503,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009849149803818997,
      "loss": 2.404,
      "step": 12624
    },
    {
      "epoch": 0.08577862574997909,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009849123575164196,
      "loss": 2.3364,
      "step": 12625
    },
    {
      "epoch": 0.08578542009657314,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009849097344264307,
      "loss": 2.2366,
      "step": 12626
    },
    {
      "epoch": 0.0857922144431672,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000984907111111934,
      "loss": 2.1908,
      "step": 12627
    },
    {
      "epoch": 0.08579900878976127,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000984904487572931,
      "loss": 2.1144,
      "step": 12628
    },
    {
      "epoch": 0.08580580313635532,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009849018638094227,
      "loss": 2.152,
      "step": 12629
    },
    {
      "epoch": 0.08581259748294938,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009848992398214104,
      "loss": 2.2556,
      "step": 12630
    },
    {
      "epoch": 0.08581939182954343,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000984896615608895,
      "loss": 2.2472,
      "step": 12631
    },
    {
      "epoch": 0.08582618617613749,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009848939911718787,
      "loss": 2.1715,
      "step": 12632
    },
    {
      "epoch": 0.08583298052273154,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009848913665103617,
      "loss": 2.2968,
      "step": 12633
    },
    {
      "epoch": 0.0858397748693256,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009848887416243454,
      "loss": 2.224,
      "step": 12634
    },
    {
      "epoch": 0.08584656921591967,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009848861165138313,
      "loss": 2.1111,
      "step": 12635
    },
    {
      "epoch": 0.08585336356251372,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009848834911788205,
      "loss": 2.3139,
      "step": 12636
    },
    {
      "epoch": 0.08586015790910778,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009848808656193141,
      "loss": 2.2074,
      "step": 12637
    },
    {
      "epoch": 0.08586695225570183,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009848782398353135,
      "loss": 2.2967,
      "step": 12638
    },
    {
      "epoch": 0.0858737466022959,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00098487561382682,
      "loss": 2.2724,
      "step": 12639
    },
    {
      "epoch": 0.08588054094888996,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009848729875938345,
      "loss": 2.3274,
      "step": 12640
    },
    {
      "epoch": 0.08588733529548401,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009848703611363583,
      "loss": 2.0482,
      "step": 12641
    },
    {
      "epoch": 0.08589412964207807,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009848677344543927,
      "loss": 2.1259,
      "step": 12642
    },
    {
      "epoch": 0.08590092398867212,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000984865107547939,
      "loss": 2.4139,
      "step": 12643
    },
    {
      "epoch": 0.08590771833526618,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009848624804169983,
      "loss": 2.1782,
      "step": 12644
    },
    {
      "epoch": 0.08591451268186025,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009848598530615717,
      "loss": 2.0483,
      "step": 12645
    },
    {
      "epoch": 0.0859213070284543,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009848572254816605,
      "loss": 2.2256,
      "step": 12646
    },
    {
      "epoch": 0.08592810137504836,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009848545976772663,
      "loss": 2.3303,
      "step": 12647
    },
    {
      "epoch": 0.08593489572164241,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009848519696483897,
      "loss": 2.3441,
      "step": 12648
    },
    {
      "epoch": 0.08594169006823647,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009848493413950321,
      "loss": 2.3574,
      "step": 12649
    },
    {
      "epoch": 0.08594848441483054,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009848467129171953,
      "loss": 2.1796,
      "step": 12650
    },
    {
      "epoch": 0.08595527876142459,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009848440842148796,
      "loss": 2.2637,
      "step": 12651
    },
    {
      "epoch": 0.08596207310801865,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009848414552880867,
      "loss": 2.2278,
      "step": 12652
    },
    {
      "epoch": 0.0859688674546127,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009848388261368178,
      "loss": 2.1757,
      "step": 12653
    },
    {
      "epoch": 0.08597566180120676,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009848361967610742,
      "loss": 2.3981,
      "step": 12654
    },
    {
      "epoch": 0.08598245614780081,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.0009848335671608568,
      "loss": 2.2702,
      "step": 12655
    },
    {
      "epoch": 0.08598925049439488,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009848309373361671,
      "loss": 2.184,
      "step": 12656
    },
    {
      "epoch": 0.08599604484098894,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009848283072870063,
      "loss": 2.1805,
      "step": 12657
    },
    {
      "epoch": 0.08600283918758299,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009848256770133754,
      "loss": 2.3248,
      "step": 12658
    },
    {
      "epoch": 0.08600963353417705,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000984823046515276,
      "loss": 2.0251,
      "step": 12659
    },
    {
      "epoch": 0.0860164278807711,
      "grad_norm": 0.140625,
      "learning_rate": 0.000984820415792709,
      "loss": 2.2211,
      "step": 12660
    },
    {
      "epoch": 0.08602322222736516,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009848177848456757,
      "loss": 2.106,
      "step": 12661
    },
    {
      "epoch": 0.08603001657395923,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009848151536741771,
      "loss": 2.3936,
      "step": 12662
    },
    {
      "epoch": 0.08603681092055328,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009848125222782149,
      "loss": 2.2004,
      "step": 12663
    },
    {
      "epoch": 0.08604360526714734,
      "grad_norm": 0.125,
      "learning_rate": 0.00098480989065779,
      "loss": 2.0168,
      "step": 12664
    },
    {
      "epoch": 0.08605039961374139,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009848072588129037,
      "loss": 2.1941,
      "step": 12665
    },
    {
      "epoch": 0.08605719396033545,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009848046267435572,
      "loss": 2.1727,
      "step": 12666
    },
    {
      "epoch": 0.08606398830692952,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009848019944497517,
      "loss": 2.2172,
      "step": 12667
    },
    {
      "epoch": 0.08607078265352357,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009847993619314884,
      "loss": 2.1682,
      "step": 12668
    },
    {
      "epoch": 0.08607757700011763,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009847967291887686,
      "loss": 2.1692,
      "step": 12669
    },
    {
      "epoch": 0.08608437134671168,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009847940962215934,
      "loss": 2.1672,
      "step": 12670
    },
    {
      "epoch": 0.08609116569330574,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009847914630299642,
      "loss": 2.1192,
      "step": 12671
    },
    {
      "epoch": 0.0860979600398998,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000984788829613882,
      "loss": 2.2624,
      "step": 12672
    },
    {
      "epoch": 0.08610475438649386,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000984786195973348,
      "loss": 2.1542,
      "step": 12673
    },
    {
      "epoch": 0.08611154873308792,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000984783562108364,
      "loss": 2.2701,
      "step": 12674
    },
    {
      "epoch": 0.08611834307968197,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009847809280189304,
      "loss": 2.1518,
      "step": 12675
    },
    {
      "epoch": 0.08612513742627603,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984778293705049,
      "loss": 2.1937,
      "step": 12676
    },
    {
      "epoch": 0.08613193177287008,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009847756591667206,
      "loss": 2.3179,
      "step": 12677
    },
    {
      "epoch": 0.08613872611946415,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009847730244039467,
      "loss": 2.0868,
      "step": 12678
    },
    {
      "epoch": 0.08614552046605821,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009847703894167285,
      "loss": 2.0602,
      "step": 12679
    },
    {
      "epoch": 0.08615231481265226,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009847677542050672,
      "loss": 2.2771,
      "step": 12680
    },
    {
      "epoch": 0.08615910915924632,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000984765118768964,
      "loss": 2.3126,
      "step": 12681
    },
    {
      "epoch": 0.08616590350584037,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00098476248310842,
      "loss": 2.2448,
      "step": 12682
    },
    {
      "epoch": 0.08617269785243444,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009847598472234366,
      "loss": 2.2345,
      "step": 12683
    },
    {
      "epoch": 0.0861794921990285,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000984757211114015,
      "loss": 2.3012,
      "step": 12684
    },
    {
      "epoch": 0.08618628654562255,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984754574780156,
      "loss": 2.3329,
      "step": 12685
    },
    {
      "epoch": 0.08619308089221661,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009847519382218614,
      "loss": 2.3084,
      "step": 12686
    },
    {
      "epoch": 0.08619987523881066,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009847493014391324,
      "loss": 2.1917,
      "step": 12687
    },
    {
      "epoch": 0.08620666958540472,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009847466644319698,
      "loss": 2.0732,
      "step": 12688
    },
    {
      "epoch": 0.08621346393199879,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009847440272003753,
      "loss": 2.2095,
      "step": 12689
    },
    {
      "epoch": 0.08622025827859284,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009847413897443498,
      "loss": 2.2759,
      "step": 12690
    },
    {
      "epoch": 0.0862270526251869,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009847387520638944,
      "loss": 2.1723,
      "step": 12691
    },
    {
      "epoch": 0.08623384697178095,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009847361141590107,
      "loss": 2.1304,
      "step": 12692
    },
    {
      "epoch": 0.08624064131837501,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009847334760296996,
      "loss": 2.2622,
      "step": 12693
    },
    {
      "epoch": 0.08624743566496908,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009847308376759625,
      "loss": 2.1871,
      "step": 12694
    },
    {
      "epoch": 0.08625423001156313,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009847281990978008,
      "loss": 2.2282,
      "step": 12695
    },
    {
      "epoch": 0.08626102435815719,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009847255602952153,
      "loss": 2.2793,
      "step": 12696
    },
    {
      "epoch": 0.08626781870475124,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009847229212682075,
      "loss": 2.2725,
      "step": 12697
    },
    {
      "epoch": 0.0862746130513453,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0009847202820167783,
      "loss": 2.3605,
      "step": 12698
    },
    {
      "epoch": 0.08628140739793937,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009847176425409295,
      "loss": 2.2804,
      "step": 12699
    },
    {
      "epoch": 0.08628820174453342,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009847150028406617,
      "loss": 2.1506,
      "step": 12700
    },
    {
      "epoch": 0.08629499609112748,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009847123629159766,
      "loss": 2.0335,
      "step": 12701
    },
    {
      "epoch": 0.08630179043772153,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000984709722766875,
      "loss": 2.1976,
      "step": 12702
    },
    {
      "epoch": 0.08630858478431559,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0009847070823933587,
      "loss": 2.0508,
      "step": 12703
    },
    {
      "epoch": 0.08631537913090964,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009847044417954285,
      "loss": 2.242,
      "step": 12704
    },
    {
      "epoch": 0.0863221734775037,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009847018009730855,
      "loss": 2.0626,
      "step": 12705
    },
    {
      "epoch": 0.08632896782409777,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009846991599263312,
      "loss": 2.1795,
      "step": 12706
    },
    {
      "epoch": 0.08633576217069182,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000984696518655167,
      "loss": 2.1658,
      "step": 12707
    },
    {
      "epoch": 0.08634255651728588,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009846938771595935,
      "loss": 1.9337,
      "step": 12708
    },
    {
      "epoch": 0.08634935086387993,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009846912354396125,
      "loss": 2.1774,
      "step": 12709
    },
    {
      "epoch": 0.086356145210474,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009846885934952248,
      "loss": 2.0664,
      "step": 12710
    },
    {
      "epoch": 0.08636293955706806,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000984685951326432,
      "loss": 2.2268,
      "step": 12711
    },
    {
      "epoch": 0.08636973390366211,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009846833089332352,
      "loss": 2.3176,
      "step": 12712
    },
    {
      "epoch": 0.08637652825025617,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009846806663156356,
      "loss": 2.134,
      "step": 12713
    },
    {
      "epoch": 0.08638332259685022,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009846780234736344,
      "loss": 2.1125,
      "step": 12714
    },
    {
      "epoch": 0.08639011694344428,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009846753804072328,
      "loss": 2.12,
      "step": 12715
    },
    {
      "epoch": 0.08639691129003835,
      "grad_norm": 0.140625,
      "learning_rate": 0.000984672737116432,
      "loss": 2.2615,
      "step": 12716
    },
    {
      "epoch": 0.0864037056366324,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009846700936012333,
      "loss": 2.1617,
      "step": 12717
    },
    {
      "epoch": 0.08641049998322646,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984667449861638,
      "loss": 2.1572,
      "step": 12718
    },
    {
      "epoch": 0.08641729432982051,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009846648058976471,
      "loss": 2.1457,
      "step": 12719
    },
    {
      "epoch": 0.08642408867641457,
      "grad_norm": 0.140625,
      "learning_rate": 0.000984662161709262,
      "loss": 2.2996,
      "step": 12720
    },
    {
      "epoch": 0.08643088302300864,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009846595172964838,
      "loss": 2.1898,
      "step": 12721
    },
    {
      "epoch": 0.08643767736960269,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000984656872659314,
      "loss": 2.2905,
      "step": 12722
    },
    {
      "epoch": 0.08644447171619675,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009846542277977533,
      "loss": 2.2853,
      "step": 12723
    },
    {
      "epoch": 0.0864512660627908,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009846515827118035,
      "loss": 2.3105,
      "step": 12724
    },
    {
      "epoch": 0.08645806040938486,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009846489374014656,
      "loss": 2.0776,
      "step": 12725
    },
    {
      "epoch": 0.08646485475597891,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009846462918667407,
      "loss": 2.1048,
      "step": 12726
    },
    {
      "epoch": 0.08647164910257298,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009846436461076302,
      "loss": 2.3824,
      "step": 12727
    },
    {
      "epoch": 0.08647844344916704,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009846410001241352,
      "loss": 2.2504,
      "step": 12728
    },
    {
      "epoch": 0.08648523779576109,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000984638353916257,
      "loss": 2.3023,
      "step": 12729
    },
    {
      "epoch": 0.08649203214235515,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009846357074839967,
      "loss": 2.0669,
      "step": 12730
    },
    {
      "epoch": 0.0864988264889492,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009846330608273557,
      "loss": 2.1059,
      "step": 12731
    },
    {
      "epoch": 0.08650562083554326,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009846304139463351,
      "loss": 2.1282,
      "step": 12732
    },
    {
      "epoch": 0.08651241518213733,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000984627766840936,
      "loss": 2.2189,
      "step": 12733
    },
    {
      "epoch": 0.08651920952873138,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009846251195111601,
      "loss": 2.215,
      "step": 12734
    },
    {
      "epoch": 0.08652600387532544,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009846224719570083,
      "loss": 2.2494,
      "step": 12735
    },
    {
      "epoch": 0.08653279822191949,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000984619824178482,
      "loss": 2.2895,
      "step": 12736
    },
    {
      "epoch": 0.08653959256851355,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009846171761755818,
      "loss": 2.3685,
      "step": 12737
    },
    {
      "epoch": 0.08654638691510762,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009846145279483096,
      "loss": 2.1922,
      "step": 12738
    },
    {
      "epoch": 0.08655318126170167,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009846118794966665,
      "loss": 2.1427,
      "step": 12739
    },
    {
      "epoch": 0.08655997560829573,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009846092308206538,
      "loss": 2.3665,
      "step": 12740
    },
    {
      "epoch": 0.08656676995488978,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009846065819202725,
      "loss": 2.2586,
      "step": 12741
    },
    {
      "epoch": 0.08657356430148384,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009846039327955237,
      "loss": 2.0796,
      "step": 12742
    },
    {
      "epoch": 0.0865803586480779,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000984601283446409,
      "loss": 2.3796,
      "step": 12743
    },
    {
      "epoch": 0.08658715299467196,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009845986338729296,
      "loss": 2.1798,
      "step": 12744
    },
    {
      "epoch": 0.08659394734126602,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009845959840750862,
      "loss": 2.0702,
      "step": 12745
    },
    {
      "epoch": 0.08660074168786007,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009845933340528808,
      "loss": 2.2731,
      "step": 12746
    },
    {
      "epoch": 0.08660753603445413,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000984590683806314,
      "loss": 2.0545,
      "step": 12747
    },
    {
      "epoch": 0.0866143303810482,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009845880333353874,
      "loss": 2.2705,
      "step": 12748
    },
    {
      "epoch": 0.08662112472764225,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000984585382640102,
      "loss": 2.3497,
      "step": 12749
    },
    {
      "epoch": 0.08662791907423631,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009845827317204593,
      "loss": 2.3144,
      "step": 12750
    },
    {
      "epoch": 0.08663471342083036,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00098458008057646,
      "loss": 2.3315,
      "step": 12751
    },
    {
      "epoch": 0.08664150776742442,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984577429208106,
      "loss": 2.2675,
      "step": 12752
    },
    {
      "epoch": 0.08664830211401847,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000984574777615398,
      "loss": 2.12,
      "step": 12753
    },
    {
      "epoch": 0.08665509646061254,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009845721257983376,
      "loss": 2.3144,
      "step": 12754
    },
    {
      "epoch": 0.0866618908072066,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009845694737569259,
      "loss": 2.1914,
      "step": 12755
    },
    {
      "epoch": 0.08666868515380065,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000984566821491164,
      "loss": 2.0911,
      "step": 12756
    },
    {
      "epoch": 0.08667547950039471,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009845641690010532,
      "loss": 2.121,
      "step": 12757
    },
    {
      "epoch": 0.08668227384698876,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009845615162865945,
      "loss": 2.0466,
      "step": 12758
    },
    {
      "epoch": 0.08668906819358282,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009845588633477897,
      "loss": 2.0459,
      "step": 12759
    },
    {
      "epoch": 0.08669586254017689,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009845562101846398,
      "loss": 2.3183,
      "step": 12760
    },
    {
      "epoch": 0.08670265688677094,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009845535567971457,
      "loss": 2.2645,
      "step": 12761
    },
    {
      "epoch": 0.086709451233365,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000984550903185309,
      "loss": 2.2074,
      "step": 12762
    },
    {
      "epoch": 0.08671624557995905,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009845482493491308,
      "loss": 2.3458,
      "step": 12763
    },
    {
      "epoch": 0.08672303992655311,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009845455952886119,
      "loss": 2.2553,
      "step": 12764
    },
    {
      "epoch": 0.08672983427314718,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009845429410037542,
      "loss": 2.2472,
      "step": 12765
    },
    {
      "epoch": 0.08673662861974123,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009845402864945589,
      "loss": 1.9542,
      "step": 12766
    },
    {
      "epoch": 0.08674342296633529,
      "grad_norm": 0.140625,
      "learning_rate": 0.000984537631761027,
      "loss": 2.2098,
      "step": 12767
    },
    {
      "epoch": 0.08675021731292934,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009845349768031594,
      "loss": 2.2972,
      "step": 12768
    },
    {
      "epoch": 0.0867570116595234,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000984532321620958,
      "loss": 2.2433,
      "step": 12769
    },
    {
      "epoch": 0.08676380600611747,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009845296662144235,
      "loss": 2.1211,
      "step": 12770
    },
    {
      "epoch": 0.08677060035271152,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009845270105835573,
      "loss": 2.313,
      "step": 12771
    },
    {
      "epoch": 0.08677739469930558,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009845243547283607,
      "loss": 2.4067,
      "step": 12772
    },
    {
      "epoch": 0.08678418904589963,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009845216986488349,
      "loss": 2.3542,
      "step": 12773
    },
    {
      "epoch": 0.08679098339249369,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000984519042344981,
      "loss": 2.3062,
      "step": 12774
    },
    {
      "epoch": 0.08679777773908774,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009845163858168003,
      "loss": 2.2015,
      "step": 12775
    },
    {
      "epoch": 0.0868045720856818,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009845137290642942,
      "loss": 2.1691,
      "step": 12776
    },
    {
      "epoch": 0.08681136643227587,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009845110720874638,
      "loss": 2.3027,
      "step": 12777
    },
    {
      "epoch": 0.08681816077886992,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009845084148863105,
      "loss": 2.1689,
      "step": 12778
    },
    {
      "epoch": 0.08682495512546398,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009845057574608348,
      "loss": 2.1091,
      "step": 12779
    },
    {
      "epoch": 0.08683174947205803,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000984503099811039,
      "loss": 2.2586,
      "step": 12780
    },
    {
      "epoch": 0.0868385438186521,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009845004419369236,
      "loss": 2.3039,
      "step": 12781
    },
    {
      "epoch": 0.08684533816524616,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009844977838384904,
      "loss": 2.2607,
      "step": 12782
    },
    {
      "epoch": 0.08685213251184021,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009844951255157399,
      "loss": 2.1555,
      "step": 12783
    },
    {
      "epoch": 0.08685892685843427,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009844924669686738,
      "loss": 2.1663,
      "step": 12784
    },
    {
      "epoch": 0.08686572120502832,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009844898081972932,
      "loss": 2.2285,
      "step": 12785
    },
    {
      "epoch": 0.08687251555162238,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009844871492015995,
      "loss": 2.1916,
      "step": 12786
    },
    {
      "epoch": 0.08687930989821645,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009844844899815936,
      "loss": 2.1812,
      "step": 12787
    },
    {
      "epoch": 0.0868861042448105,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009844818305372771,
      "loss": 2.2507,
      "step": 12788
    },
    {
      "epoch": 0.08689289859140456,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009844791708686512,
      "loss": 2.1473,
      "step": 12789
    },
    {
      "epoch": 0.08689969293799861,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009844765109757168,
      "loss": 2.0854,
      "step": 12790
    },
    {
      "epoch": 0.08690648728459267,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0009844738508584753,
      "loss": 2.2451,
      "step": 12791
    },
    {
      "epoch": 0.08691328163118674,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000984471190516928,
      "loss": 2.3055,
      "step": 12792
    },
    {
      "epoch": 0.08692007597778079,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009844685299510761,
      "loss": 2.2084,
      "step": 12793
    },
    {
      "epoch": 0.08692687032437485,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009844658691609209,
      "loss": 2.0963,
      "step": 12794
    },
    {
      "epoch": 0.0869336646709689,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009844632081464635,
      "loss": 2.1437,
      "step": 12795
    },
    {
      "epoch": 0.08694045901756296,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009844605469077053,
      "loss": 2.1239,
      "step": 12796
    },
    {
      "epoch": 0.08694725336415701,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009844578854446473,
      "loss": 2.1072,
      "step": 12797
    },
    {
      "epoch": 0.08695404771075108,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009844552237572909,
      "loss": 2.252,
      "step": 12798
    },
    {
      "epoch": 0.08696084205734514,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000984452561845637,
      "loss": 2.1812,
      "step": 12799
    },
    {
      "epoch": 0.08696763640393919,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009844498997096875,
      "loss": 2.2027,
      "step": 12800
    },
    {
      "epoch": 0.08697443075053325,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000984447237349443,
      "loss": 2.3015,
      "step": 12801
    },
    {
      "epoch": 0.0869812250971273,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000984444574764905,
      "loss": 2.1287,
      "step": 12802
    },
    {
      "epoch": 0.08698801944372136,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000984441911956075,
      "loss": 2.3704,
      "step": 12803
    },
    {
      "epoch": 0.08699481379031543,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009844392489229538,
      "loss": 2.2302,
      "step": 12804
    },
    {
      "epoch": 0.08700160813690948,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009844365856655427,
      "loss": 2.1822,
      "step": 12805
    },
    {
      "epoch": 0.08700840248350354,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000984433922183843,
      "loss": 2.2027,
      "step": 12806
    },
    {
      "epoch": 0.08701519683009759,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009844312584778561,
      "loss": 2.3136,
      "step": 12807
    },
    {
      "epoch": 0.08702199117669165,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009844285945475831,
      "loss": 2.061,
      "step": 12808
    },
    {
      "epoch": 0.08702878552328572,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000984425930393025,
      "loss": 2.0888,
      "step": 12809
    },
    {
      "epoch": 0.08703557986987977,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009844232660141834,
      "loss": 2.1628,
      "step": 12810
    },
    {
      "epoch": 0.08704237421647383,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009844206014110594,
      "loss": 2.2081,
      "step": 12811
    },
    {
      "epoch": 0.08704916856306788,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009844179365836542,
      "loss": 2.1201,
      "step": 12812
    },
    {
      "epoch": 0.08705596290966194,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009844152715319688,
      "loss": 2.3391,
      "step": 12813
    },
    {
      "epoch": 0.08706275725625601,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009844126062560049,
      "loss": 2.3024,
      "step": 12814
    },
    {
      "epoch": 0.08706955160285006,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009844099407557635,
      "loss": 2.3187,
      "step": 12815
    },
    {
      "epoch": 0.08707634594944412,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000984407275031246,
      "loss": 2.1905,
      "step": 12816
    },
    {
      "epoch": 0.08708314029603817,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009844046090824532,
      "loss": 2.1889,
      "step": 12817
    },
    {
      "epoch": 0.08708993464263223,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009844019429093866,
      "loss": 2.2539,
      "step": 12818
    },
    {
      "epoch": 0.0870967289892263,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009843992765120478,
      "loss": 2.097,
      "step": 12819
    },
    {
      "epoch": 0.08710352333582035,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009843966098904374,
      "loss": 2.2244,
      "step": 12820
    },
    {
      "epoch": 0.08711031768241441,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000984393943044557,
      "loss": 2.2966,
      "step": 12821
    },
    {
      "epoch": 0.08711711202900846,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009843912759744079,
      "loss": 2.0656,
      "step": 12822
    },
    {
      "epoch": 0.08712390637560252,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000984388608679991,
      "loss": 2.1454,
      "step": 12823
    },
    {
      "epoch": 0.08713070072219657,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000984385941161308,
      "loss": 2.1185,
      "step": 12824
    },
    {
      "epoch": 0.08713749506879064,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009843832734183595,
      "loss": 2.4364,
      "step": 12825
    },
    {
      "epoch": 0.0871442894153847,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009843806054511473,
      "loss": 2.214,
      "step": 12826
    },
    {
      "epoch": 0.08715108376197875,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009843779372596725,
      "loss": 2.2667,
      "step": 12827
    },
    {
      "epoch": 0.08715787810857281,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000984375268843936,
      "loss": 2.0837,
      "step": 12828
    },
    {
      "epoch": 0.08716467245516686,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009843726002039396,
      "loss": 2.1036,
      "step": 12829
    },
    {
      "epoch": 0.08717146680176092,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009843699313396839,
      "loss": 2.1506,
      "step": 12830
    },
    {
      "epoch": 0.08717826114835499,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009843672622511707,
      "loss": 2.2896,
      "step": 12831
    },
    {
      "epoch": 0.08718505549494904,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000984364592938401,
      "loss": 2.2869,
      "step": 12832
    },
    {
      "epoch": 0.0871918498415431,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009843619234013762,
      "loss": 2.1982,
      "step": 12833
    },
    {
      "epoch": 0.08719864418813715,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000984359253640097,
      "loss": 2.1723,
      "step": 12834
    },
    {
      "epoch": 0.08720543853473121,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009843565836545655,
      "loss": 2.2871,
      "step": 12835
    },
    {
      "epoch": 0.08721223288132528,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000984353913444782,
      "loss": 2.1181,
      "step": 12836
    },
    {
      "epoch": 0.08721902722791933,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009843512430107483,
      "loss": 2.154,
      "step": 12837
    },
    {
      "epoch": 0.08722582157451339,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009843485723524657,
      "loss": 2.1632,
      "step": 12838
    },
    {
      "epoch": 0.08723261592110744,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009843459014699351,
      "loss": 2.3683,
      "step": 12839
    },
    {
      "epoch": 0.0872394102677015,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009843432303631582,
      "loss": 2.2305,
      "step": 12840
    },
    {
      "epoch": 0.08724620461429557,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009843405590321355,
      "loss": 2.1657,
      "step": 12841
    },
    {
      "epoch": 0.08725299896088962,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984337887476869,
      "loss": 2.2159,
      "step": 12842
    },
    {
      "epoch": 0.08725979330748368,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009843352156973593,
      "loss": 2.1957,
      "step": 12843
    },
    {
      "epoch": 0.08726658765407773,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009843325436936082,
      "loss": 2.1597,
      "step": 12844
    },
    {
      "epoch": 0.08727338200067179,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009843298714656166,
      "loss": 2.2029,
      "step": 12845
    },
    {
      "epoch": 0.08728017634726584,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984327199013386,
      "loss": 2.3373,
      "step": 12846
    },
    {
      "epoch": 0.0872869706938599,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009843245263369172,
      "loss": 2.2719,
      "step": 12847
    },
    {
      "epoch": 0.08729376504045397,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000984321853436212,
      "loss": 2.1865,
      "step": 12848
    },
    {
      "epoch": 0.08730055938704802,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000984319180311271,
      "loss": 2.1862,
      "step": 12849
    },
    {
      "epoch": 0.08730735373364208,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009843165069620958,
      "loss": 2.233,
      "step": 12850
    },
    {
      "epoch": 0.08731414808023613,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009843138333886878,
      "loss": 2.4373,
      "step": 12851
    },
    {
      "epoch": 0.0873209424268302,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000984311159591048,
      "loss": 2.19,
      "step": 12852
    },
    {
      "epoch": 0.08732773677342426,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009843084855691777,
      "loss": 2.1644,
      "step": 12853
    },
    {
      "epoch": 0.08733453112001831,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009843058113230782,
      "loss": 2.2863,
      "step": 12854
    },
    {
      "epoch": 0.08734132546661237,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009843031368527504,
      "loss": 2.1387,
      "step": 12855
    },
    {
      "epoch": 0.08734811981320642,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009843004621581958,
      "loss": 2.2022,
      "step": 12856
    },
    {
      "epoch": 0.08735491415980048,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009842977872394159,
      "loss": 2.2268,
      "step": 12857
    },
    {
      "epoch": 0.08736170850639455,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009842951120964114,
      "loss": 2.2226,
      "step": 12858
    },
    {
      "epoch": 0.0873685028529886,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009842924367291842,
      "loss": 2.3835,
      "step": 12859
    },
    {
      "epoch": 0.08737529719958266,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009842897611377348,
      "loss": 2.2261,
      "step": 12860
    },
    {
      "epoch": 0.08738209154617671,
      "grad_norm": 0.125,
      "learning_rate": 0.0009842870853220649,
      "loss": 2.2801,
      "step": 12861
    },
    {
      "epoch": 0.08738888589277077,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009842844092821756,
      "loss": 2.3127,
      "step": 12862
    },
    {
      "epoch": 0.08739568023936484,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009842817330180682,
      "loss": 2.1953,
      "step": 12863
    },
    {
      "epoch": 0.08740247458595889,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000984279056529744,
      "loss": 2.1961,
      "step": 12864
    },
    {
      "epoch": 0.08740926893255295,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000984276379817204,
      "loss": 2.2547,
      "step": 12865
    },
    {
      "epoch": 0.087416063279147,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009842737028804497,
      "loss": 2.3022,
      "step": 12866
    },
    {
      "epoch": 0.08742285762574106,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009842710257194822,
      "loss": 2.1469,
      "step": 12867
    },
    {
      "epoch": 0.08742965197233511,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009842683483343028,
      "loss": 2.1908,
      "step": 12868
    },
    {
      "epoch": 0.08743644631892918,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009842656707249127,
      "loss": 2.312,
      "step": 12869
    },
    {
      "epoch": 0.08744324066552324,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000984262992891313,
      "loss": 2.2591,
      "step": 12870
    },
    {
      "epoch": 0.08745003501211729,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009842603148335052,
      "loss": 2.309,
      "step": 12871
    },
    {
      "epoch": 0.08745682935871135,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009842576365514904,
      "loss": 2.1819,
      "step": 12872
    },
    {
      "epoch": 0.0874636237053054,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.00098425495804527,
      "loss": 2.0313,
      "step": 12873
    },
    {
      "epoch": 0.08747041805189947,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000984252279314845,
      "loss": 2.197,
      "step": 12874
    },
    {
      "epoch": 0.08747721239849353,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009842496003602168,
      "loss": 2.2341,
      "step": 12875
    },
    {
      "epoch": 0.08748400674508758,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009842469211813864,
      "loss": 2.3093,
      "step": 12876
    },
    {
      "epoch": 0.08749080109168164,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009842442417783554,
      "loss": 2.2753,
      "step": 12877
    },
    {
      "epoch": 0.08749759543827569,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009842415621511248,
      "loss": 2.2398,
      "step": 12878
    },
    {
      "epoch": 0.08750438978486975,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009842388822996959,
      "loss": 2.1939,
      "step": 12879
    },
    {
      "epoch": 0.08751118413146382,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00098423620222407,
      "loss": 2.2872,
      "step": 12880
    },
    {
      "epoch": 0.08751797847805787,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009842335219242483,
      "loss": 2.2001,
      "step": 12881
    },
    {
      "epoch": 0.08752477282465193,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000984230841400232,
      "loss": 2.2506,
      "step": 12882
    },
    {
      "epoch": 0.08753156717124598,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009842281606520225,
      "loss": 2.28,
      "step": 12883
    },
    {
      "epoch": 0.08753836151784004,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009842254796796207,
      "loss": 2.2355,
      "step": 12884
    },
    {
      "epoch": 0.08754515586443411,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000984222798483028,
      "loss": 2.1605,
      "step": 12885
    },
    {
      "epoch": 0.08755195021102816,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000984220117062246,
      "loss": 2.2772,
      "step": 12886
    },
    {
      "epoch": 0.08755874455762222,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009842174354172755,
      "loss": 2.2542,
      "step": 12887
    },
    {
      "epoch": 0.08756553890421627,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009842147535481178,
      "loss": 2.2344,
      "step": 12888
    },
    {
      "epoch": 0.08757233325081033,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000984212071454774,
      "loss": 2.306,
      "step": 12889
    },
    {
      "epoch": 0.0875791275974044,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000984209389137246,
      "loss": 2.3455,
      "step": 12890
    },
    {
      "epoch": 0.08758592194399845,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009842067065955345,
      "loss": 2.3481,
      "step": 12891
    },
    {
      "epoch": 0.08759271629059251,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009842040238296408,
      "loss": 2.2173,
      "step": 12892
    },
    {
      "epoch": 0.08759951063718656,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009842013408395663,
      "loss": 2.2441,
      "step": 12893
    },
    {
      "epoch": 0.08760630498378062,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000984198657625312,
      "loss": 2.2564,
      "step": 12894
    },
    {
      "epoch": 0.08761309933037467,
      "grad_norm": 0.125,
      "learning_rate": 0.0009841959741868792,
      "loss": 2.1126,
      "step": 12895
    },
    {
      "epoch": 0.08761989367696874,
      "grad_norm": 0.140625,
      "learning_rate": 0.000984193290524269,
      "loss": 2.2184,
      "step": 12896
    },
    {
      "epoch": 0.0876266880235628,
      "grad_norm": 0.125,
      "learning_rate": 0.0009841906066374835,
      "loss": 2.2585,
      "step": 12897
    },
    {
      "epoch": 0.08763348237015685,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000984187922526523,
      "loss": 2.3451,
      "step": 12898
    },
    {
      "epoch": 0.08764027671675091,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000984185238191389,
      "loss": 2.2497,
      "step": 12899
    },
    {
      "epoch": 0.08764707106334496,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009841825536320826,
      "loss": 2.275,
      "step": 12900
    },
    {
      "epoch": 0.08765386540993902,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009841798688486054,
      "loss": 2.2153,
      "step": 12901
    },
    {
      "epoch": 0.08766065975653309,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009841771838409587,
      "loss": 2.2889,
      "step": 12902
    },
    {
      "epoch": 0.08766745410312714,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009841744986091431,
      "loss": 2.2734,
      "step": 12903
    },
    {
      "epoch": 0.0876742484497212,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009841718131531605,
      "loss": 2.1789,
      "step": 12904
    },
    {
      "epoch": 0.08768104279631525,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009841691274730118,
      "loss": 2.1505,
      "step": 12905
    },
    {
      "epoch": 0.08768783714290931,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009841664415686984,
      "loss": 2.1831,
      "step": 12906
    },
    {
      "epoch": 0.08769463148950338,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009841637554402216,
      "loss": 2.2206,
      "step": 12907
    },
    {
      "epoch": 0.08770142583609743,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009841610690875823,
      "loss": 2.198,
      "step": 12908
    },
    {
      "epoch": 0.08770822018269149,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009841583825107821,
      "loss": 2.3085,
      "step": 12909
    },
    {
      "epoch": 0.08771501452928554,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000984155695709822,
      "loss": 2.1706,
      "step": 12910
    },
    {
      "epoch": 0.0877218088758796,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009841530086847035,
      "loss": 2.2039,
      "step": 12911
    },
    {
      "epoch": 0.08772860322247367,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009841503214354278,
      "loss": 2.2552,
      "step": 12912
    },
    {
      "epoch": 0.08773539756906772,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009841476339619959,
      "loss": 2.3513,
      "step": 12913
    },
    {
      "epoch": 0.08774219191566178,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009841449462644092,
      "loss": 2.081,
      "step": 12914
    },
    {
      "epoch": 0.08774898626225583,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000984142258342669,
      "loss": 2.2758,
      "step": 12915
    },
    {
      "epoch": 0.08775578060884989,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009841395701967765,
      "loss": 2.2775,
      "step": 12916
    },
    {
      "epoch": 0.08776257495544394,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000984136881826733,
      "loss": 2.2952,
      "step": 12917
    },
    {
      "epoch": 0.087769369302038,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009841341932325394,
      "loss": 2.2206,
      "step": 12918
    },
    {
      "epoch": 0.08777616364863207,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009841315044141973,
      "loss": 2.139,
      "step": 12919
    },
    {
      "epoch": 0.08778295799522612,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000984128815371708,
      "loss": 2.2541,
      "step": 12920
    },
    {
      "epoch": 0.08778975234182018,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009841261261050727,
      "loss": 2.1247,
      "step": 12921
    },
    {
      "epoch": 0.08779654668841423,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009841234366142924,
      "loss": 2.0824,
      "step": 12922
    },
    {
      "epoch": 0.0878033410350083,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009841207468993685,
      "loss": 2.2177,
      "step": 12923
    },
    {
      "epoch": 0.08781013538160236,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009841180569603021,
      "loss": 2.2565,
      "step": 12924
    },
    {
      "epoch": 0.08781692972819641,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009841153667970947,
      "loss": 2.2254,
      "step": 12925
    },
    {
      "epoch": 0.08782372407479047,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009841126764097475,
      "loss": 2.1366,
      "step": 12926
    },
    {
      "epoch": 0.08783051842138452,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009841099857982618,
      "loss": 2.2194,
      "step": 12927
    },
    {
      "epoch": 0.08783731276797858,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009841072949626384,
      "loss": 2.1202,
      "step": 12928
    },
    {
      "epoch": 0.08784410711457265,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000984104603902879,
      "loss": 2.2951,
      "step": 12929
    },
    {
      "epoch": 0.0878509014611667,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.000984101912618985,
      "loss": 2.1518,
      "step": 12930
    },
    {
      "epoch": 0.08785769580776076,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000984099221110957,
      "loss": 2.2901,
      "step": 12931
    },
    {
      "epoch": 0.08786449015435481,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009840965293787967,
      "loss": 2.2702,
      "step": 12932
    },
    {
      "epoch": 0.08787128450094887,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009840938374225054,
      "loss": 2.2782,
      "step": 12933
    },
    {
      "epoch": 0.08787807884754294,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000984091145242084,
      "loss": 2.3734,
      "step": 12934
    },
    {
      "epoch": 0.08788487319413699,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009840884528375343,
      "loss": 2.2861,
      "step": 12935
    },
    {
      "epoch": 0.08789166754073105,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009840857602088568,
      "loss": 2.2137,
      "step": 12936
    },
    {
      "epoch": 0.0878984618873251,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009840830673560532,
      "loss": 2.1616,
      "step": 12937
    },
    {
      "epoch": 0.08790525623391916,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009840803742791249,
      "loss": 2.312,
      "step": 12938
    },
    {
      "epoch": 0.08791205058051323,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009840776809780728,
      "loss": 2.3423,
      "step": 12939
    },
    {
      "epoch": 0.08791884492710728,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009840749874528985,
      "loss": 2.368,
      "step": 12940
    },
    {
      "epoch": 0.08792563927370134,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009840722937036026,
      "loss": 2.1332,
      "step": 12941
    },
    {
      "epoch": 0.08793243362029539,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009840695997301871,
      "loss": 2.2556,
      "step": 12942
    },
    {
      "epoch": 0.08793922796688945,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009840669055326528,
      "loss": 2.3215,
      "step": 12943
    },
    {
      "epoch": 0.0879460223134835,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009840642111110013,
      "loss": 2.2296,
      "step": 12944
    },
    {
      "epoch": 0.08795281666007757,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009840615164652333,
      "loss": 2.158,
      "step": 12945
    },
    {
      "epoch": 0.08795961100667163,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009840588215953505,
      "loss": 2.0418,
      "step": 12946
    },
    {
      "epoch": 0.08796640535326568,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000984056126501354,
      "loss": 2.2593,
      "step": 12947
    },
    {
      "epoch": 0.08797319969985974,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000984053431183245,
      "loss": 2.3453,
      "step": 12948
    },
    {
      "epoch": 0.08797999404645379,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009840507356410249,
      "loss": 2.2209,
      "step": 12949
    },
    {
      "epoch": 0.08798678839304785,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009840480398746949,
      "loss": 2.2623,
      "step": 12950
    },
    {
      "epoch": 0.08799358273964192,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009840453438842562,
      "loss": 2.1541,
      "step": 12951
    },
    {
      "epoch": 0.08800037708623597,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00098404264766971,
      "loss": 2.1735,
      "step": 12952
    },
    {
      "epoch": 0.08800717143283003,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009840399512310576,
      "loss": 2.2676,
      "step": 12953
    },
    {
      "epoch": 0.08801396577942408,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009840372545683004,
      "loss": 2.2297,
      "step": 12954
    },
    {
      "epoch": 0.08802076012601814,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009840345576814392,
      "loss": 2.3724,
      "step": 12955
    },
    {
      "epoch": 0.08802755447261221,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009840318605704756,
      "loss": 2.5209,
      "step": 12956
    },
    {
      "epoch": 0.08803434881920626,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000984029163235411,
      "loss": 2.2692,
      "step": 12957
    },
    {
      "epoch": 0.08804114316580032,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009840264656762463,
      "loss": 2.1187,
      "step": 12958
    },
    {
      "epoch": 0.08804793751239437,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000984023767892983,
      "loss": 2.2098,
      "step": 12959
    },
    {
      "epoch": 0.08805473185898843,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009840210698856223,
      "loss": 2.0791,
      "step": 12960
    },
    {
      "epoch": 0.0880615262055825,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009840183716541653,
      "loss": 2.2781,
      "step": 12961
    },
    {
      "epoch": 0.08806832055217655,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009840156731986132,
      "loss": 2.2633,
      "step": 12962
    },
    {
      "epoch": 0.08807511489877061,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009840129745189675,
      "loss": 2.1859,
      "step": 12963
    },
    {
      "epoch": 0.08808190924536466,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009840102756152294,
      "loss": 2.1893,
      "step": 12964
    },
    {
      "epoch": 0.08808870359195872,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009840075764874002,
      "loss": 2.0992,
      "step": 12965
    },
    {
      "epoch": 0.08809549793855277,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000984004877135481,
      "loss": 2.1436,
      "step": 12966
    },
    {
      "epoch": 0.08810229228514684,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009840021775594726,
      "loss": 2.1837,
      "step": 12967
    },
    {
      "epoch": 0.0881090866317409,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009839994777593774,
      "loss": 2.1527,
      "step": 12968
    },
    {
      "epoch": 0.08811588097833495,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009839967777351957,
      "loss": 2.2663,
      "step": 12969
    },
    {
      "epoch": 0.08812267532492901,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983994077486929,
      "loss": 2.1366,
      "step": 12970
    },
    {
      "epoch": 0.08812946967152306,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009839913770145787,
      "loss": 2.1956,
      "step": 12971
    },
    {
      "epoch": 0.08813626401811712,
      "grad_norm": 0.158203125,
      "learning_rate": 0.000983988676318146,
      "loss": 2.42,
      "step": 12972
    },
    {
      "epoch": 0.08814305836471119,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009839859753976317,
      "loss": 2.3398,
      "step": 12973
    },
    {
      "epoch": 0.08814985271130524,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983983274253038,
      "loss": 2.2449,
      "step": 12974
    },
    {
      "epoch": 0.0881566470578993,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009839805728843651,
      "loss": 2.2449,
      "step": 12975
    },
    {
      "epoch": 0.08816344140449335,
      "grad_norm": 0.140625,
      "learning_rate": 0.000983977871291615,
      "loss": 2.1646,
      "step": 12976
    },
    {
      "epoch": 0.08817023575108741,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009839751694747888,
      "loss": 2.3795,
      "step": 12977
    },
    {
      "epoch": 0.08817703009768148,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009839724674338874,
      "loss": 2.0814,
      "step": 12978
    },
    {
      "epoch": 0.08818382444427553,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009839697651689123,
      "loss": 2.1588,
      "step": 12979
    },
    {
      "epoch": 0.08819061879086959,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009839670626798649,
      "loss": 2.2481,
      "step": 12980
    },
    {
      "epoch": 0.08819741313746364,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009839643599667462,
      "loss": 2.1337,
      "step": 12981
    },
    {
      "epoch": 0.0882042074840577,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009839616570295577,
      "loss": 2.073,
      "step": 12982
    },
    {
      "epoch": 0.08821100183065177,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009839589538683004,
      "loss": 2.3688,
      "step": 12983
    },
    {
      "epoch": 0.08821779617724582,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009839562504829755,
      "loss": 2.1325,
      "step": 12984
    },
    {
      "epoch": 0.08822459052383988,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009839535468735847,
      "loss": 2.342,
      "step": 12985
    },
    {
      "epoch": 0.08823138487043393,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009839508430401287,
      "loss": 2.2343,
      "step": 12986
    },
    {
      "epoch": 0.08823817921702799,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009839481389826092,
      "loss": 2.3032,
      "step": 12987
    },
    {
      "epoch": 0.08824497356362204,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000983945434701027,
      "loss": 2.3475,
      "step": 12988
    },
    {
      "epoch": 0.0882517679102161,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000983942730195384,
      "loss": 2.2646,
      "step": 12989
    },
    {
      "epoch": 0.08825856225681017,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009839400254656808,
      "loss": 2.3194,
      "step": 12990
    },
    {
      "epoch": 0.08826535660340422,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983937320511919,
      "loss": 2.1261,
      "step": 12991
    },
    {
      "epoch": 0.08827215094999828,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009839346153340998,
      "loss": 2.0845,
      "step": 12992
    },
    {
      "epoch": 0.08827894529659233,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009839319099322244,
      "loss": 2.1202,
      "step": 12993
    },
    {
      "epoch": 0.0882857396431864,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000983929204306294,
      "loss": 2.3282,
      "step": 12994
    },
    {
      "epoch": 0.08829253398978046,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009839264984563098,
      "loss": 2.2197,
      "step": 12995
    },
    {
      "epoch": 0.08829932833637451,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009839237923822736,
      "loss": 2.3111,
      "step": 12996
    },
    {
      "epoch": 0.08830612268296857,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009839210860841858,
      "loss": 2.2373,
      "step": 12997
    },
    {
      "epoch": 0.08831291702956262,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009839183795620483,
      "loss": 2.3405,
      "step": 12998
    },
    {
      "epoch": 0.08831971137615668,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000983915672815862,
      "loss": 2.3236,
      "step": 12999
    },
    {
      "epoch": 0.08832650572275075,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009839129658456285,
      "loss": 2.2011,
      "step": 13000
    },
    {
      "epoch": 0.0883333000693448,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009839102586513486,
      "loss": 2.1815,
      "step": 13001
    },
    {
      "epoch": 0.08834009441593886,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000983907551233024,
      "loss": 2.3247,
      "step": 13002
    },
    {
      "epoch": 0.08834688876253291,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009839048435906557,
      "loss": 2.2024,
      "step": 13003
    },
    {
      "epoch": 0.08835368310912697,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009839021357242448,
      "loss": 2.3316,
      "step": 13004
    },
    {
      "epoch": 0.08836047745572104,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000983899427633793,
      "loss": 2.4069,
      "step": 13005
    },
    {
      "epoch": 0.08836727180231509,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009838967193193013,
      "loss": 2.1313,
      "step": 13006
    },
    {
      "epoch": 0.08837406614890915,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000983894010780771,
      "loss": 2.1263,
      "step": 13007
    },
    {
      "epoch": 0.0883808604955032,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009838913020182032,
      "loss": 2.2469,
      "step": 13008
    },
    {
      "epoch": 0.08838765484209726,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009838885930315993,
      "loss": 2.1609,
      "step": 13009
    },
    {
      "epoch": 0.08839444918869133,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009838858838209605,
      "loss": 2.2271,
      "step": 13010
    },
    {
      "epoch": 0.08840124353528538,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009838831743862882,
      "loss": 2.4203,
      "step": 13011
    },
    {
      "epoch": 0.08840803788187944,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009838804647275834,
      "loss": 2.171,
      "step": 13012
    },
    {
      "epoch": 0.08841483222847349,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009838777548448476,
      "loss": 2.0616,
      "step": 13013
    },
    {
      "epoch": 0.08842162657506755,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009838750447380819,
      "loss": 2.208,
      "step": 13014
    },
    {
      "epoch": 0.0884284209216616,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009838723344072877,
      "loss": 2.2465,
      "step": 13015
    },
    {
      "epoch": 0.08843521526825567,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000983869623852466,
      "loss": 2.0256,
      "step": 13016
    },
    {
      "epoch": 0.08844200961484973,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009838669130736184,
      "loss": 2.2819,
      "step": 13017
    },
    {
      "epoch": 0.08844880396144378,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009838642020707459,
      "loss": 2.2033,
      "step": 13018
    },
    {
      "epoch": 0.08845559830803784,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009838614908438497,
      "loss": 2.349,
      "step": 13019
    },
    {
      "epoch": 0.08846239265463189,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000983858779392931,
      "loss": 2.2079,
      "step": 13020
    },
    {
      "epoch": 0.08846918700122595,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009838560677179916,
      "loss": 2.197,
      "step": 13021
    },
    {
      "epoch": 0.08847598134782002,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009838533558190323,
      "loss": 2.1357,
      "step": 13022
    },
    {
      "epoch": 0.08848277569441407,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009838506436960544,
      "loss": 2.2921,
      "step": 13023
    },
    {
      "epoch": 0.08848957004100813,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009838479313490592,
      "loss": 2.266,
      "step": 13024
    },
    {
      "epoch": 0.08849636438760218,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983845218778048,
      "loss": 2.278,
      "step": 13025
    },
    {
      "epoch": 0.08850315873419624,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000983842505983022,
      "loss": 2.1743,
      "step": 13026
    },
    {
      "epoch": 0.08850995308079031,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009838397929639825,
      "loss": 2.1411,
      "step": 13027
    },
    {
      "epoch": 0.08851674742738436,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009838370797209306,
      "loss": 2.3006,
      "step": 13028
    },
    {
      "epoch": 0.08852354177397842,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009838343662538676,
      "loss": 2.2125,
      "step": 13029
    },
    {
      "epoch": 0.08853033612057247,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009838316525627951,
      "loss": 1.9589,
      "step": 13030
    },
    {
      "epoch": 0.08853713046716653,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983828938647714,
      "loss": 2.3553,
      "step": 13031
    },
    {
      "epoch": 0.0885439248137606,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009838262245086256,
      "loss": 2.1632,
      "step": 13032
    },
    {
      "epoch": 0.08855071916035465,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000983823510145531,
      "loss": 2.3116,
      "step": 13033
    },
    {
      "epoch": 0.08855751350694871,
      "grad_norm": 0.140625,
      "learning_rate": 0.000983820795558432,
      "loss": 2.2258,
      "step": 13034
    },
    {
      "epoch": 0.08856430785354276,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009838180807473294,
      "loss": 2.3519,
      "step": 13035
    },
    {
      "epoch": 0.08857110220013682,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009838153657122247,
      "loss": 2.1032,
      "step": 13036
    },
    {
      "epoch": 0.08857789654673087,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009838126504531188,
      "loss": 2.0934,
      "step": 13037
    },
    {
      "epoch": 0.08858469089332494,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009838099349700133,
      "loss": 2.2127,
      "step": 13038
    },
    {
      "epoch": 0.088591485239919,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009838072192629092,
      "loss": 2.178,
      "step": 13039
    },
    {
      "epoch": 0.08859827958651305,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000983804503331808,
      "loss": 2.204,
      "step": 13040
    },
    {
      "epoch": 0.08860507393310711,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000983801787176711,
      "loss": 2.1563,
      "step": 13041
    },
    {
      "epoch": 0.08861186827970116,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009837990707976193,
      "loss": 2.0921,
      "step": 13042
    },
    {
      "epoch": 0.08861866262629522,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000983796354194534,
      "loss": 2.1575,
      "step": 13043
    },
    {
      "epoch": 0.08862545697288929,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009837936373674567,
      "loss": 2.3371,
      "step": 13044
    },
    {
      "epoch": 0.08863225131948334,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009837909203163884,
      "loss": 2.2081,
      "step": 13045
    },
    {
      "epoch": 0.0886390456660774,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009837882030413305,
      "loss": 2.1698,
      "step": 13046
    },
    {
      "epoch": 0.08864584001267145,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0009837854855422844,
      "loss": 2.0872,
      "step": 13047
    },
    {
      "epoch": 0.08865263435926551,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009837827678192508,
      "loss": 2.3143,
      "step": 13048
    },
    {
      "epoch": 0.08865942870585958,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009837800498722314,
      "loss": 2.0681,
      "step": 13049
    },
    {
      "epoch": 0.08866622305245363,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009837773317012276,
      "loss": 2.1471,
      "step": 13050
    },
    {
      "epoch": 0.08867301739904769,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009837746133062402,
      "loss": 2.2748,
      "step": 13051
    },
    {
      "epoch": 0.08867981174564174,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000983771894687271,
      "loss": 2.2038,
      "step": 13052
    },
    {
      "epoch": 0.0886866060922358,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009837691758443207,
      "loss": 2.2867,
      "step": 13053
    },
    {
      "epoch": 0.08869340043882987,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009837664567773907,
      "loss": 2.1856,
      "step": 13054
    },
    {
      "epoch": 0.08870019478542392,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009837637374864828,
      "loss": 2.1665,
      "step": 13055
    },
    {
      "epoch": 0.08870698913201798,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009837610179715974,
      "loss": 2.2061,
      "step": 13056
    },
    {
      "epoch": 0.08871378347861203,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009837582982327364,
      "loss": 2.2941,
      "step": 13057
    },
    {
      "epoch": 0.08872057782520609,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000983755578269901,
      "loss": 2.191,
      "step": 13058
    },
    {
      "epoch": 0.08872737217180016,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000983752858083092,
      "loss": 2.1891,
      "step": 13059
    },
    {
      "epoch": 0.0887341665183942,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009837501376723113,
      "loss": 2.3532,
      "step": 13060
    },
    {
      "epoch": 0.08874096086498827,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009837474170375595,
      "loss": 2.1642,
      "step": 13061
    },
    {
      "epoch": 0.08874775521158232,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009837446961788383,
      "loss": 2.3103,
      "step": 13062
    },
    {
      "epoch": 0.08875454955817638,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000983741975096149,
      "loss": 2.2354,
      "step": 13063
    },
    {
      "epoch": 0.08876134390477043,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009837392537894926,
      "loss": 2.2307,
      "step": 13064
    },
    {
      "epoch": 0.0887681382513645,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009837365322588705,
      "loss": 2.1378,
      "step": 13065
    },
    {
      "epoch": 0.08877493259795856,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000983733810504284,
      "loss": 2.2498,
      "step": 13066
    },
    {
      "epoch": 0.08878172694455261,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009837310885257341,
      "loss": 2.1551,
      "step": 13067
    },
    {
      "epoch": 0.08878852129114667,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009837283663232223,
      "loss": 2.26,
      "step": 13068
    },
    {
      "epoch": 0.08879531563774072,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00098372564389675,
      "loss": 2.0708,
      "step": 13069
    },
    {
      "epoch": 0.08880210998433478,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009837229212463182,
      "loss": 2.1794,
      "step": 13070
    },
    {
      "epoch": 0.08880890433092885,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009837201983719283,
      "loss": 2.2809,
      "step": 13071
    },
    {
      "epoch": 0.0888156986775229,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009837174752735812,
      "loss": 2.1484,
      "step": 13072
    },
    {
      "epoch": 0.08882249302411696,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009837147519512785,
      "loss": 2.32,
      "step": 13073
    },
    {
      "epoch": 0.08882928737071101,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009837120284050217,
      "loss": 2.2777,
      "step": 13074
    },
    {
      "epoch": 0.08883608171730507,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009837093046348117,
      "loss": 2.1663,
      "step": 13075
    },
    {
      "epoch": 0.08884287606389914,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009837065806406497,
      "loss": 2.1829,
      "step": 13076
    },
    {
      "epoch": 0.08884967041049319,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000983703856422537,
      "loss": 2.1161,
      "step": 13077
    },
    {
      "epoch": 0.08885646475708725,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009837011319804752,
      "loss": 2.21,
      "step": 13078
    },
    {
      "epoch": 0.0888632591036813,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009836984073144652,
      "loss": 2.1667,
      "step": 13079
    },
    {
      "epoch": 0.08887005345027536,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009836956824245085,
      "loss": 2.1786,
      "step": 13080
    },
    {
      "epoch": 0.08887684779686943,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009836929573106062,
      "loss": 2.2433,
      "step": 13081
    },
    {
      "epoch": 0.08888364214346348,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009836902319727595,
      "loss": 2.1454,
      "step": 13082
    },
    {
      "epoch": 0.08889043649005754,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009836875064109698,
      "loss": 2.3291,
      "step": 13083
    },
    {
      "epoch": 0.08889723083665159,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009836847806252386,
      "loss": 2.016,
      "step": 13084
    },
    {
      "epoch": 0.08890402518324565,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009836820546155665,
      "loss": 2.2233,
      "step": 13085
    },
    {
      "epoch": 0.0889108195298397,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009836793283819552,
      "loss": 2.1541,
      "step": 13086
    },
    {
      "epoch": 0.08891761387643377,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000983676601924406,
      "loss": 2.1713,
      "step": 13087
    },
    {
      "epoch": 0.08892440822302783,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009836738752429203,
      "loss": 2.113,
      "step": 13088
    },
    {
      "epoch": 0.08893120256962188,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983671148337499,
      "loss": 2.0882,
      "step": 13089
    },
    {
      "epoch": 0.08893799691621594,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009836684212081433,
      "loss": 2.0914,
      "step": 13090
    },
    {
      "epoch": 0.08894479126280999,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009836656938548548,
      "loss": 2.2037,
      "step": 13091
    },
    {
      "epoch": 0.08895158560940405,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009836629662776347,
      "loss": 2.2573,
      "step": 13092
    },
    {
      "epoch": 0.08895837995599812,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009836602384764841,
      "loss": 2.2315,
      "step": 13093
    },
    {
      "epoch": 0.08896517430259217,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009836575104514044,
      "loss": 2.2152,
      "step": 13094
    },
    {
      "epoch": 0.08897196864918623,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009836547822023969,
      "loss": 2.2691,
      "step": 13095
    },
    {
      "epoch": 0.08897876299578028,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009836520537294627,
      "loss": 2.0897,
      "step": 13096
    },
    {
      "epoch": 0.08898555734237434,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009836493250326031,
      "loss": 2.3047,
      "step": 13097
    },
    {
      "epoch": 0.08899235168896841,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009836465961118193,
      "loss": 2.2504,
      "step": 13098
    },
    {
      "epoch": 0.08899914603556246,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009836438669671129,
      "loss": 2.2899,
      "step": 13099
    },
    {
      "epoch": 0.08900594038215652,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009836411375984848,
      "loss": 2.2209,
      "step": 13100
    },
    {
      "epoch": 0.08901273472875057,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009836384080059365,
      "loss": 2.2351,
      "step": 13101
    },
    {
      "epoch": 0.08901952907534463,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983635678189469,
      "loss": 2.3055,
      "step": 13102
    },
    {
      "epoch": 0.0890263234219387,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000983632948149084,
      "loss": 2.1597,
      "step": 13103
    },
    {
      "epoch": 0.08903311776853275,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009836302178847821,
      "loss": 2.148,
      "step": 13104
    },
    {
      "epoch": 0.08903991211512681,
      "grad_norm": 0.125,
      "learning_rate": 0.0009836274873965653,
      "loss": 2.18,
      "step": 13105
    },
    {
      "epoch": 0.08904670646172086,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009836247566844345,
      "loss": 2.278,
      "step": 13106
    },
    {
      "epoch": 0.08905350080831492,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009836220257483908,
      "loss": 2.1606,
      "step": 13107
    },
    {
      "epoch": 0.08906029515490897,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009836192945884357,
      "loss": 2.2524,
      "step": 13108
    },
    {
      "epoch": 0.08906708950150304,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009836165632045705,
      "loss": 2.1212,
      "step": 13109
    },
    {
      "epoch": 0.0890738838480971,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009836138315967964,
      "loss": 2.3035,
      "step": 13110
    },
    {
      "epoch": 0.08908067819469115,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009836110997651145,
      "loss": 2.0787,
      "step": 13111
    },
    {
      "epoch": 0.08908747254128521,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009836083677095262,
      "loss": 2.1701,
      "step": 13112
    },
    {
      "epoch": 0.08909426688787926,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009836056354300328,
      "loss": 2.2121,
      "step": 13113
    },
    {
      "epoch": 0.08910106123447332,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009836029029266357,
      "loss": 2.1313,
      "step": 13114
    },
    {
      "epoch": 0.08910785558106739,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009836001701993358,
      "loss": 2.2788,
      "step": 13115
    },
    {
      "epoch": 0.08911464992766144,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009835974372481347,
      "loss": 2.3006,
      "step": 13116
    },
    {
      "epoch": 0.0891214442742555,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009835947040730333,
      "loss": 2.2577,
      "step": 13117
    },
    {
      "epoch": 0.08912823862084955,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009835919706740334,
      "loss": 2.236,
      "step": 13118
    },
    {
      "epoch": 0.08913503296744361,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009835892370511357,
      "loss": 2.1819,
      "step": 13119
    },
    {
      "epoch": 0.08914182731403768,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009835865032043417,
      "loss": 2.3099,
      "step": 13120
    },
    {
      "epoch": 0.08914862166063173,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983583769133653,
      "loss": 2.1618,
      "step": 13121
    },
    {
      "epoch": 0.08915541600722579,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009835810348390703,
      "loss": 2.1127,
      "step": 13122
    },
    {
      "epoch": 0.08916221035381984,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009835783003205954,
      "loss": 2.2884,
      "step": 13123
    },
    {
      "epoch": 0.0891690047004139,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000983575565578229,
      "loss": 2.2973,
      "step": 13124
    },
    {
      "epoch": 0.08917579904700797,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009835728306119727,
      "loss": 2.211,
      "step": 13125
    },
    {
      "epoch": 0.08918259339360202,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009835700954218278,
      "loss": 2.2095,
      "step": 13126
    },
    {
      "epoch": 0.08918938774019608,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009835673600077956,
      "loss": 2.2172,
      "step": 13127
    },
    {
      "epoch": 0.08919618208679013,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009835646243698771,
      "loss": 2.2668,
      "step": 13128
    },
    {
      "epoch": 0.08920297643338419,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009835618885080737,
      "loss": 2.1818,
      "step": 13129
    },
    {
      "epoch": 0.08920977077997826,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009835591524223867,
      "loss": 2.1204,
      "step": 13130
    },
    {
      "epoch": 0.0892165651265723,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009835564161128176,
      "loss": 2.213,
      "step": 13131
    },
    {
      "epoch": 0.08922335947316637,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009835536795793672,
      "loss": 2.2389,
      "step": 13132
    },
    {
      "epoch": 0.08923015381976042,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009835509428220372,
      "loss": 2.2029,
      "step": 13133
    },
    {
      "epoch": 0.08923694816635448,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009835482058408282,
      "loss": 2.2523,
      "step": 13134
    },
    {
      "epoch": 0.08924374251294853,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009835454686357421,
      "loss": 2.2722,
      "step": 13135
    },
    {
      "epoch": 0.0892505368595426,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009835427312067802,
      "loss": 2.26,
      "step": 13136
    },
    {
      "epoch": 0.08925733120613666,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009835399935539436,
      "loss": 2.1789,
      "step": 13137
    },
    {
      "epoch": 0.08926412555273071,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009835372556772333,
      "loss": 2.1724,
      "step": 13138
    },
    {
      "epoch": 0.08927091989932477,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009835345175766507,
      "loss": 2.2827,
      "step": 13139
    },
    {
      "epoch": 0.08927771424591882,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009835317792521974,
      "loss": 2.0657,
      "step": 13140
    },
    {
      "epoch": 0.08928450859251288,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009835290407038744,
      "loss": 2.2616,
      "step": 13141
    },
    {
      "epoch": 0.08929130293910695,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000983526301931683,
      "loss": 2.2617,
      "step": 13142
    },
    {
      "epoch": 0.089298097285701,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009835235629356246,
      "loss": 2.4039,
      "step": 13143
    },
    {
      "epoch": 0.08930489163229506,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009835208237157001,
      "loss": 2.1819,
      "step": 13144
    },
    {
      "epoch": 0.08931168597888911,
      "grad_norm": 0.140625,
      "learning_rate": 0.000983518084271911,
      "loss": 2.1531,
      "step": 13145
    },
    {
      "epoch": 0.08931848032548317,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009835153446042587,
      "loss": 2.3589,
      "step": 13146
    },
    {
      "epoch": 0.08932527467207724,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009835126047127443,
      "loss": 2.3061,
      "step": 13147
    },
    {
      "epoch": 0.08933206901867129,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000983509864597369,
      "loss": 2.2357,
      "step": 13148
    },
    {
      "epoch": 0.08933886336526535,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009835071242581342,
      "loss": 2.0933,
      "step": 13149
    },
    {
      "epoch": 0.0893456577118594,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009835043836950413,
      "loss": 2.2391,
      "step": 13150
    },
    {
      "epoch": 0.08935245205845346,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009835016429080913,
      "loss": 2.3301,
      "step": 13151
    },
    {
      "epoch": 0.08935924640504753,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009834989018972856,
      "loss": 2.0851,
      "step": 13152
    },
    {
      "epoch": 0.08936604075164158,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009834961606626253,
      "loss": 2.1896,
      "step": 13153
    },
    {
      "epoch": 0.08937283509823564,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000983493419204112,
      "loss": 2.1111,
      "step": 13154
    },
    {
      "epoch": 0.08937962944482969,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009834906775217467,
      "loss": 2.2706,
      "step": 13155
    },
    {
      "epoch": 0.08938642379142375,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000983487935615531,
      "loss": 2.1793,
      "step": 13156
    },
    {
      "epoch": 0.0893932181380178,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009834851934854656,
      "loss": 2.2156,
      "step": 13157
    },
    {
      "epoch": 0.08940001248461187,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009834824511315524,
      "loss": 2.2938,
      "step": 13158
    },
    {
      "epoch": 0.08940680683120593,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000983479708553792,
      "loss": 2.2726,
      "step": 13159
    },
    {
      "epoch": 0.08941360117779998,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009834769657521862,
      "loss": 2.2957,
      "step": 13160
    },
    {
      "epoch": 0.08942039552439404,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009834742227267362,
      "loss": 2.2317,
      "step": 13161
    },
    {
      "epoch": 0.08942718987098809,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009834714794774433,
      "loss": 2.3044,
      "step": 13162
    },
    {
      "epoch": 0.08943398421758215,
      "grad_norm": 0.125,
      "learning_rate": 0.0009834687360043084,
      "loss": 2.0152,
      "step": 13163
    },
    {
      "epoch": 0.08944077856417622,
      "grad_norm": 0.125,
      "learning_rate": 0.000983465992307333,
      "loss": 2.2868,
      "step": 13164
    },
    {
      "epoch": 0.08944757291077027,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009834632483865186,
      "loss": 2.457,
      "step": 13165
    },
    {
      "epoch": 0.08945436725736433,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009834605042418662,
      "loss": 2.1849,
      "step": 13166
    },
    {
      "epoch": 0.08946116160395838,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983457759873377,
      "loss": 2.4493,
      "step": 13167
    },
    {
      "epoch": 0.08946795595055244,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009834550152810525,
      "loss": 2.3467,
      "step": 13168
    },
    {
      "epoch": 0.08947475029714651,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009834522704648939,
      "loss": 2.0753,
      "step": 13169
    },
    {
      "epoch": 0.08948154464374056,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009834495254249023,
      "loss": 2.3745,
      "step": 13170
    },
    {
      "epoch": 0.08948833899033462,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009834467801610792,
      "loss": 2.2493,
      "step": 13171
    },
    {
      "epoch": 0.08949513333692867,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009834440346734258,
      "loss": 2.3246,
      "step": 13172
    },
    {
      "epoch": 0.08950192768352273,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009834412889619434,
      "loss": 2.1342,
      "step": 13173
    },
    {
      "epoch": 0.0895087220301168,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000983438543026633,
      "loss": 2.294,
      "step": 13174
    },
    {
      "epoch": 0.08951551637671085,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009834357968674962,
      "loss": 2.2219,
      "step": 13175
    },
    {
      "epoch": 0.08952231072330491,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009834330504845344,
      "loss": 2.1831,
      "step": 13176
    },
    {
      "epoch": 0.08952910506989896,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009834303038777485,
      "loss": 2.1635,
      "step": 13177
    },
    {
      "epoch": 0.08953589941649302,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009834275570471398,
      "loss": 2.0479,
      "step": 13178
    },
    {
      "epoch": 0.08954269376308709,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009834248099927097,
      "loss": 2.1786,
      "step": 13179
    },
    {
      "epoch": 0.08954948810968114,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009834220627144596,
      "loss": 2.2217,
      "step": 13180
    },
    {
      "epoch": 0.0895562824562752,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009834193152123904,
      "loss": 2.2639,
      "step": 13181
    },
    {
      "epoch": 0.08956307680286925,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000983416567486504,
      "loss": 2.3572,
      "step": 13182
    },
    {
      "epoch": 0.08956987114946331,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009834138195368009,
      "loss": 2.3747,
      "step": 13183
    },
    {
      "epoch": 0.08957666549605736,
      "grad_norm": 0.140625,
      "learning_rate": 0.000983411071363283,
      "loss": 2.2196,
      "step": 13184
    },
    {
      "epoch": 0.08958345984265143,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009834083229659511,
      "loss": 2.2086,
      "step": 13185
    },
    {
      "epoch": 0.08959025418924549,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009834055743448067,
      "loss": 2.147,
      "step": 13186
    },
    {
      "epoch": 0.08959704853583954,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009834028254998512,
      "loss": 2.3116,
      "step": 13187
    },
    {
      "epoch": 0.0896038428824336,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009834000764310856,
      "loss": 2.2291,
      "step": 13188
    },
    {
      "epoch": 0.08961063722902765,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009833973271385114,
      "loss": 2.2203,
      "step": 13189
    },
    {
      "epoch": 0.08961743157562171,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009833945776221297,
      "loss": 2.2927,
      "step": 13190
    },
    {
      "epoch": 0.08962422592221578,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000983391827881942,
      "loss": 2.184,
      "step": 13191
    },
    {
      "epoch": 0.08963102026880983,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009833890779179493,
      "loss": 2.1551,
      "step": 13192
    },
    {
      "epoch": 0.08963781461540389,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009833863277301529,
      "loss": 2.3068,
      "step": 13193
    },
    {
      "epoch": 0.08964460896199794,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009833835773185545,
      "loss": 2.4167,
      "step": 13194
    },
    {
      "epoch": 0.089651403308592,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009833808266831547,
      "loss": 2.2114,
      "step": 13195
    },
    {
      "epoch": 0.08965819765518607,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009833780758239552,
      "loss": 2.27,
      "step": 13196
    },
    {
      "epoch": 0.08966499200178012,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009833753247409573,
      "loss": 2.2141,
      "step": 13197
    },
    {
      "epoch": 0.08967178634837418,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000983372573434162,
      "loss": 2.1766,
      "step": 13198
    },
    {
      "epoch": 0.08967858069496823,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983369821903571,
      "loss": 2.1722,
      "step": 13199
    },
    {
      "epoch": 0.08968537504156229,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000983367070149185,
      "loss": 2.1246,
      "step": 13200
    },
    {
      "epoch": 0.08969216938815636,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009833643181710058,
      "loss": 2.2376,
      "step": 13201
    },
    {
      "epoch": 0.0896989637347504,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009833615659690345,
      "loss": 2.3199,
      "step": 13202
    },
    {
      "epoch": 0.08970575808134447,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009833588135432722,
      "loss": 2.1585,
      "step": 13203
    },
    {
      "epoch": 0.08971255242793852,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009833560608937205,
      "loss": 2.2099,
      "step": 13204
    },
    {
      "epoch": 0.08971934677453258,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009833533080203802,
      "loss": 2.1358,
      "step": 13205
    },
    {
      "epoch": 0.08972614112112663,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000983350554923253,
      "loss": 2.2436,
      "step": 13206
    },
    {
      "epoch": 0.0897329354677207,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009833478016023401,
      "loss": 2.362,
      "step": 13207
    },
    {
      "epoch": 0.08973972981431476,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009833450480576428,
      "loss": 2.1076,
      "step": 13208
    },
    {
      "epoch": 0.08974652416090881,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009833422942891622,
      "loss": 2.2347,
      "step": 13209
    },
    {
      "epoch": 0.08975331850750287,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009833395402968994,
      "loss": 2.2743,
      "step": 13210
    },
    {
      "epoch": 0.08976011285409692,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009833367860808562,
      "loss": 2.3464,
      "step": 13211
    },
    {
      "epoch": 0.08976690720069098,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009833340316410335,
      "loss": 2.2266,
      "step": 13212
    },
    {
      "epoch": 0.08977370154728505,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009833312769774328,
      "loss": 2.0852,
      "step": 13213
    },
    {
      "epoch": 0.0897804958938791,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009833285220900551,
      "loss": 2.2677,
      "step": 13214
    },
    {
      "epoch": 0.08978729024047316,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000983325766978902,
      "loss": 2.0323,
      "step": 13215
    },
    {
      "epoch": 0.08979408458706721,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009833230116439746,
      "loss": 2.1938,
      "step": 13216
    },
    {
      "epoch": 0.08980087893366127,
      "grad_norm": 0.125,
      "learning_rate": 0.000983320256085274,
      "loss": 2.1555,
      "step": 13217
    },
    {
      "epoch": 0.08980767328025534,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009833175003028018,
      "loss": 2.136,
      "step": 13218
    },
    {
      "epoch": 0.08981446762684939,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009833147442965592,
      "loss": 2.1862,
      "step": 13219
    },
    {
      "epoch": 0.08982126197344345,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009833119880665474,
      "loss": 2.1791,
      "step": 13220
    },
    {
      "epoch": 0.0898280563200375,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009833092316127675,
      "loss": 2.2093,
      "step": 13221
    },
    {
      "epoch": 0.08983485066663156,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009833064749352211,
      "loss": 2.1806,
      "step": 13222
    },
    {
      "epoch": 0.08984164501322563,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009833037180339093,
      "loss": 2.1308,
      "step": 13223
    },
    {
      "epoch": 0.08984843935981968,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009833009609088335,
      "loss": 2.2566,
      "step": 13224
    },
    {
      "epoch": 0.08985523370641374,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009832982035599949,
      "loss": 2.1357,
      "step": 13225
    },
    {
      "epoch": 0.08986202805300779,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009832954459873947,
      "loss": 2.226,
      "step": 13226
    },
    {
      "epoch": 0.08986882239960185,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009832926881910343,
      "loss": 2.2375,
      "step": 13227
    },
    {
      "epoch": 0.0898756167461959,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983289930170915,
      "loss": 2.119,
      "step": 13228
    },
    {
      "epoch": 0.08988241109278997,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009832871719270377,
      "loss": 2.2423,
      "step": 13229
    },
    {
      "epoch": 0.08988920543938403,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009832844134594044,
      "loss": 2.3939,
      "step": 13230
    },
    {
      "epoch": 0.08989599978597808,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009832816547680155,
      "loss": 2.3729,
      "step": 13231
    },
    {
      "epoch": 0.08990279413257214,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009832788958528732,
      "loss": 2.2506,
      "step": 13232
    },
    {
      "epoch": 0.08990958847916619,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983276136713978,
      "loss": 2.1386,
      "step": 13233
    },
    {
      "epoch": 0.08991638282576025,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009832733773513316,
      "loss": 2.2034,
      "step": 13234
    },
    {
      "epoch": 0.08992317717235432,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009832706177649352,
      "loss": 2.2871,
      "step": 13235
    },
    {
      "epoch": 0.08992997151894837,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009832678579547898,
      "loss": 2.2448,
      "step": 13236
    },
    {
      "epoch": 0.08993676586554243,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009832650979208972,
      "loss": 2.2178,
      "step": 13237
    },
    {
      "epoch": 0.08994356021213648,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009832623376632584,
      "loss": 2.2065,
      "step": 13238
    },
    {
      "epoch": 0.08995035455873054,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009832595771818743,
      "loss": 2.2208,
      "step": 13239
    },
    {
      "epoch": 0.08995714890532461,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000983256816476747,
      "loss": 2.3332,
      "step": 13240
    },
    {
      "epoch": 0.08996394325191866,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000983254055547877,
      "loss": 2.357,
      "step": 13241
    },
    {
      "epoch": 0.08997073759851272,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009832512943952663,
      "loss": 2.1937,
      "step": 13242
    },
    {
      "epoch": 0.08997753194510677,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009832485330189156,
      "loss": 2.2184,
      "step": 13243
    },
    {
      "epoch": 0.08998432629170083,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009832457714188264,
      "loss": 2.3177,
      "step": 13244
    },
    {
      "epoch": 0.0899911206382949,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009832430095949999,
      "loss": 2.2666,
      "step": 13245
    },
    {
      "epoch": 0.08999791498488895,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009832402475474374,
      "loss": 2.3335,
      "step": 13246
    },
    {
      "epoch": 0.09000470933148301,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00098323748527614,
      "loss": 2.1417,
      "step": 13247
    },
    {
      "epoch": 0.09001150367807706,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009832347227811096,
      "loss": 2.0914,
      "step": 13248
    },
    {
      "epoch": 0.09001829802467112,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009832319600623469,
      "loss": 2.2964,
      "step": 13249
    },
    {
      "epoch": 0.09002509237126519,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009832291971198532,
      "loss": 2.248,
      "step": 13250
    },
    {
      "epoch": 0.09003188671785924,
      "grad_norm": 0.140625,
      "learning_rate": 0.00098322643395363,
      "loss": 2.1702,
      "step": 13251
    },
    {
      "epoch": 0.0900386810644533,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009832236705636787,
      "loss": 2.2891,
      "step": 13252
    },
    {
      "epoch": 0.09004547541104735,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009832209069500003,
      "loss": 2.2678,
      "step": 13253
    },
    {
      "epoch": 0.09005226975764141,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009832181431125959,
      "loss": 2.3049,
      "step": 13254
    },
    {
      "epoch": 0.09005906410423546,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009832153790514673,
      "loss": 2.1945,
      "step": 13255
    },
    {
      "epoch": 0.09006585845082953,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009832126147666155,
      "loss": 2.3048,
      "step": 13256
    },
    {
      "epoch": 0.09007265279742359,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009832098502580416,
      "loss": 2.0808,
      "step": 13257
    },
    {
      "epoch": 0.09007944714401764,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009832070855257473,
      "loss": 2.2433,
      "step": 13258
    },
    {
      "epoch": 0.0900862414906117,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009832043205697336,
      "loss": 2.0942,
      "step": 13259
    },
    {
      "epoch": 0.09009303583720575,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009832015553900017,
      "loss": 2.2177,
      "step": 13260
    },
    {
      "epoch": 0.09009983018379981,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009831987899865534,
      "loss": 2.1383,
      "step": 13261
    },
    {
      "epoch": 0.09010662453039388,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000983196024359389,
      "loss": 2.3076,
      "step": 13262
    },
    {
      "epoch": 0.09011341887698793,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009831932585085109,
      "loss": 2.3284,
      "step": 13263
    },
    {
      "epoch": 0.09012021322358199,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009831904924339197,
      "loss": 2.2714,
      "step": 13264
    },
    {
      "epoch": 0.09012700757017604,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009831877261356169,
      "loss": 2.1321,
      "step": 13265
    },
    {
      "epoch": 0.0901338019167701,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009831849596136037,
      "loss": 2.1982,
      "step": 13266
    },
    {
      "epoch": 0.09014059626336417,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009831821928678812,
      "loss": 2.0384,
      "step": 13267
    },
    {
      "epoch": 0.09014739060995822,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983179425898451,
      "loss": 2.1982,
      "step": 13268
    },
    {
      "epoch": 0.09015418495655228,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009831766587053144,
      "loss": 2.2662,
      "step": 13269
    },
    {
      "epoch": 0.09016097930314633,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009831738912884724,
      "loss": 2.1847,
      "step": 13270
    },
    {
      "epoch": 0.0901677736497404,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009831711236479264,
      "loss": 2.162,
      "step": 13271
    },
    {
      "epoch": 0.09017456799633446,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009831683557836779,
      "loss": 2.2676,
      "step": 13272
    },
    {
      "epoch": 0.0901813623429285,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009831655876957276,
      "loss": 2.289,
      "step": 13273
    },
    {
      "epoch": 0.09018815668952257,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009831628193840776,
      "loss": 2.2553,
      "step": 13274
    },
    {
      "epoch": 0.09019495103611662,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009831600508487287,
      "loss": 2.3713,
      "step": 13275
    },
    {
      "epoch": 0.09020174538271068,
      "grad_norm": 0.2080078125,
      "learning_rate": 0.000983157282089682,
      "loss": 2.1806,
      "step": 13276
    },
    {
      "epoch": 0.09020853972930473,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009831545131069393,
      "loss": 2.2074,
      "step": 13277
    },
    {
      "epoch": 0.0902153340758988,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009831517439005013,
      "loss": 2.3205,
      "step": 13278
    },
    {
      "epoch": 0.09022212842249286,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009831489744703697,
      "loss": 2.0854,
      "step": 13279
    },
    {
      "epoch": 0.09022892276908691,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009831462048165457,
      "loss": 2.0675,
      "step": 13280
    },
    {
      "epoch": 0.09023571711568097,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009831434349390307,
      "loss": 2.1041,
      "step": 13281
    },
    {
      "epoch": 0.09024251146227502,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009831406648378256,
      "loss": 2.2529,
      "step": 13282
    },
    {
      "epoch": 0.09024930580886908,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009831378945129319,
      "loss": 2.2032,
      "step": 13283
    },
    {
      "epoch": 0.09025610015546315,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.0009831351239643512,
      "loss": 2.3645,
      "step": 13284
    },
    {
      "epoch": 0.0902628945020572,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009831323531920841,
      "loss": 2.3503,
      "step": 13285
    },
    {
      "epoch": 0.09026968884865126,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009831295821961324,
      "loss": 2.1524,
      "step": 13286
    },
    {
      "epoch": 0.09027648319524531,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009831268109764973,
      "loss": 2.2605,
      "step": 13287
    },
    {
      "epoch": 0.09028327754183937,
      "grad_norm": 0.15234375,
      "learning_rate": 0.00098312403953318,
      "loss": 2.1741,
      "step": 13288
    },
    {
      "epoch": 0.09029007188843344,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009831212678661817,
      "loss": 2.3544,
      "step": 13289
    },
    {
      "epoch": 0.09029686623502749,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000983118495975504,
      "loss": 2.1927,
      "step": 13290
    },
    {
      "epoch": 0.09030366058162155,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009831157238611478,
      "loss": 2.3188,
      "step": 13291
    },
    {
      "epoch": 0.0903104549282156,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009831129515231146,
      "loss": 2.1968,
      "step": 13292
    },
    {
      "epoch": 0.09031724927480966,
      "grad_norm": 0.3125,
      "learning_rate": 0.0009831101789614057,
      "loss": 2.2618,
      "step": 13293
    },
    {
      "epoch": 0.09032404362140373,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009831074061760223,
      "loss": 2.2637,
      "step": 13294
    },
    {
      "epoch": 0.09033083796799778,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009831046331669655,
      "loss": 2.1898,
      "step": 13295
    },
    {
      "epoch": 0.09033763231459184,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000983101859934237,
      "loss": 2.3126,
      "step": 13296
    },
    {
      "epoch": 0.09034442666118589,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000983099086477838,
      "loss": 2.1651,
      "step": 13297
    },
    {
      "epoch": 0.09035122100777995,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009830963127977694,
      "loss": 2.2772,
      "step": 13298
    },
    {
      "epoch": 0.09035801535437402,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983093538894033,
      "loss": 2.173,
      "step": 13299
    },
    {
      "epoch": 0.09036480970096807,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009830907647666296,
      "loss": 2.1925,
      "step": 13300
    },
    {
      "epoch": 0.09037160404756213,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009830879904155609,
      "loss": 2.0506,
      "step": 13301
    },
    {
      "epoch": 0.09037839839415618,
      "grad_norm": 0.140625,
      "learning_rate": 0.000983085215840828,
      "loss": 2.2543,
      "step": 13302
    },
    {
      "epoch": 0.09038519274075024,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009830824410424319,
      "loss": 2.2767,
      "step": 13303
    },
    {
      "epoch": 0.09039198708734429,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009830796660203745,
      "loss": 2.2085,
      "step": 13304
    },
    {
      "epoch": 0.09039878143393835,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009830768907746566,
      "loss": 2.3337,
      "step": 13305
    },
    {
      "epoch": 0.09040557578053242,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009830741153052797,
      "loss": 2.3126,
      "step": 13306
    },
    {
      "epoch": 0.09041237012712647,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009830713396122449,
      "loss": 2.1908,
      "step": 13307
    },
    {
      "epoch": 0.09041916447372053,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009830685636955537,
      "loss": 2.128,
      "step": 13308
    },
    {
      "epoch": 0.09042595882031458,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009830657875552072,
      "loss": 2.0641,
      "step": 13309
    },
    {
      "epoch": 0.09043275316690864,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.000983063011191207,
      "loss": 2.1384,
      "step": 13310
    },
    {
      "epoch": 0.09043954751350271,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009830602346035538,
      "loss": 2.1784,
      "step": 13311
    },
    {
      "epoch": 0.09044634186009676,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009830574577922495,
      "loss": 2.3262,
      "step": 13312
    },
    {
      "epoch": 0.09045313620669082,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000983054680757295,
      "loss": 2.1799,
      "step": 13313
    },
    {
      "epoch": 0.09045993055328487,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000983051903498692,
      "loss": 2.2495,
      "step": 13314
    },
    {
      "epoch": 0.09046672489987893,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009830491260164414,
      "loss": 2.1968,
      "step": 13315
    },
    {
      "epoch": 0.090473519246473,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009830463483105445,
      "loss": 2.2492,
      "step": 13316
    },
    {
      "epoch": 0.09048031359306705,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009830435703810025,
      "loss": 2.3312,
      "step": 13317
    },
    {
      "epoch": 0.09048710793966111,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009830407922278171,
      "loss": 2.0498,
      "step": 13318
    },
    {
      "epoch": 0.09049390228625516,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009830380138509892,
      "loss": 2.2382,
      "step": 13319
    },
    {
      "epoch": 0.09050069663284922,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009830352352505206,
      "loss": 2.0668,
      "step": 13320
    },
    {
      "epoch": 0.09050749097944329,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009830324564264119,
      "loss": 2.2706,
      "step": 13321
    },
    {
      "epoch": 0.09051428532603734,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000983029677378665,
      "loss": 2.391,
      "step": 13322
    },
    {
      "epoch": 0.0905210796726314,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009830268981072805,
      "loss": 2.3592,
      "step": 13323
    },
    {
      "epoch": 0.09052787401922545,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009830241186122603,
      "loss": 2.3519,
      "step": 13324
    },
    {
      "epoch": 0.09053466836581951,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009830213388936054,
      "loss": 2.1431,
      "step": 13325
    },
    {
      "epoch": 0.09054146271241356,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009830185589513173,
      "loss": 2.2263,
      "step": 13326
    },
    {
      "epoch": 0.09054825705900763,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000983015778785397,
      "loss": 2.2705,
      "step": 13327
    },
    {
      "epoch": 0.09055505140560169,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000983012998395846,
      "loss": 2.2294,
      "step": 13328
    },
    {
      "epoch": 0.09056184575219574,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009830102177826655,
      "loss": 2.1259,
      "step": 13329
    },
    {
      "epoch": 0.0905686400987898,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009830074369458568,
      "loss": 2.1719,
      "step": 13330
    },
    {
      "epoch": 0.09057543444538385,
      "grad_norm": 0.140625,
      "learning_rate": 0.000983004655885421,
      "loss": 2.1939,
      "step": 13331
    },
    {
      "epoch": 0.09058222879197791,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00098300187460136,
      "loss": 2.2707,
      "step": 13332
    },
    {
      "epoch": 0.09058902313857198,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009829990930936743,
      "loss": 2.1771,
      "step": 13333
    },
    {
      "epoch": 0.09059581748516603,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009829963113623657,
      "loss": 2.282,
      "step": 13334
    },
    {
      "epoch": 0.09060261183176009,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009829935294074356,
      "loss": 2.2137,
      "step": 13335
    },
    {
      "epoch": 0.09060940617835414,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009829907472288847,
      "loss": 2.1998,
      "step": 13336
    },
    {
      "epoch": 0.0906162005249482,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009829879648267147,
      "loss": 2.2698,
      "step": 13337
    },
    {
      "epoch": 0.09062299487154227,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009829851822009267,
      "loss": 2.3474,
      "step": 13338
    },
    {
      "epoch": 0.09062978921813632,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009829823993515222,
      "loss": 2.4892,
      "step": 13339
    },
    {
      "epoch": 0.09063658356473038,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009829796162785023,
      "loss": 2.0819,
      "step": 13340
    },
    {
      "epoch": 0.09064337791132443,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009829768329818686,
      "loss": 2.2643,
      "step": 13341
    },
    {
      "epoch": 0.0906501722579185,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000982974049461622,
      "loss": 2.2057,
      "step": 13342
    },
    {
      "epoch": 0.09065696660451256,
      "grad_norm": 0.123046875,
      "learning_rate": 0.000982971265717764,
      "loss": 2.1013,
      "step": 13343
    },
    {
      "epoch": 0.0906637609511066,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009829684817502958,
      "loss": 2.2612,
      "step": 13344
    },
    {
      "epoch": 0.09067055529770067,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009829656975592186,
      "loss": 2.1074,
      "step": 13345
    },
    {
      "epoch": 0.09067734964429472,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009829629131445341,
      "loss": 2.0842,
      "step": 13346
    },
    {
      "epoch": 0.09068414399088878,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009829601285062434,
      "loss": 2.2284,
      "step": 13347
    },
    {
      "epoch": 0.09069093833748283,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009829573436443474,
      "loss": 2.1796,
      "step": 13348
    },
    {
      "epoch": 0.0906977326840769,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009829545585588477,
      "loss": 2.1621,
      "step": 13349
    },
    {
      "epoch": 0.09070452703067096,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009829517732497456,
      "loss": 2.2155,
      "step": 13350
    },
    {
      "epoch": 0.09071132137726501,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009829489877170425,
      "loss": 2.1941,
      "step": 13351
    },
    {
      "epoch": 0.09071811572385907,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009829462019607393,
      "loss": 2.2995,
      "step": 13352
    },
    {
      "epoch": 0.09072491007045312,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000982943415980838,
      "loss": 2.1148,
      "step": 13353
    },
    {
      "epoch": 0.09073170441704718,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000982940629777339,
      "loss": 2.2979,
      "step": 13354
    },
    {
      "epoch": 0.09073849876364125,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009829378433502442,
      "loss": 2.1893,
      "step": 13355
    },
    {
      "epoch": 0.0907452931102353,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009829350566995547,
      "loss": 2.3454,
      "step": 13356
    },
    {
      "epoch": 0.09075208745682936,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009829322698252717,
      "loss": 2.2353,
      "step": 13357
    },
    {
      "epoch": 0.09075888180342341,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009829294827273967,
      "loss": 2.2325,
      "step": 13358
    },
    {
      "epoch": 0.09076567615001747,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009829266954059308,
      "loss": 2.2076,
      "step": 13359
    },
    {
      "epoch": 0.09077247049661154,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009829239078608756,
      "loss": 2.3388,
      "step": 13360
    },
    {
      "epoch": 0.09077926484320559,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009829211200922318,
      "loss": 2.2777,
      "step": 13361
    },
    {
      "epoch": 0.09078605918979965,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009829183321000015,
      "loss": 2.2436,
      "step": 13362
    },
    {
      "epoch": 0.0907928535363937,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009829155438841853,
      "loss": 2.4222,
      "step": 13363
    },
    {
      "epoch": 0.09079964788298776,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009829127554447847,
      "loss": 2.2734,
      "step": 13364
    },
    {
      "epoch": 0.09080644222958183,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000982909966781801,
      "loss": 2.252,
      "step": 13365
    },
    {
      "epoch": 0.09081323657617588,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009829071778952357,
      "loss": 2.2262,
      "step": 13366
    },
    {
      "epoch": 0.09082003092276994,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00098290438878509,
      "loss": 2.1797,
      "step": 13367
    },
    {
      "epoch": 0.09082682526936399,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009829015994513648,
      "loss": 2.1222,
      "step": 13368
    },
    {
      "epoch": 0.09083361961595805,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009828988098940618,
      "loss": 2.3133,
      "step": 13369
    },
    {
      "epoch": 0.09084041396255212,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009828960201131822,
      "loss": 2.3128,
      "step": 13370
    },
    {
      "epoch": 0.09084720830914617,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009828932301087273,
      "loss": 2.3051,
      "step": 13371
    },
    {
      "epoch": 0.09085400265574023,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009828904398806984,
      "loss": 2.2768,
      "step": 13372
    },
    {
      "epoch": 0.09086079700233428,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009828876494290969,
      "loss": 2.2257,
      "step": 13373
    },
    {
      "epoch": 0.09086759134892834,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009828848587539237,
      "loss": 2.1498,
      "step": 13374
    },
    {
      "epoch": 0.09087438569552239,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009828820678551805,
      "loss": 2.0283,
      "step": 13375
    },
    {
      "epoch": 0.09088118004211646,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009828792767328683,
      "loss": 2.1006,
      "step": 13376
    },
    {
      "epoch": 0.09088797438871052,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009828764853869886,
      "loss": 2.3743,
      "step": 13377
    },
    {
      "epoch": 0.09089476873530457,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009828736938175425,
      "loss": 2.3878,
      "step": 13378
    },
    {
      "epoch": 0.09090156308189863,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009828709020245315,
      "loss": 2.1531,
      "step": 13379
    },
    {
      "epoch": 0.09090835742849268,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009828681100079568,
      "loss": 2.1542,
      "step": 13380
    },
    {
      "epoch": 0.09091515177508674,
      "grad_norm": 0.140625,
      "learning_rate": 0.00098286531776782,
      "loss": 2.2425,
      "step": 13381
    },
    {
      "epoch": 0.09092194612168081,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009828625253041217,
      "loss": 2.3752,
      "step": 13382
    },
    {
      "epoch": 0.09092874046827486,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009828597326168639,
      "loss": 2.2513,
      "step": 13383
    },
    {
      "epoch": 0.09093553481486892,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009828569397060472,
      "loss": 2.135,
      "step": 13384
    },
    {
      "epoch": 0.09094232916146297,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009828541465716737,
      "loss": 2.279,
      "step": 13385
    },
    {
      "epoch": 0.09094912350805703,
      "grad_norm": 0.140625,
      "learning_rate": 0.000982851353213744,
      "loss": 2.2128,
      "step": 13386
    },
    {
      "epoch": 0.0909559178546511,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009828485596322595,
      "loss": 2.1761,
      "step": 13387
    },
    {
      "epoch": 0.09096271220124515,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000982845765827222,
      "loss": 2.1293,
      "step": 13388
    },
    {
      "epoch": 0.09096950654783921,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009828429717986323,
      "loss": 2.2403,
      "step": 13389
    },
    {
      "epoch": 0.09097630089443326,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009828401775464916,
      "loss": 2.1107,
      "step": 13390
    },
    {
      "epoch": 0.09098309524102732,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000982837383070802,
      "loss": 2.2692,
      "step": 13391
    },
    {
      "epoch": 0.09098988958762139,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009828345883715636,
      "loss": 2.0732,
      "step": 13392
    },
    {
      "epoch": 0.09099668393421544,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009828317934487787,
      "loss": 2.2138,
      "step": 13393
    },
    {
      "epoch": 0.0910034782808095,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009828289983024481,
      "loss": 2.2469,
      "step": 13394
    },
    {
      "epoch": 0.09101027262740355,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009828262029325733,
      "loss": 2.2189,
      "step": 13395
    },
    {
      "epoch": 0.09101706697399761,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009828234073391555,
      "loss": 2.118,
      "step": 13396
    },
    {
      "epoch": 0.09102386132059166,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009828206115221959,
      "loss": 2.3496,
      "step": 13397
    },
    {
      "epoch": 0.09103065566718573,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000982817815481696,
      "loss": 2.2675,
      "step": 13398
    },
    {
      "epoch": 0.09103745001377979,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000982815019217657,
      "loss": 2.2472,
      "step": 13399
    },
    {
      "epoch": 0.09104424436037384,
      "grad_norm": 0.1484375,
      "learning_rate": 0.00098281222273008,
      "loss": 2.2511,
      "step": 13400
    },
    {
      "epoch": 0.0910510387069679,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009828094260189667,
      "loss": 2.0661,
      "step": 13401
    },
    {
      "epoch": 0.09105783305356195,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000982806629084318,
      "loss": 2.2856,
      "step": 13402
    },
    {
      "epoch": 0.09106462740015601,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009828038319261354,
      "loss": 2.1243,
      "step": 13403
    },
    {
      "epoch": 0.09107142174675008,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009828010345444202,
      "loss": 2.2019,
      "step": 13404
    },
    {
      "epoch": 0.09107821609334413,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009827982369391738,
      "loss": 2.251,
      "step": 13405
    },
    {
      "epoch": 0.09108501043993819,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000982795439110397,
      "loss": 2.2569,
      "step": 13406
    },
    {
      "epoch": 0.09109180478653224,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009827926410580917,
      "loss": 2.1671,
      "step": 13407
    },
    {
      "epoch": 0.0910985991331263,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009827898427822589,
      "loss": 2.2419,
      "step": 13408
    },
    {
      "epoch": 0.09110539347972037,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009827870442829,
      "loss": 2.2961,
      "step": 13409
    },
    {
      "epoch": 0.09111218782631442,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000982784245560016,
      "loss": 2.258,
      "step": 13410
    },
    {
      "epoch": 0.09111898217290848,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009827814466136086,
      "loss": 2.1748,
      "step": 13411
    },
    {
      "epoch": 0.09112577651950253,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009827786474436789,
      "loss": 2.1972,
      "step": 13412
    },
    {
      "epoch": 0.0911325708660966,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009827758480502283,
      "loss": 2.2532,
      "step": 13413
    },
    {
      "epoch": 0.09113936521269066,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009827730484332578,
      "loss": 2.1538,
      "step": 13414
    },
    {
      "epoch": 0.0911461595592847,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009827702485927692,
      "loss": 2.2569,
      "step": 13415
    },
    {
      "epoch": 0.09115295390587877,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009827674485287632,
      "loss": 2.2044,
      "step": 13416
    },
    {
      "epoch": 0.09115974825247282,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009827646482412416,
      "loss": 2.3503,
      "step": 13417
    },
    {
      "epoch": 0.09116654259906688,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009827618477302054,
      "loss": 2.2238,
      "step": 13418
    },
    {
      "epoch": 0.09117333694566095,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009827590469956562,
      "loss": 2.2456,
      "step": 13419
    },
    {
      "epoch": 0.091180131292255,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000982756246037595,
      "loss": 2.326,
      "step": 13420
    },
    {
      "epoch": 0.09118692563884906,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009827534448560232,
      "loss": 2.301,
      "step": 13421
    },
    {
      "epoch": 0.09119371998544311,
      "grad_norm": 0.140625,
      "learning_rate": 0.000982750643450942,
      "loss": 2.2587,
      "step": 13422
    },
    {
      "epoch": 0.09120051433203717,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009827478418223527,
      "loss": 2.1495,
      "step": 13423
    },
    {
      "epoch": 0.09120730867863122,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009827450399702569,
      "loss": 2.1698,
      "step": 13424
    },
    {
      "epoch": 0.09121410302522528,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009827422378946556,
      "loss": 2.1324,
      "step": 13425
    },
    {
      "epoch": 0.09122089737181935,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009827394355955501,
      "loss": 2.0968,
      "step": 13426
    },
    {
      "epoch": 0.0912276917184134,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009827366330729418,
      "loss": 2.2138,
      "step": 13427
    },
    {
      "epoch": 0.09123448606500746,
      "grad_norm": 0.125,
      "learning_rate": 0.0009827338303268322,
      "loss": 2.1253,
      "step": 13428
    },
    {
      "epoch": 0.09124128041160151,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000982731027357222,
      "loss": 2.1729,
      "step": 13429
    },
    {
      "epoch": 0.09124807475819557,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009827282241641133,
      "loss": 2.3513,
      "step": 13430
    },
    {
      "epoch": 0.09125486910478964,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009827254207475068,
      "loss": 2.1121,
      "step": 13431
    },
    {
      "epoch": 0.09126166345138369,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009827226171074037,
      "loss": 2.215,
      "step": 13432
    },
    {
      "epoch": 0.09126845779797775,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009827198132438058,
      "loss": 2.1695,
      "step": 13433
    },
    {
      "epoch": 0.0912752521445718,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009827170091567142,
      "loss": 2.2175,
      "step": 13434
    },
    {
      "epoch": 0.09128204649116586,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00098271420484613,
      "loss": 2.2079,
      "step": 13435
    },
    {
      "epoch": 0.09128884083775993,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009827114003120548,
      "loss": 2.1919,
      "step": 13436
    },
    {
      "epoch": 0.09129563518435398,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009827085955544897,
      "loss": 2.3094,
      "step": 13437
    },
    {
      "epoch": 0.09130242953094804,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.000982705790573436,
      "loss": 2.2592,
      "step": 13438
    },
    {
      "epoch": 0.09130922387754209,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009827029853688952,
      "loss": 2.1842,
      "step": 13439
    },
    {
      "epoch": 0.09131601822413615,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009827001799408684,
      "loss": 2.2145,
      "step": 13440
    },
    {
      "epoch": 0.09132281257073022,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009826973742893567,
      "loss": 2.2516,
      "step": 13441
    },
    {
      "epoch": 0.09132960691732427,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000982694568414362,
      "loss": 2.1738,
      "step": 13442
    },
    {
      "epoch": 0.09133640126391833,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000982691762315885,
      "loss": 2.2067,
      "step": 13443
    },
    {
      "epoch": 0.09134319561051238,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009826889559939275,
      "loss": 2.2444,
      "step": 13444
    },
    {
      "epoch": 0.09134998995710644,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009826861494484902,
      "loss": 2.2519,
      "step": 13445
    },
    {
      "epoch": 0.09135678430370049,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000982683342679575,
      "loss": 2.1888,
      "step": 13446
    },
    {
      "epoch": 0.09136357865029456,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009826805356871829,
      "loss": 2.2237,
      "step": 13447
    },
    {
      "epoch": 0.09137037299688862,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009826777284713153,
      "loss": 2.3595,
      "step": 13448
    },
    {
      "epoch": 0.09137716734348267,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009826749210319733,
      "loss": 2.2908,
      "step": 13449
    },
    {
      "epoch": 0.09138396169007673,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009826721133691584,
      "loss": 2.3837,
      "step": 13450
    },
    {
      "epoch": 0.09139075603667078,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009826693054828719,
      "loss": 2.2662,
      "step": 13451
    },
    {
      "epoch": 0.09139755038326484,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000982666497373115,
      "loss": 2.0933,
      "step": 13452
    },
    {
      "epoch": 0.09140434472985891,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009826636890398893,
      "loss": 2.165,
      "step": 13453
    },
    {
      "epoch": 0.09141113907645296,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009826608804831955,
      "loss": 2.3051,
      "step": 13454
    },
    {
      "epoch": 0.09141793342304702,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009826580717030354,
      "loss": 2.1562,
      "step": 13455
    },
    {
      "epoch": 0.09142472776964107,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00098265526269941,
      "loss": 2.1518,
      "step": 13456
    },
    {
      "epoch": 0.09143152211623513,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000982652453472321,
      "loss": 2.0728,
      "step": 13457
    },
    {
      "epoch": 0.0914383164628292,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009826496440217692,
      "loss": 2.1329,
      "step": 13458
    },
    {
      "epoch": 0.09144511080942325,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009826468343477562,
      "loss": 2.2408,
      "step": 13459
    },
    {
      "epoch": 0.09145190515601731,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009826440244502831,
      "loss": 2.261,
      "step": 13460
    },
    {
      "epoch": 0.09145869950261136,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009826412143293518,
      "loss": 2.1634,
      "step": 13461
    },
    {
      "epoch": 0.09146549384920542,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009826384039849628,
      "loss": 2.2852,
      "step": 13462
    },
    {
      "epoch": 0.09147228819579949,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000982635593417118,
      "loss": 2.2048,
      "step": 13463
    },
    {
      "epoch": 0.09147908254239354,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000982632782625818,
      "loss": 2.1468,
      "step": 13464
    },
    {
      "epoch": 0.0914858768889876,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009826299716110649,
      "loss": 2.213,
      "step": 13465
    },
    {
      "epoch": 0.09149267123558165,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009826271603728598,
      "loss": 2.0989,
      "step": 13466
    },
    {
      "epoch": 0.09149946558217571,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009826243489112036,
      "loss": 2.1921,
      "step": 13467
    },
    {
      "epoch": 0.09150625992876976,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000982621537226098,
      "loss": 2.2709,
      "step": 13468
    },
    {
      "epoch": 0.09151305427536383,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000982618725317544,
      "loss": 2.3863,
      "step": 13469
    },
    {
      "epoch": 0.09151984862195789,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009826159131855433,
      "loss": 2.1693,
      "step": 13470
    },
    {
      "epoch": 0.09152664296855194,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009826131008300967,
      "loss": 2.3456,
      "step": 13471
    },
    {
      "epoch": 0.091533437315146,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009826102882512058,
      "loss": 2.2602,
      "step": 13472
    },
    {
      "epoch": 0.09154023166174005,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000982607475448872,
      "loss": 2.2308,
      "step": 13473
    },
    {
      "epoch": 0.09154702600833411,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009826046624230964,
      "loss": 2.106,
      "step": 13474
    },
    {
      "epoch": 0.09155382035492818,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009826018491738803,
      "loss": 2.3028,
      "step": 13475
    },
    {
      "epoch": 0.09156061470152223,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009825990357012254,
      "loss": 2.1623,
      "step": 13476
    },
    {
      "epoch": 0.09156740904811629,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009825962220051322,
      "loss": 2.0728,
      "step": 13477
    },
    {
      "epoch": 0.09157420339471034,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000982593408085603,
      "loss": 2.27,
      "step": 13478
    },
    {
      "epoch": 0.0915809977413044,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009825905939426383,
      "loss": 2.1726,
      "step": 13479
    },
    {
      "epoch": 0.09158779208789847,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009825877795762398,
      "loss": 2.2876,
      "step": 13480
    },
    {
      "epoch": 0.09159458643449252,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009825849649864084,
      "loss": 2.2257,
      "step": 13481
    },
    {
      "epoch": 0.09160138078108658,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000982582150173146,
      "loss": 2.2119,
      "step": 13482
    },
    {
      "epoch": 0.09160817512768063,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009825793351364534,
      "loss": 2.2479,
      "step": 13483
    },
    {
      "epoch": 0.0916149694742747,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000982576519876332,
      "loss": 2.2053,
      "step": 13484
    },
    {
      "epoch": 0.09162176382086876,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009825737043927835,
      "loss": 2.191,
      "step": 13485
    },
    {
      "epoch": 0.0916285581674628,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009825708886858089,
      "loss": 2.3301,
      "step": 13486
    },
    {
      "epoch": 0.09163535251405687,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009825680727554094,
      "loss": 2.1656,
      "step": 13487
    },
    {
      "epoch": 0.09164214686065092,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009825652566015862,
      "loss": 2.284,
      "step": 13488
    },
    {
      "epoch": 0.09164894120724498,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009825624402243413,
      "loss": 2.2053,
      "step": 13489
    },
    {
      "epoch": 0.09165573555383905,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009825596236236753,
      "loss": 2.1654,
      "step": 13490
    },
    {
      "epoch": 0.0916625299004331,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009825568067995897,
      "loss": 2.2798,
      "step": 13491
    },
    {
      "epoch": 0.09166932424702716,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009825539897520858,
      "loss": 2.1645,
      "step": 13492
    },
    {
      "epoch": 0.09167611859362121,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000982551172481165,
      "loss": 2.2528,
      "step": 13493
    },
    {
      "epoch": 0.09168291294021527,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009825483549868283,
      "loss": 2.1379,
      "step": 13494
    },
    {
      "epoch": 0.09168970728680932,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009825455372690776,
      "loss": 2.2354,
      "step": 13495
    },
    {
      "epoch": 0.09169650163340339,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009825427193279137,
      "loss": 2.1822,
      "step": 13496
    },
    {
      "epoch": 0.09170329597999745,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009825399011633381,
      "loss": 2.2705,
      "step": 13497
    },
    {
      "epoch": 0.0917100903265915,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000982537082775352,
      "loss": 2.4424,
      "step": 13498
    },
    {
      "epoch": 0.09171688467318556,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009825342641639567,
      "loss": 2.2055,
      "step": 13499
    },
    {
      "epoch": 0.09172367901977961,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009825314453291536,
      "loss": 2.2021,
      "step": 13500
    },
    {
      "epoch": 0.09173047336637367,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000982528626270944,
      "loss": 2.1412,
      "step": 13501
    },
    {
      "epoch": 0.09173726771296774,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009825258069893292,
      "loss": 2.1315,
      "step": 13502
    },
    {
      "epoch": 0.09174406205956179,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009825229874843105,
      "loss": 2.2666,
      "step": 13503
    },
    {
      "epoch": 0.09175085640615585,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000982520167755889,
      "loss": 2.1989,
      "step": 13504
    },
    {
      "epoch": 0.0917576507527499,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009825173478040666,
      "loss": 2.1258,
      "step": 13505
    },
    {
      "epoch": 0.09176444509934396,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000982514527628844,
      "loss": 2.2353,
      "step": 13506
    },
    {
      "epoch": 0.09177123944593803,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009825117072302225,
      "loss": 2.2183,
      "step": 13507
    },
    {
      "epoch": 0.09177803379253208,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009825088866082037,
      "loss": 2.1797,
      "step": 13508
    },
    {
      "epoch": 0.09178482813912614,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009825060657627888,
      "loss": 2.2485,
      "step": 13509
    },
    {
      "epoch": 0.09179162248572019,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009825032446939793,
      "loss": 2.1997,
      "step": 13510
    },
    {
      "epoch": 0.09179841683231425,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009825004234017762,
      "loss": 2.3194,
      "step": 13511
    },
    {
      "epoch": 0.09180521117890832,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000982497601886181,
      "loss": 2.3134,
      "step": 13512
    },
    {
      "epoch": 0.09181200552550237,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009824947801471949,
      "loss": 2.439,
      "step": 13513
    },
    {
      "epoch": 0.09181879987209643,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009824919581848193,
      "loss": 2.1374,
      "step": 13514
    },
    {
      "epoch": 0.09182559421869048,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009824891359990554,
      "loss": 2.3628,
      "step": 13515
    },
    {
      "epoch": 0.09183238856528454,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009824863135899046,
      "loss": 2.2195,
      "step": 13516
    },
    {
      "epoch": 0.09183918291187859,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009824834909573682,
      "loss": 2.2349,
      "step": 13517
    },
    {
      "epoch": 0.09184597725847266,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009824806681014474,
      "loss": 2.4238,
      "step": 13518
    },
    {
      "epoch": 0.09185277160506672,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009824778450221435,
      "loss": 2.1992,
      "step": 13519
    },
    {
      "epoch": 0.09185956595166077,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000982475021719458,
      "loss": 2.083,
      "step": 13520
    },
    {
      "epoch": 0.09186636029825483,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009824721981933921,
      "loss": 2.1925,
      "step": 13521
    },
    {
      "epoch": 0.09187315464484888,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009824693744439471,
      "loss": 2.0948,
      "step": 13522
    },
    {
      "epoch": 0.09187994899144294,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009824665504711244,
      "loss": 2.173,
      "step": 13523
    },
    {
      "epoch": 0.09188674333803701,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000982463726274925,
      "loss": 2.0516,
      "step": 13524
    },
    {
      "epoch": 0.09189353768463106,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009824609018553507,
      "loss": 2.3306,
      "step": 13525
    },
    {
      "epoch": 0.09190033203122512,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009824580772124023,
      "loss": 2.1274,
      "step": 13526
    },
    {
      "epoch": 0.09190712637781917,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009824552523460814,
      "loss": 2.1508,
      "step": 13527
    },
    {
      "epoch": 0.09191392072441323,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009824524272563893,
      "loss": 2.1617,
      "step": 13528
    },
    {
      "epoch": 0.0919207150710073,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009824496019433273,
      "loss": 2.278,
      "step": 13529
    },
    {
      "epoch": 0.09192750941760135,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009824467764068966,
      "loss": 2.2313,
      "step": 13530
    },
    {
      "epoch": 0.09193430376419541,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009824439506470985,
      "loss": 2.2444,
      "step": 13531
    },
    {
      "epoch": 0.09194109811078946,
      "grad_norm": 0.125,
      "learning_rate": 0.0009824411246639345,
      "loss": 2.0723,
      "step": 13532
    },
    {
      "epoch": 0.09194789245738352,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009824382984574057,
      "loss": 2.1947,
      "step": 13533
    },
    {
      "epoch": 0.09195468680397759,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009824354720275135,
      "loss": 2.0934,
      "step": 13534
    },
    {
      "epoch": 0.09196148115057164,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009824326453742593,
      "loss": 2.227,
      "step": 13535
    },
    {
      "epoch": 0.0919682754971657,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009824298184976442,
      "loss": 2.1457,
      "step": 13536
    },
    {
      "epoch": 0.09197506984375975,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009824269913976697,
      "loss": 2.2269,
      "step": 13537
    },
    {
      "epoch": 0.09198186419035381,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000982424164074337,
      "loss": 2.2692,
      "step": 13538
    },
    {
      "epoch": 0.09198865853694788,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009824213365276476,
      "loss": 2.3071,
      "step": 13539
    },
    {
      "epoch": 0.09199545288354193,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009824185087576023,
      "loss": 2.2085,
      "step": 13540
    },
    {
      "epoch": 0.09200224723013599,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000982415680764203,
      "loss": 2.246,
      "step": 13541
    },
    {
      "epoch": 0.09200904157673004,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009824128525474508,
      "loss": 2.3851,
      "step": 13542
    },
    {
      "epoch": 0.0920158359233241,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000982410024107347,
      "loss": 2.2758,
      "step": 13543
    },
    {
      "epoch": 0.09202263026991815,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009824071954438929,
      "loss": 2.2428,
      "step": 13544
    },
    {
      "epoch": 0.09202942461651221,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009824043665570897,
      "loss": 2.3008,
      "step": 13545
    },
    {
      "epoch": 0.09203621896310628,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009824015374469387,
      "loss": 2.2529,
      "step": 13546
    },
    {
      "epoch": 0.09204301330970033,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009823987081134416,
      "loss": 2.2533,
      "step": 13547
    },
    {
      "epoch": 0.09204980765629439,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009823958785565993,
      "loss": 2.2407,
      "step": 13548
    },
    {
      "epoch": 0.09205660200288844,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000982393048776413,
      "loss": 2.0672,
      "step": 13549
    },
    {
      "epoch": 0.0920633963494825,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009823902187728847,
      "loss": 2.3074,
      "step": 13550
    },
    {
      "epoch": 0.09207019069607657,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009823873885460147,
      "loss": 2.2351,
      "step": 13551
    },
    {
      "epoch": 0.09207698504267062,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009823845580958052,
      "loss": 2.124,
      "step": 13552
    },
    {
      "epoch": 0.09208377938926468,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009823817274222572,
      "loss": 2.3317,
      "step": 13553
    },
    {
      "epoch": 0.09209057373585873,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000982378896525372,
      "loss": 2.0928,
      "step": 13554
    },
    {
      "epoch": 0.0920973680824528,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000982376065405151,
      "loss": 2.21,
      "step": 13555
    },
    {
      "epoch": 0.09210416242904686,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009823732340615951,
      "loss": 2.0949,
      "step": 13556
    },
    {
      "epoch": 0.0921109567756409,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009823704024947062,
      "loss": 2.299,
      "step": 13557
    },
    {
      "epoch": 0.09211775112223497,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009823675707044852,
      "loss": 2.0953,
      "step": 13558
    },
    {
      "epoch": 0.09212454546882902,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009823647386909336,
      "loss": 2.3288,
      "step": 13559
    },
    {
      "epoch": 0.09213133981542308,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009823619064540526,
      "loss": 2.2932,
      "step": 13560
    },
    {
      "epoch": 0.09213813416201715,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009823590739938435,
      "loss": 2.3086,
      "step": 13561
    },
    {
      "epoch": 0.0921449285086112,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009823562413103079,
      "loss": 2.0924,
      "step": 13562
    },
    {
      "epoch": 0.09215172285520526,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009823534084034467,
      "loss": 2.2196,
      "step": 13563
    },
    {
      "epoch": 0.09215851720179931,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009823505752732616,
      "loss": 2.1452,
      "step": 13564
    },
    {
      "epoch": 0.09216531154839337,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009823477419197534,
      "loss": 2.2981,
      "step": 13565
    },
    {
      "epoch": 0.09217210589498742,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009823449083429241,
      "loss": 2.3206,
      "step": 13566
    },
    {
      "epoch": 0.09217890024158149,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009823420745427743,
      "loss": 2.1685,
      "step": 13567
    },
    {
      "epoch": 0.09218569458817555,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000982339240519306,
      "loss": 2.1829,
      "step": 13568
    },
    {
      "epoch": 0.0921924889347696,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009823364062725199,
      "loss": 2.2771,
      "step": 13569
    },
    {
      "epoch": 0.09219928328136366,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009823335718024176,
      "loss": 2.3363,
      "step": 13570
    },
    {
      "epoch": 0.09220607762795771,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009823307371090004,
      "loss": 2.1674,
      "step": 13571
    },
    {
      "epoch": 0.09221287197455177,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009823279021922695,
      "loss": 2.2188,
      "step": 13572
    },
    {
      "epoch": 0.09221966632114584,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009823250670522264,
      "loss": 2.1762,
      "step": 13573
    },
    {
      "epoch": 0.09222646066773989,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009823222316888724,
      "loss": 2.1646,
      "step": 13574
    },
    {
      "epoch": 0.09223325501433395,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009823193961022087,
      "loss": 2.1385,
      "step": 13575
    },
    {
      "epoch": 0.092240049360928,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009823165602922367,
      "loss": 2.2637,
      "step": 13576
    },
    {
      "epoch": 0.09224684370752206,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009823137242589576,
      "loss": 2.2421,
      "step": 13577
    },
    {
      "epoch": 0.09225363805411613,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009823108880023727,
      "loss": 2.0456,
      "step": 13578
    },
    {
      "epoch": 0.09226043240071018,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009823080515224834,
      "loss": 2.1459,
      "step": 13579
    },
    {
      "epoch": 0.09226722674730424,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009823052148192912,
      "loss": 2.1507,
      "step": 13580
    },
    {
      "epoch": 0.09227402109389829,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000982302377892797,
      "loss": 2.2647,
      "step": 13581
    },
    {
      "epoch": 0.09228081544049235,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009822995407430025,
      "loss": 2.184,
      "step": 13582
    },
    {
      "epoch": 0.09228760978708642,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009822967033699087,
      "loss": 2.2971,
      "step": 13583
    },
    {
      "epoch": 0.09229440413368047,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000982293865773517,
      "loss": 2.1666,
      "step": 13584
    },
    {
      "epoch": 0.09230119848027453,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009822910279538289,
      "loss": 2.0617,
      "step": 13585
    },
    {
      "epoch": 0.09230799282686858,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009822881899108455,
      "loss": 2.2094,
      "step": 13586
    },
    {
      "epoch": 0.09231478717346264,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009822853516445683,
      "loss": 2.2654,
      "step": 13587
    },
    {
      "epoch": 0.09232158152005669,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009822825131549984,
      "loss": 2.2111,
      "step": 13588
    },
    {
      "epoch": 0.09232837586665076,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009822796744421373,
      "loss": 2.1334,
      "step": 13589
    },
    {
      "epoch": 0.09233517021324482,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009822768355059863,
      "loss": 2.214,
      "step": 13590
    },
    {
      "epoch": 0.09234196455983887,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009822739963465466,
      "loss": 2.093,
      "step": 13591
    },
    {
      "epoch": 0.09234875890643293,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009822711569638196,
      "loss": 2.1059,
      "step": 13592
    },
    {
      "epoch": 0.09235555325302698,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009822683173578064,
      "loss": 2.1101,
      "step": 13593
    },
    {
      "epoch": 0.09236234759962104,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009822654775285087,
      "loss": 2.1939,
      "step": 13594
    },
    {
      "epoch": 0.09236914194621511,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009822626374759275,
      "loss": 2.228,
      "step": 13595
    },
    {
      "epoch": 0.09237593629280916,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009822597972000643,
      "loss": 2.3038,
      "step": 13596
    },
    {
      "epoch": 0.09238273063940322,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009822569567009202,
      "loss": 2.2413,
      "step": 13597
    },
    {
      "epoch": 0.09238952498599727,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009822541159784969,
      "loss": 2.3099,
      "step": 13598
    },
    {
      "epoch": 0.09239631933259133,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009822512750327953,
      "loss": 2.1481,
      "step": 13599
    },
    {
      "epoch": 0.0924031136791854,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009822484338638169,
      "loss": 2.1613,
      "step": 13600
    },
    {
      "epoch": 0.09240990802577945,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000982245592471563,
      "loss": 2.2969,
      "step": 13601
    },
    {
      "epoch": 0.09241670237237351,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000982242750856035,
      "loss": 2.3736,
      "step": 13602
    },
    {
      "epoch": 0.09242349671896756,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000982239909017234,
      "loss": 2.1503,
      "step": 13603
    },
    {
      "epoch": 0.09243029106556162,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009822370669551616,
      "loss": 2.1579,
      "step": 13604
    },
    {
      "epoch": 0.09243708541215569,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000982234224669819,
      "loss": 2.2199,
      "step": 13605
    },
    {
      "epoch": 0.09244387975874974,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009822313821612075,
      "loss": 2.3386,
      "step": 13606
    },
    {
      "epoch": 0.0924506741053438,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009822285394293282,
      "loss": 2.2351,
      "step": 13607
    },
    {
      "epoch": 0.09245746845193785,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009822256964741828,
      "loss": 2.022,
      "step": 13608
    },
    {
      "epoch": 0.09246426279853191,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009822228532957722,
      "loss": 2.2612,
      "step": 13609
    },
    {
      "epoch": 0.09247105714512598,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009822200098940981,
      "loss": 2.2973,
      "step": 13610
    },
    {
      "epoch": 0.09247785149172003,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009822171662691617,
      "loss": 2.3817,
      "step": 13611
    },
    {
      "epoch": 0.09248464583831409,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009822143224209642,
      "loss": 2.1599,
      "step": 13612
    },
    {
      "epoch": 0.09249144018490814,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009822114783495071,
      "loss": 2.2353,
      "step": 13613
    },
    {
      "epoch": 0.0924982345315022,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009822086340547913,
      "loss": 2.2849,
      "step": 13614
    },
    {
      "epoch": 0.09250502887809625,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009822057895368188,
      "loss": 2.2958,
      "step": 13615
    },
    {
      "epoch": 0.09251182322469031,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009822029447955904,
      "loss": 2.1344,
      "step": 13616
    },
    {
      "epoch": 0.09251861757128438,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009822000998311075,
      "loss": 2.2051,
      "step": 13617
    },
    {
      "epoch": 0.09252541191787843,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009821972546433717,
      "loss": 2.2309,
      "step": 13618
    },
    {
      "epoch": 0.09253220626447249,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000982194409232384,
      "loss": 2.2626,
      "step": 13619
    },
    {
      "epoch": 0.09253900061106654,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009821915635981457,
      "loss": 2.2659,
      "step": 13620
    },
    {
      "epoch": 0.0925457949576606,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009821887177406583,
      "loss": 2.3316,
      "step": 13621
    },
    {
      "epoch": 0.09255258930425467,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000982185871659923,
      "loss": 2.1655,
      "step": 13622
    },
    {
      "epoch": 0.09255938365084872,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009821830253559415,
      "loss": 2.2308,
      "step": 13623
    },
    {
      "epoch": 0.09256617799744278,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009821801788287145,
      "loss": 2.1764,
      "step": 13624
    },
    {
      "epoch": 0.09257297234403683,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009821773320782435,
      "loss": 2.2402,
      "step": 13625
    },
    {
      "epoch": 0.0925797666906309,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00098217448510453,
      "loss": 2.2106,
      "step": 13626
    },
    {
      "epoch": 0.09258656103722496,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009821716379075753,
      "loss": 2.1983,
      "step": 13627
    },
    {
      "epoch": 0.092593355383819,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009821687904873807,
      "loss": 2.2184,
      "step": 13628
    },
    {
      "epoch": 0.09260014973041307,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009821659428439474,
      "loss": 2.1729,
      "step": 13629
    },
    {
      "epoch": 0.09260694407700712,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009821630949772768,
      "loss": 2.2991,
      "step": 13630
    },
    {
      "epoch": 0.09261373842360118,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009821602468873703,
      "loss": 2.1581,
      "step": 13631
    },
    {
      "epoch": 0.09262053277019525,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009821573985742292,
      "loss": 2.1231,
      "step": 13632
    },
    {
      "epoch": 0.0926273271167893,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009821545500378545,
      "loss": 2.252,
      "step": 13633
    },
    {
      "epoch": 0.09263412146338336,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009821517012782478,
      "loss": 2.2081,
      "step": 13634
    },
    {
      "epoch": 0.09264091580997741,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009821488522954104,
      "loss": 2.1256,
      "step": 13635
    },
    {
      "epoch": 0.09264771015657147,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009821460030893438,
      "loss": 2.3256,
      "step": 13636
    },
    {
      "epoch": 0.09265450450316552,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000982143153660049,
      "loss": 2.2449,
      "step": 13637
    },
    {
      "epoch": 0.09266129884975959,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009821403040075275,
      "loss": 2.1482,
      "step": 13638
    },
    {
      "epoch": 0.09266809319635365,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009821374541317807,
      "loss": 2.2052,
      "step": 13639
    },
    {
      "epoch": 0.0926748875429477,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009821346040328096,
      "loss": 2.2785,
      "step": 13640
    },
    {
      "epoch": 0.09268168188954176,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009821317537106156,
      "loss": 2.1246,
      "step": 13641
    },
    {
      "epoch": 0.09268847623613581,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009821289031652003,
      "loss": 2.1425,
      "step": 13642
    },
    {
      "epoch": 0.09269527058272987,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009821260523965648,
      "loss": 2.111,
      "step": 13643
    },
    {
      "epoch": 0.09270206492932394,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009821232014047105,
      "loss": 2.3593,
      "step": 13644
    },
    {
      "epoch": 0.09270885927591799,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009821203501896385,
      "loss": 2.1225,
      "step": 13645
    },
    {
      "epoch": 0.09271565362251205,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009821174987513506,
      "loss": 2.2375,
      "step": 13646
    },
    {
      "epoch": 0.0927224479691061,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009821146470898477,
      "loss": 2.3297,
      "step": 13647
    },
    {
      "epoch": 0.09272924231570016,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009821117952051313,
      "loss": 2.306,
      "step": 13648
    },
    {
      "epoch": 0.09273603666229423,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009821089430972025,
      "loss": 2.1666,
      "step": 13649
    },
    {
      "epoch": 0.09274283100888828,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009821060907660628,
      "loss": 2.3034,
      "step": 13650
    },
    {
      "epoch": 0.09274962535548234,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009821032382117136,
      "loss": 2.1417,
      "step": 13651
    },
    {
      "epoch": 0.09275641970207639,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009821003854341562,
      "loss": 2.2431,
      "step": 13652
    },
    {
      "epoch": 0.09276321404867045,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009820975324333918,
      "loss": 2.3368,
      "step": 13653
    },
    {
      "epoch": 0.09277000839526452,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009820946792094215,
      "loss": 2.0675,
      "step": 13654
    },
    {
      "epoch": 0.09277680274185857,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009820918257622472,
      "loss": 2.2474,
      "step": 13655
    },
    {
      "epoch": 0.09278359708845263,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00098208897209187,
      "loss": 2.169,
      "step": 13656
    },
    {
      "epoch": 0.09279039143504668,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000982086118198291,
      "loss": 2.185,
      "step": 13657
    },
    {
      "epoch": 0.09279718578164074,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009820832640815116,
      "loss": 2.1182,
      "step": 13658
    },
    {
      "epoch": 0.0928039801282348,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009820804097415332,
      "loss": 2.2595,
      "step": 13659
    },
    {
      "epoch": 0.09281077447482886,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009820775551783569,
      "loss": 2.3124,
      "step": 13660
    },
    {
      "epoch": 0.09281756882142292,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009820747003919845,
      "loss": 2.154,
      "step": 13661
    },
    {
      "epoch": 0.09282436316801697,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000982071845382417,
      "loss": 2.2751,
      "step": 13662
    },
    {
      "epoch": 0.09283115751461103,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009820689901496558,
      "loss": 2.1682,
      "step": 13663
    },
    {
      "epoch": 0.09283795186120508,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000982066134693702,
      "loss": 2.3914,
      "step": 13664
    },
    {
      "epoch": 0.09284474620779914,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009820632790145572,
      "loss": 2.2431,
      "step": 13665
    },
    {
      "epoch": 0.09285154055439321,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009820604231122227,
      "loss": 2.1819,
      "step": 13666
    },
    {
      "epoch": 0.09285833490098726,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009820575669866995,
      "loss": 2.0776,
      "step": 13667
    },
    {
      "epoch": 0.09286512924758132,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009820547106379896,
      "loss": 2.2069,
      "step": 13668
    },
    {
      "epoch": 0.09287192359417537,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009820518540660935,
      "loss": 2.2619,
      "step": 13669
    },
    {
      "epoch": 0.09287871794076943,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000982048997271013,
      "loss": 2.3574,
      "step": 13670
    },
    {
      "epoch": 0.0928855122873635,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009820461402527496,
      "loss": 2.2114,
      "step": 13671
    },
    {
      "epoch": 0.09289230663395755,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009820432830113042,
      "loss": 2.186,
      "step": 13672
    },
    {
      "epoch": 0.09289910098055161,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009820404255466782,
      "loss": 2.2653,
      "step": 13673
    },
    {
      "epoch": 0.09290589532714566,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000982037567858873,
      "loss": 2.3246,
      "step": 13674
    },
    {
      "epoch": 0.09291268967373972,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00098203470994789,
      "loss": 2.312,
      "step": 13675
    },
    {
      "epoch": 0.09291948402033379,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009820318518137306,
      "loss": 2.2108,
      "step": 13676
    },
    {
      "epoch": 0.09292627836692784,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009820289934563957,
      "loss": 2.2465,
      "step": 13677
    },
    {
      "epoch": 0.0929330727135219,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009820261348758871,
      "loss": 2.2692,
      "step": 13678
    },
    {
      "epoch": 0.09293986706011595,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000982023276072206,
      "loss": 2.113,
      "step": 13679
    },
    {
      "epoch": 0.09294666140671001,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009820204170453534,
      "loss": 2.1559,
      "step": 13680
    },
    {
      "epoch": 0.09295345575330408,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000982017557795331,
      "loss": 2.1937,
      "step": 13681
    },
    {
      "epoch": 0.09296025009989813,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00098201469832214,
      "loss": 2.3204,
      "step": 13682
    },
    {
      "epoch": 0.09296704444649219,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009820118386257817,
      "loss": 2.0928,
      "step": 13683
    },
    {
      "epoch": 0.09297383879308624,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009820089787062575,
      "loss": 2.2498,
      "step": 13684
    },
    {
      "epoch": 0.0929806331396803,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009820061185635686,
      "loss": 2.2783,
      "step": 13685
    },
    {
      "epoch": 0.09298742748627435,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009820032581977162,
      "loss": 2.4756,
      "step": 13686
    },
    {
      "epoch": 0.09299422183286842,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009820003976087023,
      "loss": 2.2789,
      "step": 13687
    },
    {
      "epoch": 0.09300101617946248,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009819975367965272,
      "loss": 2.1551,
      "step": 13688
    },
    {
      "epoch": 0.09300781052605653,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009819946757611932,
      "loss": 2.1533,
      "step": 13689
    },
    {
      "epoch": 0.09301460487265059,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000981991814502701,
      "loss": 2.2497,
      "step": 13690
    },
    {
      "epoch": 0.09302139921924464,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000981988953021052,
      "loss": 2.1823,
      "step": 13691
    },
    {
      "epoch": 0.0930281935658387,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000981986091316248,
      "loss": 2.2154,
      "step": 13692
    },
    {
      "epoch": 0.09303498791243277,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009819832293882897,
      "loss": 2.1863,
      "step": 13693
    },
    {
      "epoch": 0.09304178225902682,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009819803672371787,
      "loss": 2.2021,
      "step": 13694
    },
    {
      "epoch": 0.09304857660562088,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009819775048629164,
      "loss": 2.2888,
      "step": 13695
    },
    {
      "epoch": 0.09305537095221493,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000981974642265504,
      "loss": 2.2147,
      "step": 13696
    },
    {
      "epoch": 0.093062165298809,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000981971779444943,
      "loss": 2.3068,
      "step": 13697
    },
    {
      "epoch": 0.09306895964540306,
      "grad_norm": 0.125,
      "learning_rate": 0.0009819689164012344,
      "loss": 2.169,
      "step": 13698
    },
    {
      "epoch": 0.0930757539919971,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009819660531343796,
      "loss": 2.3152,
      "step": 13699
    },
    {
      "epoch": 0.09308254833859117,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009819631896443803,
      "loss": 2.142,
      "step": 13700
    },
    {
      "epoch": 0.09308934268518522,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009819603259312375,
      "loss": 2.2308,
      "step": 13701
    },
    {
      "epoch": 0.09309613703177928,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009819574619949526,
      "loss": 2.4019,
      "step": 13702
    },
    {
      "epoch": 0.09310293137837335,
      "grad_norm": 0.125,
      "learning_rate": 0.000981954597835527,
      "loss": 2.0649,
      "step": 13703
    },
    {
      "epoch": 0.0931097257249674,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009819517334529618,
      "loss": 2.1039,
      "step": 13704
    },
    {
      "epoch": 0.09311652007156146,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009819488688472585,
      "loss": 2.0918,
      "step": 13705
    },
    {
      "epoch": 0.09312331441815551,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009819460040184182,
      "loss": 2.2871,
      "step": 13706
    },
    {
      "epoch": 0.09313010876474957,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009819431389664428,
      "loss": 2.2369,
      "step": 13707
    },
    {
      "epoch": 0.09313690311134362,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000981940273691333,
      "loss": 2.383,
      "step": 13708
    },
    {
      "epoch": 0.09314369745793769,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009819374081930904,
      "loss": 2.1596,
      "step": 13709
    },
    {
      "epoch": 0.09315049180453175,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009819345424717165,
      "loss": 2.236,
      "step": 13710
    },
    {
      "epoch": 0.0931572861511258,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009819316765272124,
      "loss": 2.2677,
      "step": 13711
    },
    {
      "epoch": 0.09316408049771986,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009819288103595793,
      "loss": 2.2701,
      "step": 13712
    },
    {
      "epoch": 0.09317087484431391,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.000981925943968819,
      "loss": 2.0384,
      "step": 13713
    },
    {
      "epoch": 0.09317766919090797,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000981923077354932,
      "loss": 2.2706,
      "step": 13714
    },
    {
      "epoch": 0.09318446353750204,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009819202105179204,
      "loss": 2.2199,
      "step": 13715
    },
    {
      "epoch": 0.09319125788409609,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009819173434577853,
      "loss": 2.2298,
      "step": 13716
    },
    {
      "epoch": 0.09319805223069015,
      "grad_norm": 0.140625,
      "learning_rate": 0.000981914476174528,
      "loss": 2.1099,
      "step": 13717
    },
    {
      "epoch": 0.0932048465772842,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00098191160866815,
      "loss": 2.3036,
      "step": 13718
    },
    {
      "epoch": 0.09321164092387826,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000981908740938652,
      "loss": 2.1205,
      "step": 13719
    },
    {
      "epoch": 0.09321843527047233,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000981905872986036,
      "loss": 2.0676,
      "step": 13720
    },
    {
      "epoch": 0.09322522961706638,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009819030048103033,
      "loss": 2.2262,
      "step": 13721
    },
    {
      "epoch": 0.09323202396366044,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009819001364114548,
      "loss": 2.252,
      "step": 13722
    },
    {
      "epoch": 0.09323881831025449,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981897267789492,
      "loss": 2.2791,
      "step": 13723
    },
    {
      "epoch": 0.09324561265684855,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009818943989444164,
      "loss": 2.3294,
      "step": 13724
    },
    {
      "epoch": 0.09325240700344262,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009818915298762294,
      "loss": 2.2241,
      "step": 13725
    },
    {
      "epoch": 0.09325920135003667,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000981888660584932,
      "loss": 2.4423,
      "step": 13726
    },
    {
      "epoch": 0.09326599569663073,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009818857910705255,
      "loss": 2.0663,
      "step": 13727
    },
    {
      "epoch": 0.09327279004322478,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009818829213330115,
      "loss": 2.3051,
      "step": 13728
    },
    {
      "epoch": 0.09327958438981884,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009818800513723912,
      "loss": 2.1695,
      "step": 13729
    },
    {
      "epoch": 0.0932863787364129,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009818771811886662,
      "loss": 2.0462,
      "step": 13730
    },
    {
      "epoch": 0.09329317308300696,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009818743107818375,
      "loss": 2.1901,
      "step": 13731
    },
    {
      "epoch": 0.09329996742960102,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009818714401519065,
      "loss": 2.1733,
      "step": 13732
    },
    {
      "epoch": 0.09330676177619507,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009818685692988744,
      "loss": 2.2048,
      "step": 13733
    },
    {
      "epoch": 0.09331355612278913,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009818656982227428,
      "loss": 2.0171,
      "step": 13734
    },
    {
      "epoch": 0.09332035046938318,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981862826923513,
      "loss": 2.3467,
      "step": 13735
    },
    {
      "epoch": 0.09332714481597724,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000981859955401186,
      "loss": 2.2716,
      "step": 13736
    },
    {
      "epoch": 0.09333393916257131,
      "grad_norm": 0.125,
      "learning_rate": 0.0009818570836557636,
      "loss": 2.1233,
      "step": 13737
    },
    {
      "epoch": 0.09334073350916536,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009818542116872466,
      "loss": 2.1886,
      "step": 13738
    },
    {
      "epoch": 0.09334752785575942,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000981851339495637,
      "loss": 2.1221,
      "step": 13739
    },
    {
      "epoch": 0.09335432220235347,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009818484670809354,
      "loss": 2.1359,
      "step": 13740
    },
    {
      "epoch": 0.09336111654894753,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009818455944431437,
      "loss": 1.9959,
      "step": 13741
    },
    {
      "epoch": 0.0933679108955416,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000981842721582263,
      "loss": 2.2069,
      "step": 13742
    },
    {
      "epoch": 0.09337470524213565,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009818398484982946,
      "loss": 2.0524,
      "step": 13743
    },
    {
      "epoch": 0.09338149958872971,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009818369751912399,
      "loss": 2.3712,
      "step": 13744
    },
    {
      "epoch": 0.09338829393532376,
      "grad_norm": 0.337890625,
      "learning_rate": 0.0009818341016611001,
      "loss": 2.3706,
      "step": 13745
    },
    {
      "epoch": 0.09339508828191782,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009818312279078767,
      "loss": 2.1547,
      "step": 13746
    },
    {
      "epoch": 0.09340188262851189,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000981828353931571,
      "loss": 2.1271,
      "step": 13747
    },
    {
      "epoch": 0.09340867697510594,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009818254797321843,
      "loss": 2.3161,
      "step": 13748
    },
    {
      "epoch": 0.0934154713217,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000981822605309718,
      "loss": 2.132,
      "step": 13749
    },
    {
      "epoch": 0.09342226566829405,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009818197306641731,
      "loss": 2.1958,
      "step": 13750
    },
    {
      "epoch": 0.09342906001488811,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009818168557955513,
      "loss": 2.2936,
      "step": 13751
    },
    {
      "epoch": 0.09343585436148218,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000981813980703854,
      "loss": 2.1734,
      "step": 13752
    },
    {
      "epoch": 0.09344264870807623,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009818111053890822,
      "loss": 2.1776,
      "step": 13753
    },
    {
      "epoch": 0.09344944305467029,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009818082298512375,
      "loss": 2.1187,
      "step": 13754
    },
    {
      "epoch": 0.09345623740126434,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000981805354090321,
      "loss": 2.2998,
      "step": 13755
    },
    {
      "epoch": 0.0934630317478584,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000981802478106334,
      "loss": 2.2975,
      "step": 13756
    },
    {
      "epoch": 0.09346982609445245,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009817996018992782,
      "loss": 2.2783,
      "step": 13757
    },
    {
      "epoch": 0.09347662044104652,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009817967254691546,
      "loss": 2.0651,
      "step": 13758
    },
    {
      "epoch": 0.09348341478764058,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009817938488159647,
      "loss": 2.1131,
      "step": 13759
    },
    {
      "epoch": 0.09349020913423463,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009817909719397099,
      "loss": 2.2696,
      "step": 13760
    },
    {
      "epoch": 0.09349700348082869,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009817880948403911,
      "loss": 2.1586,
      "step": 13761
    },
    {
      "epoch": 0.09350379782742274,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00098178521751801,
      "loss": 2.1124,
      "step": 13762
    },
    {
      "epoch": 0.0935105921740168,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000981782339972568,
      "loss": 2.3157,
      "step": 13763
    },
    {
      "epoch": 0.09351738652061087,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009817794622040664,
      "loss": 2.1554,
      "step": 13764
    },
    {
      "epoch": 0.09352418086720492,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009817765842125062,
      "loss": 2.3381,
      "step": 13765
    },
    {
      "epoch": 0.09353097521379898,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000981773705997889,
      "loss": 2.2732,
      "step": 13766
    },
    {
      "epoch": 0.09353776956039303,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009817708275602164,
      "loss": 2.1735,
      "step": 13767
    },
    {
      "epoch": 0.0935445639069871,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000981767948899489,
      "loss": 2.122,
      "step": 13768
    },
    {
      "epoch": 0.09355135825358116,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009817650700157088,
      "loss": 2.1221,
      "step": 13769
    },
    {
      "epoch": 0.0935581526001752,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009817621909088768,
      "loss": 2.326,
      "step": 13770
    },
    {
      "epoch": 0.09356494694676927,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009817593115789946,
      "loss": 2.2464,
      "step": 13771
    },
    {
      "epoch": 0.09357174129336332,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009817564320260632,
      "loss": 2.4355,
      "step": 13772
    },
    {
      "epoch": 0.09357853563995738,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009817535522500841,
      "loss": 2.0672,
      "step": 13773
    },
    {
      "epoch": 0.09358532998655145,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009817506722510588,
      "loss": 2.2808,
      "step": 13774
    },
    {
      "epoch": 0.0935921243331455,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000981747792028988,
      "loss": 2.2335,
      "step": 13775
    },
    {
      "epoch": 0.09359891867973956,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009817449115838738,
      "loss": 2.2815,
      "step": 13776
    },
    {
      "epoch": 0.09360571302633361,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009817420309157175,
      "loss": 2.2056,
      "step": 13777
    },
    {
      "epoch": 0.09361250737292767,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009817391500245199,
      "loss": 2.1794,
      "step": 13778
    },
    {
      "epoch": 0.09361930171952174,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009817362689102825,
      "loss": 2.2694,
      "step": 13779
    },
    {
      "epoch": 0.09362609606611579,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000981733387573007,
      "loss": 2.1692,
      "step": 13780
    },
    {
      "epoch": 0.09363289041270985,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009817305060126943,
      "loss": 2.1622,
      "step": 13781
    },
    {
      "epoch": 0.0936396847593039,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000981727624229346,
      "loss": 2.174,
      "step": 13782
    },
    {
      "epoch": 0.09364647910589796,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000981724742222963,
      "loss": 2.1627,
      "step": 13783
    },
    {
      "epoch": 0.09365327345249201,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009817218599935471,
      "loss": 2.2487,
      "step": 13784
    },
    {
      "epoch": 0.09366006779908607,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009817189775410996,
      "loss": 2.2026,
      "step": 13785
    },
    {
      "epoch": 0.09366686214568014,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009817160948656218,
      "loss": 2.1647,
      "step": 13786
    },
    {
      "epoch": 0.09367365649227419,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000981713211967115,
      "loss": 2.2587,
      "step": 13787
    },
    {
      "epoch": 0.09368045083886825,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009817103288455803,
      "loss": 2.2468,
      "step": 13788
    },
    {
      "epoch": 0.0936872451854623,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009817074455010194,
      "loss": 2.081,
      "step": 13789
    },
    {
      "epoch": 0.09369403953205636,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009817045619334335,
      "loss": 2.1064,
      "step": 13790
    },
    {
      "epoch": 0.09370083387865043,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009817016781428238,
      "loss": 2.0593,
      "step": 13791
    },
    {
      "epoch": 0.09370762822524448,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009816987941291917,
      "loss": 2.2545,
      "step": 13792
    },
    {
      "epoch": 0.09371442257183854,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009816959098925386,
      "loss": 2.1231,
      "step": 13793
    },
    {
      "epoch": 0.09372121691843259,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000981693025432866,
      "loss": 2.2979,
      "step": 13794
    },
    {
      "epoch": 0.09372801126502665,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009816901407501749,
      "loss": 2.1311,
      "step": 13795
    },
    {
      "epoch": 0.09373480561162072,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009816872558444669,
      "loss": 2.176,
      "step": 13796
    },
    {
      "epoch": 0.09374159995821477,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000981684370715743,
      "loss": 2.1112,
      "step": 13797
    },
    {
      "epoch": 0.09374839430480883,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981681485364005,
      "loss": 2.1944,
      "step": 13798
    },
    {
      "epoch": 0.09375518865140288,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009816785997892538,
      "loss": 2.1309,
      "step": 13799
    },
    {
      "epoch": 0.09376198299799694,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009816757139914911,
      "loss": 2.1513,
      "step": 13800
    },
    {
      "epoch": 0.093768777344591,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000981672827970718,
      "loss": 2.1793,
      "step": 13801
    },
    {
      "epoch": 0.09377557169118506,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981669941726936,
      "loss": 2.1672,
      "step": 13802
    },
    {
      "epoch": 0.09378236603777912,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009816670552601462,
      "loss": 2.3491,
      "step": 13803
    },
    {
      "epoch": 0.09378916038437317,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00098166416857035,
      "loss": 2.1086,
      "step": 13804
    },
    {
      "epoch": 0.09379595473096723,
      "grad_norm": 0.140625,
      "learning_rate": 0.000981661281657549,
      "loss": 2.0842,
      "step": 13805
    },
    {
      "epoch": 0.09380274907756128,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009816583945217443,
      "loss": 2.2724,
      "step": 13806
    },
    {
      "epoch": 0.09380954342415534,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009816555071629372,
      "loss": 2.1887,
      "step": 13807
    },
    {
      "epoch": 0.09381633777074941,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009816526195811292,
      "loss": 2.3228,
      "step": 13808
    },
    {
      "epoch": 0.09382313211734346,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009816497317763216,
      "loss": 2.1729,
      "step": 13809
    },
    {
      "epoch": 0.09382992646393752,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009816468437485156,
      "loss": 2.1963,
      "step": 13810
    },
    {
      "epoch": 0.09383672081053157,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009816439554977126,
      "loss": 2.227,
      "step": 13811
    },
    {
      "epoch": 0.09384351515712563,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009816410670239139,
      "loss": 2.3083,
      "step": 13812
    },
    {
      "epoch": 0.0938503095037197,
      "grad_norm": 0.140625,
      "learning_rate": 0.000981638178327121,
      "loss": 2.2224,
      "step": 13813
    },
    {
      "epoch": 0.09385710385031375,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009816352894073354,
      "loss": 2.2248,
      "step": 13814
    },
    {
      "epoch": 0.09386389819690781,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000981632400264558,
      "loss": 2.1817,
      "step": 13815
    },
    {
      "epoch": 0.09387069254350186,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009816295108987902,
      "loss": 2.2167,
      "step": 13816
    },
    {
      "epoch": 0.09387748689009592,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009816266213100336,
      "loss": 2.2333,
      "step": 13817
    },
    {
      "epoch": 0.09388428123668999,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009816237314982893,
      "loss": 2.1247,
      "step": 13818
    },
    {
      "epoch": 0.09389107558328404,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009816208414635589,
      "loss": 2.1063,
      "step": 13819
    },
    {
      "epoch": 0.0938978699298781,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009816179512058434,
      "loss": 2.3713,
      "step": 13820
    },
    {
      "epoch": 0.09390466427647215,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009816150607251442,
      "loss": 2.3202,
      "step": 13821
    },
    {
      "epoch": 0.09391145862306621,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000981612170021463,
      "loss": 2.1495,
      "step": 13822
    },
    {
      "epoch": 0.09391825296966028,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009816092790948007,
      "loss": 2.2669,
      "step": 13823
    },
    {
      "epoch": 0.09392504731625433,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009816063879451592,
      "loss": 2.1597,
      "step": 13824
    },
    {
      "epoch": 0.09393184166284839,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000981603496572539,
      "loss": 2.3676,
      "step": 13825
    },
    {
      "epoch": 0.09393863600944244,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009816006049769422,
      "loss": 2.2665,
      "step": 13826
    },
    {
      "epoch": 0.0939454303560365,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009815977131583697,
      "loss": 2.3447,
      "step": 13827
    },
    {
      "epoch": 0.09395222470263055,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009815948211168229,
      "loss": 2.1069,
      "step": 13828
    },
    {
      "epoch": 0.09395901904922462,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009815919288523034,
      "loss": 2.2434,
      "step": 13829
    },
    {
      "epoch": 0.09396581339581868,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009815890363648121,
      "loss": 2.2917,
      "step": 13830
    },
    {
      "epoch": 0.09397260774241273,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000981586143654351,
      "loss": 2.4084,
      "step": 13831
    },
    {
      "epoch": 0.09397940208900679,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009815832507209209,
      "loss": 2.2422,
      "step": 13832
    },
    {
      "epoch": 0.09398619643560084,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981580357564523,
      "loss": 2.3008,
      "step": 13833
    },
    {
      "epoch": 0.0939929907821949,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009815774641851592,
      "loss": 2.1861,
      "step": 13834
    },
    {
      "epoch": 0.09399978512878897,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009815745705828306,
      "loss": 2.2042,
      "step": 13835
    },
    {
      "epoch": 0.09400657947538302,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009815716767575381,
      "loss": 2.1616,
      "step": 13836
    },
    {
      "epoch": 0.09401337382197708,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009815687827092838,
      "loss": 2.0848,
      "step": 13837
    },
    {
      "epoch": 0.09402016816857113,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009815658884380686,
      "loss": 2.2058,
      "step": 13838
    },
    {
      "epoch": 0.0940269625151652,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009815629939438938,
      "loss": 2.2644,
      "step": 13839
    },
    {
      "epoch": 0.09403375686175926,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981560099226761,
      "loss": 2.3293,
      "step": 13840
    },
    {
      "epoch": 0.0940405512083533,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009815572042866714,
      "loss": 2.2213,
      "step": 13841
    },
    {
      "epoch": 0.09404734555494737,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009815543091236262,
      "loss": 2.2582,
      "step": 13842
    },
    {
      "epoch": 0.09405413990154142,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009815514137376268,
      "loss": 2.2003,
      "step": 13843
    },
    {
      "epoch": 0.09406093424813548,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000981548518128675,
      "loss": 2.3603,
      "step": 13844
    },
    {
      "epoch": 0.09406772859472955,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009815456222967713,
      "loss": 2.1233,
      "step": 13845
    },
    {
      "epoch": 0.0940745229413236,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000981542726241918,
      "loss": 2.295,
      "step": 13846
    },
    {
      "epoch": 0.09408131728791766,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009815398299641155,
      "loss": 2.1894,
      "step": 13847
    },
    {
      "epoch": 0.09408811163451171,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009815369334633657,
      "loss": 2.2768,
      "step": 13848
    },
    {
      "epoch": 0.09409490598110577,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00098153403673967,
      "loss": 2.3376,
      "step": 13849
    },
    {
      "epoch": 0.09410170032769984,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009815311397930293,
      "loss": 2.2815,
      "step": 13850
    },
    {
      "epoch": 0.09410849467429389,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009815282426234454,
      "loss": 2.119,
      "step": 13851
    },
    {
      "epoch": 0.09411528902088795,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009815253452309192,
      "loss": 2.2379,
      "step": 13852
    },
    {
      "epoch": 0.094122083367482,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009815224476154526,
      "loss": 2.3156,
      "step": 13853
    },
    {
      "epoch": 0.09412887771407606,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009815195497770463,
      "loss": 2.3088,
      "step": 13854
    },
    {
      "epoch": 0.09413567206067011,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009815166517157022,
      "loss": 2.1437,
      "step": 13855
    },
    {
      "epoch": 0.09414246640726417,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009815137534314212,
      "loss": 2.2011,
      "step": 13856
    },
    {
      "epoch": 0.09414926075385824,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009815108549242053,
      "loss": 2.0512,
      "step": 13857
    },
    {
      "epoch": 0.09415605510045229,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009815079561940549,
      "loss": 2.2708,
      "step": 13858
    },
    {
      "epoch": 0.09416284944704635,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000981505057240972,
      "loss": 2.2425,
      "step": 13859
    },
    {
      "epoch": 0.0941696437936404,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009815021580649578,
      "loss": 2.169,
      "step": 13860
    },
    {
      "epoch": 0.09417643814023446,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009814992586660138,
      "loss": 2.323,
      "step": 13861
    },
    {
      "epoch": 0.09418323248682853,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000981496359044141,
      "loss": 2.2918,
      "step": 13862
    },
    {
      "epoch": 0.09419002683342258,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000981493459199341,
      "loss": 2.2568,
      "step": 13863
    },
    {
      "epoch": 0.09419682118001664,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981490559131615,
      "loss": 2.3738,
      "step": 13864
    },
    {
      "epoch": 0.09420361552661069,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009814876588409643,
      "loss": 2.2338,
      "step": 13865
    },
    {
      "epoch": 0.09421040987320475,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009814847583273902,
      "loss": 2.2715,
      "step": 13866
    },
    {
      "epoch": 0.09421720421979882,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009814818575908945,
      "loss": 2.1906,
      "step": 13867
    },
    {
      "epoch": 0.09422399856639287,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.000981478956631478,
      "loss": 2.0941,
      "step": 13868
    },
    {
      "epoch": 0.09423079291298693,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009814760554491425,
      "loss": 2.3177,
      "step": 13869
    },
    {
      "epoch": 0.09423758725958098,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009814731540438888,
      "loss": 2.2731,
      "step": 13870
    },
    {
      "epoch": 0.09424438160617504,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009814702524157187,
      "loss": 2.2914,
      "step": 13871
    },
    {
      "epoch": 0.0942511759527691,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009814673505646335,
      "loss": 2.146,
      "step": 13872
    },
    {
      "epoch": 0.09425797029936316,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009814644484906342,
      "loss": 2.2036,
      "step": 13873
    },
    {
      "epoch": 0.09426476464595722,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009814615461937225,
      "loss": 2.1796,
      "step": 13874
    },
    {
      "epoch": 0.09427155899255127,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009814586436738997,
      "loss": 2.1656,
      "step": 13875
    },
    {
      "epoch": 0.09427835333914533,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000981455740931167,
      "loss": 2.1975,
      "step": 13876
    },
    {
      "epoch": 0.09428514768573938,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009814528379655258,
      "loss": 2.0737,
      "step": 13877
    },
    {
      "epoch": 0.09429194203233345,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009814499347769775,
      "loss": 2.1889,
      "step": 13878
    },
    {
      "epoch": 0.09429873637892751,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009814470313655233,
      "loss": 2.2913,
      "step": 13879
    },
    {
      "epoch": 0.09430553072552156,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009814441277311648,
      "loss": 2.2748,
      "step": 13880
    },
    {
      "epoch": 0.09431232507211562,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000981441223873903,
      "loss": 2.2246,
      "step": 13881
    },
    {
      "epoch": 0.09431911941870967,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009814383197937395,
      "loss": 2.2564,
      "step": 13882
    },
    {
      "epoch": 0.09432591376530373,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009814354154906756,
      "loss": 2.2736,
      "step": 13883
    },
    {
      "epoch": 0.0943327081118978,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009814325109647126,
      "loss": 2.1842,
      "step": 13884
    },
    {
      "epoch": 0.09433950245849185,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000981429606215852,
      "loss": 2.236,
      "step": 13885
    },
    {
      "epoch": 0.09434629680508591,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000981426701244095,
      "loss": 2.2865,
      "step": 13886
    },
    {
      "epoch": 0.09435309115167996,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009814237960494427,
      "loss": 2.2991,
      "step": 13887
    },
    {
      "epoch": 0.09435988549827402,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009814208906318968,
      "loss": 2.3198,
      "step": 13888
    },
    {
      "epoch": 0.09436667984486809,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009814179849914587,
      "loss": 2.1673,
      "step": 13889
    },
    {
      "epoch": 0.09437347419146214,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009814150791281296,
      "loss": 2.1148,
      "step": 13890
    },
    {
      "epoch": 0.0943802685380562,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009814121730419107,
      "loss": 1.9788,
      "step": 13891
    },
    {
      "epoch": 0.09438706288465025,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009814092667328035,
      "loss": 2.2435,
      "step": 13892
    },
    {
      "epoch": 0.09439385723124431,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009814063602008095,
      "loss": 2.0715,
      "step": 13893
    },
    {
      "epoch": 0.09440065157783838,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009814034534459295,
      "loss": 2.2362,
      "step": 13894
    },
    {
      "epoch": 0.09440744592443243,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009814005464681655,
      "loss": 2.2341,
      "step": 13895
    },
    {
      "epoch": 0.09441424027102649,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009813976392675187,
      "loss": 2.1835,
      "step": 13896
    },
    {
      "epoch": 0.09442103461762054,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.00098139473184399,
      "loss": 2.0982,
      "step": 13897
    },
    {
      "epoch": 0.0944278289642146,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009813918241975811,
      "loss": 2.2982,
      "step": 13898
    },
    {
      "epoch": 0.09443462331080867,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009813889163282933,
      "loss": 2.2635,
      "step": 13899
    },
    {
      "epoch": 0.09444141765740272,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000981386008236128,
      "loss": 2.1022,
      "step": 13900
    },
    {
      "epoch": 0.09444821200399678,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009813830999210865,
      "loss": 2.1364,
      "step": 13901
    },
    {
      "epoch": 0.09445500635059083,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009813801913831704,
      "loss": 2.2691,
      "step": 13902
    },
    {
      "epoch": 0.09446180069718489,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009813772826223807,
      "loss": 2.1672,
      "step": 13903
    },
    {
      "epoch": 0.09446859504377894,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009813743736387186,
      "loss": 2.2417,
      "step": 13904
    },
    {
      "epoch": 0.094475389390373,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009813714644321858,
      "loss": 2.1834,
      "step": 13905
    },
    {
      "epoch": 0.09448218373696707,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009813685550027836,
      "loss": 2.2479,
      "step": 13906
    },
    {
      "epoch": 0.09448897808356112,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009813656453505131,
      "loss": 2.4349,
      "step": 13907
    },
    {
      "epoch": 0.09449577243015518,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009813627354753759,
      "loss": 2.2191,
      "step": 13908
    },
    {
      "epoch": 0.09450256677674923,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009813598253773734,
      "loss": 2.2236,
      "step": 13909
    },
    {
      "epoch": 0.0945093611233433,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009813569150565068,
      "loss": 2.2749,
      "step": 13910
    },
    {
      "epoch": 0.09451615546993736,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009813540045127771,
      "loss": 2.2508,
      "step": 13911
    },
    {
      "epoch": 0.0945229498165314,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009813510937461865,
      "loss": 2.3459,
      "step": 13912
    },
    {
      "epoch": 0.09452974416312547,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009813481827567357,
      "loss": 2.2511,
      "step": 13913
    },
    {
      "epoch": 0.09453653850971952,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000981345271544426,
      "loss": 2.3036,
      "step": 13914
    },
    {
      "epoch": 0.09454333285631358,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000981342360109259,
      "loss": 2.1552,
      "step": 13915
    },
    {
      "epoch": 0.09455012720290765,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009813394484512363,
      "loss": 2.2574,
      "step": 13916
    },
    {
      "epoch": 0.0945569215495017,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009813365365703589,
      "loss": 2.2287,
      "step": 13917
    },
    {
      "epoch": 0.09456371589609576,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000981333624466628,
      "loss": 2.2159,
      "step": 13918
    },
    {
      "epoch": 0.09457051024268981,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009813307121400452,
      "loss": 2.1645,
      "step": 13919
    },
    {
      "epoch": 0.09457730458928387,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009813277995906118,
      "loss": 2.2063,
      "step": 13920
    },
    {
      "epoch": 0.09458409893587794,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009813248868183293,
      "loss": 2.2998,
      "step": 13921
    },
    {
      "epoch": 0.09459089328247199,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009813219738231986,
      "loss": 2.1728,
      "step": 13922
    },
    {
      "epoch": 0.09459768762906605,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009813190606052216,
      "loss": 2.101,
      "step": 13923
    },
    {
      "epoch": 0.0946044819756601,
      "grad_norm": 0.125,
      "learning_rate": 0.0009813161471643995,
      "loss": 2.0703,
      "step": 13924
    },
    {
      "epoch": 0.09461127632225416,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009813132335007332,
      "loss": 2.3034,
      "step": 13925
    },
    {
      "epoch": 0.09461807066884821,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009813103196142246,
      "loss": 2.1529,
      "step": 13926
    },
    {
      "epoch": 0.09462486501544227,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009813074055048747,
      "loss": 2.222,
      "step": 13927
    },
    {
      "epoch": 0.09463165936203634,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000981304491172685,
      "loss": 2.1817,
      "step": 13928
    },
    {
      "epoch": 0.09463845370863039,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000981301576617657,
      "loss": 2.203,
      "step": 13929
    },
    {
      "epoch": 0.09464524805522445,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000981298661839792,
      "loss": 2.3075,
      "step": 13930
    },
    {
      "epoch": 0.0946520424018185,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009812957468390909,
      "loss": 2.1938,
      "step": 13931
    },
    {
      "epoch": 0.09465883674841256,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009812928316155556,
      "loss": 2.2856,
      "step": 13932
    },
    {
      "epoch": 0.09466563109500663,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009812899161691872,
      "loss": 2.0924,
      "step": 13933
    },
    {
      "epoch": 0.09467242544160068,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000981287000499987,
      "loss": 2.1513,
      "step": 13934
    },
    {
      "epoch": 0.09467921978819474,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009812840846079565,
      "loss": 2.2239,
      "step": 13935
    },
    {
      "epoch": 0.09468601413478879,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000981281168493097,
      "loss": 2.2048,
      "step": 13936
    },
    {
      "epoch": 0.09469280848138285,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009812782521554098,
      "loss": 2.1849,
      "step": 13937
    },
    {
      "epoch": 0.09469960282797692,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009812753355948963,
      "loss": 2.2834,
      "step": 13938
    },
    {
      "epoch": 0.09470639717457097,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009812724188115577,
      "loss": 2.2682,
      "step": 13939
    },
    {
      "epoch": 0.09471319152116503,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009812695018053957,
      "loss": 2.2813,
      "step": 13940
    },
    {
      "epoch": 0.09471998586775908,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009812665845764113,
      "loss": 2.2335,
      "step": 13941
    },
    {
      "epoch": 0.09472678021435314,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000981263667124606,
      "loss": 2.0724,
      "step": 13942
    },
    {
      "epoch": 0.0947335745609472,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009812607494499813,
      "loss": 2.2436,
      "step": 13943
    },
    {
      "epoch": 0.09474036890754126,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009812578315525383,
      "loss": 2.3701,
      "step": 13944
    },
    {
      "epoch": 0.09474716325413532,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009812549134322784,
      "loss": 2.1758,
      "step": 13945
    },
    {
      "epoch": 0.09475395760072937,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000981251995089203,
      "loss": 2.0994,
      "step": 13946
    },
    {
      "epoch": 0.09476075194732343,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009812490765233135,
      "loss": 2.1399,
      "step": 13947
    },
    {
      "epoch": 0.09476754629391748,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000981246157734611,
      "loss": 2.1399,
      "step": 13948
    },
    {
      "epoch": 0.09477434064051155,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009812432387230972,
      "loss": 2.2599,
      "step": 13949
    },
    {
      "epoch": 0.09478113498710561,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009812403194887733,
      "loss": 2.1764,
      "step": 13950
    },
    {
      "epoch": 0.09478792933369966,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009812374000316407,
      "loss": 2.0955,
      "step": 13951
    },
    {
      "epoch": 0.09479472368029372,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009812344803517008,
      "loss": 2.1395,
      "step": 13952
    },
    {
      "epoch": 0.09480151802688777,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009812315604489545,
      "loss": 2.1014,
      "step": 13953
    },
    {
      "epoch": 0.09480831237348183,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000981228640323404,
      "loss": 2.2192,
      "step": 13954
    },
    {
      "epoch": 0.0948151067200759,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009812257199750497,
      "loss": 2.3299,
      "step": 13955
    },
    {
      "epoch": 0.09482190106666995,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009812227994038937,
      "loss": 2.1692,
      "step": 13956
    },
    {
      "epoch": 0.09482869541326401,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009812198786099367,
      "loss": 2.2345,
      "step": 13957
    },
    {
      "epoch": 0.09483548975985806,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009812169575931806,
      "loss": 2.1758,
      "step": 13958
    },
    {
      "epoch": 0.09484228410645212,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009812140363536267,
      "loss": 2.0818,
      "step": 13959
    },
    {
      "epoch": 0.09484907845304619,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000981211114891276,
      "loss": 2.1646,
      "step": 13960
    },
    {
      "epoch": 0.09485587279964024,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009812081932061303,
      "loss": 2.2132,
      "step": 13961
    },
    {
      "epoch": 0.0948626671462343,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009812052712981907,
      "loss": 2.075,
      "step": 13962
    },
    {
      "epoch": 0.09486946149282835,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009812023491674584,
      "loss": 2.1418,
      "step": 13963
    },
    {
      "epoch": 0.09487625583942241,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000981199426813935,
      "loss": 2.2342,
      "step": 13964
    },
    {
      "epoch": 0.09488305018601648,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009811965042376219,
      "loss": 2.3047,
      "step": 13965
    },
    {
      "epoch": 0.09488984453261053,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009811935814385202,
      "loss": 2.1459,
      "step": 13966
    },
    {
      "epoch": 0.09489663887920459,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009811906584166314,
      "loss": 2.0464,
      "step": 13967
    },
    {
      "epoch": 0.09490343322579864,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009811877351719568,
      "loss": 2.1357,
      "step": 13968
    },
    {
      "epoch": 0.0949102275723927,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981184811704498,
      "loss": 2.0364,
      "step": 13969
    },
    {
      "epoch": 0.09491702191898677,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000981181888014256,
      "loss": 2.1501,
      "step": 13970
    },
    {
      "epoch": 0.09492381626558082,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009811789641012323,
      "loss": 2.2728,
      "step": 13971
    },
    {
      "epoch": 0.09493061061217488,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009811760399654284,
      "loss": 2.2156,
      "step": 13972
    },
    {
      "epoch": 0.09493740495876893,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009811731156068454,
      "loss": 2.1818,
      "step": 13973
    },
    {
      "epoch": 0.09494419930536299,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009811701910254847,
      "loss": 2.4111,
      "step": 13974
    },
    {
      "epoch": 0.09495099365195704,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009811672662213477,
      "loss": 2.2011,
      "step": 13975
    },
    {
      "epoch": 0.0949577879985511,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009811643411944358,
      "loss": 2.235,
      "step": 13976
    },
    {
      "epoch": 0.09496458234514517,
      "grad_norm": 0.125,
      "learning_rate": 0.0009811614159447504,
      "loss": 2.059,
      "step": 13977
    },
    {
      "epoch": 0.09497137669173922,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009811584904722928,
      "loss": 2.2154,
      "step": 13978
    },
    {
      "epoch": 0.09497817103833328,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009811555647770641,
      "loss": 2.189,
      "step": 13979
    },
    {
      "epoch": 0.09498496538492733,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009811526388590662,
      "loss": 2.1662,
      "step": 13980
    },
    {
      "epoch": 0.0949917597315214,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009811497127183,
      "loss": 2.1899,
      "step": 13981
    },
    {
      "epoch": 0.09499855407811546,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009811467863547671,
      "loss": 2.2391,
      "step": 13982
    },
    {
      "epoch": 0.09500534842470951,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009811438597684687,
      "loss": 2.1617,
      "step": 13983
    },
    {
      "epoch": 0.09501214277130357,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000981140932959406,
      "loss": 2.0114,
      "step": 13984
    },
    {
      "epoch": 0.09501893711789762,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009811380059275808,
      "loss": 2.0855,
      "step": 13985
    },
    {
      "epoch": 0.09502573146449168,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009811350786729943,
      "loss": 2.2905,
      "step": 13986
    },
    {
      "epoch": 0.09503252581108575,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009811321511956476,
      "loss": 2.2352,
      "step": 13987
    },
    {
      "epoch": 0.0950393201576798,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009811292234955422,
      "loss": 2.1028,
      "step": 13988
    },
    {
      "epoch": 0.09504611450427386,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009811262955726795,
      "loss": 2.2531,
      "step": 13989
    },
    {
      "epoch": 0.09505290885086791,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000981123367427061,
      "loss": 2.2025,
      "step": 13990
    },
    {
      "epoch": 0.09505970319746197,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009811204390586877,
      "loss": 2.3366,
      "step": 13991
    },
    {
      "epoch": 0.09506649754405604,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009811175104675613,
      "loss": 2.2068,
      "step": 13992
    },
    {
      "epoch": 0.09507329189065009,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009811145816536828,
      "loss": 2.252,
      "step": 13993
    },
    {
      "epoch": 0.09508008623724415,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000981111652617054,
      "loss": 2.2601,
      "step": 13994
    },
    {
      "epoch": 0.0950868805838382,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009811087233576761,
      "loss": 2.1615,
      "step": 13995
    },
    {
      "epoch": 0.09509367493043226,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00098110579387555,
      "loss": 2.2294,
      "step": 13996
    },
    {
      "epoch": 0.09510046927702631,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000981102864170678,
      "loss": 2.1545,
      "step": 13997
    },
    {
      "epoch": 0.09510726362362038,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009810999342430603,
      "loss": 2.149,
      "step": 13998
    },
    {
      "epoch": 0.09511405797021444,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000981097004092699,
      "loss": 2.1622,
      "step": 13999
    },
    {
      "epoch": 0.09512085231680849,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009810940737195954,
      "loss": 2.2035,
      "step": 14000
    },
    {
      "epoch": 0.09512764666340255,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009810911431237508,
      "loss": 2.2429,
      "step": 14001
    },
    {
      "epoch": 0.0951344410099966,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009810882123051665,
      "loss": 2.1578,
      "step": 14002
    },
    {
      "epoch": 0.09514123535659066,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009810852812638438,
      "loss": 2.1885,
      "step": 14003
    },
    {
      "epoch": 0.09514802970318473,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009810823499997842,
      "loss": 2.3091,
      "step": 14004
    },
    {
      "epoch": 0.09515482404977878,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000981079418512989,
      "loss": 2.3397,
      "step": 14005
    },
    {
      "epoch": 0.09516161839637284,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009810764868034593,
      "loss": 2.1009,
      "step": 14006
    },
    {
      "epoch": 0.09516841274296689,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000981073554871197,
      "loss": 2.1387,
      "step": 14007
    },
    {
      "epoch": 0.09517520708956095,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009810706227162032,
      "loss": 2.134,
      "step": 14008
    },
    {
      "epoch": 0.09518200143615502,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000981067690338479,
      "loss": 2.1773,
      "step": 14009
    },
    {
      "epoch": 0.09518879578274907,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000981064757738026,
      "loss": 2.1669,
      "step": 14010
    },
    {
      "epoch": 0.09519559012934313,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009810618249148455,
      "loss": 2.3444,
      "step": 14011
    },
    {
      "epoch": 0.09520238447593718,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000981058891868939,
      "loss": 2.2543,
      "step": 14012
    },
    {
      "epoch": 0.09520917882253124,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009810559586003078,
      "loss": 2.2112,
      "step": 14013
    },
    {
      "epoch": 0.0952159731691253,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000981053025108953,
      "loss": 2.0912,
      "step": 14014
    },
    {
      "epoch": 0.09522276751571936,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009810500913948763,
      "loss": 2.057,
      "step": 14015
    },
    {
      "epoch": 0.09522956186231342,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000981047157458079,
      "loss": 2.2487,
      "step": 14016
    },
    {
      "epoch": 0.09523635620890747,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009810442232985623,
      "loss": 2.1969,
      "step": 14017
    },
    {
      "epoch": 0.09524315055550153,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009810412889163277,
      "loss": 2.277,
      "step": 14018
    },
    {
      "epoch": 0.0952499449020956,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009810383543113763,
      "loss": 2.1797,
      "step": 14019
    },
    {
      "epoch": 0.09525673924868965,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009810354194837098,
      "loss": 2.2312,
      "step": 14020
    },
    {
      "epoch": 0.09526353359528371,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009810324844333293,
      "loss": 2.2349,
      "step": 14021
    },
    {
      "epoch": 0.09527032794187776,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009810295491602366,
      "loss": 2.0473,
      "step": 14022
    },
    {
      "epoch": 0.09527712228847182,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009810266136644323,
      "loss": 2.252,
      "step": 14023
    },
    {
      "epoch": 0.09528391663506587,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009810236779459185,
      "loss": 2.1301,
      "step": 14024
    },
    {
      "epoch": 0.09529071098165993,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009810207420046962,
      "loss": 2.3814,
      "step": 14025
    },
    {
      "epoch": 0.095297505328254,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009810178058407668,
      "loss": 2.2352,
      "step": 14026
    },
    {
      "epoch": 0.09530429967484805,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009810148694541316,
      "loss": 2.2006,
      "step": 14027
    },
    {
      "epoch": 0.09531109402144211,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000981011932844792,
      "loss": 2.1798,
      "step": 14028
    },
    {
      "epoch": 0.09531788836803616,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009810089960127495,
      "loss": 2.0981,
      "step": 14029
    },
    {
      "epoch": 0.09532468271463022,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009810060589580053,
      "loss": 2.0798,
      "step": 14030
    },
    {
      "epoch": 0.09533147706122429,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009810031216805608,
      "loss": 2.2443,
      "step": 14031
    },
    {
      "epoch": 0.09533827140781834,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009810001841804175,
      "loss": 2.3021,
      "step": 14032
    },
    {
      "epoch": 0.0953450657544124,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009809972464575763,
      "loss": 2.2462,
      "step": 14033
    },
    {
      "epoch": 0.09535186010100645,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009809943085120393,
      "loss": 2.1497,
      "step": 14034
    },
    {
      "epoch": 0.09535865444760051,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009809913703438071,
      "loss": 2.2966,
      "step": 14035
    },
    {
      "epoch": 0.09536544879419458,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009809884319528816,
      "loss": 2.211,
      "step": 14036
    },
    {
      "epoch": 0.09537224314078863,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000980985493339264,
      "loss": 2.2865,
      "step": 14037
    },
    {
      "epoch": 0.09537903748738269,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009809825545029555,
      "loss": 2.1037,
      "step": 14038
    },
    {
      "epoch": 0.09538583183397674,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009809796154439576,
      "loss": 2.3441,
      "step": 14039
    },
    {
      "epoch": 0.0953926261805708,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009809766761622717,
      "loss": 2.0672,
      "step": 14040
    },
    {
      "epoch": 0.09539942052716487,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009809737366578992,
      "loss": 2.2927,
      "step": 14041
    },
    {
      "epoch": 0.09540621487375892,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009809707969308413,
      "loss": 2.3243,
      "step": 14042
    },
    {
      "epoch": 0.09541300922035298,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009809678569810993,
      "loss": 2.3786,
      "step": 14043
    },
    {
      "epoch": 0.09541980356694703,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000980964916808675,
      "loss": 2.1825,
      "step": 14044
    },
    {
      "epoch": 0.09542659791354109,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009809619764135692,
      "loss": 2.2277,
      "step": 14045
    },
    {
      "epoch": 0.09543339226013514,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009809590357957837,
      "loss": 2.229,
      "step": 14046
    },
    {
      "epoch": 0.0954401866067292,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009809560949553195,
      "loss": 2.2788,
      "step": 14047
    },
    {
      "epoch": 0.09544698095332327,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009809531538921782,
      "loss": 2.2244,
      "step": 14048
    },
    {
      "epoch": 0.09545377529991732,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009809502126063613,
      "loss": 2.2451,
      "step": 14049
    },
    {
      "epoch": 0.09546056964651138,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009809472710978697,
      "loss": 2.1216,
      "step": 14050
    },
    {
      "epoch": 0.09546736399310543,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000980944329366705,
      "loss": 2.1631,
      "step": 14051
    },
    {
      "epoch": 0.0954741583396995,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009809413874128689,
      "loss": 2.2533,
      "step": 14052
    },
    {
      "epoch": 0.09548095268629356,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000980938445236362,
      "loss": 2.3133,
      "step": 14053
    },
    {
      "epoch": 0.09548774703288761,
      "grad_norm": 0.125,
      "learning_rate": 0.0009809355028371864,
      "loss": 2.214,
      "step": 14054
    },
    {
      "epoch": 0.09549454137948167,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009809325602153432,
      "loss": 2.2375,
      "step": 14055
    },
    {
      "epoch": 0.09550133572607572,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009809296173708337,
      "loss": 2.2019,
      "step": 14056
    },
    {
      "epoch": 0.09550813007266978,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009809266743036594,
      "loss": 2.1248,
      "step": 14057
    },
    {
      "epoch": 0.09551492441926385,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009809237310138214,
      "loss": 2.2986,
      "step": 14058
    },
    {
      "epoch": 0.0955217187658579,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009809207875013212,
      "loss": 2.1137,
      "step": 14059
    },
    {
      "epoch": 0.09552851311245196,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009809178437661602,
      "loss": 2.1207,
      "step": 14060
    },
    {
      "epoch": 0.09553530745904601,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009809148998083399,
      "loss": 2.1754,
      "step": 14061
    },
    {
      "epoch": 0.09554210180564007,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009809119556278615,
      "loss": 2.1668,
      "step": 14062
    },
    {
      "epoch": 0.09554889615223414,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009809090112247263,
      "loss": 2.0368,
      "step": 14063
    },
    {
      "epoch": 0.09555569049882819,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009809060665989357,
      "loss": 2.2609,
      "step": 14064
    },
    {
      "epoch": 0.09556248484542225,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009809031217504912,
      "loss": 2.1244,
      "step": 14065
    },
    {
      "epoch": 0.0955692791920163,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000980900176679394,
      "loss": 2.2244,
      "step": 14066
    },
    {
      "epoch": 0.09557607353861036,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009808972313856458,
      "loss": 2.2536,
      "step": 14067
    },
    {
      "epoch": 0.09558286788520441,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009808942858692473,
      "loss": 2.3258,
      "step": 14068
    },
    {
      "epoch": 0.09558966223179848,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009808913401302005,
      "loss": 2.2361,
      "step": 14069
    },
    {
      "epoch": 0.09559645657839254,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009808883941685063,
      "loss": 2.3597,
      "step": 14070
    },
    {
      "epoch": 0.09560325092498659,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009808854479841664,
      "loss": 2.2001,
      "step": 14071
    },
    {
      "epoch": 0.09561004527158065,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009808825015771821,
      "loss": 2.1455,
      "step": 14072
    },
    {
      "epoch": 0.0956168396181747,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009808795549475547,
      "loss": 2.1727,
      "step": 14073
    },
    {
      "epoch": 0.09562363396476876,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009808766080952856,
      "loss": 2.2706,
      "step": 14074
    },
    {
      "epoch": 0.09563042831136283,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009808736610203762,
      "loss": 2.1877,
      "step": 14075
    },
    {
      "epoch": 0.09563722265795688,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009808707137228276,
      "loss": 2.286,
      "step": 14076
    },
    {
      "epoch": 0.09564401700455094,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009808677662026415,
      "loss": 2.2051,
      "step": 14077
    },
    {
      "epoch": 0.09565081135114499,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000980864818459819,
      "loss": 2.2348,
      "step": 14078
    },
    {
      "epoch": 0.09565760569773905,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000980861870494362,
      "loss": 2.0827,
      "step": 14079
    },
    {
      "epoch": 0.09566440004433312,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009808589223062709,
      "loss": 2.2695,
      "step": 14080
    },
    {
      "epoch": 0.09567119439092717,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000980855973895548,
      "loss": 2.2161,
      "step": 14081
    },
    {
      "epoch": 0.09567798873752123,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009808530252621943,
      "loss": 2.0584,
      "step": 14082
    },
    {
      "epoch": 0.09568478308411528,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009808500764062111,
      "loss": 2.1419,
      "step": 14083
    },
    {
      "epoch": 0.09569157743070934,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009808471273275998,
      "loss": 2.1701,
      "step": 14084
    },
    {
      "epoch": 0.0956983717773034,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009808441780263617,
      "loss": 2.2234,
      "step": 14085
    },
    {
      "epoch": 0.09570516612389746,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009808412285024985,
      "loss": 2.2781,
      "step": 14086
    },
    {
      "epoch": 0.09571196047049152,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.000980838278756011,
      "loss": 2.2059,
      "step": 14087
    },
    {
      "epoch": 0.09571875481708557,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009808353287869011,
      "loss": 2.3947,
      "step": 14088
    },
    {
      "epoch": 0.09572554916367963,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00098083237859517,
      "loss": 2.1132,
      "step": 14089
    },
    {
      "epoch": 0.0957323435102737,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000980829428180819,
      "loss": 2.3034,
      "step": 14090
    },
    {
      "epoch": 0.09573913785686775,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009808264775438493,
      "loss": 2.1025,
      "step": 14091
    },
    {
      "epoch": 0.09574593220346181,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009808235266842626,
      "loss": 2.1544,
      "step": 14092
    },
    {
      "epoch": 0.09575272655005586,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009808205756020602,
      "loss": 2.1564,
      "step": 14093
    },
    {
      "epoch": 0.09575952089664992,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009808176242972434,
      "loss": 2.1085,
      "step": 14094
    },
    {
      "epoch": 0.09576631524324397,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009808146727698134,
      "loss": 2.1419,
      "step": 14095
    },
    {
      "epoch": 0.09577310958983803,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009808117210197717,
      "loss": 2.225,
      "step": 14096
    },
    {
      "epoch": 0.0957799039364321,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009808087690471197,
      "loss": 2.2302,
      "step": 14097
    },
    {
      "epoch": 0.09578669828302615,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000980805816851859,
      "loss": 2.14,
      "step": 14098
    },
    {
      "epoch": 0.09579349262962021,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009808028644339906,
      "loss": 2.3067,
      "step": 14099
    },
    {
      "epoch": 0.09580028697621426,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009807999117935158,
      "loss": 2.2069,
      "step": 14100
    },
    {
      "epoch": 0.09580708132280832,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009807969589304363,
      "loss": 2.2405,
      "step": 14101
    },
    {
      "epoch": 0.09581387566940239,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009807940058447532,
      "loss": 2.3184,
      "step": 14102
    },
    {
      "epoch": 0.09582067001599644,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.000980791052536468,
      "loss": 2.0421,
      "step": 14103
    },
    {
      "epoch": 0.0958274643625905,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009807880990055822,
      "loss": 2.092,
      "step": 14104
    },
    {
      "epoch": 0.09583425870918455,
      "grad_norm": 0.125,
      "learning_rate": 0.0009807851452520969,
      "loss": 2.1676,
      "step": 14105
    },
    {
      "epoch": 0.09584105305577861,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009807821912760138,
      "loss": 2.1658,
      "step": 14106
    },
    {
      "epoch": 0.09584784740237268,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009807792370773339,
      "loss": 2.0208,
      "step": 14107
    },
    {
      "epoch": 0.09585464174896673,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009807762826560588,
      "loss": 2.249,
      "step": 14108
    },
    {
      "epoch": 0.09586143609556079,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009807733280121896,
      "loss": 2.101,
      "step": 14109
    },
    {
      "epoch": 0.09586823044215484,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000980770373145728,
      "loss": 2.1621,
      "step": 14110
    },
    {
      "epoch": 0.0958750247887489,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009807674180566752,
      "loss": 2.1345,
      "step": 14111
    },
    {
      "epoch": 0.09588181913534297,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009807644627450325,
      "loss": 2.1389,
      "step": 14112
    },
    {
      "epoch": 0.09588861348193702,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009807615072108015,
      "loss": 2.1709,
      "step": 14113
    },
    {
      "epoch": 0.09589540782853108,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009807585514539834,
      "loss": 2.3383,
      "step": 14114
    },
    {
      "epoch": 0.09590220217512513,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009807555954745796,
      "loss": 2.068,
      "step": 14115
    },
    {
      "epoch": 0.09590899652171919,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009807526392725916,
      "loss": 2.1834,
      "step": 14116
    },
    {
      "epoch": 0.09591579086831324,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009807496828480205,
      "loss": 2.2965,
      "step": 14117
    },
    {
      "epoch": 0.0959225852149073,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009807467262008679,
      "loss": 2.1843,
      "step": 14118
    },
    {
      "epoch": 0.09592937956150137,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009807437693311349,
      "loss": 2.2793,
      "step": 14119
    },
    {
      "epoch": 0.09593617390809542,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000980740812238823,
      "loss": 2.2337,
      "step": 14120
    },
    {
      "epoch": 0.09594296825468948,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000980737854923934,
      "loss": 2.2011,
      "step": 14121
    },
    {
      "epoch": 0.09594976260128353,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009807348973864686,
      "loss": 2.2877,
      "step": 14122
    },
    {
      "epoch": 0.0959565569478776,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009807319396264284,
      "loss": 2.1917,
      "step": 14123
    },
    {
      "epoch": 0.09596335129447166,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000980728981643815,
      "loss": 2.2417,
      "step": 14124
    },
    {
      "epoch": 0.09597014564106571,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009807260234386294,
      "loss": 2.0748,
      "step": 14125
    },
    {
      "epoch": 0.09597693998765977,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009807230650108734,
      "loss": 2.2803,
      "step": 14126
    },
    {
      "epoch": 0.09598373433425382,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.000980720106360548,
      "loss": 2.1582,
      "step": 14127
    },
    {
      "epoch": 0.09599052868084788,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009807171474876546,
      "loss": 2.2225,
      "step": 14128
    },
    {
      "epoch": 0.09599732302744195,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009807141883921948,
      "loss": 2.1763,
      "step": 14129
    },
    {
      "epoch": 0.096004117374036,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009807112290741698,
      "loss": 2.1381,
      "step": 14130
    },
    {
      "epoch": 0.09601091172063006,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000980708269533581,
      "loss": 2.2508,
      "step": 14131
    },
    {
      "epoch": 0.09601770606722411,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00098070530977043,
      "loss": 2.06,
      "step": 14132
    },
    {
      "epoch": 0.09602450041381817,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009807023497847176,
      "loss": 2.2814,
      "step": 14133
    },
    {
      "epoch": 0.09603129476041224,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009806993895764458,
      "loss": 2.2184,
      "step": 14134
    },
    {
      "epoch": 0.09603808910700629,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009806964291456156,
      "loss": 2.2277,
      "step": 14135
    },
    {
      "epoch": 0.09604488345360035,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009806934684922284,
      "loss": 2.1561,
      "step": 14136
    },
    {
      "epoch": 0.0960516778001944,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009806905076162858,
      "loss": 2.2006,
      "step": 14137
    },
    {
      "epoch": 0.09605847214678846,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009806875465177888,
      "loss": 2.2657,
      "step": 14138
    },
    {
      "epoch": 0.09606526649338253,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000980684585196739,
      "loss": 2.3995,
      "step": 14139
    },
    {
      "epoch": 0.09607206083997658,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000980681623653138,
      "loss": 2.1814,
      "step": 14140
    },
    {
      "epoch": 0.09607885518657064,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009806786618869866,
      "loss": 2.3912,
      "step": 14141
    },
    {
      "epoch": 0.09608564953316469,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009806756998982865,
      "loss": 2.106,
      "step": 14142
    },
    {
      "epoch": 0.09609244387975875,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009806727376870394,
      "loss": 2.108,
      "step": 14143
    },
    {
      "epoch": 0.0960992382263528,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009806697752532461,
      "loss": 2.1832,
      "step": 14144
    },
    {
      "epoch": 0.09610603257294686,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009806668125969082,
      "loss": 2.1783,
      "step": 14145
    },
    {
      "epoch": 0.09611282691954093,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009806638497180272,
      "loss": 2.1102,
      "step": 14146
    },
    {
      "epoch": 0.09611962126613498,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009806608866166043,
      "loss": 2.131,
      "step": 14147
    },
    {
      "epoch": 0.09612641561272904,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009806579232926407,
      "loss": 2.2892,
      "step": 14148
    },
    {
      "epoch": 0.09613320995932309,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009806549597461383,
      "loss": 2.2582,
      "step": 14149
    },
    {
      "epoch": 0.09614000430591715,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.000980651995977098,
      "loss": 2.14,
      "step": 14150
    },
    {
      "epoch": 0.09614679865251122,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009806490319855213,
      "loss": 2.235,
      "step": 14151
    },
    {
      "epoch": 0.09615359299910527,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009806460677714097,
      "loss": 2.0807,
      "step": 14152
    },
    {
      "epoch": 0.09616038734569933,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009806431033347646,
      "loss": 2.1722,
      "step": 14153
    },
    {
      "epoch": 0.09616718169229338,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000980640138675587,
      "loss": 2.2935,
      "step": 14154
    },
    {
      "epoch": 0.09617397603888744,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009806371737938787,
      "loss": 2.244,
      "step": 14155
    },
    {
      "epoch": 0.0961807703854815,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009806342086896406,
      "loss": 2.1657,
      "step": 14156
    },
    {
      "epoch": 0.09618756473207556,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009806312433628748,
      "loss": 2.2367,
      "step": 14157
    },
    {
      "epoch": 0.09619435907866962,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.000980628277813582,
      "loss": 2.1579,
      "step": 14158
    },
    {
      "epoch": 0.09620115342526367,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000980625312041764,
      "loss": 2.3243,
      "step": 14159
    },
    {
      "epoch": 0.09620794777185773,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009806223460474216,
      "loss": 2.1515,
      "step": 14160
    },
    {
      "epoch": 0.0962147421184518,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009806193798305568,
      "loss": 2.261,
      "step": 14161
    },
    {
      "epoch": 0.09622153646504585,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009806164133911708,
      "loss": 2.2918,
      "step": 14162
    },
    {
      "epoch": 0.09622833081163991,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000980613446729265,
      "loss": 2.2453,
      "step": 14163
    },
    {
      "epoch": 0.09623512515823396,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009806104798448405,
      "loss": 2.2538,
      "step": 14164
    },
    {
      "epoch": 0.09624191950482802,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009806075127378989,
      "loss": 2.1744,
      "step": 14165
    },
    {
      "epoch": 0.09624871385142207,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009806045454084417,
      "loss": 2.2166,
      "step": 14166
    },
    {
      "epoch": 0.09625550819801613,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00098060157785647,
      "loss": 2.0296,
      "step": 14167
    },
    {
      "epoch": 0.0962623025446102,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000980598610081985,
      "loss": 2.255,
      "step": 14168
    },
    {
      "epoch": 0.09626909689120425,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009805956420849886,
      "loss": 2.1825,
      "step": 14169
    },
    {
      "epoch": 0.09627589123779831,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000980592673865482,
      "loss": 2.1935,
      "step": 14170
    },
    {
      "epoch": 0.09628268558439236,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009805897054234664,
      "loss": 2.1921,
      "step": 14171
    },
    {
      "epoch": 0.09628947993098642,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009805867367589432,
      "loss": 2.2864,
      "step": 14172
    },
    {
      "epoch": 0.09629627427758049,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000980583767871914,
      "loss": 2.2232,
      "step": 14173
    },
    {
      "epoch": 0.09630306862417454,
      "grad_norm": 0.140625,
      "learning_rate": 0.00098058079876238,
      "loss": 2.1216,
      "step": 14174
    },
    {
      "epoch": 0.0963098629707686,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009805778294303424,
      "loss": 2.242,
      "step": 14175
    },
    {
      "epoch": 0.09631665731736265,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000980574859875803,
      "loss": 2.3514,
      "step": 14176
    },
    {
      "epoch": 0.09632345166395671,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000980571890098763,
      "loss": 2.2407,
      "step": 14177
    },
    {
      "epoch": 0.09633024601055078,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009805689200992236,
      "loss": 2.1415,
      "step": 14178
    },
    {
      "epoch": 0.09633704035714483,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009805659498771863,
      "loss": 2.19,
      "step": 14179
    },
    {
      "epoch": 0.09634383470373889,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009805629794326525,
      "loss": 2.3212,
      "step": 14180
    },
    {
      "epoch": 0.09635062905033294,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009805600087656235,
      "loss": 2.3394,
      "step": 14181
    },
    {
      "epoch": 0.096357423396927,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009805570378761009,
      "loss": 2.2383,
      "step": 14182
    },
    {
      "epoch": 0.09636421774352107,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009805540667640856,
      "loss": 2.1354,
      "step": 14183
    },
    {
      "epoch": 0.09637101209011512,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009805510954295795,
      "loss": 2.3322,
      "step": 14184
    },
    {
      "epoch": 0.09637780643670918,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009805481238725836,
      "loss": 2.2063,
      "step": 14185
    },
    {
      "epoch": 0.09638460078330323,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009805451520930996,
      "loss": 2.2148,
      "step": 14186
    },
    {
      "epoch": 0.09639139512989729,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009805421800911287,
      "loss": 2.256,
      "step": 14187
    },
    {
      "epoch": 0.09639818947649134,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009805392078666721,
      "loss": 2.2491,
      "step": 14188
    },
    {
      "epoch": 0.0964049838230854,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009805362354197314,
      "loss": 2.0707,
      "step": 14189
    },
    {
      "epoch": 0.09641177816967947,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000980533262750308,
      "loss": 2.3332,
      "step": 14190
    },
    {
      "epoch": 0.09641857251627352,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009805302898584032,
      "loss": 2.2867,
      "step": 14191
    },
    {
      "epoch": 0.09642536686286758,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009805273167440183,
      "loss": 2.2225,
      "step": 14192
    },
    {
      "epoch": 0.09643216120946163,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000980524343407155,
      "loss": 2.2849,
      "step": 14193
    },
    {
      "epoch": 0.0964389555560557,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009805213698478143,
      "loss": 2.2167,
      "step": 14194
    },
    {
      "epoch": 0.09644574990264976,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009805183960659976,
      "loss": 2.1753,
      "step": 14195
    },
    {
      "epoch": 0.09645254424924381,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009805154220617065,
      "loss": 2.2061,
      "step": 14196
    },
    {
      "epoch": 0.09645933859583787,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009805124478349423,
      "loss": 2.33,
      "step": 14197
    },
    {
      "epoch": 0.09646613294243192,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009805094733857064,
      "loss": 2.2265,
      "step": 14198
    },
    {
      "epoch": 0.09647292728902598,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000980506498714,
      "loss": 2.2999,
      "step": 14199
    },
    {
      "epoch": 0.09647972163562005,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009805035238198245,
      "loss": 2.1602,
      "step": 14200
    },
    {
      "epoch": 0.0964865159822141,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009805005487031815,
      "loss": 2.1319,
      "step": 14201
    },
    {
      "epoch": 0.09649331032880816,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009804975733640722,
      "loss": 2.1866,
      "step": 14202
    },
    {
      "epoch": 0.09650010467540221,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009804945978024982,
      "loss": 2.0446,
      "step": 14203
    },
    {
      "epoch": 0.09650689902199627,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009804916220184606,
      "loss": 2.2247,
      "step": 14204
    },
    {
      "epoch": 0.09651369336859034,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009804886460119608,
      "loss": 2.2265,
      "step": 14205
    },
    {
      "epoch": 0.09652048771518439,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009804856697830005,
      "loss": 2.184,
      "step": 14206
    },
    {
      "epoch": 0.09652728206177845,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009804826933315807,
      "loss": 2.3873,
      "step": 14207
    },
    {
      "epoch": 0.0965340764083725,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009804797166577028,
      "loss": 2.2773,
      "step": 14208
    },
    {
      "epoch": 0.09654087075496656,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009804767397613686,
      "loss": 2.1379,
      "step": 14209
    },
    {
      "epoch": 0.09654766510156063,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009804737626425788,
      "loss": 2.1993,
      "step": 14210
    },
    {
      "epoch": 0.09655445944815468,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009804707853013355,
      "loss": 2.2314,
      "step": 14211
    },
    {
      "epoch": 0.09656125379474874,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009804678077376394,
      "loss": 2.3565,
      "step": 14212
    },
    {
      "epoch": 0.09656804814134279,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009804648299514924,
      "loss": 2.1041,
      "step": 14213
    },
    {
      "epoch": 0.09657484248793685,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009804618519428957,
      "loss": 2.2274,
      "step": 14214
    },
    {
      "epoch": 0.0965816368345309,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009804588737118507,
      "loss": 2.2225,
      "step": 14215
    },
    {
      "epoch": 0.09658843118112496,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009804558952583589,
      "loss": 2.0255,
      "step": 14216
    },
    {
      "epoch": 0.09659522552771903,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009804529165824212,
      "loss": 2.2129,
      "step": 14217
    },
    {
      "epoch": 0.09660201987431308,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009804499376840393,
      "loss": 2.2408,
      "step": 14218
    },
    {
      "epoch": 0.09660881422090714,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000980446958563215,
      "loss": 2.2012,
      "step": 14219
    },
    {
      "epoch": 0.09661560856750119,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000980443979219949,
      "loss": 2.2435,
      "step": 14220
    },
    {
      "epoch": 0.09662240291409525,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009804409996542428,
      "loss": 2.4063,
      "step": 14221
    },
    {
      "epoch": 0.09662919726068932,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000980438019866098,
      "loss": 2.1853,
      "step": 14222
    },
    {
      "epoch": 0.09663599160728337,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000980435039855516,
      "loss": 2.0967,
      "step": 14223
    },
    {
      "epoch": 0.09664278595387743,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009804320596224982,
      "loss": 2.3583,
      "step": 14224
    },
    {
      "epoch": 0.09664958030047148,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009804290791670456,
      "loss": 2.153,
      "step": 14225
    },
    {
      "epoch": 0.09665637464706554,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00098042609848916,
      "loss": 2.1848,
      "step": 14226
    },
    {
      "epoch": 0.0966631689936596,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009804231175888425,
      "loss": 2.1534,
      "step": 14227
    },
    {
      "epoch": 0.09666996334025366,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009804201364660948,
      "loss": 2.144,
      "step": 14228
    },
    {
      "epoch": 0.09667675768684772,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000980417155120918,
      "loss": 2.0508,
      "step": 14229
    },
    {
      "epoch": 0.09668355203344177,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009804141735533136,
      "loss": 2.2469,
      "step": 14230
    },
    {
      "epoch": 0.09669034638003583,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009804111917632828,
      "loss": 2.1333,
      "step": 14231
    },
    {
      "epoch": 0.0966971407266299,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009804082097508273,
      "loss": 2.1547,
      "step": 14232
    },
    {
      "epoch": 0.09670393507322395,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000980405227515948,
      "loss": 2.3023,
      "step": 14233
    },
    {
      "epoch": 0.09671072941981801,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000980402245058647,
      "loss": 2.1649,
      "step": 14234
    },
    {
      "epoch": 0.09671752376641206,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000980399262378925,
      "loss": 2.2173,
      "step": 14235
    },
    {
      "epoch": 0.09672431811300612,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009803962794767839,
      "loss": 2.3549,
      "step": 14236
    },
    {
      "epoch": 0.09673111245960017,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009803932963522246,
      "loss": 2.1161,
      "step": 14237
    },
    {
      "epoch": 0.09673790680619423,
      "grad_norm": 0.125,
      "learning_rate": 0.0009803903130052487,
      "loss": 2.1107,
      "step": 14238
    },
    {
      "epoch": 0.0967447011527883,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009803873294358578,
      "loss": 2.2943,
      "step": 14239
    },
    {
      "epoch": 0.09675149549938235,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009803843456440528,
      "loss": 2.2028,
      "step": 14240
    },
    {
      "epoch": 0.09675828984597641,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009803813616298357,
      "loss": 2.2,
      "step": 14241
    },
    {
      "epoch": 0.09676508419257046,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000980378377393207,
      "loss": 2.2175,
      "step": 14242
    },
    {
      "epoch": 0.09677187853916452,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000980375392934169,
      "loss": 2.2889,
      "step": 14243
    },
    {
      "epoch": 0.09677867288575859,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009803724082527228,
      "loss": 2.2757,
      "step": 14244
    },
    {
      "epoch": 0.09678546723235264,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009803694233488697,
      "loss": 2.2907,
      "step": 14245
    },
    {
      "epoch": 0.0967922615789467,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009803664382226106,
      "loss": 2.2583,
      "step": 14246
    },
    {
      "epoch": 0.09679905592554075,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009803634528739477,
      "loss": 2.2356,
      "step": 14247
    },
    {
      "epoch": 0.09680585027213481,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009803604673028822,
      "loss": 2.3207,
      "step": 14248
    },
    {
      "epoch": 0.09681264461872888,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009803574815094149,
      "loss": 2.105,
      "step": 14249
    },
    {
      "epoch": 0.09681943896532293,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009803544954935478,
      "loss": 2.1961,
      "step": 14250
    },
    {
      "epoch": 0.09682623331191699,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000980351509255282,
      "loss": 2.2011,
      "step": 14251
    },
    {
      "epoch": 0.09683302765851104,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000980348522794619,
      "loss": 2.1024,
      "step": 14252
    },
    {
      "epoch": 0.0968398220051051,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009803455361115602,
      "loss": 2.152,
      "step": 14253
    },
    {
      "epoch": 0.09684661635169917,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000980342549206107,
      "loss": 2.1851,
      "step": 14254
    },
    {
      "epoch": 0.09685341069829322,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009803395620782604,
      "loss": 2.2053,
      "step": 14255
    },
    {
      "epoch": 0.09686020504488728,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009803365747280224,
      "loss": 2.2946,
      "step": 14256
    },
    {
      "epoch": 0.09686699939148133,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009803335871553939,
      "loss": 2.0753,
      "step": 14257
    },
    {
      "epoch": 0.09687379373807539,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009803305993603764,
      "loss": 2.1846,
      "step": 14258
    },
    {
      "epoch": 0.09688058808466946,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009803276113429715,
      "loss": 2.0582,
      "step": 14259
    },
    {
      "epoch": 0.0968873824312635,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009803246231031805,
      "loss": 2.199,
      "step": 14260
    },
    {
      "epoch": 0.09689417677785757,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009803216346410044,
      "loss": 2.1898,
      "step": 14261
    },
    {
      "epoch": 0.09690097112445162,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000980318645956445,
      "loss": 2.1432,
      "step": 14262
    },
    {
      "epoch": 0.09690776547104568,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009803156570495036,
      "loss": 2.2999,
      "step": 14263
    },
    {
      "epoch": 0.09691455981763973,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009803126679201817,
      "loss": 2.3993,
      "step": 14264
    },
    {
      "epoch": 0.0969213541642338,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009803096785684804,
      "loss": 2.2401,
      "step": 14265
    },
    {
      "epoch": 0.09692814851082786,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009803066889944013,
      "loss": 2.1989,
      "step": 14266
    },
    {
      "epoch": 0.09693494285742191,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009803036991979455,
      "loss": 2.3044,
      "step": 14267
    },
    {
      "epoch": 0.09694173720401597,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009803007091791149,
      "loss": 2.2679,
      "step": 14268
    },
    {
      "epoch": 0.09694853155061002,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009802977189379103,
      "loss": 2.1148,
      "step": 14269
    },
    {
      "epoch": 0.09695532589720408,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009802947284743336,
      "loss": 2.1704,
      "step": 14270
    },
    {
      "epoch": 0.09696212024379815,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009802917377883858,
      "loss": 2.1691,
      "step": 14271
    },
    {
      "epoch": 0.0969689145903922,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009802887468800686,
      "loss": 2.2569,
      "step": 14272
    },
    {
      "epoch": 0.09697570893698626,
      "grad_norm": 0.125,
      "learning_rate": 0.000980285755749383,
      "loss": 2.1502,
      "step": 14273
    },
    {
      "epoch": 0.09698250328358031,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009802827643963306,
      "loss": 2.1551,
      "step": 14274
    },
    {
      "epoch": 0.09698929763017437,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009802797728209129,
      "loss": 2.2591,
      "step": 14275
    },
    {
      "epoch": 0.09699609197676844,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009802767810231311,
      "loss": 2.2262,
      "step": 14276
    },
    {
      "epoch": 0.09700288632336249,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000980273789002987,
      "loss": 2.0979,
      "step": 14277
    },
    {
      "epoch": 0.09700968066995655,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000980270796760481,
      "loss": 2.4263,
      "step": 14278
    },
    {
      "epoch": 0.0970164750165506,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009802678042956157,
      "loss": 2.2638,
      "step": 14279
    },
    {
      "epoch": 0.09702326936314466,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009802648116083916,
      "loss": 2.3938,
      "step": 14280
    },
    {
      "epoch": 0.09703006370973873,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009802618186988106,
      "loss": 2.3029,
      "step": 14281
    },
    {
      "epoch": 0.09703685805633278,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009802588255668737,
      "loss": 2.3143,
      "step": 14282
    },
    {
      "epoch": 0.09704365240292684,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009802558322125825,
      "loss": 2.2386,
      "step": 14283
    },
    {
      "epoch": 0.09705044674952089,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009802528386359383,
      "loss": 2.3126,
      "step": 14284
    },
    {
      "epoch": 0.09705724109611495,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000980249844836943,
      "loss": 2.3619,
      "step": 14285
    },
    {
      "epoch": 0.097064035442709,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.000980246850815597,
      "loss": 2.0824,
      "step": 14286
    },
    {
      "epoch": 0.09707082978930306,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009802438565719024,
      "loss": 2.3073,
      "step": 14287
    },
    {
      "epoch": 0.09707762413589713,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009802408621058603,
      "loss": 2.2358,
      "step": 14288
    },
    {
      "epoch": 0.09708441848249118,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009802378674174724,
      "loss": 2.1373,
      "step": 14289
    },
    {
      "epoch": 0.09709121282908524,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009802348725067396,
      "loss": 2.3816,
      "step": 14290
    },
    {
      "epoch": 0.09709800717567929,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009802318773736638,
      "loss": 2.2482,
      "step": 14291
    },
    {
      "epoch": 0.09710480152227335,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009802288820182459,
      "loss": 2.3133,
      "step": 14292
    },
    {
      "epoch": 0.09711159586886742,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009802258864404878,
      "loss": 2.2936,
      "step": 14293
    },
    {
      "epoch": 0.09711839021546147,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009802228906403904,
      "loss": 2.3595,
      "step": 14294
    },
    {
      "epoch": 0.09712518456205553,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009802198946179555,
      "loss": 2.1339,
      "step": 14295
    },
    {
      "epoch": 0.09713197890864958,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009802168983731841,
      "loss": 2.207,
      "step": 14296
    },
    {
      "epoch": 0.09713877325524364,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009802139019060778,
      "loss": 2.1084,
      "step": 14297
    },
    {
      "epoch": 0.0971455676018377,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009802109052166382,
      "loss": 2.2043,
      "step": 14298
    },
    {
      "epoch": 0.09715236194843176,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009802079083048663,
      "loss": 2.206,
      "step": 14299
    },
    {
      "epoch": 0.09715915629502582,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009802049111707636,
      "loss": 2.265,
      "step": 14300
    },
    {
      "epoch": 0.09716595064161987,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009802019138143317,
      "loss": 2.1573,
      "step": 14301
    },
    {
      "epoch": 0.09717274498821393,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009801989162355716,
      "loss": 2.2717,
      "step": 14302
    },
    {
      "epoch": 0.097179539334808,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000980195918434485,
      "loss": 2.2276,
      "step": 14303
    },
    {
      "epoch": 0.09718633368140205,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009801929204110733,
      "loss": 2.1286,
      "step": 14304
    },
    {
      "epoch": 0.09719312802799611,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009801899221653376,
      "loss": 2.2914,
      "step": 14305
    },
    {
      "epoch": 0.09719992237459016,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009801869236972795,
      "loss": 2.332,
      "step": 14306
    },
    {
      "epoch": 0.09720671672118422,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009801839250069004,
      "loss": 2.037,
      "step": 14307
    },
    {
      "epoch": 0.09721351106777827,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009801809260942017,
      "loss": 2.1903,
      "step": 14308
    },
    {
      "epoch": 0.09722030541437233,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009801779269591848,
      "loss": 2.2108,
      "step": 14309
    },
    {
      "epoch": 0.0972270997609664,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009801749276018507,
      "loss": 2.2379,
      "step": 14310
    },
    {
      "epoch": 0.09723389410756045,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009801719280222014,
      "loss": 2.0723,
      "step": 14311
    },
    {
      "epoch": 0.09724068845415451,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009801689282202379,
      "loss": 2.2711,
      "step": 14312
    },
    {
      "epoch": 0.09724748280074856,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009801659281959618,
      "loss": 2.1994,
      "step": 14313
    },
    {
      "epoch": 0.09725427714734262,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009801629279493744,
      "loss": 2.2133,
      "step": 14314
    },
    {
      "epoch": 0.09726107149393669,
      "grad_norm": 0.154296875,
      "learning_rate": 0.000980159927480477,
      "loss": 2.4032,
      "step": 14315
    },
    {
      "epoch": 0.09726786584053074,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000980156926789271,
      "loss": 2.1713,
      "step": 14316
    },
    {
      "epoch": 0.0972746601871248,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009801539258757579,
      "loss": 2.1258,
      "step": 14317
    },
    {
      "epoch": 0.09728145453371885,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009801509247399389,
      "loss": 2.1708,
      "step": 14318
    },
    {
      "epoch": 0.09728824888031291,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000980147923381816,
      "loss": 2.1436,
      "step": 14319
    },
    {
      "epoch": 0.09729504322690698,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009801449218013896,
      "loss": 2.1482,
      "step": 14320
    },
    {
      "epoch": 0.09730183757350103,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009801419199986616,
      "loss": 2.1304,
      "step": 14321
    },
    {
      "epoch": 0.09730863192009509,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009801389179736337,
      "loss": 2.2153,
      "step": 14322
    },
    {
      "epoch": 0.09731542626668914,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009801359157263069,
      "loss": 2.1953,
      "step": 14323
    },
    {
      "epoch": 0.0973222206132832,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009801329132566826,
      "loss": 2.0726,
      "step": 14324
    },
    {
      "epoch": 0.09732901495987727,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009801299105647622,
      "loss": 2.2266,
      "step": 14325
    },
    {
      "epoch": 0.09733580930647132,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009801269076505472,
      "loss": 2.0983,
      "step": 14326
    },
    {
      "epoch": 0.09734260365306538,
      "grad_norm": 0.140625,
      "learning_rate": 0.000980123904514039,
      "loss": 2.1663,
      "step": 14327
    },
    {
      "epoch": 0.09734939799965943,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000980120901155239,
      "loss": 2.4279,
      "step": 14328
    },
    {
      "epoch": 0.09735619234625349,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009801178975741484,
      "loss": 2.4016,
      "step": 14329
    },
    {
      "epoch": 0.09736298669284756,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009801148937707688,
      "loss": 2.1142,
      "step": 14330
    },
    {
      "epoch": 0.0973697810394416,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009801118897451015,
      "loss": 2.2193,
      "step": 14331
    },
    {
      "epoch": 0.09737657538603567,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000980108885497148,
      "loss": 2.0975,
      "step": 14332
    },
    {
      "epoch": 0.09738336973262972,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009801058810269094,
      "loss": 2.3248,
      "step": 14333
    },
    {
      "epoch": 0.09739016407922378,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009801028763343873,
      "loss": 2.0296,
      "step": 14334
    },
    {
      "epoch": 0.09739695842581783,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009800998714195833,
      "loss": 2.1753,
      "step": 14335
    },
    {
      "epoch": 0.0974037527724119,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009800968662824982,
      "loss": 2.2559,
      "step": 14336
    },
    {
      "epoch": 0.09741054711900596,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000980093860923134,
      "loss": 2.2394,
      "step": 14337
    },
    {
      "epoch": 0.09741734146560001,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009800908553414917,
      "loss": 2.2662,
      "step": 14338
    },
    {
      "epoch": 0.09742413581219407,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000980087849537573,
      "loss": 2.2801,
      "step": 14339
    },
    {
      "epoch": 0.09743093015878812,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009800848435113792,
      "loss": 2.1211,
      "step": 14340
    },
    {
      "epoch": 0.09743772450538218,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009800818372629114,
      "loss": 2.2287,
      "step": 14341
    },
    {
      "epoch": 0.09744451885197625,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009800788307921713,
      "loss": 2.0314,
      "step": 14342
    },
    {
      "epoch": 0.0974513131985703,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009800758240991604,
      "loss": 2.2056,
      "step": 14343
    },
    {
      "epoch": 0.09745810754516436,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009800728171838796,
      "loss": 2.0926,
      "step": 14344
    },
    {
      "epoch": 0.09746490189175841,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009800698100463307,
      "loss": 2.1247,
      "step": 14345
    },
    {
      "epoch": 0.09747169623835247,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009800668026865148,
      "loss": 2.2023,
      "step": 14346
    },
    {
      "epoch": 0.09747849058494654,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009800637951044339,
      "loss": 2.1662,
      "step": 14347
    },
    {
      "epoch": 0.09748528493154059,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009800607873000887,
      "loss": 2.1906,
      "step": 14348
    },
    {
      "epoch": 0.09749207927813465,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009800577792734807,
      "loss": 2.2551,
      "step": 14349
    },
    {
      "epoch": 0.0974988736247287,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009800547710246117,
      "loss": 2.2555,
      "step": 14350
    },
    {
      "epoch": 0.09750566797132276,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009800517625534826,
      "loss": 2.2199,
      "step": 14351
    },
    {
      "epoch": 0.09751246231791683,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009800487538600953,
      "loss": 2.2324,
      "step": 14352
    },
    {
      "epoch": 0.09751925666451088,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009800457449444509,
      "loss": 2.1711,
      "step": 14353
    },
    {
      "epoch": 0.09752605101110494,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009800427358065507,
      "loss": 2.1971,
      "step": 14354
    },
    {
      "epoch": 0.09753284535769899,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009800397264463964,
      "loss": 2.2238,
      "step": 14355
    },
    {
      "epoch": 0.09753963970429305,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000980036716863989,
      "loss": 2.2515,
      "step": 14356
    },
    {
      "epoch": 0.0975464340508871,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00098003370705933,
      "loss": 2.2311,
      "step": 14357
    },
    {
      "epoch": 0.09755322839748116,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009800306970324214,
      "loss": 2.2414,
      "step": 14358
    },
    {
      "epoch": 0.09756002274407523,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009800276867832636,
      "loss": 2.1916,
      "step": 14359
    },
    {
      "epoch": 0.09756681709066928,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009800246763118589,
      "loss": 2.2766,
      "step": 14360
    },
    {
      "epoch": 0.09757361143726334,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009800216656182079,
      "loss": 2.2064,
      "step": 14361
    },
    {
      "epoch": 0.09758040578385739,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009800186547023124,
      "loss": 2.2024,
      "step": 14362
    },
    {
      "epoch": 0.09758720013045145,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009800156435641738,
      "loss": 2.2,
      "step": 14363
    },
    {
      "epoch": 0.09759399447704552,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009800126322037936,
      "loss": 2.167,
      "step": 14364
    },
    {
      "epoch": 0.09760078882363957,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000980009620621173,
      "loss": 2.1694,
      "step": 14365
    },
    {
      "epoch": 0.09760758317023363,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009800066088163136,
      "loss": 2.1414,
      "step": 14366
    },
    {
      "epoch": 0.09761437751682768,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009800035967892165,
      "loss": 2.1358,
      "step": 14367
    },
    {
      "epoch": 0.09762117186342174,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009800005845398833,
      "loss": 2.2502,
      "step": 14368
    },
    {
      "epoch": 0.0976279662100158,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000979997572068315,
      "loss": 2.1863,
      "step": 14369
    },
    {
      "epoch": 0.09763476055660986,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009799945593745136,
      "loss": 2.2619,
      "step": 14370
    },
    {
      "epoch": 0.09764155490320392,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009799915464584804,
      "loss": 2.0923,
      "step": 14371
    },
    {
      "epoch": 0.09764834924979797,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009799885333202163,
      "loss": 2.1485,
      "step": 14372
    },
    {
      "epoch": 0.09765514359639203,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009799855199597231,
      "loss": 2.2662,
      "step": 14373
    },
    {
      "epoch": 0.0976619379429861,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000979982506377002,
      "loss": 2.1488,
      "step": 14374
    },
    {
      "epoch": 0.09766873228958015,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009799794925720547,
      "loss": 2.4073,
      "step": 14375
    },
    {
      "epoch": 0.09767552663617421,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009799764785448823,
      "loss": 2.2474,
      "step": 14376
    },
    {
      "epoch": 0.09768232098276826,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009799734642954865,
      "loss": 2.293,
      "step": 14377
    },
    {
      "epoch": 0.09768911532936232,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009799704498238682,
      "loss": 2.2578,
      "step": 14378
    },
    {
      "epoch": 0.09769590967595639,
      "grad_norm": 0.1787109375,
      "learning_rate": 0.0009799674351300292,
      "loss": 2.2343,
      "step": 14379
    },
    {
      "epoch": 0.09770270402255044,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009799644202139709,
      "loss": 2.2321,
      "step": 14380
    },
    {
      "epoch": 0.0977094983691445,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009799614050756945,
      "loss": 2.1926,
      "step": 14381
    },
    {
      "epoch": 0.09771629271573855,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009799583897152013,
      "loss": 2.2843,
      "step": 14382
    },
    {
      "epoch": 0.09772308706233261,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009799553741324932,
      "loss": 2.1797,
      "step": 14383
    },
    {
      "epoch": 0.09772988140892666,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009799523583275708,
      "loss": 2.3618,
      "step": 14384
    },
    {
      "epoch": 0.09773667575552072,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009799493423004364,
      "loss": 2.3297,
      "step": 14385
    },
    {
      "epoch": 0.09774347010211479,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009799463260510908,
      "loss": 2.2611,
      "step": 14386
    },
    {
      "epoch": 0.09775026444870884,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009799433095795354,
      "loss": 2.1678,
      "step": 14387
    },
    {
      "epoch": 0.0977570587953029,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000979940292885772,
      "loss": 2.2005,
      "step": 14388
    },
    {
      "epoch": 0.09776385314189695,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009799372759698014,
      "loss": 2.2952,
      "step": 14389
    },
    {
      "epoch": 0.09777064748849101,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009799342588316259,
      "loss": 2.1967,
      "step": 14390
    },
    {
      "epoch": 0.09777744183508508,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009799312414712459,
      "loss": 2.2613,
      "step": 14391
    },
    {
      "epoch": 0.09778423618167913,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009799282238886634,
      "loss": 2.2259,
      "step": 14392
    },
    {
      "epoch": 0.09779103052827319,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009799252060838795,
      "loss": 2.2031,
      "step": 14393
    },
    {
      "epoch": 0.09779782487486724,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009799221880568958,
      "loss": 2.1137,
      "step": 14394
    },
    {
      "epoch": 0.0978046192214613,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009799191698077136,
      "loss": 2.3269,
      "step": 14395
    },
    {
      "epoch": 0.09781141356805537,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009799161513363345,
      "loss": 2.3579,
      "step": 14396
    },
    {
      "epoch": 0.09781820791464942,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009799131326427594,
      "loss": 2.1998,
      "step": 14397
    },
    {
      "epoch": 0.09782500226124348,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009799101137269904,
      "loss": 2.2761,
      "step": 14398
    },
    {
      "epoch": 0.09783179660783753,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009799070945890284,
      "loss": 2.0836,
      "step": 14399
    },
    {
      "epoch": 0.09783859095443159,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009799040752288748,
      "loss": 2.0893,
      "step": 14400
    },
    {
      "epoch": 0.09784538530102566,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.000979901055646531,
      "loss": 2.3858,
      "step": 14401
    },
    {
      "epoch": 0.0978521796476197,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000979898035841999,
      "loss": 2.1417,
      "step": 14402
    },
    {
      "epoch": 0.09785897399421377,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009798950158152794,
      "loss": 2.1941,
      "step": 14403
    },
    {
      "epoch": 0.09786576834080782,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009798919955663738,
      "loss": 2.2414,
      "step": 14404
    },
    {
      "epoch": 0.09787256268740188,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009798889750952838,
      "loss": 2.2168,
      "step": 14405
    },
    {
      "epoch": 0.09787935703399593,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009798859544020108,
      "loss": 2.2407,
      "step": 14406
    },
    {
      "epoch": 0.09788615138059,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000979882933486556,
      "loss": 2.2899,
      "step": 14407
    },
    {
      "epoch": 0.09789294572718406,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000979879912348921,
      "loss": 2.3534,
      "step": 14408
    },
    {
      "epoch": 0.09789974007377811,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000979876890989107,
      "loss": 2.1978,
      "step": 14409
    },
    {
      "epoch": 0.09790653442037217,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009798738694071153,
      "loss": 2.2408,
      "step": 14410
    },
    {
      "epoch": 0.09791332876696622,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009798708476029478,
      "loss": 2.2193,
      "step": 14411
    },
    {
      "epoch": 0.09792012311356028,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009798678255766055,
      "loss": 2.2411,
      "step": 14412
    },
    {
      "epoch": 0.09792691746015435,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009798648033280898,
      "loss": 2.1708,
      "step": 14413
    },
    {
      "epoch": 0.0979337118067484,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009798617808574024,
      "loss": 1.9874,
      "step": 14414
    },
    {
      "epoch": 0.09794050615334246,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009798587581645445,
      "loss": 2.1421,
      "step": 14415
    },
    {
      "epoch": 0.09794730049993651,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009798557352495175,
      "loss": 2.3813,
      "step": 14416
    },
    {
      "epoch": 0.09795409484653057,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009798527121123227,
      "loss": 2.1781,
      "step": 14417
    },
    {
      "epoch": 0.09796088919312464,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009798496887529616,
      "loss": 2.3397,
      "step": 14418
    },
    {
      "epoch": 0.09796768353971869,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009798466651714354,
      "loss": 2.252,
      "step": 14419
    },
    {
      "epoch": 0.09797447788631275,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000979843641367746,
      "loss": 2.1091,
      "step": 14420
    },
    {
      "epoch": 0.0979812722329068,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009798406173418945,
      "loss": 2.2815,
      "step": 14421
    },
    {
      "epoch": 0.09798806657950086,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000979837593093882,
      "loss": 2.1223,
      "step": 14422
    },
    {
      "epoch": 0.09799486092609493,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009798345686237106,
      "loss": 2.1508,
      "step": 14423
    },
    {
      "epoch": 0.09800165527268898,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009798315439313809,
      "loss": 2.3516,
      "step": 14424
    },
    {
      "epoch": 0.09800844961928304,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009798285190168949,
      "loss": 2.088,
      "step": 14425
    },
    {
      "epoch": 0.09801524396587709,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009798254938802539,
      "loss": 2.1849,
      "step": 14426
    },
    {
      "epoch": 0.09802203831247115,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000979822468521459,
      "loss": 2.2022,
      "step": 14427
    },
    {
      "epoch": 0.0980288326590652,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000979819442940512,
      "loss": 2.2994,
      "step": 14428
    },
    {
      "epoch": 0.09803562700565926,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000979816417137414,
      "loss": 2.2778,
      "step": 14429
    },
    {
      "epoch": 0.09804242135225333,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009798133911121665,
      "loss": 2.2083,
      "step": 14430
    },
    {
      "epoch": 0.09804921569884738,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009798103648647708,
      "loss": 2.314,
      "step": 14431
    },
    {
      "epoch": 0.09805601004544144,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009798073383952284,
      "loss": 2.2482,
      "step": 14432
    },
    {
      "epoch": 0.09806280439203549,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009798043117035407,
      "loss": 2.1835,
      "step": 14433
    },
    {
      "epoch": 0.09806959873862955,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009798012847897094,
      "loss": 2.2551,
      "step": 14434
    },
    {
      "epoch": 0.09807639308522362,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009797982576537352,
      "loss": 2.2324,
      "step": 14435
    },
    {
      "epoch": 0.09808318743181767,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009797952302956201,
      "loss": 2.1766,
      "step": 14436
    },
    {
      "epoch": 0.09808998177841173,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009797922027153655,
      "loss": 2.1367,
      "step": 14437
    },
    {
      "epoch": 0.09809677612500578,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009797891749129724,
      "loss": 2.1339,
      "step": 14438
    },
    {
      "epoch": 0.09810357047159984,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009797861468884424,
      "loss": 2.2874,
      "step": 14439
    },
    {
      "epoch": 0.09811036481819391,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009797831186417769,
      "loss": 2.1719,
      "step": 14440
    },
    {
      "epoch": 0.09811715916478796,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009797800901729773,
      "loss": 2.4011,
      "step": 14441
    },
    {
      "epoch": 0.09812395351138202,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009797770614820452,
      "loss": 2.0899,
      "step": 14442
    },
    {
      "epoch": 0.09813074785797607,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009797740325689815,
      "loss": 2.3565,
      "step": 14443
    },
    {
      "epoch": 0.09813754220457013,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009797710034337881,
      "loss": 2.2577,
      "step": 14444
    },
    {
      "epoch": 0.0981443365511642,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009797679740764663,
      "loss": 2.1521,
      "step": 14445
    },
    {
      "epoch": 0.09815113089775825,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009797649444970174,
      "loss": 2.4056,
      "step": 14446
    },
    {
      "epoch": 0.09815792524435231,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009797619146954429,
      "loss": 2.2171,
      "step": 14447
    },
    {
      "epoch": 0.09816471959094636,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000979758884671744,
      "loss": 2.2618,
      "step": 14448
    },
    {
      "epoch": 0.09817151393754042,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009797558544259223,
      "loss": 2.1779,
      "step": 14449
    },
    {
      "epoch": 0.09817830828413449,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000979752823957979,
      "loss": 2.1505,
      "step": 14450
    },
    {
      "epoch": 0.09818510263072854,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009797497932679158,
      "loss": 2.1804,
      "step": 14451
    },
    {
      "epoch": 0.0981918969773226,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000979746762355734,
      "loss": 2.2143,
      "step": 14452
    },
    {
      "epoch": 0.09819869132391665,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009797437312214347,
      "loss": 2.2087,
      "step": 14453
    },
    {
      "epoch": 0.09820548567051071,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009797406998650199,
      "loss": 2.3146,
      "step": 14454
    },
    {
      "epoch": 0.09821228001710476,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009797376682864902,
      "loss": 2.2498,
      "step": 14455
    },
    {
      "epoch": 0.09821907436369882,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009797346364858479,
      "loss": 2.177,
      "step": 14456
    },
    {
      "epoch": 0.09822586871029289,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009797316044630937,
      "loss": 2.2285,
      "step": 14457
    },
    {
      "epoch": 0.09823266305688694,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009797285722182294,
      "loss": 2.2036,
      "step": 14458
    },
    {
      "epoch": 0.098239457403481,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009797255397512561,
      "loss": 2.2012,
      "step": 14459
    },
    {
      "epoch": 0.09824625175007505,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009797225070621756,
      "loss": 2.1214,
      "step": 14460
    },
    {
      "epoch": 0.09825304609666911,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009797194741509889,
      "loss": 2.1246,
      "step": 14461
    },
    {
      "epoch": 0.09825984044326318,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009797164410176977,
      "loss": 2.1236,
      "step": 14462
    },
    {
      "epoch": 0.09826663478985723,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009797134076623033,
      "loss": 2.1215,
      "step": 14463
    },
    {
      "epoch": 0.09827342913645129,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000979710374084807,
      "loss": 2.1647,
      "step": 14464
    },
    {
      "epoch": 0.09828022348304534,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009797073402852104,
      "loss": 2.3408,
      "step": 14465
    },
    {
      "epoch": 0.0982870178296394,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009797043062635146,
      "loss": 2.1619,
      "step": 14466
    },
    {
      "epoch": 0.09829381217623347,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009797012720197213,
      "loss": 2.2299,
      "step": 14467
    },
    {
      "epoch": 0.09830060652282752,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009796982375538318,
      "loss": 2.2439,
      "step": 14468
    },
    {
      "epoch": 0.09830740086942158,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009796952028658476,
      "loss": 2.1223,
      "step": 14469
    },
    {
      "epoch": 0.09831419521601563,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00097969216795577,
      "loss": 2.2771,
      "step": 14470
    },
    {
      "epoch": 0.09832098956260969,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009796891328236006,
      "loss": 2.1374,
      "step": 14471
    },
    {
      "epoch": 0.09832778390920376,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009796860974693406,
      "loss": 2.3045,
      "step": 14472
    },
    {
      "epoch": 0.0983345782557978,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000979683061892991,
      "loss": 2.1824,
      "step": 14473
    },
    {
      "epoch": 0.09834137260239187,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000979680026094554,
      "loss": 2.2314,
      "step": 14474
    },
    {
      "epoch": 0.09834816694898592,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009796769900740306,
      "loss": 2.3044,
      "step": 14475
    },
    {
      "epoch": 0.09835496129557998,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009796739538314222,
      "loss": 2.2955,
      "step": 14476
    },
    {
      "epoch": 0.09836175564217403,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009796709173667303,
      "loss": 2.2276,
      "step": 14477
    },
    {
      "epoch": 0.0983685499887681,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000979667880679956,
      "loss": 2.1258,
      "step": 14478
    },
    {
      "epoch": 0.09837534433536216,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009796648437711015,
      "loss": 2.1489,
      "step": 14479
    },
    {
      "epoch": 0.09838213868195621,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009796618066401671,
      "loss": 2.1306,
      "step": 14480
    },
    {
      "epoch": 0.09838893302855027,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009796587692871552,
      "loss": 2.3227,
      "step": 14481
    },
    {
      "epoch": 0.09839572737514432,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009796557317120667,
      "loss": 2.228,
      "step": 14482
    },
    {
      "epoch": 0.09840252172173838,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000979652693914903,
      "loss": 2.2397,
      "step": 14483
    },
    {
      "epoch": 0.09840931606833245,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009796496558956659,
      "loss": 2.2417,
      "step": 14484
    },
    {
      "epoch": 0.0984161104149265,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000979646617654356,
      "loss": 2.2523,
      "step": 14485
    },
    {
      "epoch": 0.09842290476152056,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009796435791909757,
      "loss": 2.1494,
      "step": 14486
    },
    {
      "epoch": 0.09842969910811461,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0009796405405055255,
      "loss": 2.0575,
      "step": 14487
    },
    {
      "epoch": 0.09843649345470867,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009796375015980075,
      "loss": 2.302,
      "step": 14488
    },
    {
      "epoch": 0.09844328780130274,
      "grad_norm": 0.125,
      "learning_rate": 0.0009796344624684228,
      "loss": 2.0729,
      "step": 14489
    },
    {
      "epoch": 0.09845008214789679,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009796314231167729,
      "loss": 2.16,
      "step": 14490
    },
    {
      "epoch": 0.09845687649449085,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000979628383543059,
      "loss": 2.2993,
      "step": 14491
    },
    {
      "epoch": 0.0984636708410849,
      "grad_norm": 0.125,
      "learning_rate": 0.0009796253437472828,
      "loss": 2.1373,
      "step": 14492
    },
    {
      "epoch": 0.09847046518767896,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009796223037294453,
      "loss": 2.2796,
      "step": 14493
    },
    {
      "epoch": 0.09847725953427303,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009796192634895485,
      "loss": 2.2146,
      "step": 14494
    },
    {
      "epoch": 0.09848405388086708,
      "grad_norm": 0.125,
      "learning_rate": 0.0009796162230275933,
      "loss": 2.0667,
      "step": 14495
    },
    {
      "epoch": 0.09849084822746114,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009796131823435813,
      "loss": 2.1079,
      "step": 14496
    },
    {
      "epoch": 0.09849764257405519,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000979610141437514,
      "loss": 2.234,
      "step": 14497
    },
    {
      "epoch": 0.09850443692064925,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009796071003093925,
      "loss": 2.1939,
      "step": 14498
    },
    {
      "epoch": 0.0985112312672433,
      "grad_norm": 0.1826171875,
      "learning_rate": 0.0009796040589592186,
      "loss": 2.2596,
      "step": 14499
    },
    {
      "epoch": 0.09851802561383737,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009796010173869934,
      "loss": 2.1713,
      "step": 14500
    },
    {
      "epoch": 0.09852481996043143,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009795979755927184,
      "loss": 2.2419,
      "step": 14501
    },
    {
      "epoch": 0.09853161430702548,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000979594933576395,
      "loss": 2.1815,
      "step": 14502
    },
    {
      "epoch": 0.09853840865361954,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000979591891338025,
      "loss": 2.2992,
      "step": 14503
    },
    {
      "epoch": 0.09854520300021359,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009795888488776093,
      "loss": 2.2305,
      "step": 14504
    },
    {
      "epoch": 0.09855199734680765,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0009795858061951493,
      "loss": 2.1657,
      "step": 14505
    },
    {
      "epoch": 0.09855879169340172,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009795827632906467,
      "loss": 2.1282,
      "step": 14506
    },
    {
      "epoch": 0.09856558603999577,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009795797201641026,
      "loss": 2.3101,
      "step": 14507
    },
    {
      "epoch": 0.09857238038658983,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009795766768155188,
      "loss": 2.2627,
      "step": 14508
    },
    {
      "epoch": 0.09857917473318388,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009795736332448963,
      "loss": 2.228,
      "step": 14509
    },
    {
      "epoch": 0.09858596907977794,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000979570589452237,
      "loss": 2.287,
      "step": 14510
    },
    {
      "epoch": 0.09859276342637201,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009795675454375417,
      "loss": 2.4002,
      "step": 14511
    },
    {
      "epoch": 0.09859955777296606,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009795645012008122,
      "loss": 2.306,
      "step": 14512
    },
    {
      "epoch": 0.09860635211956012,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009795614567420499,
      "loss": 2.3535,
      "step": 14513
    },
    {
      "epoch": 0.09861314646615417,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009795584120612562,
      "loss": 2.1631,
      "step": 14514
    },
    {
      "epoch": 0.09861994081274823,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009795553671584324,
      "loss": 2.2536,
      "step": 14515
    },
    {
      "epoch": 0.0986267351593423,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009795523220335798,
      "loss": 2.3639,
      "step": 14516
    },
    {
      "epoch": 0.09863352950593635,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009795492766867002,
      "loss": 2.2761,
      "step": 14517
    },
    {
      "epoch": 0.09864032385253041,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009795462311177948,
      "loss": 2.2619,
      "step": 14518
    },
    {
      "epoch": 0.09864711819912446,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009795431853268648,
      "loss": 2.1148,
      "step": 14519
    },
    {
      "epoch": 0.09865391254571852,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009795401393139119,
      "loss": 2.2221,
      "step": 14520
    },
    {
      "epoch": 0.09866070689231259,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009795370930789373,
      "loss": 2.1377,
      "step": 14521
    },
    {
      "epoch": 0.09866750123890664,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009795340466219426,
      "loss": 2.154,
      "step": 14522
    },
    {
      "epoch": 0.0986742955855007,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009795309999429292,
      "loss": 2.1859,
      "step": 14523
    },
    {
      "epoch": 0.09868108993209475,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009795279530418984,
      "loss": 2.2095,
      "step": 14524
    },
    {
      "epoch": 0.09868788427868881,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009795249059188517,
      "loss": 2.2877,
      "step": 14525
    },
    {
      "epoch": 0.09869467862528286,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009795218585737903,
      "loss": 2.1243,
      "step": 14526
    },
    {
      "epoch": 0.09870147297187692,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009795188110067162,
      "loss": 2.1975,
      "step": 14527
    },
    {
      "epoch": 0.09870826731847099,
      "grad_norm": 0.154296875,
      "learning_rate": 0.00097951576321763,
      "loss": 2.3243,
      "step": 14528
    },
    {
      "epoch": 0.09871506166506504,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009795127152065334,
      "loss": 2.1755,
      "step": 14529
    },
    {
      "epoch": 0.0987218560116591,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009795096669734282,
      "loss": 2.1744,
      "step": 14530
    },
    {
      "epoch": 0.09872865035825315,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009795066185183156,
      "loss": 2.1439,
      "step": 14531
    },
    {
      "epoch": 0.09873544470484721,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009795035698411966,
      "loss": 2.2274,
      "step": 14532
    },
    {
      "epoch": 0.09874223905144128,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000979500520942073,
      "loss": 2.3097,
      "step": 14533
    },
    {
      "epoch": 0.09874903339803533,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009794974718209464,
      "loss": 2.1862,
      "step": 14534
    },
    {
      "epoch": 0.09875582774462939,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000979494422477818,
      "loss": 2.1279,
      "step": 14535
    },
    {
      "epoch": 0.09876262209122344,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009794913729126889,
      "loss": 2.1534,
      "step": 14536
    },
    {
      "epoch": 0.0987694164378175,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000979488323125561,
      "loss": 2.2054,
      "step": 14537
    },
    {
      "epoch": 0.09877621078441157,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009794852731164353,
      "loss": 2.199,
      "step": 14538
    },
    {
      "epoch": 0.09878300513100562,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009794822228853136,
      "loss": 2.2911,
      "step": 14539
    },
    {
      "epoch": 0.09878979947759968,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009794791724321969,
      "loss": 2.1424,
      "step": 14540
    },
    {
      "epoch": 0.09879659382419373,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000979476121757087,
      "loss": 2.1944,
      "step": 14541
    },
    {
      "epoch": 0.09880338817078779,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009794730708599852,
      "loss": 2.3172,
      "step": 14542
    },
    {
      "epoch": 0.09881018251738186,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009794700197408928,
      "loss": 2.2674,
      "step": 14543
    },
    {
      "epoch": 0.0988169768639759,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000979466968399811,
      "loss": 2.2892,
      "step": 14544
    },
    {
      "epoch": 0.09882377121056997,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000979463916836742,
      "loss": 2.0533,
      "step": 14545
    },
    {
      "epoch": 0.09883056555716402,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009794608650516865,
      "loss": 2.2475,
      "step": 14546
    },
    {
      "epoch": 0.09883735990375808,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009794578130446461,
      "loss": 2.0682,
      "step": 14547
    },
    {
      "epoch": 0.09884415425035213,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009794547608156221,
      "loss": 2.2473,
      "step": 14548
    },
    {
      "epoch": 0.0988509485969462,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009794517083646162,
      "loss": 2.1029,
      "step": 14549
    },
    {
      "epoch": 0.09885774294354026,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009794486556916296,
      "loss": 2.1805,
      "step": 14550
    },
    {
      "epoch": 0.09886453729013431,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000979445602796664,
      "loss": 2.2691,
      "step": 14551
    },
    {
      "epoch": 0.09887133163672837,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009794425496797202,
      "loss": 2.2049,
      "step": 14552
    },
    {
      "epoch": 0.09887812598332242,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009794394963408002,
      "loss": 2.1191,
      "step": 14553
    },
    {
      "epoch": 0.09888492032991648,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009794364427799052,
      "loss": 2.0734,
      "step": 14554
    },
    {
      "epoch": 0.09889171467651055,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009794333889970367,
      "loss": 2.2954,
      "step": 14555
    },
    {
      "epoch": 0.0988985090231046,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009794303349921958,
      "loss": 2.1338,
      "step": 14556
    },
    {
      "epoch": 0.09890530336969866,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009794272807653844,
      "loss": 2.4789,
      "step": 14557
    },
    {
      "epoch": 0.09891209771629271,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009794242263166035,
      "loss": 2.2945,
      "step": 14558
    },
    {
      "epoch": 0.09891889206288677,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009794211716458547,
      "loss": 2.3742,
      "step": 14559
    },
    {
      "epoch": 0.09892568640948084,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009794181167531394,
      "loss": 2.2113,
      "step": 14560
    },
    {
      "epoch": 0.09893248075607489,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000979415061638459,
      "loss": 2.2016,
      "step": 14561
    },
    {
      "epoch": 0.09893927510266895,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000979412006301815,
      "loss": 2.3892,
      "step": 14562
    },
    {
      "epoch": 0.098946069449263,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009794089507432087,
      "loss": 2.2716,
      "step": 14563
    },
    {
      "epoch": 0.09895286379585706,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009794058949626416,
      "loss": 2.1703,
      "step": 14564
    },
    {
      "epoch": 0.09895965814245113,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000979402838960115,
      "loss": 2.1143,
      "step": 14565
    },
    {
      "epoch": 0.09896645248904518,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009793997827356304,
      "loss": 2.2766,
      "step": 14566
    },
    {
      "epoch": 0.09897324683563924,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009793967262891892,
      "loss": 2.2135,
      "step": 14567
    },
    {
      "epoch": 0.09898004118223329,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009793936696207927,
      "loss": 2.2395,
      "step": 14568
    },
    {
      "epoch": 0.09898683552882735,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009793906127304427,
      "loss": 2.289,
      "step": 14569
    },
    {
      "epoch": 0.09899362987542142,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00097938755561814,
      "loss": 2.4163,
      "step": 14570
    },
    {
      "epoch": 0.09900042422201547,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009793844982838866,
      "loss": 2.3379,
      "step": 14571
    },
    {
      "epoch": 0.09900721856860953,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009793814407276836,
      "loss": 2.1501,
      "step": 14572
    },
    {
      "epoch": 0.09901401291520358,
      "grad_norm": 0.125,
      "learning_rate": 0.0009793783829495325,
      "loss": 2.2452,
      "step": 14573
    },
    {
      "epoch": 0.09902080726179764,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009793753249494348,
      "loss": 2.2772,
      "step": 14574
    },
    {
      "epoch": 0.09902760160839169,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0009793722667273917,
      "loss": 2.1458,
      "step": 14575
    },
    {
      "epoch": 0.09903439595498575,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009793692082834047,
      "loss": 2.287,
      "step": 14576
    },
    {
      "epoch": 0.09904119030157982,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009793661496174753,
      "loss": 2.1195,
      "step": 14577
    },
    {
      "epoch": 0.09904798464817387,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009793630907296048,
      "loss": 2.1884,
      "step": 14578
    },
    {
      "epoch": 0.09905477899476793,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009793600316197948,
      "loss": 1.9976,
      "step": 14579
    },
    {
      "epoch": 0.09906157334136198,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009793569722880465,
      "loss": 2.27,
      "step": 14580
    },
    {
      "epoch": 0.09906836768795604,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009793539127343615,
      "loss": 2.3123,
      "step": 14581
    },
    {
      "epoch": 0.09907516203455011,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000979350852958741,
      "loss": 2.3663,
      "step": 14582
    },
    {
      "epoch": 0.09908195638114416,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009793477929611867,
      "loss": 2.1045,
      "step": 14583
    },
    {
      "epoch": 0.09908875072773822,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009793447327417,
      "loss": 2.3321,
      "step": 14584
    },
    {
      "epoch": 0.09909554507433227,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000979341672300282,
      "loss": 2.2487,
      "step": 14585
    },
    {
      "epoch": 0.09910233942092633,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009793386116369342,
      "loss": 2.2934,
      "step": 14586
    },
    {
      "epoch": 0.0991091337675204,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009793355507516583,
      "loss": 2.3778,
      "step": 14587
    },
    {
      "epoch": 0.09911592811411445,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009793324896444555,
      "loss": 2.3111,
      "step": 14588
    },
    {
      "epoch": 0.09912272246070851,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009793294283153271,
      "loss": 2.2421,
      "step": 14589
    },
    {
      "epoch": 0.09912951680730256,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000979326366764275,
      "loss": 2.2094,
      "step": 14590
    },
    {
      "epoch": 0.09913631115389662,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009793233049913,
      "loss": 2.2139,
      "step": 14591
    },
    {
      "epoch": 0.09914310550049069,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000979320242996404,
      "loss": 2.2709,
      "step": 14592
    },
    {
      "epoch": 0.09914989984708474,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000979317180779588,
      "loss": 2.1493,
      "step": 14593
    },
    {
      "epoch": 0.0991566941936788,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009793141183408536,
      "loss": 2.2212,
      "step": 14594
    },
    {
      "epoch": 0.09916348854027285,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009793110556802025,
      "loss": 2.2608,
      "step": 14595
    },
    {
      "epoch": 0.09917028288686691,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009793079927976358,
      "loss": 2.0288,
      "step": 14596
    },
    {
      "epoch": 0.09917707723346096,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000979304929693155,
      "loss": 2.1949,
      "step": 14597
    },
    {
      "epoch": 0.09918387158005502,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009793018663667614,
      "loss": 2.2069,
      "step": 14598
    },
    {
      "epoch": 0.09919066592664909,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009792988028184567,
      "loss": 2.207,
      "step": 14599
    },
    {
      "epoch": 0.09919746027324314,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000979295739048242,
      "loss": 2.2374,
      "step": 14600
    },
    {
      "epoch": 0.0992042546198372,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009792926750561192,
      "loss": 2.113,
      "step": 14601
    },
    {
      "epoch": 0.09921104896643125,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009792896108420889,
      "loss": 2.2436,
      "step": 14602
    },
    {
      "epoch": 0.09921784331302531,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009792865464061532,
      "loss": 2.1828,
      "step": 14603
    },
    {
      "epoch": 0.09922463765961938,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009792834817483134,
      "loss": 2.1707,
      "step": 14604
    },
    {
      "epoch": 0.09923143200621343,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009792804168685708,
      "loss": 2.1431,
      "step": 14605
    },
    {
      "epoch": 0.09923822635280749,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000979277351766927,
      "loss": 2.1851,
      "step": 14606
    },
    {
      "epoch": 0.09924502069940154,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000979274286443383,
      "loss": 2.3255,
      "step": 14607
    },
    {
      "epoch": 0.0992518150459956,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009792712208979407,
      "loss": 2.2484,
      "step": 14608
    },
    {
      "epoch": 0.09925860939258967,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009792681551306012,
      "loss": 2.3094,
      "step": 14609
    },
    {
      "epoch": 0.09926540373918372,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009792650891413662,
      "loss": 2.1535,
      "step": 14610
    },
    {
      "epoch": 0.09927219808577778,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009792620229302368,
      "loss": 2.1364,
      "step": 14611
    },
    {
      "epoch": 0.09927899243237183,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009792589564972145,
      "loss": 2.2612,
      "step": 14612
    },
    {
      "epoch": 0.09928578677896589,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000979255889842301,
      "loss": 2.0356,
      "step": 14613
    },
    {
      "epoch": 0.09929258112555996,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009792528229654975,
      "loss": 2.0982,
      "step": 14614
    },
    {
      "epoch": 0.099299375472154,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009792497558668053,
      "loss": 2.3108,
      "step": 14615
    },
    {
      "epoch": 0.09930616981874807,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009792466885462261,
      "loss": 2.2735,
      "step": 14616
    },
    {
      "epoch": 0.09931296416534212,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000979243621003761,
      "loss": 2.2971,
      "step": 14617
    },
    {
      "epoch": 0.09931975851193618,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009792405532394117,
      "loss": 2.212,
      "step": 14618
    },
    {
      "epoch": 0.09932655285853023,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009792374852531796,
      "loss": 2.0665,
      "step": 14619
    },
    {
      "epoch": 0.0993333472051243,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000979234417045066,
      "loss": 2.2668,
      "step": 14620
    },
    {
      "epoch": 0.09934014155171836,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009792313486150723,
      "loss": 2.2515,
      "step": 14621
    },
    {
      "epoch": 0.09934693589831241,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009792282799632,
      "loss": 2.23,
      "step": 14622
    },
    {
      "epoch": 0.09935373024490647,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0009792252110894506,
      "loss": 2.1589,
      "step": 14623
    },
    {
      "epoch": 0.09936052459150052,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009792221419938254,
      "loss": 2.1287,
      "step": 14624
    },
    {
      "epoch": 0.09936731893809458,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009792190726763256,
      "loss": 2.2827,
      "step": 14625
    },
    {
      "epoch": 0.09937411328468865,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000979216003136953,
      "loss": 2.0841,
      "step": 14626
    },
    {
      "epoch": 0.0993809076312827,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009792129333757091,
      "loss": 2.2557,
      "step": 14627
    },
    {
      "epoch": 0.09938770197787676,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009792098633925948,
      "loss": 2.2023,
      "step": 14628
    },
    {
      "epoch": 0.09939449632447081,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000979206793187612,
      "loss": 2.2443,
      "step": 14629
    },
    {
      "epoch": 0.09940129067106487,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000979203722760762,
      "loss": 2.2358,
      "step": 14630
    },
    {
      "epoch": 0.09940808501765894,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000979200652112046,
      "loss": 2.2111,
      "step": 14631
    },
    {
      "epoch": 0.09941487936425299,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009791975812414658,
      "loss": 2.1752,
      "step": 14632
    },
    {
      "epoch": 0.09942167371084705,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009791945101490223,
      "loss": 2.2315,
      "step": 14633
    },
    {
      "epoch": 0.0994284680574411,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009791914388347174,
      "loss": 2.1724,
      "step": 14634
    },
    {
      "epoch": 0.09943526240403516,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009791883672985524,
      "loss": 2.2547,
      "step": 14635
    },
    {
      "epoch": 0.09944205675062923,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009791852955405287,
      "loss": 2.1267,
      "step": 14636
    },
    {
      "epoch": 0.09944885109722328,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009791822235606475,
      "loss": 2.1513,
      "step": 14637
    },
    {
      "epoch": 0.09945564544381734,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009791791513589108,
      "loss": 2.196,
      "step": 14638
    },
    {
      "epoch": 0.09946243979041139,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0009791760789353194,
      "loss": 2.1191,
      "step": 14639
    },
    {
      "epoch": 0.09946923413700545,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000979173006289875,
      "loss": 2.2884,
      "step": 14640
    },
    {
      "epoch": 0.09947602848359952,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000979169933422579,
      "loss": 2.1575,
      "step": 14641
    },
    {
      "epoch": 0.09948282283019357,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009791668603334328,
      "loss": 2.1705,
      "step": 14642
    },
    {
      "epoch": 0.09948961717678763,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009791637870224378,
      "loss": 2.2028,
      "step": 14643
    },
    {
      "epoch": 0.09949641152338168,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009791607134895955,
      "loss": 2.2674,
      "step": 14644
    },
    {
      "epoch": 0.09950320586997574,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009791576397349072,
      "loss": 2.1466,
      "step": 14645
    },
    {
      "epoch": 0.09951000021656979,
      "grad_norm": 0.125,
      "learning_rate": 0.0009791545657583745,
      "loss": 2.1314,
      "step": 14646
    },
    {
      "epoch": 0.09951679456316385,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009791514915599988,
      "loss": 2.246,
      "step": 14647
    },
    {
      "epoch": 0.09952358890975792,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009791484171397813,
      "loss": 2.0604,
      "step": 14648
    },
    {
      "epoch": 0.09953038325635197,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009791453424977236,
      "loss": 2.0723,
      "step": 14649
    },
    {
      "epoch": 0.09953717760294603,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009791422676338271,
      "loss": 2.3026,
      "step": 14650
    },
    {
      "epoch": 0.09954397194954008,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009791391925480933,
      "loss": 2.2934,
      "step": 14651
    },
    {
      "epoch": 0.09955076629613414,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009791361172405235,
      "loss": 2.1004,
      "step": 14652
    },
    {
      "epoch": 0.09955756064272821,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009791330417111192,
      "loss": 2.1914,
      "step": 14653
    },
    {
      "epoch": 0.09956435498932226,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009791299659598818,
      "loss": 2.2075,
      "step": 14654
    },
    {
      "epoch": 0.09957114933591632,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009791268899868128,
      "loss": 2.0931,
      "step": 14655
    },
    {
      "epoch": 0.09957794368251037,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009791238137919133,
      "loss": 2.1214,
      "step": 14656
    },
    {
      "epoch": 0.09958473802910443,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009791207373751852,
      "loss": 2.2356,
      "step": 14657
    },
    {
      "epoch": 0.0995915323756985,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009791176607366297,
      "loss": 2.2804,
      "step": 14658
    },
    {
      "epoch": 0.09959832672229255,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000979114583876248,
      "loss": 2.299,
      "step": 14659
    },
    {
      "epoch": 0.09960512106888661,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000979111506794042,
      "loss": 2.1217,
      "step": 14660
    },
    {
      "epoch": 0.09961191541548066,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009791084294900126,
      "loss": 2.2608,
      "step": 14661
    },
    {
      "epoch": 0.09961870976207472,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009791053519641617,
      "loss": 2.1123,
      "step": 14662
    },
    {
      "epoch": 0.09962550410866879,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009791022742164904,
      "loss": 2.2725,
      "step": 14663
    },
    {
      "epoch": 0.09963229845526284,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009790991962470002,
      "loss": 2.1713,
      "step": 14664
    },
    {
      "epoch": 0.0996390928018569,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009790961180556927,
      "loss": 2.3152,
      "step": 14665
    },
    {
      "epoch": 0.09964588714845095,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000979093039642569,
      "loss": 2.1448,
      "step": 14666
    },
    {
      "epoch": 0.09965268149504501,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000979089961007631,
      "loss": 2.2312,
      "step": 14667
    },
    {
      "epoch": 0.09965947584163906,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009790868821508796,
      "loss": 2.1287,
      "step": 14668
    },
    {
      "epoch": 0.09966627018823312,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009790838030723165,
      "loss": 1.9558,
      "step": 14669
    },
    {
      "epoch": 0.09967306453482719,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009790807237719432,
      "loss": 2.2273,
      "step": 14670
    },
    {
      "epoch": 0.09967985888142124,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009790776442497608,
      "loss": 2.1885,
      "step": 14671
    },
    {
      "epoch": 0.0996866532280153,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009790745645057713,
      "loss": 2.2407,
      "step": 14672
    },
    {
      "epoch": 0.09969344757460935,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009790714845399756,
      "loss": 2.1274,
      "step": 14673
    },
    {
      "epoch": 0.09970024192120341,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009790684043523752,
      "loss": 2.2384,
      "step": 14674
    },
    {
      "epoch": 0.09970703626779748,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009790653239429717,
      "loss": 2.0537,
      "step": 14675
    },
    {
      "epoch": 0.09971383061439153,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009790622433117663,
      "loss": 2.2311,
      "step": 14676
    },
    {
      "epoch": 0.09972062496098559,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009790591624587609,
      "loss": 2.2705,
      "step": 14677
    },
    {
      "epoch": 0.09972741930757964,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009790560813839563,
      "loss": 2.3119,
      "step": 14678
    },
    {
      "epoch": 0.0997342136541737,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009790530000873543,
      "loss": 2.2913,
      "step": 14679
    },
    {
      "epoch": 0.09974100800076777,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009790499185689562,
      "loss": 2.0741,
      "step": 14680
    },
    {
      "epoch": 0.09974780234736182,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009790468368287635,
      "loss": 2.2419,
      "step": 14681
    },
    {
      "epoch": 0.09975459669395588,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009790437548667778,
      "loss": 2.21,
      "step": 14682
    },
    {
      "epoch": 0.09976139104054993,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.000979040672683,
      "loss": 2.2315,
      "step": 14683
    },
    {
      "epoch": 0.09976818538714399,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009790375902774323,
      "loss": 2.1248,
      "step": 14684
    },
    {
      "epoch": 0.09977497973373806,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009790345076500753,
      "loss": 2.391,
      "step": 14685
    },
    {
      "epoch": 0.0997817740803321,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009790314248009312,
      "loss": 2.1526,
      "step": 14686
    },
    {
      "epoch": 0.09978856842692617,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009790283417300007,
      "loss": 2.0937,
      "step": 14687
    },
    {
      "epoch": 0.09979536277352022,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009790252584372856,
      "loss": 2.1956,
      "step": 14688
    },
    {
      "epoch": 0.09980215712011428,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009790221749227874,
      "loss": 2.221,
      "step": 14689
    },
    {
      "epoch": 0.09980895146670835,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009790190911865074,
      "loss": 2.2908,
      "step": 14690
    },
    {
      "epoch": 0.0998157458133024,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000979016007228447,
      "loss": 2.2412,
      "step": 14691
    },
    {
      "epoch": 0.09982254015989646,
      "grad_norm": 0.125,
      "learning_rate": 0.0009790129230486078,
      "loss": 2.1503,
      "step": 14692
    },
    {
      "epoch": 0.09982933450649051,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000979009838646991,
      "loss": 2.2054,
      "step": 14693
    },
    {
      "epoch": 0.09983612885308457,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009790067540235983,
      "loss": 2.2226,
      "step": 14694
    },
    {
      "epoch": 0.09984292319967862,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009790036691784308,
      "loss": 2.2749,
      "step": 14695
    },
    {
      "epoch": 0.09984971754627268,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009790005841114901,
      "loss": 2.2768,
      "step": 14696
    },
    {
      "epoch": 0.09985651189286675,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009789974988227777,
      "loss": 2.2185,
      "step": 14697
    },
    {
      "epoch": 0.0998633062394608,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009789944133122947,
      "loss": 2.3563,
      "step": 14698
    },
    {
      "epoch": 0.09987010058605486,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000978991327580043,
      "loss": 2.2026,
      "step": 14699
    },
    {
      "epoch": 0.09987689493264891,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009789882416260237,
      "loss": 2.1488,
      "step": 14700
    },
    {
      "epoch": 0.09988368927924297,
      "grad_norm": 0.197265625,
      "learning_rate": 0.0009789851554502385,
      "loss": 2.2982,
      "step": 14701
    },
    {
      "epoch": 0.09989048362583704,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009789820690526887,
      "loss": 2.1659,
      "step": 14702
    },
    {
      "epoch": 0.09989727797243109,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009789789824333757,
      "loss": 2.2624,
      "step": 14703
    },
    {
      "epoch": 0.09990407231902515,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009789758955923006,
      "loss": 2.1814,
      "step": 14704
    },
    {
      "epoch": 0.0999108666656192,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009789728085294654,
      "loss": 2.2182,
      "step": 14705
    },
    {
      "epoch": 0.09991766101221326,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009789697212448712,
      "loss": 2.246,
      "step": 14706
    },
    {
      "epoch": 0.09992445535880733,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009789666337385195,
      "loss": 2.3212,
      "step": 14707
    },
    {
      "epoch": 0.09993124970540138,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009789635460104118,
      "loss": 2.2824,
      "step": 14708
    },
    {
      "epoch": 0.09993804405199544,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009789604580605496,
      "loss": 2.2547,
      "step": 14709
    },
    {
      "epoch": 0.09994483839858949,
      "grad_norm": 0.140625,
      "learning_rate": 0.000978957369888934,
      "loss": 2.2226,
      "step": 14710
    },
    {
      "epoch": 0.09995163274518355,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009789542814955666,
      "loss": 2.099,
      "step": 14711
    },
    {
      "epoch": 0.09995842709177762,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000978951192880449,
      "loss": 2.2224,
      "step": 14712
    },
    {
      "epoch": 0.09996522143837167,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009789481040435824,
      "loss": 2.1478,
      "step": 14713
    },
    {
      "epoch": 0.09997201578496573,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009789450149849683,
      "loss": 2.14,
      "step": 14714
    },
    {
      "epoch": 0.09997881013155978,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009789419257046083,
      "loss": 2.2823,
      "step": 14715
    },
    {
      "epoch": 0.09998560447815384,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009789388362025037,
      "loss": 2.152,
      "step": 14716
    },
    {
      "epoch": 0.09999239882474789,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009789357464786556,
      "loss": 2.3542,
      "step": 14717
    },
    {
      "epoch": 0.09999919317134195,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000978932656533066,
      "loss": 2.2006,
      "step": 14718
    },
    {
      "epoch": 0.10000598751793602,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000978929566365736,
      "loss": 2.1756,
      "step": 14719
    },
    {
      "epoch": 0.10001278186453007,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009789264759766671,
      "loss": 2.13,
      "step": 14720
    },
    {
      "epoch": 0.10001957621112413,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009789233853658608,
      "loss": 2.2571,
      "step": 14721
    },
    {
      "epoch": 0.10002637055771818,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009789202945333182,
      "loss": 2.2516,
      "step": 14722
    },
    {
      "epoch": 0.10003316490431224,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009789172034790413,
      "loss": 2.0657,
      "step": 14723
    },
    {
      "epoch": 0.10003995925090631,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009789141122030312,
      "loss": 2.2288,
      "step": 14724
    },
    {
      "epoch": 0.10004675359750036,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000978911020705289,
      "loss": 2.1847,
      "step": 14725
    },
    {
      "epoch": 0.10005354794409442,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009789079289858169,
      "loss": 2.2836,
      "step": 14726
    },
    {
      "epoch": 0.10006034229068847,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009789048370446158,
      "loss": 2.2186,
      "step": 14727
    },
    {
      "epoch": 0.10006713663728253,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009789017448816873,
      "loss": 2.266,
      "step": 14728
    },
    {
      "epoch": 0.1000739309838766,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009788986524970325,
      "loss": 2.2335,
      "step": 14729
    },
    {
      "epoch": 0.10008072533047065,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009788955598906534,
      "loss": 2.2224,
      "step": 14730
    },
    {
      "epoch": 0.10008751967706471,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000978892467062551,
      "loss": 2.2874,
      "step": 14731
    },
    {
      "epoch": 0.10009431402365876,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000978889374012727,
      "loss": 2.1761,
      "step": 14732
    },
    {
      "epoch": 0.10010110837025282,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009788862807411827,
      "loss": 2.1278,
      "step": 14733
    },
    {
      "epoch": 0.10010790271684689,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009788831872479195,
      "loss": 2.191,
      "step": 14734
    },
    {
      "epoch": 0.10011469706344094,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009788800935329389,
      "loss": 2.1986,
      "step": 14735
    },
    {
      "epoch": 0.100121491410035,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009788769995962424,
      "loss": 2.1589,
      "step": 14736
    },
    {
      "epoch": 0.10012828575662905,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009788739054378312,
      "loss": 2.2922,
      "step": 14737
    },
    {
      "epoch": 0.10013508010322311,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009788708110577066,
      "loss": 2.2331,
      "step": 14738
    },
    {
      "epoch": 0.10014187444981716,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009788677164558708,
      "loss": 2.2309,
      "step": 14739
    },
    {
      "epoch": 0.10014866879641122,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009788646216323245,
      "loss": 2.2802,
      "step": 14740
    },
    {
      "epoch": 0.10015546314300529,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009788615265870694,
      "loss": 2.1925,
      "step": 14741
    },
    {
      "epoch": 0.10016225748959934,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000978858431320107,
      "loss": 2.3212,
      "step": 14742
    },
    {
      "epoch": 0.1001690518361934,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009788553358314384,
      "loss": 2.2996,
      "step": 14743
    },
    {
      "epoch": 0.10017584618278745,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009788522401210655,
      "loss": 2.271,
      "step": 14744
    },
    {
      "epoch": 0.10018264052938151,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009788491441889895,
      "loss": 2.2029,
      "step": 14745
    },
    {
      "epoch": 0.10018943487597558,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009788460480352117,
      "loss": 2.2523,
      "step": 14746
    },
    {
      "epoch": 0.10019622922256963,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009788429516597336,
      "loss": 2.1848,
      "step": 14747
    },
    {
      "epoch": 0.10020302356916369,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000978839855062557,
      "loss": 2.3477,
      "step": 14748
    },
    {
      "epoch": 0.10020981791575774,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009788367582436828,
      "loss": 2.1693,
      "step": 14749
    },
    {
      "epoch": 0.1002166122623518,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009788336612031126,
      "loss": 2.1031,
      "step": 14750
    },
    {
      "epoch": 0.10022340660894587,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009788305639408482,
      "loss": 2.3522,
      "step": 14751
    },
    {
      "epoch": 0.10023020095553992,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009788274664568905,
      "loss": 2.338,
      "step": 14752
    },
    {
      "epoch": 0.10023699530213398,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009788243687512412,
      "loss": 2.1196,
      "step": 14753
    },
    {
      "epoch": 0.10024378964872803,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009788212708239019,
      "loss": 2.2485,
      "step": 14754
    },
    {
      "epoch": 0.10025058399532209,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009788181726748735,
      "loss": 2.3042,
      "step": 14755
    },
    {
      "epoch": 0.10025737834191616,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009788150743041582,
      "loss": 2.3679,
      "step": 14756
    },
    {
      "epoch": 0.1002641726885102,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009788119757117566,
      "loss": 2.1703,
      "step": 14757
    },
    {
      "epoch": 0.10027096703510427,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009788088768976709,
      "loss": 2.1495,
      "step": 14758
    },
    {
      "epoch": 0.10027776138169832,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000978805777861902,
      "loss": 2.3007,
      "step": 14759
    },
    {
      "epoch": 0.10028455572829238,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009788026786044515,
      "loss": 2.1724,
      "step": 14760
    },
    {
      "epoch": 0.10029135007488645,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009787995791253209,
      "loss": 2.1809,
      "step": 14761
    },
    {
      "epoch": 0.1002981444214805,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009787964794245115,
      "loss": 2.0279,
      "step": 14762
    },
    {
      "epoch": 0.10030493876807456,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009787933795020248,
      "loss": 2.0705,
      "step": 14763
    },
    {
      "epoch": 0.10031173311466861,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009787902793578621,
      "loss": 2.1155,
      "step": 14764
    },
    {
      "epoch": 0.10031852746126267,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009787871789920255,
      "loss": 2.1393,
      "step": 14765
    },
    {
      "epoch": 0.10032532180785672,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009787840784045156,
      "loss": 2.1279,
      "step": 14766
    },
    {
      "epoch": 0.10033211615445078,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009787809775953342,
      "loss": 2.0271,
      "step": 14767
    },
    {
      "epoch": 0.10033891050104485,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009787778765644826,
      "loss": 2.3049,
      "step": 14768
    },
    {
      "epoch": 0.1003457048476389,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009787747753119624,
      "loss": 2.2912,
      "step": 14769
    },
    {
      "epoch": 0.10035249919423296,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000978771673837775,
      "loss": 2.0069,
      "step": 14770
    },
    {
      "epoch": 0.10035929354082701,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000978768572141922,
      "loss": 2.218,
      "step": 14771
    },
    {
      "epoch": 0.10036608788742107,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009787654702244043,
      "loss": 2.1793,
      "step": 14772
    },
    {
      "epoch": 0.10037288223401514,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009787623680852238,
      "loss": 2.2061,
      "step": 14773
    },
    {
      "epoch": 0.10037967658060919,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000978759265724382,
      "loss": 2.1532,
      "step": 14774
    },
    {
      "epoch": 0.10038647092720325,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00097875616314188,
      "loss": 2.1405,
      "step": 14775
    },
    {
      "epoch": 0.1003932652737973,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009787530603377194,
      "loss": 2.0369,
      "step": 14776
    },
    {
      "epoch": 0.10040005962039136,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009787499573119015,
      "loss": 2.2525,
      "step": 14777
    },
    {
      "epoch": 0.10040685396698543,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009787468540644279,
      "loss": 2.1256,
      "step": 14778
    },
    {
      "epoch": 0.10041364831357948,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009787437505953,
      "loss": 2.2093,
      "step": 14779
    },
    {
      "epoch": 0.10042044266017354,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009787406469045192,
      "loss": 2.2837,
      "step": 14780
    },
    {
      "epoch": 0.10042723700676759,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000978737542992087,
      "loss": 2.1986,
      "step": 14781
    },
    {
      "epoch": 0.10043403135336165,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000978734438858005,
      "loss": 2.1696,
      "step": 14782
    },
    {
      "epoch": 0.10044082569995572,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009787313345022742,
      "loss": 2.2387,
      "step": 14783
    },
    {
      "epoch": 0.10044762004654977,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009787282299248965,
      "loss": 2.297,
      "step": 14784
    },
    {
      "epoch": 0.10045441439314383,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000978725125125873,
      "loss": 2.1274,
      "step": 14785
    },
    {
      "epoch": 0.10046120873973788,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009787220201052053,
      "loss": 2.1687,
      "step": 14786
    },
    {
      "epoch": 0.10046800308633194,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009787189148628946,
      "loss": 2.2018,
      "step": 14787
    },
    {
      "epoch": 0.10047479743292599,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009787158093989426,
      "loss": 2.2178,
      "step": 14788
    },
    {
      "epoch": 0.10048159177952005,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009787127037133507,
      "loss": 2.1438,
      "step": 14789
    },
    {
      "epoch": 0.10048838612611412,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009787095978061205,
      "loss": 2.1376,
      "step": 14790
    },
    {
      "epoch": 0.10049518047270817,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000978706491677253,
      "loss": 2.1719,
      "step": 14791
    },
    {
      "epoch": 0.10050197481930223,
      "grad_norm": 0.125,
      "learning_rate": 0.00097870338532675,
      "loss": 2.1196,
      "step": 14792
    },
    {
      "epoch": 0.10050876916589628,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009787002787546128,
      "loss": 2.1401,
      "step": 14793
    },
    {
      "epoch": 0.10051556351249034,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009786971719608429,
      "loss": 2.2584,
      "step": 14794
    },
    {
      "epoch": 0.10052235785908441,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009786940649454416,
      "loss": 2.1395,
      "step": 14795
    },
    {
      "epoch": 0.10052915220567846,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009786909577084103,
      "loss": 2.2617,
      "step": 14796
    },
    {
      "epoch": 0.10053594655227252,
      "grad_norm": 0.123046875,
      "learning_rate": 0.000978687850249751,
      "loss": 2.0866,
      "step": 14797
    },
    {
      "epoch": 0.10054274089886657,
      "grad_norm": 0.125,
      "learning_rate": 0.0009786847425694643,
      "loss": 2.1474,
      "step": 14798
    },
    {
      "epoch": 0.10054953524546063,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009786816346675521,
      "loss": 2.2385,
      "step": 14799
    },
    {
      "epoch": 0.1005563295920547,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000978678526544016,
      "loss": 2.1525,
      "step": 14800
    },
    {
      "epoch": 0.10056312393864875,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009786754181988573,
      "loss": 2.1504,
      "step": 14801
    },
    {
      "epoch": 0.10056991828524281,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000978672309632077,
      "loss": 2.368,
      "step": 14802
    },
    {
      "epoch": 0.10057671263183686,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009786692008436773,
      "loss": 2.2316,
      "step": 14803
    },
    {
      "epoch": 0.10058350697843092,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000978666091833659,
      "loss": 2.1746,
      "step": 14804
    },
    {
      "epoch": 0.10059030132502499,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978662982602024,
      "loss": 2.1207,
      "step": 14805
    },
    {
      "epoch": 0.10059709567161904,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009786598731487733,
      "loss": 2.0742,
      "step": 14806
    },
    {
      "epoch": 0.1006038900182131,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009786567634739085,
      "loss": 2.2344,
      "step": 14807
    },
    {
      "epoch": 0.10061068436480715,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009786536535774312,
      "loss": 2.0624,
      "step": 14808
    },
    {
      "epoch": 0.10061747871140121,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009786505434593428,
      "loss": 2.2364,
      "step": 14809
    },
    {
      "epoch": 0.10062427305799528,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009786474331196448,
      "loss": 2.3602,
      "step": 14810
    },
    {
      "epoch": 0.10063106740458932,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009786443225583384,
      "loss": 2.1068,
      "step": 14811
    },
    {
      "epoch": 0.10063786175118339,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009786412117754253,
      "loss": 2.2215,
      "step": 14812
    },
    {
      "epoch": 0.10064465609777744,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009786381007709066,
      "loss": 2.4064,
      "step": 14813
    },
    {
      "epoch": 0.1006514504443715,
      "grad_norm": 0.125,
      "learning_rate": 0.000978634989544784,
      "loss": 2.1352,
      "step": 14814
    },
    {
      "epoch": 0.10065824479096555,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009786318780970588,
      "loss": 2.3403,
      "step": 14815
    },
    {
      "epoch": 0.10066503913755961,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009786287664277327,
      "loss": 2.2307,
      "step": 14816
    },
    {
      "epoch": 0.10067183348415368,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000978625654536807,
      "loss": 2.3944,
      "step": 14817
    },
    {
      "epoch": 0.10067862783074773,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000978622542424283,
      "loss": 2.2732,
      "step": 14818
    },
    {
      "epoch": 0.10068542217734179,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009786194300901623,
      "loss": 2.1368,
      "step": 14819
    },
    {
      "epoch": 0.10069221652393584,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009786163175344463,
      "loss": 2.1857,
      "step": 14820
    },
    {
      "epoch": 0.1006990108705299,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009786132047571363,
      "loss": 2.2347,
      "step": 14821
    },
    {
      "epoch": 0.10070580521712397,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000978610091758234,
      "loss": 2.2319,
      "step": 14822
    },
    {
      "epoch": 0.10071259956371802,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009786069785377408,
      "loss": 2.1959,
      "step": 14823
    },
    {
      "epoch": 0.10071939391031208,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009786038650956579,
      "loss": 2.2713,
      "step": 14824
    },
    {
      "epoch": 0.10072618825690613,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000978600751431987,
      "loss": 2.3133,
      "step": 14825
    },
    {
      "epoch": 0.10073298260350019,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009785976375467293,
      "loss": 2.1666,
      "step": 14826
    },
    {
      "epoch": 0.10073977695009426,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009785945234398864,
      "loss": 2.2969,
      "step": 14827
    },
    {
      "epoch": 0.1007465712966883,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009785914091114598,
      "loss": 2.2085,
      "step": 14828
    },
    {
      "epoch": 0.10075336564328237,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009785882945614508,
      "loss": 2.2256,
      "step": 14829
    },
    {
      "epoch": 0.10076015998987642,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000978585179789861,
      "loss": 2.2511,
      "step": 14830
    },
    {
      "epoch": 0.10076695433647048,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009785820647966916,
      "loss": 2.1804,
      "step": 14831
    },
    {
      "epoch": 0.10077374868306455,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009785789495819445,
      "loss": 2.1664,
      "step": 14832
    },
    {
      "epoch": 0.1007805430296586,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009785758341456206,
      "loss": 2.1299,
      "step": 14833
    },
    {
      "epoch": 0.10078733737625266,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009785727184877216,
      "loss": 2.2428,
      "step": 14834
    },
    {
      "epoch": 0.10079413172284671,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000978569602608249,
      "loss": 2.1135,
      "step": 14835
    },
    {
      "epoch": 0.10080092606944077,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009785664865072038,
      "loss": 2.224,
      "step": 14836
    },
    {
      "epoch": 0.10080772041603482,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009785633701845883,
      "loss": 2.2322,
      "step": 14837
    },
    {
      "epoch": 0.10081451476262888,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009785602536404032,
      "loss": 2.2356,
      "step": 14838
    },
    {
      "epoch": 0.10082130910922295,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009785571368746502,
      "loss": 2.101,
      "step": 14839
    },
    {
      "epoch": 0.100828103455817,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009785540198873307,
      "loss": 2.3651,
      "step": 14840
    },
    {
      "epoch": 0.10083489780241106,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009785509026784462,
      "loss": 2.2616,
      "step": 14841
    },
    {
      "epoch": 0.10084169214900511,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000978547785247998,
      "loss": 2.2816,
      "step": 14842
    },
    {
      "epoch": 0.10084848649559917,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009785446675959879,
      "loss": 2.1542,
      "step": 14843
    },
    {
      "epoch": 0.10085528084219324,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000978541549722417,
      "loss": 2.2146,
      "step": 14844
    },
    {
      "epoch": 0.10086207518878729,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978538431627287,
      "loss": 2.4126,
      "step": 14845
    },
    {
      "epoch": 0.10086886953538135,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000978535313310599,
      "loss": 2.2243,
      "step": 14846
    },
    {
      "epoch": 0.1008756638819754,
      "grad_norm": 0.125,
      "learning_rate": 0.0009785321947723548,
      "loss": 2.2134,
      "step": 14847
    },
    {
      "epoch": 0.10088245822856946,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009785290760125553,
      "loss": 2.2469,
      "step": 14848
    },
    {
      "epoch": 0.10088925257516353,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009785259570312026,
      "loss": 2.2338,
      "step": 14849
    },
    {
      "epoch": 0.10089604692175758,
      "grad_norm": 0.162109375,
      "learning_rate": 0.000978522837828298,
      "loss": 2.2882,
      "step": 14850
    },
    {
      "epoch": 0.10090284126835164,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009785197184038427,
      "loss": 2.2826,
      "step": 14851
    },
    {
      "epoch": 0.10090963561494569,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0009785165987578383,
      "loss": 2.3438,
      "step": 14852
    },
    {
      "epoch": 0.10091642996153975,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978513478890286,
      "loss": 2.1131,
      "step": 14853
    },
    {
      "epoch": 0.10092322430813382,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009785103588011875,
      "loss": 2.2369,
      "step": 14854
    },
    {
      "epoch": 0.10093001865472787,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009785072384905441,
      "loss": 2.2066,
      "step": 14855
    },
    {
      "epoch": 0.10093681300132193,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009785041179583578,
      "loss": 2.1236,
      "step": 14856
    },
    {
      "epoch": 0.10094360734791598,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009785009972046292,
      "loss": 2.2062,
      "step": 14857
    },
    {
      "epoch": 0.10095040169451004,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009784978762293602,
      "loss": 2.1463,
      "step": 14858
    },
    {
      "epoch": 0.10095719604110409,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009784947550325522,
      "loss": 2.242,
      "step": 14859
    },
    {
      "epoch": 0.10096399038769815,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009784916336142064,
      "loss": 2.3004,
      "step": 14860
    },
    {
      "epoch": 0.10097078473429222,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009784885119743247,
      "loss": 2.2206,
      "step": 14861
    },
    {
      "epoch": 0.10097757908088627,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009784853901129082,
      "loss": 2.203,
      "step": 14862
    },
    {
      "epoch": 0.10098437342748033,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009784822680299586,
      "loss": 2.2944,
      "step": 14863
    },
    {
      "epoch": 0.10099116777407438,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009784791457254772,
      "loss": 2.1979,
      "step": 14864
    },
    {
      "epoch": 0.10099796212066844,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009784760231994652,
      "loss": 2.2495,
      "step": 14865
    },
    {
      "epoch": 0.10100475646726251,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009784729004519244,
      "loss": 2.1782,
      "step": 14866
    },
    {
      "epoch": 0.10101155081385656,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009784697774828561,
      "loss": 2.3709,
      "step": 14867
    },
    {
      "epoch": 0.10101834516045062,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000978466654292262,
      "loss": 2.1711,
      "step": 14868
    },
    {
      "epoch": 0.10102513950704467,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978463530880143,
      "loss": 2.2418,
      "step": 14869
    },
    {
      "epoch": 0.10103193385363873,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000978460407246501,
      "loss": 2.2225,
      "step": 14870
    },
    {
      "epoch": 0.1010387282002328,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009784572833913374,
      "loss": 2.1475,
      "step": 14871
    },
    {
      "epoch": 0.10104552254682685,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009784541593146536,
      "loss": 2.1895,
      "step": 14872
    },
    {
      "epoch": 0.10105231689342091,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009784510350164509,
      "loss": 2.2101,
      "step": 14873
    },
    {
      "epoch": 0.10105911124001496,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009784479104967309,
      "loss": 2.2848,
      "step": 14874
    },
    {
      "epoch": 0.10106590558660902,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000978444785755495,
      "loss": 2.2123,
      "step": 14875
    },
    {
      "epoch": 0.10107269993320309,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009784416607927443,
      "loss": 2.2359,
      "step": 14876
    },
    {
      "epoch": 0.10107949427979714,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978438535608481,
      "loss": 2.3575,
      "step": 14877
    },
    {
      "epoch": 0.1010862886263912,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009784354102027062,
      "loss": 2.2735,
      "step": 14878
    },
    {
      "epoch": 0.10109308297298525,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009784322845754212,
      "loss": 2.2086,
      "step": 14879
    },
    {
      "epoch": 0.10109987731957931,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009784291587266274,
      "loss": 2.28,
      "step": 14880
    },
    {
      "epoch": 0.10110667166617338,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009784260326563265,
      "loss": 2.1682,
      "step": 14881
    },
    {
      "epoch": 0.10111346601276743,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009784229063645197,
      "loss": 2.0347,
      "step": 14882
    },
    {
      "epoch": 0.10112026035936149,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009784197798512087,
      "loss": 2.1294,
      "step": 14883
    },
    {
      "epoch": 0.10112705470595554,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000978416653116395,
      "loss": 2.3309,
      "step": 14884
    },
    {
      "epoch": 0.1011338490525496,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009784135261600795,
      "loss": 2.109,
      "step": 14885
    },
    {
      "epoch": 0.10114064339914365,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009784103989822644,
      "loss": 2.1054,
      "step": 14886
    },
    {
      "epoch": 0.10114743774573771,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009784072715829505,
      "loss": 2.3582,
      "step": 14887
    },
    {
      "epoch": 0.10115423209233178,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009784041439621395,
      "loss": 2.1491,
      "step": 14888
    },
    {
      "epoch": 0.10116102643892583,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000978401016119833,
      "loss": 2.1205,
      "step": 14889
    },
    {
      "epoch": 0.10116782078551989,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009783978880560323,
      "loss": 2.1535,
      "step": 14890
    },
    {
      "epoch": 0.10117461513211394,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009783947597707388,
      "loss": 2.0949,
      "step": 14891
    },
    {
      "epoch": 0.101181409478708,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000978391631263954,
      "loss": 2.2438,
      "step": 14892
    },
    {
      "epoch": 0.10118820382530207,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009783885025356795,
      "loss": 2.3649,
      "step": 14893
    },
    {
      "epoch": 0.10119499817189612,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009783853735859166,
      "loss": 2.2082,
      "step": 14894
    },
    {
      "epoch": 0.10120179251849018,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009783822444146665,
      "loss": 2.315,
      "step": 14895
    },
    {
      "epoch": 0.10120858686508423,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009783791150219311,
      "loss": 2.2642,
      "step": 14896
    },
    {
      "epoch": 0.10121538121167829,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009783759854077117,
      "loss": 2.2938,
      "step": 14897
    },
    {
      "epoch": 0.10122217555827236,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009783728555720096,
      "loss": 2.3049,
      "step": 14898
    },
    {
      "epoch": 0.1012289699048664,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009783697255148264,
      "loss": 2.2989,
      "step": 14899
    },
    {
      "epoch": 0.10123576425146047,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009783665952361636,
      "loss": 2.1322,
      "step": 14900
    },
    {
      "epoch": 0.10124255859805452,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009783634647360223,
      "loss": 2.108,
      "step": 14901
    },
    {
      "epoch": 0.10124935294464858,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009783603340144044,
      "loss": 2.371,
      "step": 14902
    },
    {
      "epoch": 0.10125614729124265,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000978357203071311,
      "loss": 2.1283,
      "step": 14903
    },
    {
      "epoch": 0.1012629416378367,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009783540719067439,
      "loss": 2.4268,
      "step": 14904
    },
    {
      "epoch": 0.10126973598443076,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000978350940520704,
      "loss": 2.204,
      "step": 14905
    },
    {
      "epoch": 0.10127653033102481,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009783478089131934,
      "loss": 2.1087,
      "step": 14906
    },
    {
      "epoch": 0.10128332467761887,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009783446770842133,
      "loss": 2.2531,
      "step": 14907
    },
    {
      "epoch": 0.10129011902421292,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009783415450337648,
      "loss": 2.0677,
      "step": 14908
    },
    {
      "epoch": 0.10129691337080698,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00097833841276185,
      "loss": 2.0783,
      "step": 14909
    },
    {
      "epoch": 0.10130370771740105,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009783352802684698,
      "loss": 2.1368,
      "step": 14910
    },
    {
      "epoch": 0.1013105020639951,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000978332147553626,
      "loss": 2.0515,
      "step": 14911
    },
    {
      "epoch": 0.10131729641058916,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009783290146173197,
      "loss": 2.1521,
      "step": 14912
    },
    {
      "epoch": 0.10132409075718321,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009783258814595526,
      "loss": 2.3131,
      "step": 14913
    },
    {
      "epoch": 0.10133088510377727,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009783227480803262,
      "loss": 2.2403,
      "step": 14914
    },
    {
      "epoch": 0.10133767945037134,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009783196144796417,
      "loss": 2.1988,
      "step": 14915
    },
    {
      "epoch": 0.10134447379696539,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978316480657501,
      "loss": 2.2043,
      "step": 14916
    },
    {
      "epoch": 0.10135126814355945,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009783133466139049,
      "loss": 2.3377,
      "step": 14917
    },
    {
      "epoch": 0.1013580624901535,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009783102123488553,
      "loss": 2.1254,
      "step": 14918
    },
    {
      "epoch": 0.10136485683674756,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009783070778623536,
      "loss": 2.2747,
      "step": 14919
    },
    {
      "epoch": 0.10137165118334163,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009783039431544012,
      "loss": 2.0605,
      "step": 14920
    },
    {
      "epoch": 0.10137844552993568,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009783008082249996,
      "loss": 2.153,
      "step": 14921
    },
    {
      "epoch": 0.10138523987652974,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009782976730741502,
      "loss": 2.1263,
      "step": 14922
    },
    {
      "epoch": 0.10139203422312379,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009782945377018544,
      "loss": 2.2178,
      "step": 14923
    },
    {
      "epoch": 0.10139882856971785,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009782914021081138,
      "loss": 2.3326,
      "step": 14924
    },
    {
      "epoch": 0.10140562291631192,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009782882662929294,
      "loss": 2.2749,
      "step": 14925
    },
    {
      "epoch": 0.10141241726290597,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009782851302563035,
      "loss": 2.2383,
      "step": 14926
    },
    {
      "epoch": 0.10141921160950003,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000978281993998237,
      "loss": 2.325,
      "step": 14927
    },
    {
      "epoch": 0.10142600595609408,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009782788575187311,
      "loss": 2.22,
      "step": 14928
    },
    {
      "epoch": 0.10143280030268814,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009782757208177879,
      "loss": 2.1121,
      "step": 14929
    },
    {
      "epoch": 0.1014395946492822,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009782725838954085,
      "loss": 2.2865,
      "step": 14930
    },
    {
      "epoch": 0.10144638899587625,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009782694467515942,
      "loss": 2.1319,
      "step": 14931
    },
    {
      "epoch": 0.10145318334247032,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009782663093863466,
      "loss": 2.3399,
      "step": 14932
    },
    {
      "epoch": 0.10145997768906437,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009782631717996674,
      "loss": 2.1289,
      "step": 14933
    },
    {
      "epoch": 0.10146677203565843,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009782600339915577,
      "loss": 2.3076,
      "step": 14934
    },
    {
      "epoch": 0.10147356638225248,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000978256895962019,
      "loss": 2.2465,
      "step": 14935
    },
    {
      "epoch": 0.10148036072884654,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978253757711053,
      "loss": 2.1727,
      "step": 14936
    },
    {
      "epoch": 0.10148715507544061,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009782506192386612,
      "loss": 2.322,
      "step": 14937
    },
    {
      "epoch": 0.10149394942203466,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009782474805448444,
      "loss": 2.3417,
      "step": 14938
    },
    {
      "epoch": 0.10150074376862872,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009782443416296047,
      "loss": 2.0755,
      "step": 14939
    },
    {
      "epoch": 0.10150753811522277,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009782412024929433,
      "loss": 2.2658,
      "step": 14940
    },
    {
      "epoch": 0.10151433246181683,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009782380631348617,
      "loss": 2.119,
      "step": 14941
    },
    {
      "epoch": 0.1015211268084109,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009782349235553617,
      "loss": 2.1519,
      "step": 14942
    },
    {
      "epoch": 0.10152792115500495,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000978231783754444,
      "loss": 2.2548,
      "step": 14943
    },
    {
      "epoch": 0.10153471550159901,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009782286437321106,
      "loss": 2.2508,
      "step": 14944
    },
    {
      "epoch": 0.10154150984819306,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009782255034883626,
      "loss": 2.169,
      "step": 14945
    },
    {
      "epoch": 0.10154830419478712,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009782223630232022,
      "loss": 2.3081,
      "step": 14946
    },
    {
      "epoch": 0.10155509854138119,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009782192223366298,
      "loss": 2.1937,
      "step": 14947
    },
    {
      "epoch": 0.10156189288797524,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009782160814286475,
      "loss": 2.2867,
      "step": 14948
    },
    {
      "epoch": 0.1015686872345693,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009782129402992569,
      "loss": 2.3176,
      "step": 14949
    },
    {
      "epoch": 0.10157548158116335,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000978209798948459,
      "loss": 2.314,
      "step": 14950
    },
    {
      "epoch": 0.10158227592775741,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009782066573762556,
      "loss": 2.2247,
      "step": 14951
    },
    {
      "epoch": 0.10158907027435148,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000978203515582648,
      "loss": 2.2647,
      "step": 14952
    },
    {
      "epoch": 0.10159586462094553,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009782003735676371,
      "loss": 2.2002,
      "step": 14953
    },
    {
      "epoch": 0.10160265896753959,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009781972313312255,
      "loss": 2.1961,
      "step": 14954
    },
    {
      "epoch": 0.10160945331413364,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000978194088873414,
      "loss": 2.228,
      "step": 14955
    },
    {
      "epoch": 0.1016162476607277,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000978190946194204,
      "loss": 2.2235,
      "step": 14956
    },
    {
      "epoch": 0.10162304200732175,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009781878032935972,
      "loss": 2.1268,
      "step": 14957
    },
    {
      "epoch": 0.10162983635391581,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009781846601715947,
      "loss": 2.2177,
      "step": 14958
    },
    {
      "epoch": 0.10163663070050988,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009781815168281981,
      "loss": 2.0759,
      "step": 14959
    },
    {
      "epoch": 0.10164342504710393,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009781783732634092,
      "loss": 2.2268,
      "step": 14960
    },
    {
      "epoch": 0.10165021939369799,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009781752294772293,
      "loss": 2.2873,
      "step": 14961
    },
    {
      "epoch": 0.10165701374029204,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009781720854696596,
      "loss": 2.0984,
      "step": 14962
    },
    {
      "epoch": 0.1016638080868861,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009781689412407017,
      "loss": 2.1773,
      "step": 14963
    },
    {
      "epoch": 0.10167060243348017,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000978165796790357,
      "loss": 2.3135,
      "step": 14964
    },
    {
      "epoch": 0.10167739678007422,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009781626521186272,
      "loss": 2.0812,
      "step": 14965
    },
    {
      "epoch": 0.10168419112666828,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009781595072255135,
      "loss": 2.2268,
      "step": 14966
    },
    {
      "epoch": 0.10169098547326233,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009781563621110172,
      "loss": 2.168,
      "step": 14967
    },
    {
      "epoch": 0.1016977798198564,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009781532167751402,
      "loss": 2.1924,
      "step": 14968
    },
    {
      "epoch": 0.10170457416645046,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009781500712178837,
      "loss": 2.2328,
      "step": 14969
    },
    {
      "epoch": 0.1017113685130445,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009781469254392492,
      "loss": 2.205,
      "step": 14970
    },
    {
      "epoch": 0.10171816285963857,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009781437794392383,
      "loss": 2.1705,
      "step": 14971
    },
    {
      "epoch": 0.10172495720623262,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000978140633217852,
      "loss": 2.2949,
      "step": 14972
    },
    {
      "epoch": 0.10173175155282668,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009781374867750923,
      "loss": 2.3328,
      "step": 14973
    },
    {
      "epoch": 0.10173854589942075,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009781343401109603,
      "loss": 2.1381,
      "step": 14974
    },
    {
      "epoch": 0.1017453402460148,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009781311932254577,
      "loss": 2.3415,
      "step": 14975
    },
    {
      "epoch": 0.10175213459260886,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000978128046118586,
      "loss": 2.041,
      "step": 14976
    },
    {
      "epoch": 0.10175892893920291,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000978124898790346,
      "loss": 2.3207,
      "step": 14977
    },
    {
      "epoch": 0.10176572328579697,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009781217512407402,
      "loss": 2.2955,
      "step": 14978
    },
    {
      "epoch": 0.10177251763239102,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009781186034697692,
      "loss": 2.2621,
      "step": 14979
    },
    {
      "epoch": 0.10177931197898508,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009781154554774347,
      "loss": 2.3178,
      "step": 14980
    },
    {
      "epoch": 0.10178610632557915,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009781123072637383,
      "loss": 2.1459,
      "step": 14981
    },
    {
      "epoch": 0.1017929006721732,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009781091588286816,
      "loss": 2.2616,
      "step": 14982
    },
    {
      "epoch": 0.10179969501876726,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009781060101722657,
      "loss": 2.0922,
      "step": 14983
    },
    {
      "epoch": 0.10180648936536131,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009781028612944921,
      "loss": 2.2826,
      "step": 14984
    },
    {
      "epoch": 0.10181328371195537,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009780997121953624,
      "loss": 2.2265,
      "step": 14985
    },
    {
      "epoch": 0.10182007805854944,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009780965628748781,
      "loss": 2.1826,
      "step": 14986
    },
    {
      "epoch": 0.10182687240514349,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009780934133330405,
      "loss": 2.1309,
      "step": 14987
    },
    {
      "epoch": 0.10183366675173755,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000978090263569851,
      "loss": 2.1288,
      "step": 14988
    },
    {
      "epoch": 0.1018404610983316,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009780871135853114,
      "loss": 2.1206,
      "step": 14989
    },
    {
      "epoch": 0.10184725544492566,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000978083963379423,
      "loss": 2.2257,
      "step": 14990
    },
    {
      "epoch": 0.10185404979151973,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000978080812952187,
      "loss": 2.1843,
      "step": 14991
    },
    {
      "epoch": 0.10186084413811378,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009780776623036051,
      "loss": 2.1341,
      "step": 14992
    },
    {
      "epoch": 0.10186763848470784,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009780745114336788,
      "loss": 2.2049,
      "step": 14993
    },
    {
      "epoch": 0.10187443283130189,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009780713603424096,
      "loss": 2.3079,
      "step": 14994
    },
    {
      "epoch": 0.10188122717789595,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009780682090297986,
      "loss": 2.003,
      "step": 14995
    },
    {
      "epoch": 0.10188802152449002,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009780650574958477,
      "loss": 2.3478,
      "step": 14996
    },
    {
      "epoch": 0.10189481587108407,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000978061905740558,
      "loss": 2.187,
      "step": 14997
    },
    {
      "epoch": 0.10190161021767813,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009780587537639312,
      "loss": 2.1546,
      "step": 14998
    },
    {
      "epoch": 0.10190840456427218,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009780556015659685,
      "loss": 2.1976,
      "step": 14999
    },
    {
      "epoch": 0.10191519891086624,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009780524491466718,
      "loss": 2.144,
      "step": 15000
    },
    {
      "epoch": 0.1019219932574603,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000978049296506042,
      "loss": 2.0921,
      "step": 15001
    },
    {
      "epoch": 0.10192878760405436,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009780461436440812,
      "loss": 2.0828,
      "step": 15002
    },
    {
      "epoch": 0.10193558195064842,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009780429905607904,
      "loss": 2.3062,
      "step": 15003
    },
    {
      "epoch": 0.10194237629724247,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000978039837256171,
      "loss": 2.2512,
      "step": 15004
    },
    {
      "epoch": 0.10194917064383653,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009780366837302248,
      "loss": 2.3128,
      "step": 15005
    },
    {
      "epoch": 0.10195596499043058,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000978033529982953,
      "loss": 2.2454,
      "step": 15006
    },
    {
      "epoch": 0.10196275933702464,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009780303760143574,
      "loss": 2.2729,
      "step": 15007
    },
    {
      "epoch": 0.10196955368361871,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000978027221824439,
      "loss": 2.2381,
      "step": 15008
    },
    {
      "epoch": 0.10197634803021276,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009780240674131995,
      "loss": 2.3914,
      "step": 15009
    },
    {
      "epoch": 0.10198314237680682,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009780209127806404,
      "loss": 2.2273,
      "step": 15010
    },
    {
      "epoch": 0.10198993672340087,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000978017757926763,
      "loss": 2.3035,
      "step": 15011
    },
    {
      "epoch": 0.10199673106999493,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000978014602851569,
      "loss": 2.2441,
      "step": 15012
    },
    {
      "epoch": 0.102003525416589,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009780114475550594,
      "loss": 2.0555,
      "step": 15013
    },
    {
      "epoch": 0.10201031976318305,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009780082920372365,
      "loss": 2.1405,
      "step": 15014
    },
    {
      "epoch": 0.10201711410977711,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009780051362981007,
      "loss": 2.2898,
      "step": 15015
    },
    {
      "epoch": 0.10202390845637116,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009780019803376544,
      "loss": 2.1187,
      "step": 15016
    },
    {
      "epoch": 0.10203070280296522,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009779988241558984,
      "loss": 2.2503,
      "step": 15017
    },
    {
      "epoch": 0.10203749714955929,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009779956677528345,
      "loss": 2.0722,
      "step": 15018
    },
    {
      "epoch": 0.10204429149615334,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009779925111284642,
      "loss": 2.274,
      "step": 15019
    },
    {
      "epoch": 0.1020510858427474,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009779893542827887,
      "loss": 2.1471,
      "step": 15020
    },
    {
      "epoch": 0.10205788018934145,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009779861972158098,
      "loss": 2.0844,
      "step": 15021
    },
    {
      "epoch": 0.10206467453593551,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009779830399275285,
      "loss": 2.1989,
      "step": 15022
    },
    {
      "epoch": 0.10207146888252958,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009779798824179468,
      "loss": 2.0769,
      "step": 15023
    },
    {
      "epoch": 0.10207826322912363,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009779767246870658,
      "loss": 2.2774,
      "step": 15024
    },
    {
      "epoch": 0.10208505757571769,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000977973566734887,
      "loss": 2.1211,
      "step": 15025
    },
    {
      "epoch": 0.10209185192231174,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000977970408561412,
      "loss": 2.2619,
      "step": 15026
    },
    {
      "epoch": 0.1020986462689058,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009779672501666421,
      "loss": 2.2418,
      "step": 15027
    },
    {
      "epoch": 0.10210544061549985,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009779640915505788,
      "loss": 2.2451,
      "step": 15028
    },
    {
      "epoch": 0.10211223496209391,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009779609327132237,
      "loss": 2.2831,
      "step": 15029
    },
    {
      "epoch": 0.10211902930868798,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009779577736545781,
      "loss": 2.1554,
      "step": 15030
    },
    {
      "epoch": 0.10212582365528203,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009779546143746437,
      "loss": 2.3971,
      "step": 15031
    },
    {
      "epoch": 0.10213261800187609,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009779514548734216,
      "loss": 2.1298,
      "step": 15032
    },
    {
      "epoch": 0.10213941234847014,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009779482951509134,
      "loss": 2.2877,
      "step": 15033
    },
    {
      "epoch": 0.1021462066950642,
      "grad_norm": 0.125,
      "learning_rate": 0.000977945135207121,
      "loss": 2.0678,
      "step": 15034
    },
    {
      "epoch": 0.10215300104165827,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009779419750420451,
      "loss": 2.1735,
      "step": 15035
    },
    {
      "epoch": 0.10215979538825232,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009779388146556878,
      "loss": 2.1774,
      "step": 15036
    },
    {
      "epoch": 0.10216658973484638,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00097793565404805,
      "loss": 2.0637,
      "step": 15037
    },
    {
      "epoch": 0.10217338408144043,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009779324932191339,
      "loss": 2.2441,
      "step": 15038
    },
    {
      "epoch": 0.1021801784280345,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009779293321689401,
      "loss": 2.1864,
      "step": 15039
    },
    {
      "epoch": 0.10218697277462856,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009779261708974707,
      "loss": 2.2806,
      "step": 15040
    },
    {
      "epoch": 0.1021937671212226,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000977923009404727,
      "loss": 2.2032,
      "step": 15041
    },
    {
      "epoch": 0.10220056146781667,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009779198476907102,
      "loss": 2.1093,
      "step": 15042
    },
    {
      "epoch": 0.10220735581441072,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009779166857554222,
      "loss": 2.0737,
      "step": 15043
    },
    {
      "epoch": 0.10221415016100478,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009779135235988643,
      "loss": 2.186,
      "step": 15044
    },
    {
      "epoch": 0.10222094450759885,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000977910361221038,
      "loss": 2.3101,
      "step": 15045
    },
    {
      "epoch": 0.1022277388541929,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009779071986219443,
      "loss": 2.2605,
      "step": 15046
    },
    {
      "epoch": 0.10223453320078696,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009779040358015854,
      "loss": 2.2235,
      "step": 15047
    },
    {
      "epoch": 0.10224132754738101,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009779008727599623,
      "loss": 2.0346,
      "step": 15048
    },
    {
      "epoch": 0.10224812189397507,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009778977094970765,
      "loss": 2.2822,
      "step": 15049
    },
    {
      "epoch": 0.10225491624056914,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009778945460129295,
      "loss": 2.1173,
      "step": 15050
    },
    {
      "epoch": 0.10226171058716318,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000977891382307523,
      "loss": 2.1159,
      "step": 15051
    },
    {
      "epoch": 0.10226850493375725,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000977888218380858,
      "loss": 2.3659,
      "step": 15052
    },
    {
      "epoch": 0.1022752992803513,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009778850542329365,
      "loss": 2.1232,
      "step": 15053
    },
    {
      "epoch": 0.10228209362694536,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009778818898637596,
      "loss": 2.1922,
      "step": 15054
    },
    {
      "epoch": 0.10228888797353941,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009778787252733288,
      "loss": 2.3926,
      "step": 15055
    },
    {
      "epoch": 0.10229568232013347,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009778755604616456,
      "loss": 2.3725,
      "step": 15056
    },
    {
      "epoch": 0.10230247666672754,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009778723954287115,
      "loss": 1.9893,
      "step": 15057
    },
    {
      "epoch": 0.10230927101332159,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009778692301745282,
      "loss": 2.2139,
      "step": 15058
    },
    {
      "epoch": 0.10231606535991565,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009778660646990968,
      "loss": 2.2063,
      "step": 15059
    },
    {
      "epoch": 0.1023228597065097,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009778628990024189,
      "loss": 2.1506,
      "step": 15060
    },
    {
      "epoch": 0.10232965405310376,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009778597330844959,
      "loss": 2.1971,
      "step": 15061
    },
    {
      "epoch": 0.10233644839969783,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009778565669453293,
      "loss": 2.1593,
      "step": 15062
    },
    {
      "epoch": 0.10234324274629188,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009778534005849206,
      "loss": 2.1407,
      "step": 15063
    },
    {
      "epoch": 0.10235003709288594,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009778502340032711,
      "loss": 2.1577,
      "step": 15064
    },
    {
      "epoch": 0.10235683143947999,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009778470672003828,
      "loss": 2.2292,
      "step": 15065
    },
    {
      "epoch": 0.10236362578607405,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009778439001762564,
      "loss": 2.2089,
      "step": 15066
    },
    {
      "epoch": 0.10237042013266812,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009778407329308941,
      "loss": 2.1573,
      "step": 15067
    },
    {
      "epoch": 0.10237721447926217,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009778375654642968,
      "loss": 2.2158,
      "step": 15068
    },
    {
      "epoch": 0.10238400882585623,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009778343977764662,
      "loss": 2.285,
      "step": 15069
    },
    {
      "epoch": 0.10239080317245028,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009778312298674037,
      "loss": 2.3914,
      "step": 15070
    },
    {
      "epoch": 0.10239759751904434,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009778280617371111,
      "loss": 2.0327,
      "step": 15071
    },
    {
      "epoch": 0.1024043918656384,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009778248933855894,
      "loss": 2.2023,
      "step": 15072
    },
    {
      "epoch": 0.10241118621223246,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009778217248128402,
      "loss": 2.3329,
      "step": 15073
    },
    {
      "epoch": 0.10241798055882652,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009778185560188652,
      "loss": 2.1883,
      "step": 15074
    },
    {
      "epoch": 0.10242477490542057,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009778153870036654,
      "loss": 2.1584,
      "step": 15075
    },
    {
      "epoch": 0.10243156925201463,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009778122177672426,
      "loss": 2.318,
      "step": 15076
    },
    {
      "epoch": 0.10243836359860868,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009778090483095983,
      "loss": 2.2501,
      "step": 15077
    },
    {
      "epoch": 0.10244515794520274,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000977805878630734,
      "loss": 2.1817,
      "step": 15078
    },
    {
      "epoch": 0.10245195229179681,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009778027087306508,
      "loss": 2.1842,
      "step": 15079
    },
    {
      "epoch": 0.10245874663839086,
      "grad_norm": 0.125,
      "learning_rate": 0.0009777995386093508,
      "loss": 2.1148,
      "step": 15080
    },
    {
      "epoch": 0.10246554098498492,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000977796368266835,
      "loss": 2.1739,
      "step": 15081
    },
    {
      "epoch": 0.10247233533157897,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009777931977031048,
      "loss": 2.212,
      "step": 15082
    },
    {
      "epoch": 0.10247912967817303,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009777900269181618,
      "loss": 2.2304,
      "step": 15083
    },
    {
      "epoch": 0.1024859240247671,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009777868559120075,
      "loss": 2.1244,
      "step": 15084
    },
    {
      "epoch": 0.10249271837136115,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009777836846846435,
      "loss": 2.087,
      "step": 15085
    },
    {
      "epoch": 0.10249951271795521,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009777805132360712,
      "loss": 2.2091,
      "step": 15086
    },
    {
      "epoch": 0.10250630706454926,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009777773415662915,
      "loss": 2.1945,
      "step": 15087
    },
    {
      "epoch": 0.10251310141114332,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009777741696753068,
      "loss": 2.1536,
      "step": 15088
    },
    {
      "epoch": 0.10251989575773739,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009777709975631182,
      "loss": 2.2757,
      "step": 15089
    },
    {
      "epoch": 0.10252669010433144,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000977767825229727,
      "loss": 2.1926,
      "step": 15090
    },
    {
      "epoch": 0.1025334844509255,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009777646526751347,
      "loss": 2.1699,
      "step": 15091
    },
    {
      "epoch": 0.10254027879751955,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009777614798993429,
      "loss": 2.0959,
      "step": 15092
    },
    {
      "epoch": 0.10254707314411361,
      "grad_norm": 0.140625,
      "learning_rate": 0.000977758306902353,
      "loss": 2.2968,
      "step": 15093
    },
    {
      "epoch": 0.10255386749070768,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009777551336841665,
      "loss": 2.1778,
      "step": 15094
    },
    {
      "epoch": 0.10256066183730173,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000977751960244785,
      "loss": 2.1818,
      "step": 15095
    },
    {
      "epoch": 0.10256745618389579,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009777487865842093,
      "loss": 2.0943,
      "step": 15096
    },
    {
      "epoch": 0.10257425053048984,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009777456127024419,
      "loss": 2.2066,
      "step": 15097
    },
    {
      "epoch": 0.1025810448770839,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009777424385994835,
      "loss": 2.1977,
      "step": 15098
    },
    {
      "epoch": 0.10258783922367795,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000977739264275336,
      "loss": 2.3803,
      "step": 15099
    },
    {
      "epoch": 0.10259463357027201,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009777360897300007,
      "loss": 2.1426,
      "step": 15100
    },
    {
      "epoch": 0.10260142791686608,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000977732914963479,
      "loss": 2.2524,
      "step": 15101
    },
    {
      "epoch": 0.10260822226346013,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009777297399757724,
      "loss": 2.1712,
      "step": 15102
    },
    {
      "epoch": 0.10261501661005419,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009777265647668825,
      "loss": 2.0554,
      "step": 15103
    },
    {
      "epoch": 0.10262181095664824,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009777233893368104,
      "loss": 2.306,
      "step": 15104
    },
    {
      "epoch": 0.1026286053032423,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009777202136855582,
      "loss": 2.1729,
      "step": 15105
    },
    {
      "epoch": 0.10263539964983637,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009777170378131269,
      "loss": 2.3687,
      "step": 15106
    },
    {
      "epoch": 0.10264219399643042,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000977713861719518,
      "loss": 2.1718,
      "step": 15107
    },
    {
      "epoch": 0.10264898834302448,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000977710685404733,
      "loss": 2.1879,
      "step": 15108
    },
    {
      "epoch": 0.10265578268961853,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009777075088687737,
      "loss": 2.2589,
      "step": 15109
    },
    {
      "epoch": 0.1026625770362126,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009777043321116412,
      "loss": 2.2643,
      "step": 15110
    },
    {
      "epoch": 0.10266937138280666,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000977701155133337,
      "loss": 2.188,
      "step": 15111
    },
    {
      "epoch": 0.1026761657294007,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009776979779338626,
      "loss": 2.1918,
      "step": 15112
    },
    {
      "epoch": 0.10268296007599477,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009776948005132196,
      "loss": 2.2628,
      "step": 15113
    },
    {
      "epoch": 0.10268975442258882,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009776916228714095,
      "loss": 2.0042,
      "step": 15114
    },
    {
      "epoch": 0.10269654876918288,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009776884450084335,
      "loss": 2.2359,
      "step": 15115
    },
    {
      "epoch": 0.10270334311577695,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009776852669242933,
      "loss": 2.1617,
      "step": 15116
    },
    {
      "epoch": 0.102710137462371,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00097768208861899,
      "loss": 2.1803,
      "step": 15117
    },
    {
      "epoch": 0.10271693180896506,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009776789100925257,
      "loss": 2.1189,
      "step": 15118
    },
    {
      "epoch": 0.10272372615555911,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009776757313449014,
      "loss": 2.2347,
      "step": 15119
    },
    {
      "epoch": 0.10273052050215317,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009776725523761188,
      "loss": 2.2273,
      "step": 15120
    },
    {
      "epoch": 0.10273731484874724,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009776693731861792,
      "loss": 2.2406,
      "step": 15121
    },
    {
      "epoch": 0.10274410919534128,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009776661937750844,
      "loss": 2.2791,
      "step": 15122
    },
    {
      "epoch": 0.10275090354193535,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009776630141428353,
      "loss": 2.2254,
      "step": 15123
    },
    {
      "epoch": 0.1027576978885294,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009776598342894338,
      "loss": 2.2829,
      "step": 15124
    },
    {
      "epoch": 0.10276449223512346,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009776566542148814,
      "loss": 2.2044,
      "step": 15125
    },
    {
      "epoch": 0.10277128658171751,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009776534739191793,
      "loss": 2.2653,
      "step": 15126
    },
    {
      "epoch": 0.10277808092831157,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009776502934023292,
      "loss": 2.0192,
      "step": 15127
    },
    {
      "epoch": 0.10278487527490564,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009776471126643324,
      "loss": 2.3147,
      "step": 15128
    },
    {
      "epoch": 0.10279166962149969,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009776439317051907,
      "loss": 2.298,
      "step": 15129
    },
    {
      "epoch": 0.10279846396809375,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009776407505249051,
      "loss": 2.2357,
      "step": 15130
    },
    {
      "epoch": 0.1028052583146878,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009776375691234772,
      "loss": 2.2151,
      "step": 15131
    },
    {
      "epoch": 0.10281205266128186,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000977634387500909,
      "loss": 2.138,
      "step": 15132
    },
    {
      "epoch": 0.10281884700787593,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009776312056572012,
      "loss": 2.2464,
      "step": 15133
    },
    {
      "epoch": 0.10282564135446998,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009776280235923559,
      "loss": 2.1991,
      "step": 15134
    },
    {
      "epoch": 0.10283243570106404,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000977624841306374,
      "loss": 2.1276,
      "step": 15135
    },
    {
      "epoch": 0.10283923004765809,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009776216587992575,
      "loss": 2.2877,
      "step": 15136
    },
    {
      "epoch": 0.10284602439425215,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009776184760710075,
      "loss": 2.1299,
      "step": 15137
    },
    {
      "epoch": 0.10285281874084622,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009776152931216258,
      "loss": 2.1261,
      "step": 15138
    },
    {
      "epoch": 0.10285961308744027,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009776121099511134,
      "loss": 2.205,
      "step": 15139
    },
    {
      "epoch": 0.10286640743403433,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009776089265594724,
      "loss": 2.1833,
      "step": 15140
    },
    {
      "epoch": 0.10287320178062838,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009776057429467036,
      "loss": 2.1739,
      "step": 15141
    },
    {
      "epoch": 0.10287999612722244,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000977602559112809,
      "loss": 2.1567,
      "step": 15142
    },
    {
      "epoch": 0.1028867904738165,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00097759937505779,
      "loss": 2.2692,
      "step": 15143
    },
    {
      "epoch": 0.10289358482041056,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009775961907816477,
      "loss": 2.1478,
      "step": 15144
    },
    {
      "epoch": 0.10290037916700462,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009775930062843842,
      "loss": 2.0824,
      "step": 15145
    },
    {
      "epoch": 0.10290717351359867,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009775898215660002,
      "loss": 2.2416,
      "step": 15146
    },
    {
      "epoch": 0.10291396786019273,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009775866366264977,
      "loss": 2.3462,
      "step": 15147
    },
    {
      "epoch": 0.10292076220678678,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009775834514658783,
      "loss": 2.2026,
      "step": 15148
    },
    {
      "epoch": 0.10292755655338084,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000977580266084143,
      "loss": 2.3149,
      "step": 15149
    },
    {
      "epoch": 0.10293435089997491,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009775770804812938,
      "loss": 2.1139,
      "step": 15150
    },
    {
      "epoch": 0.10294114524656896,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009775738946573315,
      "loss": 2.13,
      "step": 15151
    },
    {
      "epoch": 0.10294793959316302,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009775707086122584,
      "loss": 2.1375,
      "step": 15152
    },
    {
      "epoch": 0.10295473393975707,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009775675223460752,
      "loss": 1.975,
      "step": 15153
    },
    {
      "epoch": 0.10296152828635113,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009775643358587837,
      "loss": 2.2882,
      "step": 15154
    },
    {
      "epoch": 0.1029683226329452,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009775611491503855,
      "loss": 2.1145,
      "step": 15155
    },
    {
      "epoch": 0.10297511697953925,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009775579622208822,
      "loss": 2.0937,
      "step": 15156
    },
    {
      "epoch": 0.10298191132613331,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009775547750702747,
      "loss": 2.1761,
      "step": 15157
    },
    {
      "epoch": 0.10298870567272736,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009775515876985648,
      "loss": 2.2449,
      "step": 15158
    },
    {
      "epoch": 0.10299550001932142,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009775484001057543,
      "loss": 2.1698,
      "step": 15159
    },
    {
      "epoch": 0.10300229436591549,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000977545212291844,
      "loss": 2.1852,
      "step": 15160
    },
    {
      "epoch": 0.10300908871250954,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000977542024256836,
      "loss": 2.1526,
      "step": 15161
    },
    {
      "epoch": 0.1030158830591036,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009775388360007313,
      "loss": 2.304,
      "step": 15162
    },
    {
      "epoch": 0.10302267740569765,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009775356475235318,
      "loss": 2.2221,
      "step": 15163
    },
    {
      "epoch": 0.10302947175229171,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009775324588252386,
      "loss": 2.1994,
      "step": 15164
    },
    {
      "epoch": 0.10303626609888578,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009775292699058535,
      "loss": 2.0717,
      "step": 15165
    },
    {
      "epoch": 0.10304306044547983,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009775260807653777,
      "loss": 2.0677,
      "step": 15166
    },
    {
      "epoch": 0.10304985479207389,
      "grad_norm": 0.140625,
      "learning_rate": 0.000977522891403813,
      "loss": 2.3078,
      "step": 15167
    },
    {
      "epoch": 0.10305664913866794,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009775197018211605,
      "loss": 2.2932,
      "step": 15168
    },
    {
      "epoch": 0.103063443485262,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009775165120174218,
      "loss": 2.2203,
      "step": 15169
    },
    {
      "epoch": 0.10307023783185607,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009775133219925985,
      "loss": 2.2975,
      "step": 15170
    },
    {
      "epoch": 0.10307703217845011,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009775101317466921,
      "loss": 2.089,
      "step": 15171
    },
    {
      "epoch": 0.10308382652504418,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009775069412797039,
      "loss": 2.2168,
      "step": 15172
    },
    {
      "epoch": 0.10309062087163823,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009775037505916355,
      "loss": 2.1926,
      "step": 15173
    },
    {
      "epoch": 0.10309741521823229,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009775005596824881,
      "loss": 2.0549,
      "step": 15174
    },
    {
      "epoch": 0.10310420956482634,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009774973685522637,
      "loss": 2.3173,
      "step": 15175
    },
    {
      "epoch": 0.1031110039114204,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009774941772009635,
      "loss": 2.1911,
      "step": 15176
    },
    {
      "epoch": 0.10311779825801447,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009774909856285889,
      "loss": 2.2507,
      "step": 15177
    },
    {
      "epoch": 0.10312459260460852,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009774877938351415,
      "loss": 2.1517,
      "step": 15178
    },
    {
      "epoch": 0.10313138695120258,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009774846018206225,
      "loss": 2.2422,
      "step": 15179
    },
    {
      "epoch": 0.10313818129779663,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009774814095850337,
      "loss": 2.2702,
      "step": 15180
    },
    {
      "epoch": 0.1031449756443907,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009774782171283766,
      "loss": 2.1592,
      "step": 15181
    },
    {
      "epoch": 0.10315176999098476,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009774750244506524,
      "loss": 2.177,
      "step": 15182
    },
    {
      "epoch": 0.1031585643375788,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000977471831551863,
      "loss": 2.1589,
      "step": 15183
    },
    {
      "epoch": 0.10316535868417287,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009774686384320093,
      "loss": 2.3572,
      "step": 15184
    },
    {
      "epoch": 0.10317215303076692,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009774654450910934,
      "loss": 2.2279,
      "step": 15185
    },
    {
      "epoch": 0.10317894737736098,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009774622515291162,
      "loss": 2.2414,
      "step": 15186
    },
    {
      "epoch": 0.10318574172395505,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009774590577460796,
      "loss": 2.228,
      "step": 15187
    },
    {
      "epoch": 0.1031925360705491,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000977455863741985,
      "loss": 2.2756,
      "step": 15188
    },
    {
      "epoch": 0.10319933041714316,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009774526695168336,
      "loss": 2.2999,
      "step": 15189
    },
    {
      "epoch": 0.10320612476373721,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009774494750706274,
      "loss": 2.1867,
      "step": 15190
    },
    {
      "epoch": 0.10321291911033127,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009774462804033673,
      "loss": 2.1658,
      "step": 15191
    },
    {
      "epoch": 0.10321971345692534,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000977443085515055,
      "loss": 2.3803,
      "step": 15192
    },
    {
      "epoch": 0.10322650780351939,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009774398904056922,
      "loss": 2.1834,
      "step": 15193
    },
    {
      "epoch": 0.10323330215011345,
      "grad_norm": 0.13671875,
      "learning_rate": 0.00097743669507528,
      "loss": 2.238,
      "step": 15194
    },
    {
      "epoch": 0.1032400964967075,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009774334995238203,
      "loss": 2.1965,
      "step": 15195
    },
    {
      "epoch": 0.10324689084330156,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009774303037513143,
      "loss": 2.1533,
      "step": 15196
    },
    {
      "epoch": 0.10325368518989561,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009774271077577635,
      "loss": 2.1505,
      "step": 15197
    },
    {
      "epoch": 0.10326047953648967,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009774239115431696,
      "loss": 2.2246,
      "step": 15198
    },
    {
      "epoch": 0.10326727388308374,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009774207151075336,
      "loss": 2.2025,
      "step": 15199
    },
    {
      "epoch": 0.10327406822967779,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009774175184508575,
      "loss": 2.2297,
      "step": 15200
    },
    {
      "epoch": 0.10328086257627185,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009774143215731425,
      "loss": 2.2404,
      "step": 15201
    },
    {
      "epoch": 0.1032876569228659,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00097741112447439,
      "loss": 2.2055,
      "step": 15202
    },
    {
      "epoch": 0.10329445126945996,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000977407927154602,
      "loss": 2.2158,
      "step": 15203
    },
    {
      "epoch": 0.10330124561605403,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009774047296137792,
      "loss": 2.0749,
      "step": 15204
    },
    {
      "epoch": 0.10330803996264808,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009774015318519238,
      "loss": 2.1698,
      "step": 15205
    },
    {
      "epoch": 0.10331483430924214,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009773983338690366,
      "loss": 2.1532,
      "step": 15206
    },
    {
      "epoch": 0.10332162865583619,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009773951356651198,
      "loss": 2.2634,
      "step": 15207
    },
    {
      "epoch": 0.10332842300243025,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009773919372401743,
      "loss": 2.1875,
      "step": 15208
    },
    {
      "epoch": 0.10333521734902432,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000977388738594202,
      "loss": 2.0916,
      "step": 15209
    },
    {
      "epoch": 0.10334201169561837,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000977385539727204,
      "loss": 2.256,
      "step": 15210
    },
    {
      "epoch": 0.10334880604221243,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009773823406391822,
      "loss": 2.3135,
      "step": 15211
    },
    {
      "epoch": 0.10335560038880648,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009773791413301377,
      "loss": 2.1207,
      "step": 15212
    },
    {
      "epoch": 0.10336239473540054,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000977375941800072,
      "loss": 2.1047,
      "step": 15213
    },
    {
      "epoch": 0.1033691890819946,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000977372742048987,
      "loss": 2.0317,
      "step": 15214
    },
    {
      "epoch": 0.10337598342858866,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009773695420768834,
      "loss": 2.2725,
      "step": 15215
    },
    {
      "epoch": 0.10338277777518272,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009773663418837636,
      "loss": 2.171,
      "step": 15216
    },
    {
      "epoch": 0.10338957212177677,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009773631414696287,
      "loss": 2.3293,
      "step": 15217
    },
    {
      "epoch": 0.10339636646837083,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00097735994083448,
      "loss": 2.2354,
      "step": 15218
    },
    {
      "epoch": 0.10340316081496488,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000977356739978319,
      "loss": 2.2007,
      "step": 15219
    },
    {
      "epoch": 0.10340995516155894,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009773535389011475,
      "loss": 2.2303,
      "step": 15220
    },
    {
      "epoch": 0.10341674950815301,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009773503376029667,
      "loss": 2.2113,
      "step": 15221
    },
    {
      "epoch": 0.10342354385474706,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.000977347136083778,
      "loss": 2.0284,
      "step": 15222
    },
    {
      "epoch": 0.10343033820134112,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009773439343435833,
      "loss": 2.2033,
      "step": 15223
    },
    {
      "epoch": 0.10343713254793517,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009773407323823838,
      "loss": 2.1381,
      "step": 15224
    },
    {
      "epoch": 0.10344392689452923,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000977337530200181,
      "loss": 2.2212,
      "step": 15225
    },
    {
      "epoch": 0.1034507212411233,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009773343277969763,
      "loss": 2.1286,
      "step": 15226
    },
    {
      "epoch": 0.10345751558771735,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009773311251727714,
      "loss": 2.2068,
      "step": 15227
    },
    {
      "epoch": 0.10346430993431141,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009773279223275675,
      "loss": 2.2442,
      "step": 15228
    },
    {
      "epoch": 0.10347110428090546,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009773247192613663,
      "loss": 2.246,
      "step": 15229
    },
    {
      "epoch": 0.10347789862749952,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009773215159741694,
      "loss": 2.2222,
      "step": 15230
    },
    {
      "epoch": 0.10348469297409359,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009773183124659778,
      "loss": 2.2491,
      "step": 15231
    },
    {
      "epoch": 0.10349148732068764,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009773151087367937,
      "loss": 2.2384,
      "step": 15232
    },
    {
      "epoch": 0.1034982816672817,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009773119047866178,
      "loss": 2.2136,
      "step": 15233
    },
    {
      "epoch": 0.10350507601387575,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009773087006154522,
      "loss": 2.21,
      "step": 15234
    },
    {
      "epoch": 0.10351187036046981,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000977305496223298,
      "loss": 1.9996,
      "step": 15235
    },
    {
      "epoch": 0.10351866470706388,
      "grad_norm": 0.140625,
      "learning_rate": 0.000977302291610157,
      "loss": 2.1883,
      "step": 15236
    },
    {
      "epoch": 0.10352545905365793,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009772990867760302,
      "loss": 2.2344,
      "step": 15237
    },
    {
      "epoch": 0.10353225340025199,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009772958817209196,
      "loss": 2.18,
      "step": 15238
    },
    {
      "epoch": 0.10353904774684604,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009772926764448266,
      "loss": 2.1498,
      "step": 15239
    },
    {
      "epoch": 0.1035458420934401,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009772894709477525,
      "loss": 2.2759,
      "step": 15240
    },
    {
      "epoch": 0.10355263644003417,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009772862652296988,
      "loss": 2.2094,
      "step": 15241
    },
    {
      "epoch": 0.10355943078662821,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000977283059290667,
      "loss": 2.2097,
      "step": 15242
    },
    {
      "epoch": 0.10356622513322228,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009772798531306585,
      "loss": 2.152,
      "step": 15243
    },
    {
      "epoch": 0.10357301947981633,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009772766467496749,
      "loss": 2.3383,
      "step": 15244
    },
    {
      "epoch": 0.10357981382641039,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009772734401477178,
      "loss": 2.1921,
      "step": 15245
    },
    {
      "epoch": 0.10358660817300444,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009772702333247887,
      "loss": 2.3675,
      "step": 15246
    },
    {
      "epoch": 0.1035934025195985,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000977267026280889,
      "loss": 2.2602,
      "step": 15247
    },
    {
      "epoch": 0.10360019686619257,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009772638190160199,
      "loss": 2.2088,
      "step": 15248
    },
    {
      "epoch": 0.10360699121278662,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009772606115301832,
      "loss": 2.2071,
      "step": 15249
    },
    {
      "epoch": 0.10361378555938068,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009772574038233802,
      "loss": 2.0586,
      "step": 15250
    },
    {
      "epoch": 0.10362057990597473,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009772541958956128,
      "loss": 2.0643,
      "step": 15251
    },
    {
      "epoch": 0.1036273742525688,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009772509877468817,
      "loss": 2.2858,
      "step": 15252
    },
    {
      "epoch": 0.10363416859916286,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009772477793771892,
      "loss": 2.2608,
      "step": 15253
    },
    {
      "epoch": 0.1036409629457569,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009772445707865365,
      "loss": 2.2656,
      "step": 15254
    },
    {
      "epoch": 0.10364775729235097,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009772413619749249,
      "loss": 2.1978,
      "step": 15255
    },
    {
      "epoch": 0.10365455163894502,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000977238152942356,
      "loss": 2.1514,
      "step": 15256
    },
    {
      "epoch": 0.10366134598553908,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009772349436888313,
      "loss": 2.2037,
      "step": 15257
    },
    {
      "epoch": 0.10366814033213315,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009772317342143524,
      "loss": 2.1478,
      "step": 15258
    },
    {
      "epoch": 0.1036749346787272,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009772285245189206,
      "loss": 2.0761,
      "step": 15259
    },
    {
      "epoch": 0.10368172902532126,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009772253146025376,
      "loss": 2.234,
      "step": 15260
    },
    {
      "epoch": 0.10368852337191531,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009772221044652045,
      "loss": 2.2352,
      "step": 15261
    },
    {
      "epoch": 0.10369531771850937,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009772188941069232,
      "loss": 2.2478,
      "step": 15262
    },
    {
      "epoch": 0.10370211206510344,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000977215683527695,
      "loss": 2.2616,
      "step": 15263
    },
    {
      "epoch": 0.10370890641169749,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009772124727275213,
      "loss": 2.0673,
      "step": 15264
    },
    {
      "epoch": 0.10371570075829155,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009772092617064039,
      "loss": 2.3206,
      "step": 15265
    },
    {
      "epoch": 0.1037224951048856,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000977206050464344,
      "loss": 2.0108,
      "step": 15266
    },
    {
      "epoch": 0.10372928945147966,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009772028390013432,
      "loss": 2.1917,
      "step": 15267
    },
    {
      "epoch": 0.10373608379807371,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009771996273174029,
      "loss": 2.2669,
      "step": 15268
    },
    {
      "epoch": 0.10374287814466777,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009771964154125245,
      "loss": 2.1223,
      "step": 15269
    },
    {
      "epoch": 0.10374967249126184,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009771932032867098,
      "loss": 2.1966,
      "step": 15270
    },
    {
      "epoch": 0.10375646683785589,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009771899909399601,
      "loss": 2.1918,
      "step": 15271
    },
    {
      "epoch": 0.10376326118444995,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.000977186778372277,
      "loss": 2.0863,
      "step": 15272
    },
    {
      "epoch": 0.103770055531044,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000977183565583662,
      "loss": 2.0703,
      "step": 15273
    },
    {
      "epoch": 0.10377684987763806,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009771803525741162,
      "loss": 2.2052,
      "step": 15274
    },
    {
      "epoch": 0.10378364422423213,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009771771393436413,
      "loss": 2.047,
      "step": 15275
    },
    {
      "epoch": 0.10379043857082618,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000977173925892239,
      "loss": 2.0703,
      "step": 15276
    },
    {
      "epoch": 0.10379723291742024,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009771707122199107,
      "loss": 2.1945,
      "step": 15277
    },
    {
      "epoch": 0.10380402726401429,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009771674983266576,
      "loss": 2.2811,
      "step": 15278
    },
    {
      "epoch": 0.10381082161060835,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009771642842124817,
      "loss": 2.3121,
      "step": 15279
    },
    {
      "epoch": 0.10381761595720242,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000977161069877384,
      "loss": 2.1986,
      "step": 15280
    },
    {
      "epoch": 0.10382441030379647,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009771578553213664,
      "loss": 2.2632,
      "step": 15281
    },
    {
      "epoch": 0.10383120465039053,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00097715464054443,
      "loss": 2.1023,
      "step": 15282
    },
    {
      "epoch": 0.10383799899698458,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009771514255465766,
      "loss": 2.2003,
      "step": 15283
    },
    {
      "epoch": 0.10384479334357864,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009771482103278076,
      "loss": 2.0476,
      "step": 15284
    },
    {
      "epoch": 0.1038515876901727,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009771449948881245,
      "loss": 2.0874,
      "step": 15285
    },
    {
      "epoch": 0.10385838203676676,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009771417792275284,
      "loss": 2.2153,
      "step": 15286
    },
    {
      "epoch": 0.10386517638336082,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009771385633460213,
      "loss": 2.2626,
      "step": 15287
    },
    {
      "epoch": 0.10387197072995487,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009771353472436046,
      "loss": 2.2659,
      "step": 15288
    },
    {
      "epoch": 0.10387876507654893,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009771321309202797,
      "loss": 2.2606,
      "step": 15289
    },
    {
      "epoch": 0.103885559423143,
      "grad_norm": 0.140625,
      "learning_rate": 0.000977128914376048,
      "loss": 2.1453,
      "step": 15290
    },
    {
      "epoch": 0.10389235376973704,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000977125697610911,
      "loss": 2.1439,
      "step": 15291
    },
    {
      "epoch": 0.10389914811633111,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009771224806248704,
      "loss": 2.2059,
      "step": 15292
    },
    {
      "epoch": 0.10390594246292516,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009771192634179275,
      "loss": 2.2906,
      "step": 15293
    },
    {
      "epoch": 0.10391273680951922,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000977116045990084,
      "loss": 2.2792,
      "step": 15294
    },
    {
      "epoch": 0.10391953115611327,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000977112828341341,
      "loss": 2.2002,
      "step": 15295
    },
    {
      "epoch": 0.10392632550270733,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009771096104717006,
      "loss": 2.2135,
      "step": 15296
    },
    {
      "epoch": 0.1039331198493014,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009771063923811635,
      "loss": 2.0598,
      "step": 15297
    },
    {
      "epoch": 0.10393991419589545,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000977103174069732,
      "loss": 2.271,
      "step": 15298
    },
    {
      "epoch": 0.10394670854248951,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000977099955537407,
      "loss": 2.1386,
      "step": 15299
    },
    {
      "epoch": 0.10395350288908356,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00097709673678419,
      "loss": 2.2762,
      "step": 15300
    },
    {
      "epoch": 0.10396029723567762,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000977093517810083,
      "loss": 2.3082,
      "step": 15301
    },
    {
      "epoch": 0.10396709158227169,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000977090298615087,
      "loss": 2.2693,
      "step": 15302
    },
    {
      "epoch": 0.10397388592886574,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009770870791992038,
      "loss": 2.2232,
      "step": 15303
    },
    {
      "epoch": 0.1039806802754598,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009770838595624346,
      "loss": 2.2263,
      "step": 15304
    },
    {
      "epoch": 0.10398747462205385,
      "grad_norm": 0.125,
      "learning_rate": 0.000977080639704781,
      "loss": 2.1214,
      "step": 15305
    },
    {
      "epoch": 0.10399426896864791,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009770774196262448,
      "loss": 2.1827,
      "step": 15306
    },
    {
      "epoch": 0.10400106331524198,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.000977074199326827,
      "loss": 1.9726,
      "step": 15307
    },
    {
      "epoch": 0.10400785766183603,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009770709788065293,
      "loss": 2.0933,
      "step": 15308
    },
    {
      "epoch": 0.10401465200843009,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009770677580653534,
      "loss": 2.0369,
      "step": 15309
    },
    {
      "epoch": 0.10402144635502414,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009770645371033003,
      "loss": 2.2532,
      "step": 15310
    },
    {
      "epoch": 0.1040282407016182,
      "grad_norm": 0.140625,
      "learning_rate": 0.000977061315920372,
      "loss": 2.3016,
      "step": 15311
    },
    {
      "epoch": 0.10403503504821227,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.00097705809451657,
      "loss": 2.2711,
      "step": 15312
    },
    {
      "epoch": 0.10404182939480631,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000977054872891895,
      "loss": 2.3084,
      "step": 15313
    },
    {
      "epoch": 0.10404862374140038,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009770516510463497,
      "loss": 2.1544,
      "step": 15314
    },
    {
      "epoch": 0.10405541808799443,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009770484289799346,
      "loss": 2.2924,
      "step": 15315
    },
    {
      "epoch": 0.10406221243458849,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009770452066926514,
      "loss": 2.3797,
      "step": 15316
    },
    {
      "epoch": 0.10406900678118254,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000977041984184502,
      "loss": 2.2702,
      "step": 15317
    },
    {
      "epoch": 0.1040758011277766,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009770387614554876,
      "loss": 2.1691,
      "step": 15318
    },
    {
      "epoch": 0.10408259547437067,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009770355385056095,
      "loss": 2.0162,
      "step": 15319
    },
    {
      "epoch": 0.10408938982096472,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009770323153348696,
      "loss": 2.0517,
      "step": 15320
    },
    {
      "epoch": 0.10409618416755878,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009770290919432693,
      "loss": 2.241,
      "step": 15321
    },
    {
      "epoch": 0.10410297851415283,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009770258683308099,
      "loss": 2.0294,
      "step": 15322
    },
    {
      "epoch": 0.1041097728607469,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000977022644497493,
      "loss": 2.1873,
      "step": 15323
    },
    {
      "epoch": 0.10411656720734096,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009770194204433201,
      "loss": 2.3306,
      "step": 15324
    },
    {
      "epoch": 0.104123361553935,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009770161961682926,
      "loss": 2.0596,
      "step": 15325
    },
    {
      "epoch": 0.10413015590052907,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000977012971672412,
      "loss": 2.2842,
      "step": 15326
    },
    {
      "epoch": 0.10413695024712312,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00097700974695568,
      "loss": 2.1565,
      "step": 15327
    },
    {
      "epoch": 0.10414374459371718,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009770065220180981,
      "loss": 2.0694,
      "step": 15328
    },
    {
      "epoch": 0.10415053894031125,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009770032968596674,
      "loss": 2.1166,
      "step": 15329
    },
    {
      "epoch": 0.1041573332869053,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009770000714803898,
      "loss": 2.2989,
      "step": 15330
    },
    {
      "epoch": 0.10416412763349936,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009769968458802666,
      "loss": 2.4485,
      "step": 15331
    },
    {
      "epoch": 0.10417092198009341,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009769936200592993,
      "loss": 2.293,
      "step": 15332
    },
    {
      "epoch": 0.10417771632668747,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009769903940174893,
      "loss": 2.217,
      "step": 15333
    },
    {
      "epoch": 0.10418451067328154,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009769871677548384,
      "loss": 2.0705,
      "step": 15334
    },
    {
      "epoch": 0.10419130501987559,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009769839412713478,
      "loss": 2.1932,
      "step": 15335
    },
    {
      "epoch": 0.10419809936646965,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009769807145670192,
      "loss": 2.277,
      "step": 15336
    },
    {
      "epoch": 0.1042048937130637,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000976977487641854,
      "loss": 2.0968,
      "step": 15337
    },
    {
      "epoch": 0.10421168805965776,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009769742604958536,
      "loss": 2.16,
      "step": 15338
    },
    {
      "epoch": 0.10421848240625181,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009769710331290196,
      "loss": 2.1224,
      "step": 15339
    },
    {
      "epoch": 0.10422527675284587,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009769678055413535,
      "loss": 2.2645,
      "step": 15340
    },
    {
      "epoch": 0.10423207109943994,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009769645777328567,
      "loss": 2.1398,
      "step": 15341
    },
    {
      "epoch": 0.10423886544603399,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009769613497035308,
      "loss": 2.0925,
      "step": 15342
    },
    {
      "epoch": 0.10424565979262805,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009769581214533773,
      "loss": 2.2216,
      "step": 15343
    },
    {
      "epoch": 0.1042524541392221,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009769548929823978,
      "loss": 2.2999,
      "step": 15344
    },
    {
      "epoch": 0.10425924848581616,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009769516642905934,
      "loss": 2.2693,
      "step": 15345
    },
    {
      "epoch": 0.10426604283241023,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009769484353779662,
      "loss": 2.2352,
      "step": 15346
    },
    {
      "epoch": 0.10427283717900428,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009769452062445168,
      "loss": 2.2939,
      "step": 15347
    },
    {
      "epoch": 0.10427963152559834,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009769419768902476,
      "loss": 2.1564,
      "step": 15348
    },
    {
      "epoch": 0.10428642587219239,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009769387473151597,
      "loss": 2.0557,
      "step": 15349
    },
    {
      "epoch": 0.10429322021878645,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009769355175192546,
      "loss": 2.2275,
      "step": 15350
    },
    {
      "epoch": 0.10430001456538052,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000976932287502534,
      "loss": 2.1679,
      "step": 15351
    },
    {
      "epoch": 0.10430680891197457,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000976929057264999,
      "loss": 2.1622,
      "step": 15352
    },
    {
      "epoch": 0.10431360325856863,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009769258268066512,
      "loss": 2.2481,
      "step": 15353
    },
    {
      "epoch": 0.10432039760516268,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009769225961274923,
      "loss": 2.1846,
      "step": 15354
    },
    {
      "epoch": 0.10432719195175674,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000976919365227524,
      "loss": 2.3325,
      "step": 15355
    },
    {
      "epoch": 0.1043339862983508,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009769161341067472,
      "loss": 2.2326,
      "step": 15356
    },
    {
      "epoch": 0.10434078064494486,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009769129027651638,
      "loss": 2.3203,
      "step": 15357
    },
    {
      "epoch": 0.10434757499153892,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009769096712027752,
      "loss": 2.1589,
      "step": 15358
    },
    {
      "epoch": 0.10435436933813297,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009769064394195828,
      "loss": 2.2419,
      "step": 15359
    },
    {
      "epoch": 0.10436116368472703,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009769032074155883,
      "loss": 2.0915,
      "step": 15360
    },
    {
      "epoch": 0.1043679580313211,
      "grad_norm": 0.140625,
      "learning_rate": 0.000976899975190793,
      "loss": 2.2647,
      "step": 15361
    },
    {
      "epoch": 0.10437475237791514,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009768967427451986,
      "loss": 2.2341,
      "step": 15362
    },
    {
      "epoch": 0.10438154672450921,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009768935100788062,
      "loss": 2.181,
      "step": 15363
    },
    {
      "epoch": 0.10438834107110326,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009768902771916178,
      "loss": 2.2936,
      "step": 15364
    },
    {
      "epoch": 0.10439513541769732,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009768870440836346,
      "loss": 2.1937,
      "step": 15365
    },
    {
      "epoch": 0.10440192976429137,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000976883810754858,
      "loss": 2.1424,
      "step": 15366
    },
    {
      "epoch": 0.10440872411088543,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.00097688057720529,
      "loss": 2.3453,
      "step": 15367
    },
    {
      "epoch": 0.1044155184574795,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009768773434349317,
      "loss": 2.1803,
      "step": 15368
    },
    {
      "epoch": 0.10442231280407355,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009768741094437845,
      "loss": 2.2662,
      "step": 15369
    },
    {
      "epoch": 0.10442910715066761,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009768708752318502,
      "loss": 2.1336,
      "step": 15370
    },
    {
      "epoch": 0.10443590149726166,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009768676407991299,
      "loss": 2.287,
      "step": 15371
    },
    {
      "epoch": 0.10444269584385572,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009768644061456256,
      "loss": 2.1899,
      "step": 15372
    },
    {
      "epoch": 0.10444949019044979,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009768611712713383,
      "loss": 2.2426,
      "step": 15373
    },
    {
      "epoch": 0.10445628453704384,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009768579361762702,
      "loss": 2.1766,
      "step": 15374
    },
    {
      "epoch": 0.1044630788836379,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000976854700860422,
      "loss": 2.2431,
      "step": 15375
    },
    {
      "epoch": 0.10446987323023195,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009768514653237957,
      "loss": 2.1136,
      "step": 15376
    },
    {
      "epoch": 0.10447666757682601,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009768482295663924,
      "loss": 2.2058,
      "step": 15377
    },
    {
      "epoch": 0.10448346192342008,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000976844993588214,
      "loss": 2.1929,
      "step": 15378
    },
    {
      "epoch": 0.10449025627001413,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009768417573892618,
      "loss": 2.2064,
      "step": 15379
    },
    {
      "epoch": 0.10449705061660819,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009768385209695373,
      "loss": 2.3907,
      "step": 15380
    },
    {
      "epoch": 0.10450384496320224,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009768352843290422,
      "loss": 2.272,
      "step": 15381
    },
    {
      "epoch": 0.1045106393097963,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009768320474677777,
      "loss": 2.2777,
      "step": 15382
    },
    {
      "epoch": 0.10451743365639037,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009768288103857454,
      "loss": 2.17,
      "step": 15383
    },
    {
      "epoch": 0.10452422800298442,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009768255730829467,
      "loss": 2.2873,
      "step": 15384
    },
    {
      "epoch": 0.10453102234957848,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009768223355593835,
      "loss": 2.0961,
      "step": 15385
    },
    {
      "epoch": 0.10453781669617253,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000976819097815057,
      "loss": 2.1837,
      "step": 15386
    },
    {
      "epoch": 0.10454461104276659,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009768158598499686,
      "loss": 2.1152,
      "step": 15387
    },
    {
      "epoch": 0.10455140538936064,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0009768126216641198,
      "loss": 2.1384,
      "step": 15388
    },
    {
      "epoch": 0.1045581997359547,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009768093832575125,
      "loss": 2.2412,
      "step": 15389
    },
    {
      "epoch": 0.10456499408254877,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009768061446301477,
      "loss": 2.0619,
      "step": 15390
    },
    {
      "epoch": 0.10457178842914282,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009768029057820274,
      "loss": 2.0433,
      "step": 15391
    },
    {
      "epoch": 0.10457858277573688,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009767996667131526,
      "loss": 2.2584,
      "step": 15392
    },
    {
      "epoch": 0.10458537712233093,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000976796427423525,
      "loss": 2.3423,
      "step": 15393
    },
    {
      "epoch": 0.104592171468925,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009767931879131463,
      "loss": 2.1931,
      "step": 15394
    },
    {
      "epoch": 0.10459896581551906,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009767899481820175,
      "loss": 2.2835,
      "step": 15395
    },
    {
      "epoch": 0.1046057601621131,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009767867082301407,
      "loss": 2.1714,
      "step": 15396
    },
    {
      "epoch": 0.10461255450870717,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000976783468057517,
      "loss": 2.0942,
      "step": 15397
    },
    {
      "epoch": 0.10461934885530122,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009767802276641482,
      "loss": 2.2483,
      "step": 15398
    },
    {
      "epoch": 0.10462614320189528,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009767769870500354,
      "loss": 2.1849,
      "step": 15399
    },
    {
      "epoch": 0.10463293754848935,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009767737462151804,
      "loss": 2.1697,
      "step": 15400
    },
    {
      "epoch": 0.1046397318950834,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009767705051595846,
      "loss": 2.1764,
      "step": 15401
    },
    {
      "epoch": 0.10464652624167746,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009767672638832496,
      "loss": 2.2275,
      "step": 15402
    },
    {
      "epoch": 0.10465332058827151,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009767640223861767,
      "loss": 2.2796,
      "step": 15403
    },
    {
      "epoch": 0.10466011493486557,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009767607806683675,
      "loss": 2.2995,
      "step": 15404
    },
    {
      "epoch": 0.10466690928145964,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009767575387298237,
      "loss": 2.195,
      "step": 15405
    },
    {
      "epoch": 0.10467370362805369,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009767542965705466,
      "loss": 2.1157,
      "step": 15406
    },
    {
      "epoch": 0.10468049797464775,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009767510541905376,
      "loss": 2.2595,
      "step": 15407
    },
    {
      "epoch": 0.1046872923212418,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009767478115897984,
      "loss": 2.204,
      "step": 15408
    },
    {
      "epoch": 0.10469408666783586,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009767445687683305,
      "loss": 2.1771,
      "step": 15409
    },
    {
      "epoch": 0.10470088101442993,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009767413257261352,
      "loss": 2.0394,
      "step": 15410
    },
    {
      "epoch": 0.10470767536102397,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009767380824632142,
      "loss": 2.1517,
      "step": 15411
    },
    {
      "epoch": 0.10471446970761804,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000976734838979569,
      "loss": 2.2512,
      "step": 15412
    },
    {
      "epoch": 0.10472126405421209,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009767315952752009,
      "loss": 2.2283,
      "step": 15413
    },
    {
      "epoch": 0.10472805840080615,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009767283513501116,
      "loss": 2.3592,
      "step": 15414
    },
    {
      "epoch": 0.1047348527474002,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009767251072043026,
      "loss": 2.1275,
      "step": 15415
    },
    {
      "epoch": 0.10474164709399426,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009767218628377752,
      "loss": 2.1835,
      "step": 15416
    },
    {
      "epoch": 0.10474844144058833,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009767186182505312,
      "loss": 2.0401,
      "step": 15417
    },
    {
      "epoch": 0.10475523578718238,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009767153734425717,
      "loss": 2.2107,
      "step": 15418
    },
    {
      "epoch": 0.10476203013377644,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009767121284138987,
      "loss": 2.1051,
      "step": 15419
    },
    {
      "epoch": 0.10476882448037049,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009767088831645135,
      "loss": 2.1592,
      "step": 15420
    },
    {
      "epoch": 0.10477561882696455,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009767056376944173,
      "loss": 2.1421,
      "step": 15421
    },
    {
      "epoch": 0.10478241317355862,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000976702392003612,
      "loss": 2.3833,
      "step": 15422
    },
    {
      "epoch": 0.10478920752015267,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009766991460920988,
      "loss": 2.2372,
      "step": 15423
    },
    {
      "epoch": 0.10479600186674673,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009766958999598797,
      "loss": 2.3419,
      "step": 15424
    },
    {
      "epoch": 0.10480279621334078,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009766926536069556,
      "loss": 2.2504,
      "step": 15425
    },
    {
      "epoch": 0.10480959055993484,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009766894070333283,
      "loss": 2.2249,
      "step": 15426
    },
    {
      "epoch": 0.1048163849065289,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009766861602389992,
      "loss": 2.1901,
      "step": 15427
    },
    {
      "epoch": 0.10482317925312296,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009766829132239702,
      "loss": 2.2037,
      "step": 15428
    },
    {
      "epoch": 0.10482997359971702,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000976679665988242,
      "loss": 2.3112,
      "step": 15429
    },
    {
      "epoch": 0.10483676794631107,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000976676418531817,
      "loss": 2.2431,
      "step": 15430
    },
    {
      "epoch": 0.10484356229290513,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000976673170854696,
      "loss": 2.1159,
      "step": 15431
    },
    {
      "epoch": 0.1048503566394992,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000976669922956881,
      "loss": 2.2284,
      "step": 15432
    },
    {
      "epoch": 0.10485715098609324,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000976666674838373,
      "loss": 2.165,
      "step": 15433
    },
    {
      "epoch": 0.10486394533268731,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000976663426499174,
      "loss": 2.1535,
      "step": 15434
    },
    {
      "epoch": 0.10487073967928136,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009766601779392851,
      "loss": 2.2194,
      "step": 15435
    },
    {
      "epoch": 0.10487753402587542,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009766569291587083,
      "loss": 2.3496,
      "step": 15436
    },
    {
      "epoch": 0.10488432837246947,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009766536801574448,
      "loss": 2.2757,
      "step": 15437
    },
    {
      "epoch": 0.10489112271906353,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000976650430935496,
      "loss": 2.1345,
      "step": 15438
    },
    {
      "epoch": 0.1048979170656576,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009766471814928635,
      "loss": 2.1521,
      "step": 15439
    },
    {
      "epoch": 0.10490471141225165,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009766439318295488,
      "loss": 2.1567,
      "step": 15440
    },
    {
      "epoch": 0.10491150575884571,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009766406819455534,
      "loss": 2.1675,
      "step": 15441
    },
    {
      "epoch": 0.10491830010543976,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009766374318408788,
      "loss": 2.1425,
      "step": 15442
    },
    {
      "epoch": 0.10492509445203382,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009766341815155267,
      "loss": 2.2182,
      "step": 15443
    },
    {
      "epoch": 0.10493188879862789,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009766309309694984,
      "loss": 2.1223,
      "step": 15444
    },
    {
      "epoch": 0.10493868314522194,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009766276802027954,
      "loss": 2.3082,
      "step": 15445
    },
    {
      "epoch": 0.104945477491816,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009766244292154192,
      "loss": 2.2437,
      "step": 15446
    },
    {
      "epoch": 0.10495227183841005,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009766211780073712,
      "loss": 2.0837,
      "step": 15447
    },
    {
      "epoch": 0.10495906618500411,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009766179265786534,
      "loss": 2.3303,
      "step": 15448
    },
    {
      "epoch": 0.10496586053159818,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009766146749292668,
      "loss": 2.1984,
      "step": 15449
    },
    {
      "epoch": 0.10497265487819223,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009766114230592132,
      "loss": 2.1612,
      "step": 15450
    },
    {
      "epoch": 0.10497944922478629,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009766081709684938,
      "loss": 2.1683,
      "step": 15451
    },
    {
      "epoch": 0.10498624357138034,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009766049186571103,
      "loss": 2.1857,
      "step": 15452
    },
    {
      "epoch": 0.1049930379179744,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000976601666125064,
      "loss": 2.0945,
      "step": 15453
    },
    {
      "epoch": 0.10499983226456847,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000976598413372357,
      "loss": 2.2369,
      "step": 15454
    },
    {
      "epoch": 0.10500662661116252,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009765951603989901,
      "loss": 2.1682,
      "step": 15455
    },
    {
      "epoch": 0.10501342095775658,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009765919072049654,
      "loss": 2.3911,
      "step": 15456
    },
    {
      "epoch": 0.10502021530435063,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009765886537902839,
      "loss": 2.1049,
      "step": 15457
    },
    {
      "epoch": 0.10502700965094469,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009765854001549472,
      "loss": 2.1429,
      "step": 15458
    },
    {
      "epoch": 0.10503380399753874,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009765821462989573,
      "loss": 2.1756,
      "step": 15459
    },
    {
      "epoch": 0.1050405983441328,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000976578892222315,
      "loss": 2.1787,
      "step": 15460
    },
    {
      "epoch": 0.10504739269072687,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009765756379250222,
      "loss": 2.1694,
      "step": 15461
    },
    {
      "epoch": 0.10505418703732092,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009765723834070804,
      "loss": 2.1336,
      "step": 15462
    },
    {
      "epoch": 0.10506098138391498,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000976569128668491,
      "loss": 2.0819,
      "step": 15463
    },
    {
      "epoch": 0.10506777573050903,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009765658737092558,
      "loss": 2.0664,
      "step": 15464
    },
    {
      "epoch": 0.1050745700771031,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009765626185293758,
      "loss": 2.2521,
      "step": 15465
    },
    {
      "epoch": 0.10508136442369716,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009765593631288527,
      "loss": 2.2947,
      "step": 15466
    },
    {
      "epoch": 0.1050881587702912,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009765561075076884,
      "loss": 2.202,
      "step": 15467
    },
    {
      "epoch": 0.10509495311688527,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009765528516658836,
      "loss": 2.1686,
      "step": 15468
    },
    {
      "epoch": 0.10510174746347932,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009765495956034406,
      "loss": 2.123,
      "step": 15469
    },
    {
      "epoch": 0.10510854181007338,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009765463393203607,
      "loss": 2.1323,
      "step": 15470
    },
    {
      "epoch": 0.10511533615666745,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009765430828166452,
      "loss": 2.2302,
      "step": 15471
    },
    {
      "epoch": 0.1051221305032615,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009765398260922957,
      "loss": 2.0961,
      "step": 15472
    },
    {
      "epoch": 0.10512892484985556,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009765365691473138,
      "loss": 2.1092,
      "step": 15473
    },
    {
      "epoch": 0.10513571919644961,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009765333119817008,
      "loss": 2.1344,
      "step": 15474
    },
    {
      "epoch": 0.10514251354304367,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009765300545954584,
      "loss": 2.1237,
      "step": 15475
    },
    {
      "epoch": 0.10514930788963774,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000976526796988588,
      "loss": 2.1839,
      "step": 15476
    },
    {
      "epoch": 0.10515610223623179,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009765235391610913,
      "loss": 2.2582,
      "step": 15477
    },
    {
      "epoch": 0.10516289658282585,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009765202811129695,
      "loss": 2.1241,
      "step": 15478
    },
    {
      "epoch": 0.1051696909294199,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009765170228442244,
      "loss": 2.1687,
      "step": 15479
    },
    {
      "epoch": 0.10517648527601396,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009765137643548574,
      "loss": 2.3086,
      "step": 15480
    },
    {
      "epoch": 0.10518327962260803,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009765105056448697,
      "loss": 2.2688,
      "step": 15481
    },
    {
      "epoch": 0.10519007396920207,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009765072467142635,
      "loss": 2.129,
      "step": 15482
    },
    {
      "epoch": 0.10519686831579614,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009765039875630397,
      "loss": 2.3285,
      "step": 15483
    },
    {
      "epoch": 0.10520366266239019,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009765007281912,
      "loss": 2.2257,
      "step": 15484
    },
    {
      "epoch": 0.10521045700898425,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000976497468598746,
      "loss": 2.0194,
      "step": 15485
    },
    {
      "epoch": 0.1052172513555783,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.000976494208785679,
      "loss": 2.112,
      "step": 15486
    },
    {
      "epoch": 0.10522404570217236,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009764909487520008,
      "loss": 2.1749,
      "step": 15487
    },
    {
      "epoch": 0.10523084004876643,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009764876884977128,
      "loss": 2.2083,
      "step": 15488
    },
    {
      "epoch": 0.10523763439536048,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009764844280228162,
      "loss": 2.3639,
      "step": 15489
    },
    {
      "epoch": 0.10524442874195454,
      "grad_norm": 0.125,
      "learning_rate": 0.0009764811673273131,
      "loss": 2.0296,
      "step": 15490
    },
    {
      "epoch": 0.10525122308854859,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009764779064112043,
      "loss": 2.3029,
      "step": 15491
    },
    {
      "epoch": 0.10525801743514265,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000976474645274492,
      "loss": 2.2373,
      "step": 15492
    },
    {
      "epoch": 0.10526481178173672,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009764713839171773,
      "loss": 2.2015,
      "step": 15493
    },
    {
      "epoch": 0.10527160612833077,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009764681223392619,
      "loss": 2.2863,
      "step": 15494
    },
    {
      "epoch": 0.10527840047492483,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009764648605407471,
      "loss": 2.1447,
      "step": 15495
    },
    {
      "epoch": 0.10528519482151888,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009764615985216346,
      "loss": 2.2259,
      "step": 15496
    },
    {
      "epoch": 0.10529198916811294,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009764583362819258,
      "loss": 2.0496,
      "step": 15497
    },
    {
      "epoch": 0.105298783514707,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009764550738216222,
      "loss": 2.0945,
      "step": 15498
    },
    {
      "epoch": 0.10530557786130106,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009764518111407255,
      "loss": 2.2209,
      "step": 15499
    },
    {
      "epoch": 0.10531237220789512,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000976448548239237,
      "loss": 2.0685,
      "step": 15500
    },
    {
      "epoch": 0.10531916655448917,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009764452851171583,
      "loss": 2.1942,
      "step": 15501
    },
    {
      "epoch": 0.10532596090108323,
      "grad_norm": 0.125,
      "learning_rate": 0.000976442021774491,
      "loss": 2.0803,
      "step": 15502
    },
    {
      "epoch": 0.1053327552476773,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009764387582112362,
      "loss": 2.1498,
      "step": 15503
    },
    {
      "epoch": 0.10533954959427135,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009764354944273959,
      "loss": 2.1636,
      "step": 15504
    },
    {
      "epoch": 0.10534634394086541,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009764322304229716,
      "loss": 2.1494,
      "step": 15505
    },
    {
      "epoch": 0.10535313828745946,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009764289661979644,
      "loss": 2.1856,
      "step": 15506
    },
    {
      "epoch": 0.10535993263405352,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009764257017523761,
      "loss": 2.26,
      "step": 15507
    },
    {
      "epoch": 0.10536672698064757,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009764224370862083,
      "loss": 2.1007,
      "step": 15508
    },
    {
      "epoch": 0.10537352132724163,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009764191721994622,
      "loss": 2.2029,
      "step": 15509
    },
    {
      "epoch": 0.1053803156738357,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009764159070921395,
      "loss": 2.296,
      "step": 15510
    },
    {
      "epoch": 0.10538711002042975,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009764126417642419,
      "loss": 2.3256,
      "step": 15511
    },
    {
      "epoch": 0.10539390436702381,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009764093762157704,
      "loss": 2.1848,
      "step": 15512
    },
    {
      "epoch": 0.10540069871361786,
      "grad_norm": 0.357421875,
      "learning_rate": 0.000976406110446727,
      "loss": 2.2909,
      "step": 15513
    },
    {
      "epoch": 0.10540749306021192,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000976402844457113,
      "loss": 2.2047,
      "step": 15514
    },
    {
      "epoch": 0.10541428740680599,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009763995782469299,
      "loss": 1.9631,
      "step": 15515
    },
    {
      "epoch": 0.10542108175340004,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009763963118161794,
      "loss": 2.2347,
      "step": 15516
    },
    {
      "epoch": 0.1054278760999941,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009763930451648628,
      "loss": 2.2477,
      "step": 15517
    },
    {
      "epoch": 0.10543467044658815,
      "grad_norm": 0.1708984375,
      "learning_rate": 0.0009763897782929816,
      "loss": 2.0951,
      "step": 15518
    },
    {
      "epoch": 0.10544146479318221,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009763865112005374,
      "loss": 2.2568,
      "step": 15519
    },
    {
      "epoch": 0.10544825913977628,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009763832438875319,
      "loss": 2.2711,
      "step": 15520
    },
    {
      "epoch": 0.10545505348637033,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009763799763539663,
      "loss": 2.1047,
      "step": 15521
    },
    {
      "epoch": 0.10546184783296439,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000976376708599842,
      "loss": 2.1089,
      "step": 15522
    },
    {
      "epoch": 0.10546864217955844,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009763734406251611,
      "loss": 2.1873,
      "step": 15523
    },
    {
      "epoch": 0.1054754365261525,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009763701724299246,
      "loss": 2.3665,
      "step": 15524
    },
    {
      "epoch": 0.10548223087274657,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009763669040141342,
      "loss": 2.1436,
      "step": 15525
    },
    {
      "epoch": 0.10548902521934062,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009763636353777913,
      "loss": 2.1244,
      "step": 15526
    },
    {
      "epoch": 0.10549581956593468,
      "grad_norm": 0.1767578125,
      "learning_rate": 0.0009763603665208975,
      "loss": 2.3546,
      "step": 15527
    },
    {
      "epoch": 0.10550261391252873,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009763570974434545,
      "loss": 2.2019,
      "step": 15528
    },
    {
      "epoch": 0.10550940825912279,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009763538281454636,
      "loss": 2.0137,
      "step": 15529
    },
    {
      "epoch": 0.10551620260571686,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009763505586269261,
      "loss": 2.2892,
      "step": 15530
    },
    {
      "epoch": 0.1055229969523109,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000976347288887844,
      "loss": 2.1373,
      "step": 15531
    },
    {
      "epoch": 0.10552979129890497,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009763440189282185,
      "loss": 2.249,
      "step": 15532
    },
    {
      "epoch": 0.10553658564549902,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009763407487480511,
      "loss": 2.1298,
      "step": 15533
    },
    {
      "epoch": 0.10554337999209308,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009763374783473435,
      "loss": 2.3569,
      "step": 15534
    },
    {
      "epoch": 0.10555017433868713,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009763342077260969,
      "loss": 2.2737,
      "step": 15535
    },
    {
      "epoch": 0.1055569686852812,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009763309368843132,
      "loss": 2.2067,
      "step": 15536
    },
    {
      "epoch": 0.10556376303187526,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009763276658219937,
      "loss": 2.1258,
      "step": 15537
    },
    {
      "epoch": 0.1055705573784693,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009763243945391401,
      "loss": 2.3258,
      "step": 15538
    },
    {
      "epoch": 0.10557735172506337,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009763211230357536,
      "loss": 2.1341,
      "step": 15539
    },
    {
      "epoch": 0.10558414607165742,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009763178513118358,
      "loss": 2.1178,
      "step": 15540
    },
    {
      "epoch": 0.10559094041825148,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009763145793673884,
      "loss": 2.1706,
      "step": 15541
    },
    {
      "epoch": 0.10559773476484555,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009763113072024128,
      "loss": 2.1944,
      "step": 15542
    },
    {
      "epoch": 0.1056045291114396,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009763080348169107,
      "loss": 2.2269,
      "step": 15543
    },
    {
      "epoch": 0.10561132345803366,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009763047622108832,
      "loss": 2.1129,
      "step": 15544
    },
    {
      "epoch": 0.10561811780462771,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000976301489384332,
      "loss": 2.1339,
      "step": 15545
    },
    {
      "epoch": 0.10562491215122177,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009762982163372589,
      "loss": 2.2628,
      "step": 15546
    },
    {
      "epoch": 0.10563170649781584,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009762949430696651,
      "loss": 2.1788,
      "step": 15547
    },
    {
      "epoch": 0.10563850084440989,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009762916695815522,
      "loss": 2.2193,
      "step": 15548
    },
    {
      "epoch": 0.10564529519100395,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009762883958729218,
      "loss": 2.21,
      "step": 15549
    },
    {
      "epoch": 0.105652089537598,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009762851219437752,
      "loss": 2.206,
      "step": 15550
    },
    {
      "epoch": 0.10565888388419206,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000976281847794114,
      "loss": 2.3608,
      "step": 15551
    },
    {
      "epoch": 0.10566567823078613,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009762785734239399,
      "loss": 2.1522,
      "step": 15552
    },
    {
      "epoch": 0.10567247257738017,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009762752988332541,
      "loss": 2.2386,
      "step": 15553
    },
    {
      "epoch": 0.10567926692397424,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009762720240220586,
      "loss": 2.2594,
      "step": 15554
    },
    {
      "epoch": 0.10568606127056829,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009762687489903544,
      "loss": 2.3131,
      "step": 15555
    },
    {
      "epoch": 0.10569285561716235,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009762654737381434,
      "loss": 2.0606,
      "step": 15556
    },
    {
      "epoch": 0.1056996499637564,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009762621982654267,
      "loss": 2.141,
      "step": 15557
    },
    {
      "epoch": 0.10570644431035046,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009762589225722063,
      "loss": 2.4572,
      "step": 15558
    },
    {
      "epoch": 0.10571323865694453,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009762556466584834,
      "loss": 2.3474,
      "step": 15559
    },
    {
      "epoch": 0.10572003300353858,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009762523705242594,
      "loss": 2.2824,
      "step": 15560
    },
    {
      "epoch": 0.10572682735013264,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009762490941695362,
      "loss": 2.3552,
      "step": 15561
    },
    {
      "epoch": 0.10573362169672669,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009762458175943152,
      "loss": 2.0924,
      "step": 15562
    },
    {
      "epoch": 0.10574041604332075,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009762425407985976,
      "loss": 2.2495,
      "step": 15563
    },
    {
      "epoch": 0.10574721038991482,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009762392637823853,
      "loss": 2.234,
      "step": 15564
    },
    {
      "epoch": 0.10575400473650887,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009762359865456798,
      "loss": 2.2152,
      "step": 15565
    },
    {
      "epoch": 0.10576079908310293,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009762327090884824,
      "loss": 2.2025,
      "step": 15566
    },
    {
      "epoch": 0.10576759342969698,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009762294314107947,
      "loss": 2.1795,
      "step": 15567
    },
    {
      "epoch": 0.10577438777629104,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009762261535126183,
      "loss": 2.0654,
      "step": 15568
    },
    {
      "epoch": 0.1057811821228851,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009762228753939545,
      "loss": 2.1373,
      "step": 15569
    },
    {
      "epoch": 0.10578797646947916,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009762195970548051,
      "loss": 2.3191,
      "step": 15570
    },
    {
      "epoch": 0.10579477081607322,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009762163184951713,
      "loss": 2.4166,
      "step": 15571
    },
    {
      "epoch": 0.10580156516266727,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000976213039715055,
      "loss": 2.1435,
      "step": 15572
    },
    {
      "epoch": 0.10580835950926133,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009762097607144576,
      "loss": 2.2832,
      "step": 15573
    },
    {
      "epoch": 0.1058151538558554,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009762064814933803,
      "loss": 2.2675,
      "step": 15574
    },
    {
      "epoch": 0.10582194820244945,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000976203202051825,
      "loss": 2.2798,
      "step": 15575
    },
    {
      "epoch": 0.10582874254904351,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009761999223897931,
      "loss": 2.3422,
      "step": 15576
    },
    {
      "epoch": 0.10583553689563756,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.000976196642507286,
      "loss": 2.1307,
      "step": 15577
    },
    {
      "epoch": 0.10584233124223162,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009761933624043053,
      "loss": 2.1878,
      "step": 15578
    },
    {
      "epoch": 0.10584912558882567,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009761900820808527,
      "loss": 2.195,
      "step": 15579
    },
    {
      "epoch": 0.10585591993541973,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009761868015369293,
      "loss": 2.2688,
      "step": 15580
    },
    {
      "epoch": 0.1058627142820138,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009761835207725371,
      "loss": 2.1525,
      "step": 15581
    },
    {
      "epoch": 0.10586950862860785,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009761802397876773,
      "loss": 2.1649,
      "step": 15582
    },
    {
      "epoch": 0.10587630297520191,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009761769585823514,
      "loss": 2.1846,
      "step": 15583
    },
    {
      "epoch": 0.10588309732179596,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009761736771565613,
      "loss": 2.1493,
      "step": 15584
    },
    {
      "epoch": 0.10588989166839002,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000976170395510308,
      "loss": 2.1869,
      "step": 15585
    },
    {
      "epoch": 0.10589668601498409,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009761671136435933,
      "loss": 2.1608,
      "step": 15586
    },
    {
      "epoch": 0.10590348036157814,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009761638315564187,
      "loss": 2.2525,
      "step": 15587
    },
    {
      "epoch": 0.1059102747081722,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009761605492487858,
      "loss": 2.207,
      "step": 15588
    },
    {
      "epoch": 0.10591706905476625,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009761572667206959,
      "loss": 2.3205,
      "step": 15589
    },
    {
      "epoch": 0.10592386340136031,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009761539839721507,
      "loss": 2.2524,
      "step": 15590
    },
    {
      "epoch": 0.10593065774795438,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009761507010031519,
      "loss": 2.1937,
      "step": 15591
    },
    {
      "epoch": 0.10593745209454843,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009761474178137004,
      "loss": 2.1803,
      "step": 15592
    },
    {
      "epoch": 0.10594424644114249,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009761441344037983,
      "loss": 2.3273,
      "step": 15593
    },
    {
      "epoch": 0.10595104078773654,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009761408507734468,
      "loss": 2.2616,
      "step": 15594
    },
    {
      "epoch": 0.1059578351343306,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009761375669226478,
      "loss": 2.0986,
      "step": 15595
    },
    {
      "epoch": 0.10596462948092467,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009761342828514023,
      "loss": 2.3073,
      "step": 15596
    },
    {
      "epoch": 0.10597142382751872,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009761309985597122,
      "loss": 2.1874,
      "step": 15597
    },
    {
      "epoch": 0.10597821817411278,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009761277140475788,
      "loss": 2.1512,
      "step": 15598
    },
    {
      "epoch": 0.10598501252070683,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009761244293150038,
      "loss": 2.0945,
      "step": 15599
    },
    {
      "epoch": 0.10599180686730089,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009761211443619888,
      "loss": 2.2449,
      "step": 15600
    },
    {
      "epoch": 0.10599860121389496,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009761178591885349,
      "loss": 2.2796,
      "step": 15601
    },
    {
      "epoch": 0.106005395560489,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009761145737946441,
      "loss": 2.1511,
      "step": 15602
    },
    {
      "epoch": 0.10601218990708307,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009761112881803176,
      "loss": 2.2315,
      "step": 15603
    },
    {
      "epoch": 0.10601898425367712,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009761080023455571,
      "loss": 2.2299,
      "step": 15604
    },
    {
      "epoch": 0.10602577860027118,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000976104716290364,
      "loss": 2.2052,
      "step": 15605
    },
    {
      "epoch": 0.10603257294686523,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009761014300147399,
      "loss": 2.2476,
      "step": 15606
    },
    {
      "epoch": 0.1060393672934593,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009760981435186862,
      "loss": 2.2086,
      "step": 15607
    },
    {
      "epoch": 0.10604616164005336,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009760948568022046,
      "loss": 2.1892,
      "step": 15608
    },
    {
      "epoch": 0.10605295598664741,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009760915698652966,
      "loss": 2.3342,
      "step": 15609
    },
    {
      "epoch": 0.10605975033324147,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009760882827079635,
      "loss": 2.2488,
      "step": 15610
    },
    {
      "epoch": 0.10606654467983552,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009760849953302071,
      "loss": 2.1158,
      "step": 15611
    },
    {
      "epoch": 0.10607333902642958,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009760817077320289,
      "loss": 2.2407,
      "step": 15612
    },
    {
      "epoch": 0.10608013337302365,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009760784199134303,
      "loss": 2.1304,
      "step": 15613
    },
    {
      "epoch": 0.1060869277196177,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009760751318744126,
      "loss": 2.2379,
      "step": 15614
    },
    {
      "epoch": 0.10609372206621176,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009760718436149778,
      "loss": 2.2049,
      "step": 15615
    },
    {
      "epoch": 0.10610051641280581,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009760685551351272,
      "loss": 2.1191,
      "step": 15616
    },
    {
      "epoch": 0.10610731075939987,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009760652664348622,
      "loss": 2.2598,
      "step": 15617
    },
    {
      "epoch": 0.10611410510599394,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009760619775141846,
      "loss": 2.196,
      "step": 15618
    },
    {
      "epoch": 0.10612089945258799,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009760586883730955,
      "loss": 2.4381,
      "step": 15619
    },
    {
      "epoch": 0.10612769379918205,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009760553990115969,
      "loss": 2.2002,
      "step": 15620
    },
    {
      "epoch": 0.1061344881457761,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009760521094296901,
      "loss": 2.1103,
      "step": 15621
    },
    {
      "epoch": 0.10614128249237016,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009760488196273765,
      "loss": 2.2876,
      "step": 15622
    },
    {
      "epoch": 0.10614807683896423,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009760455296046579,
      "loss": 2.1455,
      "step": 15623
    },
    {
      "epoch": 0.10615487118555827,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009760422393615356,
      "loss": 2.1523,
      "step": 15624
    },
    {
      "epoch": 0.10616166553215234,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009760389488980112,
      "loss": 2.3323,
      "step": 15625
    },
    {
      "epoch": 0.10616845987874639,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009760356582140861,
      "loss": 2.2872,
      "step": 15626
    },
    {
      "epoch": 0.10617525422534045,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009760323673097622,
      "loss": 2.3303,
      "step": 15627
    },
    {
      "epoch": 0.1061820485719345,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009760290761850407,
      "loss": 2.3892,
      "step": 15628
    },
    {
      "epoch": 0.10618884291852856,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009760257848399231,
      "loss": 2.1935,
      "step": 15629
    },
    {
      "epoch": 0.10619563726512263,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000976022493274411,
      "loss": 2.3072,
      "step": 15630
    },
    {
      "epoch": 0.10620243161171668,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009760192014885061,
      "loss": 2.2789,
      "step": 15631
    },
    {
      "epoch": 0.10620922595831074,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009760159094822096,
      "loss": 2.1465,
      "step": 15632
    },
    {
      "epoch": 0.10621602030490479,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009760126172555232,
      "loss": 2.247,
      "step": 15633
    },
    {
      "epoch": 0.10622281465149885,
      "grad_norm": 0.189453125,
      "learning_rate": 0.0009760093248084486,
      "loss": 2.206,
      "step": 15634
    },
    {
      "epoch": 0.10622960899809292,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000976006032140987,
      "loss": 2.2288,
      "step": 15635
    },
    {
      "epoch": 0.10623640334468697,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.00097600273925314,
      "loss": 2.2179,
      "step": 15636
    },
    {
      "epoch": 0.10624319769128103,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009759994461449093,
      "loss": 2.0961,
      "step": 15637
    },
    {
      "epoch": 0.10624999203787508,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009759961528162964,
      "loss": 2.2376,
      "step": 15638
    },
    {
      "epoch": 0.10625678638446914,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009759928592673027,
      "loss": 2.1973,
      "step": 15639
    },
    {
      "epoch": 0.1062635807310632,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009759895654979296,
      "loss": 2.1179,
      "step": 15640
    },
    {
      "epoch": 0.10627037507765726,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009759862715081789,
      "loss": 2.1963,
      "step": 15641
    },
    {
      "epoch": 0.10627716942425132,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000975982977298052,
      "loss": 2.1759,
      "step": 15642
    },
    {
      "epoch": 0.10628396377084537,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009759796828675505,
      "loss": 2.1292,
      "step": 15643
    },
    {
      "epoch": 0.10629075811743943,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009759763882166756,
      "loss": 2.1709,
      "step": 15644
    },
    {
      "epoch": 0.1062975524640335,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009759730933454294,
      "loss": 2.269,
      "step": 15645
    },
    {
      "epoch": 0.10630434681062755,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009759697982538129,
      "loss": 2.1854,
      "step": 15646
    },
    {
      "epoch": 0.10631114115722161,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009759665029418281,
      "loss": 2.3093,
      "step": 15647
    },
    {
      "epoch": 0.10631793550381566,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009759632074094759,
      "loss": 2.169,
      "step": 15648
    },
    {
      "epoch": 0.10632472985040972,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009759599116567585,
      "loss": 2.044,
      "step": 15649
    },
    {
      "epoch": 0.10633152419700378,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000975956615683677,
      "loss": 2.2646,
      "step": 15650
    },
    {
      "epoch": 0.10633831854359783,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009759533194902331,
      "loss": 2.1656,
      "step": 15651
    },
    {
      "epoch": 0.1063451128901919,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009759500230764282,
      "loss": 2.2022,
      "step": 15652
    },
    {
      "epoch": 0.10635190723678595,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009759467264422639,
      "loss": 2.3332,
      "step": 15653
    },
    {
      "epoch": 0.10635870158338001,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009759434295877418,
      "loss": 2.1028,
      "step": 15654
    },
    {
      "epoch": 0.10636549592997406,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009759401325128634,
      "loss": 2.1823,
      "step": 15655
    },
    {
      "epoch": 0.10637229027656812,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00097593683521763,
      "loss": 2.2038,
      "step": 15656
    },
    {
      "epoch": 0.10637908462316219,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009759335377020433,
      "loss": 2.0999,
      "step": 15657
    },
    {
      "epoch": 0.10638587896975624,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000975930239966105,
      "loss": 2.1035,
      "step": 15658
    },
    {
      "epoch": 0.1063926733163503,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009759269420098163,
      "loss": 2.2664,
      "step": 15659
    },
    {
      "epoch": 0.10639946766294435,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009759236438331789,
      "loss": 2.2743,
      "step": 15660
    },
    {
      "epoch": 0.10640626200953841,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009759203454361944,
      "loss": 2.0417,
      "step": 15661
    },
    {
      "epoch": 0.10641305635613248,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009759170468188642,
      "loss": 2.1412,
      "step": 15662
    },
    {
      "epoch": 0.10641985070272653,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009759137479811898,
      "loss": 2.1314,
      "step": 15663
    },
    {
      "epoch": 0.10642664504932059,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009759104489231729,
      "loss": 2.2659,
      "step": 15664
    },
    {
      "epoch": 0.10643343939591464,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009759071496448149,
      "loss": 2.2397,
      "step": 15665
    },
    {
      "epoch": 0.1064402337425087,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009759038501461173,
      "loss": 2.2381,
      "step": 15666
    },
    {
      "epoch": 0.10644702808910277,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009759005504270815,
      "loss": 2.3317,
      "step": 15667
    },
    {
      "epoch": 0.10645382243569682,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009758972504877096,
      "loss": 2.2941,
      "step": 15668
    },
    {
      "epoch": 0.10646061678229088,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009758939503280024,
      "loss": 2.2099,
      "step": 15669
    },
    {
      "epoch": 0.10646741112888493,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009758906499479619,
      "loss": 2.1606,
      "step": 15670
    },
    {
      "epoch": 0.10647420547547899,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009758873493475894,
      "loss": 2.1202,
      "step": 15671
    },
    {
      "epoch": 0.10648099982207306,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009758840485268865,
      "loss": 2.2528,
      "step": 15672
    },
    {
      "epoch": 0.1064877941686671,
      "grad_norm": 0.125,
      "learning_rate": 0.0009758807474858548,
      "loss": 2.2466,
      "step": 15673
    },
    {
      "epoch": 0.10649458851526117,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009758774462244958,
      "loss": 2.0288,
      "step": 15674
    },
    {
      "epoch": 0.10650138286185522,
      "grad_norm": 0.140625,
      "learning_rate": 0.000975874144742811,
      "loss": 2.1106,
      "step": 15675
    },
    {
      "epoch": 0.10650817720844928,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009758708430408019,
      "loss": 2.0507,
      "step": 15676
    },
    {
      "epoch": 0.10651497155504333,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009758675411184701,
      "loss": 2.2795,
      "step": 15677
    },
    {
      "epoch": 0.1065217659016374,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000975864238975817,
      "loss": 2.0846,
      "step": 15678
    },
    {
      "epoch": 0.10652856024823146,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009758609366128442,
      "loss": 2.1093,
      "step": 15679
    },
    {
      "epoch": 0.10653535459482551,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009758576340295533,
      "loss": 2.0446,
      "step": 15680
    },
    {
      "epoch": 0.10654214894141957,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009758543312259458,
      "loss": 2.1873,
      "step": 15681
    },
    {
      "epoch": 0.10654894328801362,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009758510282020232,
      "loss": 2.1341,
      "step": 15682
    },
    {
      "epoch": 0.10655573763460768,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009758477249577868,
      "loss": 2.124,
      "step": 15683
    },
    {
      "epoch": 0.10656253198120175,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009758444214932387,
      "loss": 2.2194,
      "step": 15684
    },
    {
      "epoch": 0.1065693263277958,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009758411178083799,
      "loss": 2.2204,
      "step": 15685
    },
    {
      "epoch": 0.10657612067438986,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009758378139032121,
      "loss": 2.081,
      "step": 15686
    },
    {
      "epoch": 0.10658291502098391,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000975834509777737,
      "loss": 2.1497,
      "step": 15687
    },
    {
      "epoch": 0.10658970936757797,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009758312054319558,
      "loss": 2.2306,
      "step": 15688
    },
    {
      "epoch": 0.10659650371417204,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009758279008658704,
      "loss": 2.1214,
      "step": 15689
    },
    {
      "epoch": 0.10660329806076609,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009758245960794819,
      "loss": 2.2077,
      "step": 15690
    },
    {
      "epoch": 0.10661009240736015,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009758212910727922,
      "loss": 2.236,
      "step": 15691
    },
    {
      "epoch": 0.1066168867539542,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009758179858458027,
      "loss": 2.4256,
      "step": 15692
    },
    {
      "epoch": 0.10662368110054826,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000975814680398515,
      "loss": 2.3014,
      "step": 15693
    },
    {
      "epoch": 0.10663047544714233,
      "grad_norm": 0.125,
      "learning_rate": 0.0009758113747309304,
      "loss": 2.0602,
      "step": 15694
    },
    {
      "epoch": 0.10663726979373638,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009758080688430506,
      "loss": 2.1574,
      "step": 15695
    },
    {
      "epoch": 0.10664406414033044,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009758047627348771,
      "loss": 2.2426,
      "step": 15696
    },
    {
      "epoch": 0.10665085848692449,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009758014564064117,
      "loss": 2.1302,
      "step": 15697
    },
    {
      "epoch": 0.10665765283351855,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009757981498576554,
      "loss": 2.1197,
      "step": 15698
    },
    {
      "epoch": 0.1066644471801126,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009757948430886102,
      "loss": 2.1823,
      "step": 15699
    },
    {
      "epoch": 0.10667124152670666,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009757915360992774,
      "loss": 2.1765,
      "step": 15700
    },
    {
      "epoch": 0.10667803587330073,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009757882288896584,
      "loss": 2.176,
      "step": 15701
    },
    {
      "epoch": 0.10668483021989478,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009757849214597551,
      "loss": 2.1061,
      "step": 15702
    },
    {
      "epoch": 0.10669162456648884,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009757816138095687,
      "loss": 2.1832,
      "step": 15703
    },
    {
      "epoch": 0.10669841891308289,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009757783059391012,
      "loss": 2.2082,
      "step": 15704
    },
    {
      "epoch": 0.10670521325967695,
      "grad_norm": 0.173828125,
      "learning_rate": 0.0009757749978483534,
      "loss": 2.1593,
      "step": 15705
    },
    {
      "epoch": 0.10671200760627102,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009757716895373275,
      "loss": 2.3202,
      "step": 15706
    },
    {
      "epoch": 0.10671880195286507,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009757683810060245,
      "loss": 2.2155,
      "step": 15707
    },
    {
      "epoch": 0.10672559629945913,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009757650722544464,
      "loss": 2.3147,
      "step": 15708
    },
    {
      "epoch": 0.10673239064605318,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009757617632825944,
      "loss": 2.2659,
      "step": 15709
    },
    {
      "epoch": 0.10673918499264724,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009757584540904703,
      "loss": 2.2829,
      "step": 15710
    },
    {
      "epoch": 0.1067459793392413,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009757551446780753,
      "loss": 2.2313,
      "step": 15711
    },
    {
      "epoch": 0.10675277368583536,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009757518350454114,
      "loss": 2.2505,
      "step": 15712
    },
    {
      "epoch": 0.10675956803242942,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009757485251924796,
      "loss": 2.2626,
      "step": 15713
    },
    {
      "epoch": 0.10676636237902347,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009757452151192818,
      "loss": 2.1952,
      "step": 15714
    },
    {
      "epoch": 0.10677315672561753,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009757419048258193,
      "loss": 2.0217,
      "step": 15715
    },
    {
      "epoch": 0.1067799510722116,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009757385943120939,
      "loss": 2.2923,
      "step": 15716
    },
    {
      "epoch": 0.10678674541880565,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009757352835781068,
      "loss": 2.06,
      "step": 15717
    },
    {
      "epoch": 0.10679353976539971,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009757319726238599,
      "loss": 2.0097,
      "step": 15718
    },
    {
      "epoch": 0.10680033411199376,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009757286614493545,
      "loss": 2.2705,
      "step": 15719
    },
    {
      "epoch": 0.10680712845858782,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009757253500545922,
      "loss": 2.1728,
      "step": 15720
    },
    {
      "epoch": 0.10681392280518189,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009757220384395745,
      "loss": 2.2465,
      "step": 15721
    },
    {
      "epoch": 0.10682071715177593,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.000975718726604303,
      "loss": 2.1543,
      "step": 15722
    },
    {
      "epoch": 0.10682751149837,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009757154145487791,
      "loss": 2.1932,
      "step": 15723
    },
    {
      "epoch": 0.10683430584496405,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009757121022730044,
      "loss": 2.1608,
      "step": 15724
    },
    {
      "epoch": 0.10684110019155811,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009757087897769805,
      "loss": 2.2986,
      "step": 15725
    },
    {
      "epoch": 0.10684789453815216,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009757054770607089,
      "loss": 2.2082,
      "step": 15726
    },
    {
      "epoch": 0.10685468888474622,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009757021641241911,
      "loss": 2.0744,
      "step": 15727
    },
    {
      "epoch": 0.10686148323134029,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009756988509674287,
      "loss": 2.1965,
      "step": 15728
    },
    {
      "epoch": 0.10686827757793434,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009756955375904231,
      "loss": 2.0443,
      "step": 15729
    },
    {
      "epoch": 0.1068750719245284,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009756922239931759,
      "loss": 2.4006,
      "step": 15730
    },
    {
      "epoch": 0.10688186627112245,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009756889101756888,
      "loss": 2.12,
      "step": 15731
    },
    {
      "epoch": 0.10688866061771651,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000975685596137963,
      "loss": 2.268,
      "step": 15732
    },
    {
      "epoch": 0.10689545496431058,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009756822818800004,
      "loss": 2.1808,
      "step": 15733
    },
    {
      "epoch": 0.10690224931090463,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009756789674018023,
      "loss": 2.3329,
      "step": 15734
    },
    {
      "epoch": 0.10690904365749869,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009756756527033703,
      "loss": 2.1087,
      "step": 15735
    },
    {
      "epoch": 0.10691583800409274,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009756723377847058,
      "loss": 2.1939,
      "step": 15736
    },
    {
      "epoch": 0.1069226323506868,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009756690226458105,
      "loss": 2.2637,
      "step": 15737
    },
    {
      "epoch": 0.10692942669728087,
      "grad_norm": 0.140625,
      "learning_rate": 0.000975665707286686,
      "loss": 2.2095,
      "step": 15738
    },
    {
      "epoch": 0.10693622104387492,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009756623917073337,
      "loss": 2.1946,
      "step": 15739
    },
    {
      "epoch": 0.10694301539046898,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009756590759077552,
      "loss": 2.1662,
      "step": 15740
    },
    {
      "epoch": 0.10694980973706303,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009756557598879519,
      "loss": 2.1057,
      "step": 15741
    },
    {
      "epoch": 0.10695660408365709,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009756524436479255,
      "loss": 2.1563,
      "step": 15742
    },
    {
      "epoch": 0.10696339843025116,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009756491271876775,
      "loss": 2.2981,
      "step": 15743
    },
    {
      "epoch": 0.1069701927768452,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009756458105072092,
      "loss": 2.2031,
      "step": 15744
    },
    {
      "epoch": 0.10697698712343927,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009756424936065227,
      "loss": 2.2021,
      "step": 15745
    },
    {
      "epoch": 0.10698378147003332,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009756391764856189,
      "loss": 2.2896,
      "step": 15746
    },
    {
      "epoch": 0.10699057581662738,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009756358591444997,
      "loss": 2.1128,
      "step": 15747
    },
    {
      "epoch": 0.10699737016322143,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009756325415831666,
      "loss": 2.2174,
      "step": 15748
    },
    {
      "epoch": 0.1070041645098155,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000975629223801621,
      "loss": 2.2035,
      "step": 15749
    },
    {
      "epoch": 0.10701095885640956,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009756259057998645,
      "loss": 2.1819,
      "step": 15750
    },
    {
      "epoch": 0.10701775320300361,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009756225875778988,
      "loss": 2.1815,
      "step": 15751
    },
    {
      "epoch": 0.10702454754959767,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009756192691357252,
      "loss": 2.1968,
      "step": 15752
    },
    {
      "epoch": 0.10703134189619172,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009756159504733453,
      "loss": 2.1826,
      "step": 15753
    },
    {
      "epoch": 0.10703813624278578,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009756126315907608,
      "loss": 2.1781,
      "step": 15754
    },
    {
      "epoch": 0.10704493058937985,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000975609312487973,
      "loss": 2.215,
      "step": 15755
    },
    {
      "epoch": 0.1070517249359739,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009756059931649835,
      "loss": 2.1529,
      "step": 15756
    },
    {
      "epoch": 0.10705851928256796,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000975602673621794,
      "loss": 2.276,
      "step": 15757
    },
    {
      "epoch": 0.10706531362916201,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009755993538584059,
      "loss": 2.3282,
      "step": 15758
    },
    {
      "epoch": 0.10707210797575607,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009755960338748206,
      "loss": 2.0453,
      "step": 15759
    },
    {
      "epoch": 0.10707890232235014,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009755927136710399,
      "loss": 2.2054,
      "step": 15760
    },
    {
      "epoch": 0.10708569666894419,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009755893932470652,
      "loss": 2.2237,
      "step": 15761
    },
    {
      "epoch": 0.10709249101553825,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000975586072602898,
      "loss": 2.2428,
      "step": 15762
    },
    {
      "epoch": 0.1070992853621323,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00097558275173854,
      "loss": 2.0975,
      "step": 15763
    },
    {
      "epoch": 0.10710607970872636,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009755794306539927,
      "loss": 2.171,
      "step": 15764
    },
    {
      "epoch": 0.10711287405532043,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009755761093492574,
      "loss": 2.1806,
      "step": 15765
    },
    {
      "epoch": 0.10711966840191448,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009755727878243359,
      "loss": 2.1011,
      "step": 15766
    },
    {
      "epoch": 0.10712646274850854,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009755694660792297,
      "loss": 2.1316,
      "step": 15767
    },
    {
      "epoch": 0.10713325709510259,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009755661441139403,
      "loss": 2.2513,
      "step": 15768
    },
    {
      "epoch": 0.10714005144169665,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009755628219284691,
      "loss": 2.2527,
      "step": 15769
    },
    {
      "epoch": 0.10714684578829071,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009755594995228178,
      "loss": 2.1836,
      "step": 15770
    },
    {
      "epoch": 0.10715364013488476,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009755561768969881,
      "loss": 2.2142,
      "step": 15771
    },
    {
      "epoch": 0.10716043448147883,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009755528540509811,
      "loss": 2.2203,
      "step": 15772
    },
    {
      "epoch": 0.10716722882807288,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009755495309847986,
      "loss": 2.1699,
      "step": 15773
    },
    {
      "epoch": 0.10717402317466694,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009755462076984423,
      "loss": 2.2135,
      "step": 15774
    },
    {
      "epoch": 0.10718081752126099,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009755428841919134,
      "loss": 2.3509,
      "step": 15775
    },
    {
      "epoch": 0.10718761186785505,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009755395604652136,
      "loss": 2.1413,
      "step": 15776
    },
    {
      "epoch": 0.10719440621444912,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009755362365183445,
      "loss": 2.208,
      "step": 15777
    },
    {
      "epoch": 0.10720120056104317,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009755329123513074,
      "loss": 2.1704,
      "step": 15778
    },
    {
      "epoch": 0.10720799490763723,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009755295879641043,
      "loss": 2.3247,
      "step": 15779
    },
    {
      "epoch": 0.10721478925423128,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009755262633567364,
      "loss": 2.4128,
      "step": 15780
    },
    {
      "epoch": 0.10722158360082534,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009755229385292051,
      "loss": 2.1222,
      "step": 15781
    },
    {
      "epoch": 0.1072283779474194,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009755196134815123,
      "loss": 2.2422,
      "step": 15782
    },
    {
      "epoch": 0.10723517229401346,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009755162882136593,
      "loss": 2.1321,
      "step": 15783
    },
    {
      "epoch": 0.10724196664060752,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009755129627256477,
      "loss": 2.0697,
      "step": 15784
    },
    {
      "epoch": 0.10724876098720157,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009755096370174791,
      "loss": 2.1548,
      "step": 15785
    },
    {
      "epoch": 0.10725555533379563,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009755063110891549,
      "loss": 2.1232,
      "step": 15786
    },
    {
      "epoch": 0.1072623496803897,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009755029849406768,
      "loss": 2.211,
      "step": 15787
    },
    {
      "epoch": 0.10726914402698375,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009754996585720463,
      "loss": 2.2649,
      "step": 15788
    },
    {
      "epoch": 0.10727593837357781,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009754963319832648,
      "loss": 2.1413,
      "step": 15789
    },
    {
      "epoch": 0.10728273272017186,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009754930051743341,
      "loss": 2.272,
      "step": 15790
    },
    {
      "epoch": 0.10728952706676592,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009754896781452555,
      "loss": 2.2903,
      "step": 15791
    },
    {
      "epoch": 0.10729632141335999,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009754863508960306,
      "loss": 2.2925,
      "step": 15792
    },
    {
      "epoch": 0.10730311575995403,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000975483023426661,
      "loss": 2.09,
      "step": 15793
    },
    {
      "epoch": 0.1073099101065481,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009754796957371482,
      "loss": 2.2224,
      "step": 15794
    },
    {
      "epoch": 0.10731670445314215,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009754763678274938,
      "loss": 2.214,
      "step": 15795
    },
    {
      "epoch": 0.10732349879973621,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009754730396976992,
      "loss": 2.3422,
      "step": 15796
    },
    {
      "epoch": 0.10733029314633026,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009754697113477661,
      "loss": 2.3548,
      "step": 15797
    },
    {
      "epoch": 0.10733708749292432,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000975466382777696,
      "loss": 2.2018,
      "step": 15798
    },
    {
      "epoch": 0.10734388183951839,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009754630539874905,
      "loss": 2.3683,
      "step": 15799
    },
    {
      "epoch": 0.10735067618611244,
      "grad_norm": 0.125,
      "learning_rate": 0.0009754597249771508,
      "loss": 2.1669,
      "step": 15800
    },
    {
      "epoch": 0.1073574705327065,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009754563957466789,
      "loss": 2.2062,
      "step": 15801
    },
    {
      "epoch": 0.10736426487930055,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009754530662960761,
      "loss": 2.2554,
      "step": 15802
    },
    {
      "epoch": 0.10737105922589461,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009754497366253438,
      "loss": 2.1166,
      "step": 15803
    },
    {
      "epoch": 0.10737785357248868,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000975446406734484,
      "loss": 2.3428,
      "step": 15804
    },
    {
      "epoch": 0.10738464791908273,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009754430766234979,
      "loss": 2.3277,
      "step": 15805
    },
    {
      "epoch": 0.10739144226567679,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000975439746292387,
      "loss": 2.187,
      "step": 15806
    },
    {
      "epoch": 0.10739823661227084,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009754364157411531,
      "loss": 2.1607,
      "step": 15807
    },
    {
      "epoch": 0.1074050309588649,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009754330849697974,
      "loss": 2.177,
      "step": 15808
    },
    {
      "epoch": 0.10741182530545897,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009754297539783217,
      "loss": 2.1608,
      "step": 15809
    },
    {
      "epoch": 0.10741861965205302,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009754264227667276,
      "loss": 2.3782,
      "step": 15810
    },
    {
      "epoch": 0.10742541399864708,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009754230913350164,
      "loss": 2.2685,
      "step": 15811
    },
    {
      "epoch": 0.10743220834524113,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009754197596831898,
      "loss": 2.3286,
      "step": 15812
    },
    {
      "epoch": 0.10743900269183519,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009754164278112492,
      "loss": 2.2015,
      "step": 15813
    },
    {
      "epoch": 0.10744579703842926,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009754130957191963,
      "loss": 2.2431,
      "step": 15814
    },
    {
      "epoch": 0.1074525913850233,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009754097634070327,
      "loss": 2.2639,
      "step": 15815
    },
    {
      "epoch": 0.10745938573161737,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009754064308747597,
      "loss": 2.2045,
      "step": 15816
    },
    {
      "epoch": 0.10746618007821142,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000975403098122379,
      "loss": 2.1312,
      "step": 15817
    },
    {
      "epoch": 0.10747297442480548,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009753997651498922,
      "loss": 2.4155,
      "step": 15818
    },
    {
      "epoch": 0.10747976877139953,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009753964319573007,
      "loss": 2.2652,
      "step": 15819
    },
    {
      "epoch": 0.1074865631179936,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000975393098544606,
      "loss": 2.1635,
      "step": 15820
    },
    {
      "epoch": 0.10749335746458766,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009753897649118099,
      "loss": 2.2696,
      "step": 15821
    },
    {
      "epoch": 0.10750015181118171,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009753864310589137,
      "loss": 2.3611,
      "step": 15822
    },
    {
      "epoch": 0.10750694615777577,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000975383096985919,
      "loss": 2.1125,
      "step": 15823
    },
    {
      "epoch": 0.10751374050436982,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009753797626928275,
      "loss": 2.0656,
      "step": 15824
    },
    {
      "epoch": 0.10752053485096388,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009753764281796404,
      "loss": 2.1519,
      "step": 15825
    },
    {
      "epoch": 0.10752732919755795,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009753730934463598,
      "loss": 2.3443,
      "step": 15826
    },
    {
      "epoch": 0.107534123544152,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009753697584929868,
      "loss": 2.2458,
      "step": 15827
    },
    {
      "epoch": 0.10754091789074606,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009753664233195229,
      "loss": 2.203,
      "step": 15828
    },
    {
      "epoch": 0.10754771223734011,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009753630879259698,
      "loss": 2.2188,
      "step": 15829
    },
    {
      "epoch": 0.10755450658393417,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009753597523123292,
      "loss": 2.2457,
      "step": 15830
    },
    {
      "epoch": 0.10756130093052824,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009753564164786024,
      "loss": 2.2082,
      "step": 15831
    },
    {
      "epoch": 0.10756809527712229,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009753530804247909,
      "loss": 2.1412,
      "step": 15832
    },
    {
      "epoch": 0.10757488962371635,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009753497441508965,
      "loss": 2.1674,
      "step": 15833
    },
    {
      "epoch": 0.1075816839703104,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009753464076569207,
      "loss": 2.1437,
      "step": 15834
    },
    {
      "epoch": 0.10758847831690446,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009753430709428648,
      "loss": 2.1192,
      "step": 15835
    },
    {
      "epoch": 0.10759527266349853,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009753397340087306,
      "loss": 2.1892,
      "step": 15836
    },
    {
      "epoch": 0.10760206701009258,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009753363968545196,
      "loss": 2.2375,
      "step": 15837
    },
    {
      "epoch": 0.10760886135668664,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009753330594802332,
      "loss": 2.1595,
      "step": 15838
    },
    {
      "epoch": 0.10761565570328069,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009753297218858731,
      "loss": 2.2362,
      "step": 15839
    },
    {
      "epoch": 0.10762245004987475,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009753263840714408,
      "loss": 2.0908,
      "step": 15840
    },
    {
      "epoch": 0.10762924439646882,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009753230460369378,
      "loss": 2.3692,
      "step": 15841
    },
    {
      "epoch": 0.10763603874306286,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009753197077823657,
      "loss": 2.1818,
      "step": 15842
    },
    {
      "epoch": 0.10764283308965693,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009753163693077259,
      "loss": 2.1809,
      "step": 15843
    },
    {
      "epoch": 0.10764962743625098,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009753130306130203,
      "loss": 2.2459,
      "step": 15844
    },
    {
      "epoch": 0.10765642178284504,
      "grad_norm": 0.14453125,
      "learning_rate": 0.00097530969169825,
      "loss": 2.3002,
      "step": 15845
    },
    {
      "epoch": 0.10766321612943909,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009753063525634169,
      "loss": 2.0466,
      "step": 15846
    },
    {
      "epoch": 0.10767001047603315,
      "grad_norm": 0.125,
      "learning_rate": 0.0009753030132085223,
      "loss": 2.1744,
      "step": 15847
    },
    {
      "epoch": 0.10767680482262722,
      "grad_norm": 0.140625,
      "learning_rate": 0.000975299673633568,
      "loss": 2.2118,
      "step": 15848
    },
    {
      "epoch": 0.10768359916922127,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009752963338385552,
      "loss": 2.1049,
      "step": 15849
    },
    {
      "epoch": 0.10769039351581533,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009752929938234857,
      "loss": 2.0628,
      "step": 15850
    },
    {
      "epoch": 0.10769718786240938,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009752896535883612,
      "loss": 2.2033,
      "step": 15851
    },
    {
      "epoch": 0.10770398220900344,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009752863131331828,
      "loss": 2.0915,
      "step": 15852
    },
    {
      "epoch": 0.1077107765555975,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009752829724579525,
      "loss": 2.1552,
      "step": 15853
    },
    {
      "epoch": 0.10771757090219156,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009752796315626715,
      "loss": 2.3372,
      "step": 15854
    },
    {
      "epoch": 0.10772436524878562,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009752762904473415,
      "loss": 2.2288,
      "step": 15855
    },
    {
      "epoch": 0.10773115959537967,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000975272949111964,
      "loss": 2.2579,
      "step": 15856
    },
    {
      "epoch": 0.10773795394197373,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009752696075565406,
      "loss": 2.2717,
      "step": 15857
    },
    {
      "epoch": 0.1077447482885678,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009752662657810728,
      "loss": 2.318,
      "step": 15858
    },
    {
      "epoch": 0.10775154263516185,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009752629237855621,
      "loss": 2.2054,
      "step": 15859
    },
    {
      "epoch": 0.10775833698175591,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009752595815700103,
      "loss": 2.2366,
      "step": 15860
    },
    {
      "epoch": 0.10776513132834996,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009752562391344186,
      "loss": 2.221,
      "step": 15861
    },
    {
      "epoch": 0.10777192567494402,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009752528964787888,
      "loss": 2.3422,
      "step": 15862
    },
    {
      "epoch": 0.10777872002153809,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009752495536031223,
      "loss": 2.2659,
      "step": 15863
    },
    {
      "epoch": 0.10778551436813213,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009752462105074207,
      "loss": 2.0847,
      "step": 15864
    },
    {
      "epoch": 0.1077923087147262,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009752428671916856,
      "loss": 2.2187,
      "step": 15865
    },
    {
      "epoch": 0.10779910306132025,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009752395236559186,
      "loss": 2.1689,
      "step": 15866
    },
    {
      "epoch": 0.10780589740791431,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000975236179900121,
      "loss": 2.1325,
      "step": 15867
    },
    {
      "epoch": 0.10781269175450836,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009752328359242945,
      "loss": 2.2272,
      "step": 15868
    },
    {
      "epoch": 0.10781948610110242,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009752294917284407,
      "loss": 2.1,
      "step": 15869
    },
    {
      "epoch": 0.10782628044769649,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000975226147312561,
      "loss": 2.1994,
      "step": 15870
    },
    {
      "epoch": 0.10783307479429054,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009752228026766572,
      "loss": 2.2058,
      "step": 15871
    },
    {
      "epoch": 0.1078398691408846,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009752194578207305,
      "loss": 2.2913,
      "step": 15872
    },
    {
      "epoch": 0.10784666348747865,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009752161127447827,
      "loss": 2.1602,
      "step": 15873
    },
    {
      "epoch": 0.10785345783407271,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009752127674488154,
      "loss": 2.2377,
      "step": 15874
    },
    {
      "epoch": 0.10786025218066678,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00097520942193283,
      "loss": 2.232,
      "step": 15875
    },
    {
      "epoch": 0.10786704652726083,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000975206076196828,
      "loss": 2.2308,
      "step": 15876
    },
    {
      "epoch": 0.10787384087385489,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009752027302408111,
      "loss": 2.1512,
      "step": 15877
    },
    {
      "epoch": 0.10788063522044894,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009751993840647808,
      "loss": 2.1989,
      "step": 15878
    },
    {
      "epoch": 0.107887429567043,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009751960376687386,
      "loss": 2.1215,
      "step": 15879
    },
    {
      "epoch": 0.10789422391363707,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000975192691052686,
      "loss": 2.2978,
      "step": 15880
    },
    {
      "epoch": 0.10790101826023112,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009751893442166248,
      "loss": 2.0738,
      "step": 15881
    },
    {
      "epoch": 0.10790781260682518,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009751859971605563,
      "loss": 1.9898,
      "step": 15882
    },
    {
      "epoch": 0.10791460695341923,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009751826498844822,
      "loss": 2.1433,
      "step": 15883
    },
    {
      "epoch": 0.10792140130001329,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009751793023884039,
      "loss": 2.2044,
      "step": 15884
    },
    {
      "epoch": 0.10792819564660736,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000975175954672323,
      "loss": 2.1953,
      "step": 15885
    },
    {
      "epoch": 0.1079349899932014,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009751726067362412,
      "loss": 2.2573,
      "step": 15886
    },
    {
      "epoch": 0.10794178433979547,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009751692585801597,
      "loss": 2.1818,
      "step": 15887
    },
    {
      "epoch": 0.10794857868638952,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009751659102040806,
      "loss": 2.2037,
      "step": 15888
    },
    {
      "epoch": 0.10795537303298358,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009751625616080049,
      "loss": 2.4393,
      "step": 15889
    },
    {
      "epoch": 0.10796216737957764,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009751592127919344,
      "loss": 2.3238,
      "step": 15890
    },
    {
      "epoch": 0.1079689617261717,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009751558637558707,
      "loss": 2.1834,
      "step": 15891
    },
    {
      "epoch": 0.10797575607276576,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009751525144998152,
      "loss": 2.1362,
      "step": 15892
    },
    {
      "epoch": 0.10798255041935981,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009751491650237696,
      "loss": 2.2163,
      "step": 15893
    },
    {
      "epoch": 0.10798934476595387,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009751458153277355,
      "loss": 2.3409,
      "step": 15894
    },
    {
      "epoch": 0.10799613911254792,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009751424654117142,
      "loss": 2.2475,
      "step": 15895
    },
    {
      "epoch": 0.10800293345914198,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009751391152757074,
      "loss": 2.257,
      "step": 15896
    },
    {
      "epoch": 0.10800972780573605,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009751357649197166,
      "loss": 2.0401,
      "step": 15897
    },
    {
      "epoch": 0.1080165221523301,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009751324143437434,
      "loss": 2.2049,
      "step": 15898
    },
    {
      "epoch": 0.10802331649892416,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009751290635477895,
      "loss": 2.2052,
      "step": 15899
    },
    {
      "epoch": 0.10803011084551821,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000975125712531856,
      "loss": 2.0972,
      "step": 15900
    },
    {
      "epoch": 0.10803690519211227,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009751223612959449,
      "loss": 2.1783,
      "step": 15901
    },
    {
      "epoch": 0.10804369953870634,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009751190098400576,
      "loss": 2.2426,
      "step": 15902
    },
    {
      "epoch": 0.10805049388530039,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009751156581641956,
      "loss": 2.134,
      "step": 15903
    },
    {
      "epoch": 0.10805728823189445,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009751123062683606,
      "loss": 2.1427,
      "step": 15904
    },
    {
      "epoch": 0.1080640825784885,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000975108954152554,
      "loss": 2.2768,
      "step": 15905
    },
    {
      "epoch": 0.10807087692508256,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009751056018167773,
      "loss": 2.2311,
      "step": 15906
    },
    {
      "epoch": 0.10807767127167663,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009751022492610322,
      "loss": 2.2215,
      "step": 15907
    },
    {
      "epoch": 0.10808446561827068,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009750988964853202,
      "loss": 2.1346,
      "step": 15908
    },
    {
      "epoch": 0.10809125996486474,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000975095543489643,
      "loss": 2.255,
      "step": 15909
    },
    {
      "epoch": 0.10809805431145879,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000975092190274002,
      "loss": 2.2363,
      "step": 15910
    },
    {
      "epoch": 0.10810484865805285,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009750888368383985,
      "loss": 2.3065,
      "step": 15911
    },
    {
      "epoch": 0.10811164300464692,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009750854831828347,
      "loss": 2.328,
      "step": 15912
    },
    {
      "epoch": 0.10811843735124096,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009750821293073114,
      "loss": 2.0912,
      "step": 15913
    },
    {
      "epoch": 0.10812523169783503,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009750787752118306,
      "loss": 2.1556,
      "step": 15914
    },
    {
      "epoch": 0.10813202604442908,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009750754208963939,
      "loss": 2.1902,
      "step": 15915
    },
    {
      "epoch": 0.10813882039102314,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009750720663610026,
      "loss": 2.1474,
      "step": 15916
    },
    {
      "epoch": 0.10814561473761719,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009750687116056586,
      "loss": 2.1792,
      "step": 15917
    },
    {
      "epoch": 0.10815240908421125,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009750653566303632,
      "loss": 2.1981,
      "step": 15918
    },
    {
      "epoch": 0.10815920343080532,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009750620014351177,
      "loss": 2.1431,
      "step": 15919
    },
    {
      "epoch": 0.10816599777739937,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009750586460199242,
      "loss": 2.1277,
      "step": 15920
    },
    {
      "epoch": 0.10817279212399343,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009750552903847838,
      "loss": 2.182,
      "step": 15921
    },
    {
      "epoch": 0.10817958647058748,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009750519345296984,
      "loss": 2.0369,
      "step": 15922
    },
    {
      "epoch": 0.10818638081718154,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009750485784546694,
      "loss": 2.2882,
      "step": 15923
    },
    {
      "epoch": 0.1081931751637756,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009750452221596983,
      "loss": 2.2093,
      "step": 15924
    },
    {
      "epoch": 0.10819996951036966,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009750418656447867,
      "loss": 2.1965,
      "step": 15925
    },
    {
      "epoch": 0.10820676385696372,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009750385089099361,
      "loss": 2.192,
      "step": 15926
    },
    {
      "epoch": 0.10821355820355777,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009750351519551483,
      "loss": 2.3448,
      "step": 15927
    },
    {
      "epoch": 0.10822035255015183,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009750317947804245,
      "loss": 2.1983,
      "step": 15928
    },
    {
      "epoch": 0.1082271468967459,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009750284373857666,
      "loss": 2.2292,
      "step": 15929
    },
    {
      "epoch": 0.10823394124333995,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009750250797711757,
      "loss": 2.2363,
      "step": 15930
    },
    {
      "epoch": 0.10824073558993401,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000975021721936654,
      "loss": 1.9802,
      "step": 15931
    },
    {
      "epoch": 0.10824752993652806,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009750183638822023,
      "loss": 2.2011,
      "step": 15932
    },
    {
      "epoch": 0.10825432428312212,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009750150056078229,
      "loss": 2.2274,
      "step": 15933
    },
    {
      "epoch": 0.10826111862971619,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009750116471135168,
      "loss": 2.1673,
      "step": 15934
    },
    {
      "epoch": 0.10826791297631023,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009750082883992858,
      "loss": 2.1473,
      "step": 15935
    },
    {
      "epoch": 0.1082747073229043,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009750049294651313,
      "loss": 2.1885,
      "step": 15936
    },
    {
      "epoch": 0.10828150166949835,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009750015703110551,
      "loss": 2.1309,
      "step": 15937
    },
    {
      "epoch": 0.10828829601609241,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009749982109370585,
      "loss": 2.2258,
      "step": 15938
    },
    {
      "epoch": 0.10829509036268646,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009749948513431434,
      "loss": 2.1419,
      "step": 15939
    },
    {
      "epoch": 0.10830188470928052,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000974991491529311,
      "loss": 2.2214,
      "step": 15940
    },
    {
      "epoch": 0.10830867905587459,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009749881314955629,
      "loss": 2.1488,
      "step": 15941
    },
    {
      "epoch": 0.10831547340246864,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009749847712419009,
      "loss": 2.1314,
      "step": 15942
    },
    {
      "epoch": 0.1083222677490627,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009749814107683261,
      "loss": 2.3151,
      "step": 15943
    },
    {
      "epoch": 0.10832906209565675,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009749780500748407,
      "loss": 2.0806,
      "step": 15944
    },
    {
      "epoch": 0.10833585644225081,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009749746891614457,
      "loss": 2.3096,
      "step": 15945
    },
    {
      "epoch": 0.10834265078884488,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009749713280281429,
      "loss": 2.1395,
      "step": 15946
    },
    {
      "epoch": 0.10834944513543893,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000974967966674934,
      "loss": 2.3041,
      "step": 15947
    },
    {
      "epoch": 0.10835623948203299,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009749646051018201,
      "loss": 2.1848,
      "step": 15948
    },
    {
      "epoch": 0.10836303382862704,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009749612433088032,
      "loss": 2.0945,
      "step": 15949
    },
    {
      "epoch": 0.1083698281752211,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009749578812958846,
      "loss": 2.1978,
      "step": 15950
    },
    {
      "epoch": 0.10837662252181517,
      "grad_norm": 0.140625,
      "learning_rate": 0.000974954519063066,
      "loss": 2.306,
      "step": 15951
    },
    {
      "epoch": 0.10838341686840922,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000974951156610349,
      "loss": 2.2242,
      "step": 15952
    },
    {
      "epoch": 0.10839021121500328,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009749477939377349,
      "loss": 2.2768,
      "step": 15953
    },
    {
      "epoch": 0.10839700556159733,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009749444310452255,
      "loss": 2.1915,
      "step": 15954
    },
    {
      "epoch": 0.10840379990819139,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009749410679328222,
      "loss": 2.1949,
      "step": 15955
    },
    {
      "epoch": 0.10841059425478546,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009749377046005267,
      "loss": 2.0142,
      "step": 15956
    },
    {
      "epoch": 0.1084173886013795,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009749343410483404,
      "loss": 2.1856,
      "step": 15957
    },
    {
      "epoch": 0.10842418294797357,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009749309772762652,
      "loss": 2.1497,
      "step": 15958
    },
    {
      "epoch": 0.10843097729456762,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009749276132843021,
      "loss": 2.2578,
      "step": 15959
    },
    {
      "epoch": 0.10843777164116168,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000974924249072453,
      "loss": 2.2165,
      "step": 15960
    },
    {
      "epoch": 0.10844456598775574,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009749208846407196,
      "loss": 2.2224,
      "step": 15961
    },
    {
      "epoch": 0.1084513603343498,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009749175199891032,
      "loss": 2.1802,
      "step": 15962
    },
    {
      "epoch": 0.10845815468094386,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009749141551176053,
      "loss": 2.1039,
      "step": 15963
    },
    {
      "epoch": 0.10846494902753791,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009749107900262278,
      "loss": 2.1592,
      "step": 15964
    },
    {
      "epoch": 0.10847174337413197,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009749074247149719,
      "loss": 2.3228,
      "step": 15965
    },
    {
      "epoch": 0.10847853772072602,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009749040591838393,
      "loss": 2.1603,
      "step": 15966
    },
    {
      "epoch": 0.10848533206732008,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009749006934328315,
      "loss": 2.1943,
      "step": 15967
    },
    {
      "epoch": 0.10849212641391415,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009748973274619504,
      "loss": 2.1056,
      "step": 15968
    },
    {
      "epoch": 0.1084989207605082,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009748939612711971,
      "loss": 2.1465,
      "step": 15969
    },
    {
      "epoch": 0.10850571510710226,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009748905948605733,
      "loss": 2.154,
      "step": 15970
    },
    {
      "epoch": 0.10851250945369631,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009748872282300807,
      "loss": 2.4043,
      "step": 15971
    },
    {
      "epoch": 0.10851930380029037,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009748838613797207,
      "loss": 2.1395,
      "step": 15972
    },
    {
      "epoch": 0.10852609814688444,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009748804943094948,
      "loss": 2.2735,
      "step": 15973
    },
    {
      "epoch": 0.10853289249347849,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.000974877127019405,
      "loss": 2.4223,
      "step": 15974
    },
    {
      "epoch": 0.10853968684007255,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009748737595094522,
      "loss": 2.2518,
      "step": 15975
    },
    {
      "epoch": 0.1085464811866666,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009748703917796384,
      "loss": 2.1392,
      "step": 15976
    },
    {
      "epoch": 0.10855327553326066,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000974867023829965,
      "loss": 2.0938,
      "step": 15977
    },
    {
      "epoch": 0.10856006987985473,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009748636556604338,
      "loss": 2.1217,
      "step": 15978
    },
    {
      "epoch": 0.10856686422644878,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009748602872710462,
      "loss": 2.2049,
      "step": 15979
    },
    {
      "epoch": 0.10857365857304284,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009748569186618035,
      "loss": 2.2323,
      "step": 15980
    },
    {
      "epoch": 0.10858045291963689,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009748535498327076,
      "loss": 2.3101,
      "step": 15981
    },
    {
      "epoch": 0.10858724726623095,
      "grad_norm": 0.126953125,
      "learning_rate": 0.00097485018078376,
      "loss": 2.1835,
      "step": 15982
    },
    {
      "epoch": 0.10859404161282502,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000974846811514962,
      "loss": 2.3687,
      "step": 15983
    },
    {
      "epoch": 0.10860083595941906,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009748434420263156,
      "loss": 2.225,
      "step": 15984
    },
    {
      "epoch": 0.10860763030601313,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009748400723178221,
      "loss": 2.2912,
      "step": 15985
    },
    {
      "epoch": 0.10861442465260718,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000974836702389483,
      "loss": 2.3097,
      "step": 15986
    },
    {
      "epoch": 0.10862121899920124,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009748333322413,
      "loss": 2.2779,
      "step": 15987
    },
    {
      "epoch": 0.10862801334579529,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009748299618732746,
      "loss": 2.2296,
      "step": 15988
    },
    {
      "epoch": 0.10863480769238935,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009748265912854084,
      "loss": 2.2732,
      "step": 15989
    },
    {
      "epoch": 0.10864160203898342,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009748232204777029,
      "loss": 2.2118,
      "step": 15990
    },
    {
      "epoch": 0.10864839638557747,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009748198494501597,
      "loss": 2.1917,
      "step": 15991
    },
    {
      "epoch": 0.10865519073217153,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009748164782027803,
      "loss": 2.2522,
      "step": 15992
    },
    {
      "epoch": 0.10866198507876558,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009748131067355663,
      "loss": 2.1518,
      "step": 15993
    },
    {
      "epoch": 0.10866877942535964,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009748097350485194,
      "loss": 2.0585,
      "step": 15994
    },
    {
      "epoch": 0.1086755737719537,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000974806363141641,
      "loss": 2.3418,
      "step": 15995
    },
    {
      "epoch": 0.10868236811854776,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009748029910149328,
      "loss": 2.2308,
      "step": 15996
    },
    {
      "epoch": 0.10868916246514182,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009747996186683961,
      "loss": 2.3128,
      "step": 15997
    },
    {
      "epoch": 0.10869595681173587,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009747962461020325,
      "loss": 2.2914,
      "step": 15998
    },
    {
      "epoch": 0.10870275115832993,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009747928733158438,
      "loss": 2.2563,
      "step": 15999
    },
    {
      "epoch": 0.108709545504924,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009747895003098315,
      "loss": 2.1853,
      "step": 16000
    },
    {
      "epoch": 0.10871633985151805,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009747861270839971,
      "loss": 2.1777,
      "step": 16001
    },
    {
      "epoch": 0.10872313419811211,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009747827536383423,
      "loss": 2.0304,
      "step": 16002
    },
    {
      "epoch": 0.10872992854470616,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009747793799728683,
      "loss": 2.2291,
      "step": 16003
    },
    {
      "epoch": 0.10873672289130022,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000974776006087577,
      "loss": 2.212,
      "step": 16004
    },
    {
      "epoch": 0.10874351723789429,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009747726319824698,
      "loss": 2.2509,
      "step": 16005
    },
    {
      "epoch": 0.10875031158448834,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009747692576575483,
      "loss": 2.154,
      "step": 16006
    },
    {
      "epoch": 0.1087571059310824,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009747658831128142,
      "loss": 2.2092,
      "step": 16007
    },
    {
      "epoch": 0.10876390027767645,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009747625083482688,
      "loss": 2.2145,
      "step": 16008
    },
    {
      "epoch": 0.10877069462427051,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009747591333639138,
      "loss": 2.2096,
      "step": 16009
    },
    {
      "epoch": 0.10877748897086457,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009747557581597508,
      "loss": 2.2556,
      "step": 16010
    },
    {
      "epoch": 0.10878428331745862,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009747523827357814,
      "loss": 2.1572,
      "step": 16011
    },
    {
      "epoch": 0.10879107766405269,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000974749007092007,
      "loss": 2.0991,
      "step": 16012
    },
    {
      "epoch": 0.10879787201064674,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009747456312284293,
      "loss": 2.1165,
      "step": 16013
    },
    {
      "epoch": 0.1088046663572408,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009747422551450497,
      "loss": 2.1577,
      "step": 16014
    },
    {
      "epoch": 0.10881146070383485,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009747388788418701,
      "loss": 2.2272,
      "step": 16015
    },
    {
      "epoch": 0.10881825505042891,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009747355023188918,
      "loss": 2.2064,
      "step": 16016
    },
    {
      "epoch": 0.10882504939702298,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009747321255761161,
      "loss": 2.28,
      "step": 16017
    },
    {
      "epoch": 0.10883184374361703,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009747287486135452,
      "loss": 2.2652,
      "step": 16018
    },
    {
      "epoch": 0.10883863809021109,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009747253714311803,
      "loss": 2.3227,
      "step": 16019
    },
    {
      "epoch": 0.10884543243680514,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009747219940290228,
      "loss": 2.3017,
      "step": 16020
    },
    {
      "epoch": 0.1088522267833992,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009747186164070746,
      "loss": 2.1522,
      "step": 16021
    },
    {
      "epoch": 0.10885902112999327,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009747152385653371,
      "loss": 2.2535,
      "step": 16022
    },
    {
      "epoch": 0.10886581547658732,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009747118605038119,
      "loss": 2.1237,
      "step": 16023
    },
    {
      "epoch": 0.10887260982318138,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009747084822225006,
      "loss": 2.1824,
      "step": 16024
    },
    {
      "epoch": 0.10887940416977543,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009747051037214045,
      "loss": 2.0993,
      "step": 16025
    },
    {
      "epoch": 0.10888619851636949,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009747017250005257,
      "loss": 2.0477,
      "step": 16026
    },
    {
      "epoch": 0.10889299286296356,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009746983460598652,
      "loss": 2.26,
      "step": 16027
    },
    {
      "epoch": 0.1088997872095576,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009746949668994249,
      "loss": 2.2169,
      "step": 16028
    },
    {
      "epoch": 0.10890658155615167,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009746915875192063,
      "loss": 2.1176,
      "step": 16029
    },
    {
      "epoch": 0.10891337590274572,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009746882079192108,
      "loss": 2.1459,
      "step": 16030
    },
    {
      "epoch": 0.10892017024933978,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009746848280994402,
      "loss": 2.3961,
      "step": 16031
    },
    {
      "epoch": 0.10892696459593385,
      "grad_norm": 0.140625,
      "learning_rate": 0.000974681448059896,
      "loss": 2.2803,
      "step": 16032
    },
    {
      "epoch": 0.1089337589425279,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009746780678005798,
      "loss": 2.188,
      "step": 16033
    },
    {
      "epoch": 0.10894055328912196,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009746746873214929,
      "loss": 2.3494,
      "step": 16034
    },
    {
      "epoch": 0.10894734763571601,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009746713066226371,
      "loss": 2.2262,
      "step": 16035
    },
    {
      "epoch": 0.10895414198231007,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009746679257040141,
      "loss": 2.2165,
      "step": 16036
    },
    {
      "epoch": 0.10896093632890412,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000974664544565625,
      "loss": 2.1552,
      "step": 16037
    },
    {
      "epoch": 0.10896773067549818,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000974661163207472,
      "loss": 2.3008,
      "step": 16038
    },
    {
      "epoch": 0.10897452502209225,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009746577816295561,
      "loss": 2.1502,
      "step": 16039
    },
    {
      "epoch": 0.1089813193686863,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009746543998318791,
      "loss": 2.148,
      "step": 16040
    },
    {
      "epoch": 0.10898811371528036,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009746510178144425,
      "loss": 2.1975,
      "step": 16041
    },
    {
      "epoch": 0.10899490806187441,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009746476355772481,
      "loss": 2.0337,
      "step": 16042
    },
    {
      "epoch": 0.10900170240846847,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009746442531202973,
      "loss": 2.2806,
      "step": 16043
    },
    {
      "epoch": 0.10900849675506254,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009746408704435915,
      "loss": 2.2421,
      "step": 16044
    },
    {
      "epoch": 0.10901529110165659,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009746374875471324,
      "loss": 2.0774,
      "step": 16045
    },
    {
      "epoch": 0.10902208544825065,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009746341044309217,
      "loss": 2.2155,
      "step": 16046
    },
    {
      "epoch": 0.1090288797948447,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009746307210949607,
      "loss": 2.1953,
      "step": 16047
    },
    {
      "epoch": 0.10903567414143876,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009746273375392514,
      "loss": 2.0894,
      "step": 16048
    },
    {
      "epoch": 0.10904246848803283,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009746239537637949,
      "loss": 2.2452,
      "step": 16049
    },
    {
      "epoch": 0.10904926283462688,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009746205697685929,
      "loss": 2.2184,
      "step": 16050
    },
    {
      "epoch": 0.10905605718122094,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009746171855536472,
      "loss": 2.1364,
      "step": 16051
    },
    {
      "epoch": 0.10906285152781499,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000974613801118959,
      "loss": 2.1986,
      "step": 16052
    },
    {
      "epoch": 0.10906964587440905,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009746104164645301,
      "loss": 2.34,
      "step": 16053
    },
    {
      "epoch": 0.10907644022100312,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000974607031590362,
      "loss": 2.2464,
      "step": 16054
    },
    {
      "epoch": 0.10908323456759716,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009746036464964564,
      "loss": 2.227,
      "step": 16055
    },
    {
      "epoch": 0.10909002891419123,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009746002611828147,
      "loss": 2.0844,
      "step": 16056
    },
    {
      "epoch": 0.10909682326078528,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009745968756494385,
      "loss": 2.2374,
      "step": 16057
    },
    {
      "epoch": 0.10910361760737934,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009745934898963294,
      "loss": 2.2344,
      "step": 16058
    },
    {
      "epoch": 0.10911041195397339,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009745901039234887,
      "loss": 2.1481,
      "step": 16059
    },
    {
      "epoch": 0.10911720630056745,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009745867177309187,
      "loss": 2.2006,
      "step": 16060
    },
    {
      "epoch": 0.10912400064716152,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009745833313186201,
      "loss": 2.3288,
      "step": 16061
    },
    {
      "epoch": 0.10913079499375557,
      "grad_norm": 0.125,
      "learning_rate": 0.0009745799446865951,
      "loss": 2.0459,
      "step": 16062
    },
    {
      "epoch": 0.10913758934034963,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009745765578348449,
      "loss": 2.2078,
      "step": 16063
    },
    {
      "epoch": 0.10914438368694368,
      "grad_norm": 0.119140625,
      "learning_rate": 0.0009745731707633712,
      "loss": 2.1254,
      "step": 16064
    },
    {
      "epoch": 0.10915117803353774,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009745697834721756,
      "loss": 2.0992,
      "step": 16065
    },
    {
      "epoch": 0.1091579723801318,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009745663959612596,
      "loss": 2.0878,
      "step": 16066
    },
    {
      "epoch": 0.10916476672672586,
      "grad_norm": 0.125,
      "learning_rate": 0.0009745630082306249,
      "loss": 2.1314,
      "step": 16067
    },
    {
      "epoch": 0.10917156107331992,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009745596202802728,
      "loss": 2.0479,
      "step": 16068
    },
    {
      "epoch": 0.10917835541991397,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009745562321102052,
      "loss": 2.1907,
      "step": 16069
    },
    {
      "epoch": 0.10918514976650803,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009745528437204234,
      "loss": 2.2548,
      "step": 16070
    },
    {
      "epoch": 0.1091919441131021,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009745494551109291,
      "loss": 2.2647,
      "step": 16071
    },
    {
      "epoch": 0.10919873845969615,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009745460662817238,
      "loss": 2.1802,
      "step": 16072
    },
    {
      "epoch": 0.10920553280629021,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009745426772328092,
      "loss": 2.2185,
      "step": 16073
    },
    {
      "epoch": 0.10921232715288426,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009745392879641867,
      "loss": 2.161,
      "step": 16074
    },
    {
      "epoch": 0.10921912149947832,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009745358984758581,
      "loss": 2.024,
      "step": 16075
    },
    {
      "epoch": 0.10922591584607239,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009745325087678246,
      "loss": 2.1734,
      "step": 16076
    },
    {
      "epoch": 0.10923271019266644,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009745291188400882,
      "loss": 2.2326,
      "step": 16077
    },
    {
      "epoch": 0.1092395045392605,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009745257286926501,
      "loss": 2.1867,
      "step": 16078
    },
    {
      "epoch": 0.10924629888585455,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000974522338325512,
      "loss": 2.17,
      "step": 16079
    },
    {
      "epoch": 0.10925309323244861,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009745189477386757,
      "loss": 2.1026,
      "step": 16080
    },
    {
      "epoch": 0.10925988757904267,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009745155569321425,
      "loss": 2.3005,
      "step": 16081
    },
    {
      "epoch": 0.10926668192563672,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000974512165905914,
      "loss": 2.1999,
      "step": 16082
    },
    {
      "epoch": 0.10927347627223079,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009745087746599917,
      "loss": 2.1227,
      "step": 16083
    },
    {
      "epoch": 0.10928027061882484,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009745053831943773,
      "loss": 2.2821,
      "step": 16084
    },
    {
      "epoch": 0.1092870649654189,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009745019915090724,
      "loss": 2.3182,
      "step": 16085
    },
    {
      "epoch": 0.10929385931201295,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009744985996040786,
      "loss": 2.1429,
      "step": 16086
    },
    {
      "epoch": 0.10930065365860701,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009744952074793973,
      "loss": 2.1553,
      "step": 16087
    },
    {
      "epoch": 0.10930744800520108,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009744918151350302,
      "loss": 2.116,
      "step": 16088
    },
    {
      "epoch": 0.10931424235179513,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009744884225709788,
      "loss": 2.3131,
      "step": 16089
    },
    {
      "epoch": 0.10932103669838919,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009744850297872447,
      "loss": 2.0685,
      "step": 16090
    },
    {
      "epoch": 0.10932783104498324,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009744816367838295,
      "loss": 2.0812,
      "step": 16091
    },
    {
      "epoch": 0.1093346253915773,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009744782435607347,
      "loss": 2.226,
      "step": 16092
    },
    {
      "epoch": 0.10934141973817137,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000974474850117962,
      "loss": 2.2049,
      "step": 16093
    },
    {
      "epoch": 0.10934821408476542,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009744714564555128,
      "loss": 2.2331,
      "step": 16094
    },
    {
      "epoch": 0.10935500843135948,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009744680625733888,
      "loss": 2.0743,
      "step": 16095
    },
    {
      "epoch": 0.10936180277795353,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009744646684715914,
      "loss": 2.1816,
      "step": 16096
    },
    {
      "epoch": 0.10936859712454759,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009744612741501225,
      "loss": 2.0905,
      "step": 16097
    },
    {
      "epoch": 0.10937539147114166,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009744578796089834,
      "loss": 2.2804,
      "step": 16098
    },
    {
      "epoch": 0.1093821858177357,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009744544848481756,
      "loss": 2.2315,
      "step": 16099
    },
    {
      "epoch": 0.10938898016432977,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000974451089867701,
      "loss": 2.1825,
      "step": 16100
    },
    {
      "epoch": 0.10939577451092382,
      "grad_norm": 0.125,
      "learning_rate": 0.0009744476946675609,
      "loss": 2.2745,
      "step": 16101
    },
    {
      "epoch": 0.10940256885751788,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009744442992477569,
      "loss": 2.1555,
      "step": 16102
    },
    {
      "epoch": 0.10940936320411195,
      "grad_norm": 0.125,
      "learning_rate": 0.0009744409036082908,
      "loss": 2.1614,
      "step": 16103
    },
    {
      "epoch": 0.109416157550706,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009744375077491639,
      "loss": 2.1112,
      "step": 16104
    },
    {
      "epoch": 0.10942295189730006,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009744341116703778,
      "loss": 2.1817,
      "step": 16105
    },
    {
      "epoch": 0.10942974624389411,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009744307153719343,
      "loss": 2.1918,
      "step": 16106
    },
    {
      "epoch": 0.10943654059048817,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009744273188538347,
      "loss": 2.1259,
      "step": 16107
    },
    {
      "epoch": 0.10944333493708222,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009744239221160806,
      "loss": 2.0977,
      "step": 16108
    },
    {
      "epoch": 0.10945012928367628,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000974420525158674,
      "loss": 2.0617,
      "step": 16109
    },
    {
      "epoch": 0.10945692363027035,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009744171279816159,
      "loss": 2.4135,
      "step": 16110
    },
    {
      "epoch": 0.1094637179768644,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009744137305849081,
      "loss": 2.1462,
      "step": 16111
    },
    {
      "epoch": 0.10947051232345846,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009744103329685521,
      "loss": 2.2212,
      "step": 16112
    },
    {
      "epoch": 0.10947730667005251,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009744069351325498,
      "loss": 2.2102,
      "step": 16113
    },
    {
      "epoch": 0.10948410101664657,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009744035370769024,
      "loss": 2.1152,
      "step": 16114
    },
    {
      "epoch": 0.10949089536324064,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009744001388016115,
      "loss": 2.3186,
      "step": 16115
    },
    {
      "epoch": 0.10949768970983469,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009743967403066789,
      "loss": 2.2116,
      "step": 16116
    },
    {
      "epoch": 0.10950448405642875,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000974393341592106,
      "loss": 2.1361,
      "step": 16117
    },
    {
      "epoch": 0.1095112784030228,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009743899426578943,
      "loss": 2.1611,
      "step": 16118
    },
    {
      "epoch": 0.10951807274961686,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009743865435040457,
      "loss": 2.0748,
      "step": 16119
    },
    {
      "epoch": 0.10952486709621093,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0009743831441305615,
      "loss": 2.1786,
      "step": 16120
    },
    {
      "epoch": 0.10953166144280498,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009743797445374434,
      "loss": 2.2198,
      "step": 16121
    },
    {
      "epoch": 0.10953845578939904,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009743763447246929,
      "loss": 2.2561,
      "step": 16122
    },
    {
      "epoch": 0.10954525013599309,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009743729446923115,
      "loss": 2.1975,
      "step": 16123
    },
    {
      "epoch": 0.10955204448258715,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009743695444403009,
      "loss": 2.2446,
      "step": 16124
    },
    {
      "epoch": 0.10955883882918122,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009743661439686628,
      "loss": 2.0756,
      "step": 16125
    },
    {
      "epoch": 0.10956563317577526,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009743627432773984,
      "loss": 2.2212,
      "step": 16126
    },
    {
      "epoch": 0.10957242752236933,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009743593423665095,
      "loss": 2.2531,
      "step": 16127
    },
    {
      "epoch": 0.10957922186896338,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009743559412359978,
      "loss": 2.209,
      "step": 16128
    },
    {
      "epoch": 0.10958601621555744,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009743525398858646,
      "loss": 2.1831,
      "step": 16129
    },
    {
      "epoch": 0.10959281056215149,
      "grad_norm": 0.125,
      "learning_rate": 0.0009743491383161117,
      "loss": 2.2254,
      "step": 16130
    },
    {
      "epoch": 0.10959960490874555,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009743457365267405,
      "loss": 2.2503,
      "step": 16131
    },
    {
      "epoch": 0.10960639925533962,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009743423345177528,
      "loss": 2.2137,
      "step": 16132
    },
    {
      "epoch": 0.10961319360193367,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009743389322891499,
      "loss": 2.2656,
      "step": 16133
    },
    {
      "epoch": 0.10961998794852773,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009743355298409336,
      "loss": 2.2692,
      "step": 16134
    },
    {
      "epoch": 0.10962678229512178,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009743321271731053,
      "loss": 2.1566,
      "step": 16135
    },
    {
      "epoch": 0.10963357664171584,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009743287242856669,
      "loss": 2.2867,
      "step": 16136
    },
    {
      "epoch": 0.10964037098830991,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009743253211786195,
      "loss": 2.349,
      "step": 16137
    },
    {
      "epoch": 0.10964716533490396,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009743219178519649,
      "loss": 2.1211,
      "step": 16138
    },
    {
      "epoch": 0.10965395968149802,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009743185143057049,
      "loss": 2.1521,
      "step": 16139
    },
    {
      "epoch": 0.10966075402809207,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009743151105398406,
      "loss": 2.1514,
      "step": 16140
    },
    {
      "epoch": 0.10966754837468613,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009743117065543741,
      "loss": 2.2092,
      "step": 16141
    },
    {
      "epoch": 0.1096743427212802,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009743083023493064,
      "loss": 2.1423,
      "step": 16142
    },
    {
      "epoch": 0.10968113706787425,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009743048979246398,
      "loss": 2.2981,
      "step": 16143
    },
    {
      "epoch": 0.10968793141446831,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009743014932803751,
      "loss": 2.2802,
      "step": 16144
    },
    {
      "epoch": 0.10969472576106236,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009742980884165144,
      "loss": 2.2175,
      "step": 16145
    },
    {
      "epoch": 0.10970152010765642,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009742946833330593,
      "loss": 2.1373,
      "step": 16146
    },
    {
      "epoch": 0.10970831445425049,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009742912780300109,
      "loss": 2.1828,
      "step": 16147
    },
    {
      "epoch": 0.10971510880084454,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009742878725073712,
      "loss": 2.1833,
      "step": 16148
    },
    {
      "epoch": 0.1097219031474386,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009742844667651416,
      "loss": 2.1357,
      "step": 16149
    },
    {
      "epoch": 0.10972869749403265,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009742810608033239,
      "loss": 2.2858,
      "step": 16150
    },
    {
      "epoch": 0.10973549184062671,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009742776546219193,
      "loss": 2.3212,
      "step": 16151
    },
    {
      "epoch": 0.10974228618722077,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009742742482209297,
      "loss": 2.2424,
      "step": 16152
    },
    {
      "epoch": 0.10974908053381482,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009742708416003565,
      "loss": 2.0932,
      "step": 16153
    },
    {
      "epoch": 0.10975587488040889,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009742674347602014,
      "loss": 2.1618,
      "step": 16154
    },
    {
      "epoch": 0.10976266922700294,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009742640277004658,
      "loss": 2.1427,
      "step": 16155
    },
    {
      "epoch": 0.109769463573597,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009742606204211515,
      "loss": 2.1551,
      "step": 16156
    },
    {
      "epoch": 0.10977625792019105,
      "grad_norm": 0.125,
      "learning_rate": 0.00097425721292226,
      "loss": 2.0806,
      "step": 16157
    },
    {
      "epoch": 0.10978305226678511,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009742538052037928,
      "loss": 2.2028,
      "step": 16158
    },
    {
      "epoch": 0.10978984661337918,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009742503972657515,
      "loss": 2.1571,
      "step": 16159
    },
    {
      "epoch": 0.10979664095997323,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009742469891081379,
      "loss": 2.2471,
      "step": 16160
    },
    {
      "epoch": 0.10980343530656729,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009742435807309531,
      "loss": 2.2819,
      "step": 16161
    },
    {
      "epoch": 0.10981022965316134,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000974240172134199,
      "loss": 2.3732,
      "step": 16162
    },
    {
      "epoch": 0.1098170239997554,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009742367633178773,
      "loss": 2.3239,
      "step": 16163
    },
    {
      "epoch": 0.10982381834634947,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009742333542819892,
      "loss": 2.2672,
      "step": 16164
    },
    {
      "epoch": 0.10983061269294352,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009742299450265366,
      "loss": 2.3491,
      "step": 16165
    },
    {
      "epoch": 0.10983740703953758,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000974226535551521,
      "loss": 2.3044,
      "step": 16166
    },
    {
      "epoch": 0.10984420138613163,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000974223125856944,
      "loss": 2.1405,
      "step": 16167
    },
    {
      "epoch": 0.10985099573272569,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000974219715942807,
      "loss": 2.2129,
      "step": 16168
    },
    {
      "epoch": 0.10985779007931976,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009742163058091118,
      "loss": 2.1527,
      "step": 16169
    },
    {
      "epoch": 0.1098645844259138,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0009742128954558598,
      "loss": 2.0263,
      "step": 16170
    },
    {
      "epoch": 0.10987137877250787,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009742094848830527,
      "loss": 2.1377,
      "step": 16171
    },
    {
      "epoch": 0.10987817311910192,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000974206074090692,
      "loss": 2.3315,
      "step": 16172
    },
    {
      "epoch": 0.10988496746569598,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009742026630787795,
      "loss": 2.2566,
      "step": 16173
    },
    {
      "epoch": 0.10989176181229005,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009741992518473164,
      "loss": 2.1304,
      "step": 16174
    },
    {
      "epoch": 0.1098985561588841,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009741958403963047,
      "loss": 2.2978,
      "step": 16175
    },
    {
      "epoch": 0.10990535050547816,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009741924287257456,
      "loss": 2.1298,
      "step": 16176
    },
    {
      "epoch": 0.10991214485207221,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009741890168356408,
      "loss": 2.2988,
      "step": 16177
    },
    {
      "epoch": 0.10991893919866627,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009741856047259921,
      "loss": 2.134,
      "step": 16178
    },
    {
      "epoch": 0.10992573354526032,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009741821923968008,
      "loss": 2.1174,
      "step": 16179
    },
    {
      "epoch": 0.10993252789185438,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009741787798480685,
      "loss": 2.2043,
      "step": 16180
    },
    {
      "epoch": 0.10993932223844845,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009741753670797971,
      "loss": 2.1487,
      "step": 16181
    },
    {
      "epoch": 0.1099461165850425,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009741719540919877,
      "loss": 2.1603,
      "step": 16182
    },
    {
      "epoch": 0.10995291093163656,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009741685408846422,
      "loss": 2.1805,
      "step": 16183
    },
    {
      "epoch": 0.10995970527823061,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000974165127457762,
      "loss": 2.2552,
      "step": 16184
    },
    {
      "epoch": 0.10996649962482467,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000974161713811349,
      "loss": 2.2089,
      "step": 16185
    },
    {
      "epoch": 0.10997329397141874,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009741582999454044,
      "loss": 2.3124,
      "step": 16186
    },
    {
      "epoch": 0.10998008831801279,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009741548858599301,
      "loss": 2.2865,
      "step": 16187
    },
    {
      "epoch": 0.10998688266460685,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009741514715549273,
      "loss": 2.0976,
      "step": 16188
    },
    {
      "epoch": 0.1099936770112009,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000974148057030398,
      "loss": 2.1381,
      "step": 16189
    },
    {
      "epoch": 0.11000047135779496,
      "grad_norm": 0.125,
      "learning_rate": 0.0009741446422863436,
      "loss": 2.1139,
      "step": 16190
    },
    {
      "epoch": 0.11000726570438903,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009741412273227654,
      "loss": 2.2073,
      "step": 16191
    },
    {
      "epoch": 0.11001406005098308,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009741378121396656,
      "loss": 2.0806,
      "step": 16192
    },
    {
      "epoch": 0.11002085439757714,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009741343967370452,
      "loss": 2.2597,
      "step": 16193
    },
    {
      "epoch": 0.11002764874417119,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000974130981114906,
      "loss": 2.0138,
      "step": 16194
    },
    {
      "epoch": 0.11003444309076525,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009741275652732497,
      "loss": 2.2193,
      "step": 16195
    },
    {
      "epoch": 0.11004123743735932,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009741241492120778,
      "loss": 2.2147,
      "step": 16196
    },
    {
      "epoch": 0.11004803178395337,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009741207329313917,
      "loss": 2.0708,
      "step": 16197
    },
    {
      "epoch": 0.11005482613054743,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009741173164311931,
      "loss": 2.149,
      "step": 16198
    },
    {
      "epoch": 0.11006162047714148,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009741138997114839,
      "loss": 2.2267,
      "step": 16199
    },
    {
      "epoch": 0.11006841482373554,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009741104827722652,
      "loss": 2.1752,
      "step": 16200
    },
    {
      "epoch": 0.1100752091703296,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009741070656135389,
      "loss": 2.1609,
      "step": 16201
    },
    {
      "epoch": 0.11008200351692365,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009741036482353063,
      "loss": 2.2317,
      "step": 16202
    },
    {
      "epoch": 0.11008879786351772,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009741002306375693,
      "loss": 2.2537,
      "step": 16203
    },
    {
      "epoch": 0.11009559221011177,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009740968128203291,
      "loss": 2.0771,
      "step": 16204
    },
    {
      "epoch": 0.11010238655670583,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009740933947835878,
      "loss": 2.1791,
      "step": 16205
    },
    {
      "epoch": 0.11010918090329988,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009740899765273465,
      "loss": 2.0999,
      "step": 16206
    },
    {
      "epoch": 0.11011597524989394,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009740865580516071,
      "loss": 2.2407,
      "step": 16207
    },
    {
      "epoch": 0.11012276959648801,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009740831393563709,
      "loss": 2.112,
      "step": 16208
    },
    {
      "epoch": 0.11012956394308206,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009740797204416398,
      "loss": 2.1901,
      "step": 16209
    },
    {
      "epoch": 0.11013635828967612,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009740763013074152,
      "loss": 2.2099,
      "step": 16210
    },
    {
      "epoch": 0.11014315263627017,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009740728819536986,
      "loss": 2.1277,
      "step": 16211
    },
    {
      "epoch": 0.11014994698286423,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009740694623804918,
      "loss": 2.0407,
      "step": 16212
    },
    {
      "epoch": 0.1101567413294583,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009740660425877963,
      "loss": 2.2552,
      "step": 16213
    },
    {
      "epoch": 0.11016353567605235,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009740626225756135,
      "loss": 2.1635,
      "step": 16214
    },
    {
      "epoch": 0.11017033002264641,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009740592023439453,
      "loss": 2.2125,
      "step": 16215
    },
    {
      "epoch": 0.11017712436924046,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009740557818927931,
      "loss": 2.1662,
      "step": 16216
    },
    {
      "epoch": 0.11018391871583452,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009740523612221583,
      "loss": 2.2503,
      "step": 16217
    },
    {
      "epoch": 0.11019071306242859,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000974048940332043,
      "loss": 2.241,
      "step": 16218
    },
    {
      "epoch": 0.11019750740902264,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009740455192224482,
      "loss": 2.2697,
      "step": 16219
    },
    {
      "epoch": 0.1102043017556167,
      "grad_norm": 0.140625,
      "learning_rate": 0.000974042097893376,
      "loss": 2.3287,
      "step": 16220
    },
    {
      "epoch": 0.11021109610221075,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009740386763448277,
      "loss": 2.1498,
      "step": 16221
    },
    {
      "epoch": 0.11021789044880481,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009740352545768048,
      "loss": 2.0444,
      "step": 16222
    },
    {
      "epoch": 0.11022468479539888,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000974031832589309,
      "loss": 2.2117,
      "step": 16223
    },
    {
      "epoch": 0.11023147914199292,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000974028410382342,
      "loss": 2.2292,
      "step": 16224
    },
    {
      "epoch": 0.11023827348858699,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009740249879559053,
      "loss": 2.3959,
      "step": 16225
    },
    {
      "epoch": 0.11024506783518104,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009740215653100005,
      "loss": 2.1887,
      "step": 16226
    },
    {
      "epoch": 0.1102518621817751,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009740181424446289,
      "loss": 2.0115,
      "step": 16227
    },
    {
      "epoch": 0.11025865652836915,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009740147193597926,
      "loss": 2.1836,
      "step": 16228
    },
    {
      "epoch": 0.11026545087496321,
      "grad_norm": 0.125,
      "learning_rate": 0.0009740112960554927,
      "loss": 2.199,
      "step": 16229
    },
    {
      "epoch": 0.11027224522155728,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009740078725317312,
      "loss": 2.1454,
      "step": 16230
    },
    {
      "epoch": 0.11027903956815133,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009740044487885094,
      "loss": 2.2141,
      "step": 16231
    },
    {
      "epoch": 0.11028583391474539,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009740010248258291,
      "loss": 2.2084,
      "step": 16232
    },
    {
      "epoch": 0.11029262826133944,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009739976006436916,
      "loss": 2.017,
      "step": 16233
    },
    {
      "epoch": 0.1102994226079335,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009739941762420987,
      "loss": 2.2979,
      "step": 16234
    },
    {
      "epoch": 0.11030621695452757,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009739907516210518,
      "loss": 2.1576,
      "step": 16235
    },
    {
      "epoch": 0.11031301130112162,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009739873267805528,
      "loss": 2.3342,
      "step": 16236
    },
    {
      "epoch": 0.11031980564771568,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009739839017206031,
      "loss": 2.1828,
      "step": 16237
    },
    {
      "epoch": 0.11032659999430973,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009739804764412041,
      "loss": 2.1736,
      "step": 16238
    },
    {
      "epoch": 0.11033339434090379,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009739770509423578,
      "loss": 2.1617,
      "step": 16239
    },
    {
      "epoch": 0.11034018868749786,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009739736252240654,
      "loss": 2.2252,
      "step": 16240
    },
    {
      "epoch": 0.1103469830340919,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009739701992863288,
      "loss": 2.1711,
      "step": 16241
    },
    {
      "epoch": 0.11035377738068597,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009739667731291493,
      "loss": 2.2402,
      "step": 16242
    },
    {
      "epoch": 0.11036057172728002,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009739633467525286,
      "loss": 2.2805,
      "step": 16243
    },
    {
      "epoch": 0.11036736607387408,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009739599201564684,
      "loss": 2.1447,
      "step": 16244
    },
    {
      "epoch": 0.11037416042046815,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009739564933409702,
      "loss": 2.2779,
      "step": 16245
    },
    {
      "epoch": 0.1103809547670622,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009739530663060354,
      "loss": 2.1177,
      "step": 16246
    },
    {
      "epoch": 0.11038774911365626,
      "grad_norm": 0.140625,
      "learning_rate": 0.000973949639051666,
      "loss": 2.05,
      "step": 16247
    },
    {
      "epoch": 0.11039454346025031,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009739462115778633,
      "loss": 2.1457,
      "step": 16248
    },
    {
      "epoch": 0.11040133780684437,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009739427838846289,
      "loss": 2.1458,
      "step": 16249
    },
    {
      "epoch": 0.11040813215343842,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009739393559719643,
      "loss": 2.1449,
      "step": 16250
    },
    {
      "epoch": 0.11041492650003248,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009739359278398714,
      "loss": 2.2068,
      "step": 16251
    },
    {
      "epoch": 0.11042172084662655,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009739324994883513,
      "loss": 2.1161,
      "step": 16252
    },
    {
      "epoch": 0.1104285151932206,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009739290709174063,
      "loss": 2.0545,
      "step": 16253
    },
    {
      "epoch": 0.11043530953981466,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009739256421270372,
      "loss": 2.2861,
      "step": 16254
    },
    {
      "epoch": 0.11044210388640871,
      "grad_norm": 0.125,
      "learning_rate": 0.0009739222131172463,
      "loss": 2.2534,
      "step": 16255
    },
    {
      "epoch": 0.11044889823300277,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009739187838880346,
      "loss": 2.0583,
      "step": 16256
    },
    {
      "epoch": 0.11045569257959684,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000973915354439404,
      "loss": 2.0434,
      "step": 16257
    },
    {
      "epoch": 0.11046248692619089,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009739119247713561,
      "loss": 2.1193,
      "step": 16258
    },
    {
      "epoch": 0.11046928127278495,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009739084948838923,
      "loss": 2.2365,
      "step": 16259
    },
    {
      "epoch": 0.110476075619379,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009739050647770144,
      "loss": 2.1128,
      "step": 16260
    },
    {
      "epoch": 0.11048286996597306,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009739016344507238,
      "loss": 2.1704,
      "step": 16261
    },
    {
      "epoch": 0.11048966431256713,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009738982039050223,
      "loss": 2.0664,
      "step": 16262
    },
    {
      "epoch": 0.11049645865916118,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009738947731399112,
      "loss": 2.2075,
      "step": 16263
    },
    {
      "epoch": 0.11050325300575524,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009738913421553923,
      "loss": 2.2969,
      "step": 16264
    },
    {
      "epoch": 0.11051004735234929,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009738879109514674,
      "loss": 2.1596,
      "step": 16265
    },
    {
      "epoch": 0.11051684169894335,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009738844795281374,
      "loss": 2.196,
      "step": 16266
    },
    {
      "epoch": 0.11052363604553742,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009738810478854045,
      "loss": 2.3,
      "step": 16267
    },
    {
      "epoch": 0.11053043039213147,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009738776160232703,
      "loss": 2.2277,
      "step": 16268
    },
    {
      "epoch": 0.11053722473872553,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000973874183941736,
      "loss": 2.0781,
      "step": 16269
    },
    {
      "epoch": 0.11054401908531958,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009738707516408034,
      "loss": 2.2098,
      "step": 16270
    },
    {
      "epoch": 0.11055081343191364,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009738673191204741,
      "loss": 2.1879,
      "step": 16271
    },
    {
      "epoch": 0.1105576077785077,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009738638863807497,
      "loss": 2.1649,
      "step": 16272
    },
    {
      "epoch": 0.11056440212510175,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009738604534216316,
      "loss": 2.0773,
      "step": 16273
    },
    {
      "epoch": 0.11057119647169582,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009738570202431217,
      "loss": 2.3022,
      "step": 16274
    },
    {
      "epoch": 0.11057799081828987,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009738535868452214,
      "loss": 2.2519,
      "step": 16275
    },
    {
      "epoch": 0.11058478516488393,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009738501532279322,
      "loss": 2.221,
      "step": 16276
    },
    {
      "epoch": 0.11059157951147798,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000973846719391256,
      "loss": 2.2634,
      "step": 16277
    },
    {
      "epoch": 0.11059837385807204,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009738432853351942,
      "loss": 2.2203,
      "step": 16278
    },
    {
      "epoch": 0.11060516820466611,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009738398510597483,
      "loss": 2.1963,
      "step": 16279
    },
    {
      "epoch": 0.11061196255126016,
      "grad_norm": 0.130859375,
      "learning_rate": 0.00097383641656492,
      "loss": 2.1905,
      "step": 16280
    },
    {
      "epoch": 0.11061875689785422,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009738329818507108,
      "loss": 2.1152,
      "step": 16281
    },
    {
      "epoch": 0.11062555124444827,
      "grad_norm": 0.125,
      "learning_rate": 0.0009738295469171224,
      "loss": 2.0831,
      "step": 16282
    },
    {
      "epoch": 0.11063234559104233,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009738261117641564,
      "loss": 2.191,
      "step": 16283
    },
    {
      "epoch": 0.1106391399376364,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009738226763918144,
      "loss": 2.2915,
      "step": 16284
    },
    {
      "epoch": 0.11064593428423045,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009738192408000979,
      "loss": 2.1551,
      "step": 16285
    },
    {
      "epoch": 0.11065272863082451,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009738158049890084,
      "loss": 2.257,
      "step": 16286
    },
    {
      "epoch": 0.11065952297741856,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009738123689585476,
      "loss": 2.2555,
      "step": 16287
    },
    {
      "epoch": 0.11066631732401262,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009738089327087173,
      "loss": 2.2128,
      "step": 16288
    },
    {
      "epoch": 0.11067311167060669,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009738054962395188,
      "loss": 2.1112,
      "step": 16289
    },
    {
      "epoch": 0.11067990601720074,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009738020595509538,
      "loss": 2.3159,
      "step": 16290
    },
    {
      "epoch": 0.1106867003637948,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009737986226430237,
      "loss": 2.2056,
      "step": 16291
    },
    {
      "epoch": 0.11069349471038885,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009737951855157306,
      "loss": 2.1438,
      "step": 16292
    },
    {
      "epoch": 0.11070028905698291,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009737917481690754,
      "loss": 2.1308,
      "step": 16293
    },
    {
      "epoch": 0.11070708340357698,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009737883106030603,
      "loss": 2.2771,
      "step": 16294
    },
    {
      "epoch": 0.11071387775017102,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009737848728176866,
      "loss": 2.1829,
      "step": 16295
    },
    {
      "epoch": 0.11072067209676509,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009737814348129559,
      "loss": 2.1755,
      "step": 16296
    },
    {
      "epoch": 0.11072746644335914,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009737779965888699,
      "loss": 2.3002,
      "step": 16297
    },
    {
      "epoch": 0.1107342607899532,
      "grad_norm": 0.1328125,
      "learning_rate": 0.00097377455814543,
      "loss": 2.185,
      "step": 16298
    },
    {
      "epoch": 0.11074105513654725,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009737711194826381,
      "loss": 2.1987,
      "step": 16299
    },
    {
      "epoch": 0.11074784948314131,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009737676806004954,
      "loss": 2.2132,
      "step": 16300
    },
    {
      "epoch": 0.11075464382973538,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009737642414990037,
      "loss": 2.2356,
      "step": 16301
    },
    {
      "epoch": 0.11076143817632943,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009737608021781648,
      "loss": 2.1131,
      "step": 16302
    },
    {
      "epoch": 0.11076823252292349,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009737573626379799,
      "loss": 2.2083,
      "step": 16303
    },
    {
      "epoch": 0.11077502686951754,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009737539228784509,
      "loss": 2.2057,
      "step": 16304
    },
    {
      "epoch": 0.1107818212161116,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009737504828995792,
      "loss": 2.1153,
      "step": 16305
    },
    {
      "epoch": 0.11078861556270567,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009737470427013664,
      "loss": 2.0776,
      "step": 16306
    },
    {
      "epoch": 0.11079540990929972,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009737436022838142,
      "loss": 2.2788,
      "step": 16307
    },
    {
      "epoch": 0.11080220425589378,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009737401616469241,
      "loss": 2.1938,
      "step": 16308
    },
    {
      "epoch": 0.11080899860248783,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009737367207906977,
      "loss": 2.2997,
      "step": 16309
    },
    {
      "epoch": 0.11081579294908189,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009737332797151369,
      "loss": 2.1864,
      "step": 16310
    },
    {
      "epoch": 0.11082258729567596,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009737298384202426,
      "loss": 2.3907,
      "step": 16311
    },
    {
      "epoch": 0.11082938164227,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000973726396906017,
      "loss": 2.2314,
      "step": 16312
    },
    {
      "epoch": 0.11083617598886407,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009737229551724616,
      "loss": 2.1406,
      "step": 16313
    },
    {
      "epoch": 0.11084297033545812,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000973719513219578,
      "loss": 2.1678,
      "step": 16314
    },
    {
      "epoch": 0.11084976468205218,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009737160710473674,
      "loss": 2.2303,
      "step": 16315
    },
    {
      "epoch": 0.11085655902864625,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009737126286558319,
      "loss": 2.3822,
      "step": 16316
    },
    {
      "epoch": 0.1108633533752403,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009737091860449728,
      "loss": 2.2322,
      "step": 16317
    },
    {
      "epoch": 0.11087014772183436,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009737057432147917,
      "loss": 2.1559,
      "step": 16318
    },
    {
      "epoch": 0.11087694206842841,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009737023001652905,
      "loss": 2.1241,
      "step": 16319
    },
    {
      "epoch": 0.11088373641502247,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009736988568964704,
      "loss": 2.2633,
      "step": 16320
    },
    {
      "epoch": 0.11089053076161653,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009736954134083332,
      "loss": 2.317,
      "step": 16321
    },
    {
      "epoch": 0.11089732510821058,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009736919697008805,
      "loss": 2.1529,
      "step": 16322
    },
    {
      "epoch": 0.11090411945480465,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009736885257741137,
      "loss": 2.1081,
      "step": 16323
    },
    {
      "epoch": 0.1109109138013987,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009736850816280347,
      "loss": 2.3225,
      "step": 16324
    },
    {
      "epoch": 0.11091770814799276,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009736816372626449,
      "loss": 2.1053,
      "step": 16325
    },
    {
      "epoch": 0.11092450249458681,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009736781926779459,
      "loss": 2.1566,
      "step": 16326
    },
    {
      "epoch": 0.11093129684118087,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009736747478739394,
      "loss": 2.1913,
      "step": 16327
    },
    {
      "epoch": 0.11093809118777494,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009736713028506268,
      "loss": 2.0817,
      "step": 16328
    },
    {
      "epoch": 0.11094488553436899,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009736678576080098,
      "loss": 2.2191,
      "step": 16329
    },
    {
      "epoch": 0.11095167988096305,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009736644121460901,
      "loss": 2.2461,
      "step": 16330
    },
    {
      "epoch": 0.1109584742275571,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009736609664648693,
      "loss": 2.1033,
      "step": 16331
    },
    {
      "epoch": 0.11096526857415116,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009736575205643487,
      "loss": 2.1929,
      "step": 16332
    },
    {
      "epoch": 0.11097206292074523,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009736540744445302,
      "loss": 2.1837,
      "step": 16333
    },
    {
      "epoch": 0.11097885726733928,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009736506281054154,
      "loss": 2.3024,
      "step": 16334
    },
    {
      "epoch": 0.11098565161393334,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009736471815470056,
      "loss": 2.1549,
      "step": 16335
    },
    {
      "epoch": 0.11099244596052739,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009736437347693026,
      "loss": 2.2131,
      "step": 16336
    },
    {
      "epoch": 0.11099924030712145,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009736402877723082,
      "loss": 2.1893,
      "step": 16337
    },
    {
      "epoch": 0.11100603465371552,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009736368405560236,
      "loss": 2.1526,
      "step": 16338
    },
    {
      "epoch": 0.11101282900030957,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009736333931204506,
      "loss": 2.1761,
      "step": 16339
    },
    {
      "epoch": 0.11101962334690363,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009736299454655909,
      "loss": 2.2195,
      "step": 16340
    },
    {
      "epoch": 0.11102641769349768,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009736264975914457,
      "loss": 2.1552,
      "step": 16341
    },
    {
      "epoch": 0.11103321204009174,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000973623049498017,
      "loss": 2.0688,
      "step": 16342
    },
    {
      "epoch": 0.1110400063866858,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009736196011853062,
      "loss": 2.137,
      "step": 16343
    },
    {
      "epoch": 0.11104680073327985,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000973616152653315,
      "loss": 2.1236,
      "step": 16344
    },
    {
      "epoch": 0.11105359507987392,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000973612703902045,
      "loss": 2.2001,
      "step": 16345
    },
    {
      "epoch": 0.11106038942646797,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009736092549314976,
      "loss": 2.1924,
      "step": 16346
    },
    {
      "epoch": 0.11106718377306203,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009736058057416748,
      "loss": 2.2232,
      "step": 16347
    },
    {
      "epoch": 0.11107397811965608,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009736023563325778,
      "loss": 2.1542,
      "step": 16348
    },
    {
      "epoch": 0.11108077246625014,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009735989067042082,
      "loss": 2.325,
      "step": 16349
    },
    {
      "epoch": 0.11108756681284421,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009735954568565679,
      "loss": 2.2455,
      "step": 16350
    },
    {
      "epoch": 0.11109436115943826,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009735920067896584,
      "loss": 2.2242,
      "step": 16351
    },
    {
      "epoch": 0.11110115550603232,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009735885565034809,
      "loss": 2.1002,
      "step": 16352
    },
    {
      "epoch": 0.11110794985262637,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009735851059980376,
      "loss": 2.0142,
      "step": 16353
    },
    {
      "epoch": 0.11111474419922043,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009735816552733299,
      "loss": 2.3967,
      "step": 16354
    },
    {
      "epoch": 0.1111215385458145,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009735782043293592,
      "loss": 2.3098,
      "step": 16355
    },
    {
      "epoch": 0.11112833289240855,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009735747531661272,
      "loss": 2.2686,
      "step": 16356
    },
    {
      "epoch": 0.11113512723900261,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009735713017836356,
      "loss": 2.3356,
      "step": 16357
    },
    {
      "epoch": 0.11114192158559666,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009735678501818859,
      "loss": 2.3233,
      "step": 16358
    },
    {
      "epoch": 0.11114871593219072,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009735643983608797,
      "loss": 2.1868,
      "step": 16359
    },
    {
      "epoch": 0.11115551027878479,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009735609463206185,
      "loss": 2.2098,
      "step": 16360
    },
    {
      "epoch": 0.11116230462537884,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009735574940611041,
      "loss": 2.3195,
      "step": 16361
    },
    {
      "epoch": 0.1111690989719729,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0009735540415823381,
      "loss": 2.1794,
      "step": 16362
    },
    {
      "epoch": 0.11117589331856695,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009735505888843218,
      "loss": 2.186,
      "step": 16363
    },
    {
      "epoch": 0.11118268766516101,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009735471359670571,
      "loss": 2.1918,
      "step": 16364
    },
    {
      "epoch": 0.11118948201175508,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009735436828305455,
      "loss": 2.1934,
      "step": 16365
    },
    {
      "epoch": 0.11119627635834912,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009735402294747887,
      "loss": 2.2189,
      "step": 16366
    },
    {
      "epoch": 0.11120307070494319,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009735367758997882,
      "loss": 2.3004,
      "step": 16367
    },
    {
      "epoch": 0.11120986505153724,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009735333221055454,
      "loss": 2.0935,
      "step": 16368
    },
    {
      "epoch": 0.1112166593981313,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009735298680920623,
      "loss": 2.1295,
      "step": 16369
    },
    {
      "epoch": 0.11122345374472535,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0009735264138593403,
      "loss": 2.2746,
      "step": 16370
    },
    {
      "epoch": 0.11123024809131941,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009735229594073809,
      "loss": 2.2349,
      "step": 16371
    },
    {
      "epoch": 0.11123704243791348,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009735195047361857,
      "loss": 2.2557,
      "step": 16372
    },
    {
      "epoch": 0.11124383678450753,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009735160498457567,
      "loss": 2.0899,
      "step": 16373
    },
    {
      "epoch": 0.11125063113110159,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000973512594736095,
      "loss": 2.2386,
      "step": 16374
    },
    {
      "epoch": 0.11125742547769564,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009735091394072025,
      "loss": 2.1422,
      "step": 16375
    },
    {
      "epoch": 0.1112642198242897,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009735056838590805,
      "loss": 2.2578,
      "step": 16376
    },
    {
      "epoch": 0.11127101417088377,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009735022280917309,
      "loss": 2.1358,
      "step": 16377
    },
    {
      "epoch": 0.11127780851747782,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009734987721051553,
      "loss": 2.0236,
      "step": 16378
    },
    {
      "epoch": 0.11128460286407188,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009734953158993552,
      "loss": 2.147,
      "step": 16379
    },
    {
      "epoch": 0.11129139721066593,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000973491859474332,
      "loss": 2.1124,
      "step": 16380
    },
    {
      "epoch": 0.11129819155725999,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009734884028300876,
      "loss": 2.1814,
      "step": 16381
    },
    {
      "epoch": 0.11130498590385406,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009734849459666235,
      "loss": 2.0563,
      "step": 16382
    },
    {
      "epoch": 0.1113117802504481,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009734814888839412,
      "loss": 2.1012,
      "step": 16383
    },
    {
      "epoch": 0.11131857459704217,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009734780315820426,
      "loss": 2.2598,
      "step": 16384
    },
    {
      "epoch": 0.11132536894363622,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000973474574060929,
      "loss": 2.2683,
      "step": 16385
    },
    {
      "epoch": 0.11133216329023028,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009734711163206021,
      "loss": 2.1858,
      "step": 16386
    },
    {
      "epoch": 0.11133895763682435,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009734676583610636,
      "loss": 2.1814,
      "step": 16387
    },
    {
      "epoch": 0.1113457519834184,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009734642001823149,
      "loss": 2.2455,
      "step": 16388
    },
    {
      "epoch": 0.11135254633001246,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009734607417843576,
      "loss": 2.1693,
      "step": 16389
    },
    {
      "epoch": 0.11135934067660651,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009734572831671935,
      "loss": 2.0477,
      "step": 16390
    },
    {
      "epoch": 0.11136613502320057,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009734538243308242,
      "loss": 2.1043,
      "step": 16391
    },
    {
      "epoch": 0.11137292936979463,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009734503652752511,
      "loss": 2.1242,
      "step": 16392
    },
    {
      "epoch": 0.11137972371638868,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009734469060004759,
      "loss": 2.169,
      "step": 16393
    },
    {
      "epoch": 0.11138651806298275,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009734434465065003,
      "loss": 2.339,
      "step": 16394
    },
    {
      "epoch": 0.1113933124095768,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009734399867933256,
      "loss": 2.1807,
      "step": 16395
    },
    {
      "epoch": 0.11140010675617086,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000973436526860954,
      "loss": 2.1818,
      "step": 16396
    },
    {
      "epoch": 0.11140690110276491,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009734330667093865,
      "loss": 2.0756,
      "step": 16397
    },
    {
      "epoch": 0.11141369544935897,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009734296063386248,
      "loss": 2.1718,
      "step": 16398
    },
    {
      "epoch": 0.11142048979595304,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009734261457486707,
      "loss": 2.119,
      "step": 16399
    },
    {
      "epoch": 0.11142728414254709,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009734226849395258,
      "loss": 2.2722,
      "step": 16400
    },
    {
      "epoch": 0.11143407848914115,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009734192239111917,
      "loss": 2.1398,
      "step": 16401
    },
    {
      "epoch": 0.1114408728357352,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009734157626636697,
      "loss": 2.095,
      "step": 16402
    },
    {
      "epoch": 0.11144766718232926,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009734123011969617,
      "loss": 2.2846,
      "step": 16403
    },
    {
      "epoch": 0.11145446152892333,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009734088395110692,
      "loss": 2.2738,
      "step": 16404
    },
    {
      "epoch": 0.11146125587551738,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009734053776059939,
      "loss": 2.154,
      "step": 16405
    },
    {
      "epoch": 0.11146805022211144,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009734019154817374,
      "loss": 2.1956,
      "step": 16406
    },
    {
      "epoch": 0.11147484456870549,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009733984531383011,
      "loss": 2.2549,
      "step": 16407
    },
    {
      "epoch": 0.11148163891529955,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009733949905756869,
      "loss": 2.2972,
      "step": 16408
    },
    {
      "epoch": 0.11148843326189362,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000973391527793896,
      "loss": 2.2134,
      "step": 16409
    },
    {
      "epoch": 0.11149522760848767,
      "grad_norm": 0.125,
      "learning_rate": 0.0009733880647929305,
      "loss": 2.0234,
      "step": 16410
    },
    {
      "epoch": 0.11150202195508173,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009733846015727918,
      "loss": 2.2074,
      "step": 16411
    },
    {
      "epoch": 0.11150881630167578,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009733811381334812,
      "loss": 2.2133,
      "step": 16412
    },
    {
      "epoch": 0.11151561064826984,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009733776744750007,
      "loss": 2.3508,
      "step": 16413
    },
    {
      "epoch": 0.1115224049948639,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009733742105973518,
      "loss": 2.1167,
      "step": 16414
    },
    {
      "epoch": 0.11152919934145795,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000973370746500536,
      "loss": 2.1811,
      "step": 16415
    },
    {
      "epoch": 0.11153599368805202,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.000973367282184555,
      "loss": 2.1762,
      "step": 16416
    },
    {
      "epoch": 0.11154278803464607,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009733638176494104,
      "loss": 2.1783,
      "step": 16417
    },
    {
      "epoch": 0.11154958238124013,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009733603528951036,
      "loss": 2.1927,
      "step": 16418
    },
    {
      "epoch": 0.11155637672783418,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009733568879216366,
      "loss": 2.1483,
      "step": 16419
    },
    {
      "epoch": 0.11156317107442824,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009733534227290107,
      "loss": 2.174,
      "step": 16420
    },
    {
      "epoch": 0.11156996542102231,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009733499573172276,
      "loss": 2.1447,
      "step": 16421
    },
    {
      "epoch": 0.11157675976761636,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009733464916862888,
      "loss": 2.19,
      "step": 16422
    },
    {
      "epoch": 0.11158355411421042,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009733430258361961,
      "loss": 2.4179,
      "step": 16423
    },
    {
      "epoch": 0.11159034846080447,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009733395597669513,
      "loss": 2.2347,
      "step": 16424
    },
    {
      "epoch": 0.11159714280739853,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009733360934785552,
      "loss": 2.1133,
      "step": 16425
    },
    {
      "epoch": 0.1116039371539926,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009733326269710101,
      "loss": 2.0579,
      "step": 16426
    },
    {
      "epoch": 0.11161073150058665,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009733291602443175,
      "loss": 2.0952,
      "step": 16427
    },
    {
      "epoch": 0.11161752584718071,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009733256932984789,
      "loss": 2.2716,
      "step": 16428
    },
    {
      "epoch": 0.11162432019377476,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009733222261334959,
      "loss": 2.2696,
      "step": 16429
    },
    {
      "epoch": 0.11163111454036882,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009733187587493701,
      "loss": 2.224,
      "step": 16430
    },
    {
      "epoch": 0.11163790888696289,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009733152911461032,
      "loss": 2.0636,
      "step": 16431
    },
    {
      "epoch": 0.11164470323355694,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009733118233236968,
      "loss": 2.1944,
      "step": 16432
    },
    {
      "epoch": 0.111651497580151,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009733083552821524,
      "loss": 2.1933,
      "step": 16433
    },
    {
      "epoch": 0.11165829192674505,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009733048870214716,
      "loss": 2.1624,
      "step": 16434
    },
    {
      "epoch": 0.11166508627333911,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009733014185416561,
      "loss": 2.1584,
      "step": 16435
    },
    {
      "epoch": 0.11167188061993318,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009732979498427074,
      "loss": 2.0792,
      "step": 16436
    },
    {
      "epoch": 0.11167867496652722,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009732944809246274,
      "loss": 2.2196,
      "step": 16437
    },
    {
      "epoch": 0.11168546931312129,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009732910117874173,
      "loss": 2.3243,
      "step": 16438
    },
    {
      "epoch": 0.11169226365971534,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000973287542431079,
      "loss": 2.1513,
      "step": 16439
    },
    {
      "epoch": 0.1116990580063094,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009732840728556138,
      "loss": 2.3327,
      "step": 16440
    },
    {
      "epoch": 0.11170585235290346,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009732806030610236,
      "loss": 2.3021,
      "step": 16441
    },
    {
      "epoch": 0.11171264669949751,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009732771330473099,
      "loss": 2.223,
      "step": 16442
    },
    {
      "epoch": 0.11171944104609158,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009732736628144744,
      "loss": 2.081,
      "step": 16443
    },
    {
      "epoch": 0.11172623539268563,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009732701923625184,
      "loss": 2.1417,
      "step": 16444
    },
    {
      "epoch": 0.11173302973927969,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000973266721691444,
      "loss": 2.17,
      "step": 16445
    },
    {
      "epoch": 0.11173982408587374,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009732632508012524,
      "loss": 2.1747,
      "step": 16446
    },
    {
      "epoch": 0.1117466184324678,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009732597796919453,
      "loss": 2.1749,
      "step": 16447
    },
    {
      "epoch": 0.11175341277906187,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009732563083635244,
      "loss": 2.2142,
      "step": 16448
    },
    {
      "epoch": 0.11176020712565592,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009732528368159914,
      "loss": 2.1119,
      "step": 16449
    },
    {
      "epoch": 0.11176700147224998,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009732493650493476,
      "loss": 2.1988,
      "step": 16450
    },
    {
      "epoch": 0.11177379581884403,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009732458930635948,
      "loss": 2.1671,
      "step": 16451
    },
    {
      "epoch": 0.11178059016543809,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009732424208587345,
      "loss": 2.1433,
      "step": 16452
    },
    {
      "epoch": 0.11178738451203216,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009732389484347684,
      "loss": 2.2728,
      "step": 16453
    },
    {
      "epoch": 0.1117941788586262,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009732354757916982,
      "loss": 2.154,
      "step": 16454
    },
    {
      "epoch": 0.11180097320522027,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009732320029295254,
      "loss": 2.235,
      "step": 16455
    },
    {
      "epoch": 0.11180776755181432,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009732285298482515,
      "loss": 2.252,
      "step": 16456
    },
    {
      "epoch": 0.11181456189840838,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009732250565478782,
      "loss": 2.1636,
      "step": 16457
    },
    {
      "epoch": 0.11182135624500245,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009732215830284073,
      "loss": 2.1365,
      "step": 16458
    },
    {
      "epoch": 0.1118281505915965,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.00097321810928984,
      "loss": 2.3481,
      "step": 16459
    },
    {
      "epoch": 0.11183494493819056,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009732146353321784,
      "loss": 2.2121,
      "step": 16460
    },
    {
      "epoch": 0.11184173928478461,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009732111611554237,
      "loss": 2.2886,
      "step": 16461
    },
    {
      "epoch": 0.11184853363137867,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009732076867595776,
      "loss": 2.2186,
      "step": 16462
    },
    {
      "epoch": 0.11185532797797273,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009732042121446418,
      "loss": 2.1374,
      "step": 16463
    },
    {
      "epoch": 0.11186212232456678,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000973200737310618,
      "loss": 2.233,
      "step": 16464
    },
    {
      "epoch": 0.11186891667116085,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009731972622575076,
      "loss": 2.1956,
      "step": 16465
    },
    {
      "epoch": 0.1118757110177549,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009731937869853122,
      "loss": 2.2921,
      "step": 16466
    },
    {
      "epoch": 0.11188250536434896,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009731903114940337,
      "loss": 2.3089,
      "step": 16467
    },
    {
      "epoch": 0.11188929971094301,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009731868357836733,
      "loss": 2.2307,
      "step": 16468
    },
    {
      "epoch": 0.11189609405753707,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000973183359854233,
      "loss": 2.0696,
      "step": 16469
    },
    {
      "epoch": 0.11190288840413114,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009731798837057142,
      "loss": 2.1242,
      "step": 16470
    },
    {
      "epoch": 0.11190968275072519,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009731764073381184,
      "loss": 2.0921,
      "step": 16471
    },
    {
      "epoch": 0.11191647709731925,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009731729307514474,
      "loss": 2.2228,
      "step": 16472
    },
    {
      "epoch": 0.1119232714439133,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009731694539457028,
      "loss": 2.2967,
      "step": 16473
    },
    {
      "epoch": 0.11193006579050736,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009731659769208862,
      "loss": 2.2318,
      "step": 16474
    },
    {
      "epoch": 0.11193686013710143,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009731624996769992,
      "loss": 2.225,
      "step": 16475
    },
    {
      "epoch": 0.11194365448369548,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009731590222140431,
      "loss": 2.2269,
      "step": 16476
    },
    {
      "epoch": 0.11195044883028954,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009731555445320201,
      "loss": 2.0965,
      "step": 16477
    },
    {
      "epoch": 0.11195724317688359,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009731520666309315,
      "loss": 2.2091,
      "step": 16478
    },
    {
      "epoch": 0.11196403752347765,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009731485885107789,
      "loss": 2.345,
      "step": 16479
    },
    {
      "epoch": 0.11197083187007172,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009731451101715638,
      "loss": 2.3555,
      "step": 16480
    },
    {
      "epoch": 0.11197762621666577,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009731416316132881,
      "loss": 2.2694,
      "step": 16481
    },
    {
      "epoch": 0.11198442056325983,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009731381528359531,
      "loss": 2.1976,
      "step": 16482
    },
    {
      "epoch": 0.11199121490985388,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009731346738395606,
      "loss": 2.3296,
      "step": 16483
    },
    {
      "epoch": 0.11199800925644794,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009731311946241122,
      "loss": 2.1526,
      "step": 16484
    },
    {
      "epoch": 0.112004803603042,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009731277151896095,
      "loss": 2.1671,
      "step": 16485
    },
    {
      "epoch": 0.11201159794963605,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009731242355360541,
      "loss": 2.2033,
      "step": 16486
    },
    {
      "epoch": 0.11201839229623012,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009731207556634476,
      "loss": 2.2116,
      "step": 16487
    },
    {
      "epoch": 0.11202518664282417,
      "grad_norm": 0.181640625,
      "learning_rate": 0.0009731172755717915,
      "loss": 2.309,
      "step": 16488
    },
    {
      "epoch": 0.11203198098941823,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009731137952610875,
      "loss": 2.2849,
      "step": 16489
    },
    {
      "epoch": 0.11203877533601228,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009731103147313375,
      "loss": 2.212,
      "step": 16490
    },
    {
      "epoch": 0.11204556968260634,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009731068339825426,
      "loss": 2.2404,
      "step": 16491
    },
    {
      "epoch": 0.11205236402920041,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009731033530147048,
      "loss": 2.4385,
      "step": 16492
    },
    {
      "epoch": 0.11205915837579446,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009730998718278256,
      "loss": 2.2024,
      "step": 16493
    },
    {
      "epoch": 0.11206595272238852,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009730963904219064,
      "loss": 2.212,
      "step": 16494
    },
    {
      "epoch": 0.11207274706898257,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009730929087969491,
      "loss": 2.2721,
      "step": 16495
    },
    {
      "epoch": 0.11207954141557663,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009730894269529553,
      "loss": 2.2826,
      "step": 16496
    },
    {
      "epoch": 0.1120863357621707,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009730859448899264,
      "loss": 2.1496,
      "step": 16497
    },
    {
      "epoch": 0.11209313010876475,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009730824626078641,
      "loss": 2.1508,
      "step": 16498
    },
    {
      "epoch": 0.11209992445535881,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009730789801067701,
      "loss": 2.1675,
      "step": 16499
    },
    {
      "epoch": 0.11210671880195286,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000973075497386646,
      "loss": 2.2351,
      "step": 16500
    },
    {
      "epoch": 0.11211351314854692,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009730720144474934,
      "loss": 2.1314,
      "step": 16501
    },
    {
      "epoch": 0.11212030749514099,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009730685312893138,
      "loss": 2.133,
      "step": 16502
    },
    {
      "epoch": 0.11212710184173504,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009730650479121088,
      "loss": 2.1777,
      "step": 16503
    },
    {
      "epoch": 0.1121338961883291,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009730615643158803,
      "loss": 2.1495,
      "step": 16504
    },
    {
      "epoch": 0.11214069053492315,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009730580805006296,
      "loss": 2.2651,
      "step": 16505
    },
    {
      "epoch": 0.11214748488151721,
      "grad_norm": 0.125,
      "learning_rate": 0.0009730545964663584,
      "loss": 2.1138,
      "step": 16506
    },
    {
      "epoch": 0.11215427922811128,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009730511122130685,
      "loss": 2.2654,
      "step": 16507
    },
    {
      "epoch": 0.11216107357470533,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009730476277407612,
      "loss": 2.2527,
      "step": 16508
    },
    {
      "epoch": 0.11216786792129939,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009730441430494383,
      "loss": 2.0559,
      "step": 16509
    },
    {
      "epoch": 0.11217466226789344,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009730406581391015,
      "loss": 2.2129,
      "step": 16510
    },
    {
      "epoch": 0.1121814566144875,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009730371730097523,
      "loss": 2.1559,
      "step": 16511
    },
    {
      "epoch": 0.11218825096108156,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009730336876613921,
      "loss": 2.177,
      "step": 16512
    },
    {
      "epoch": 0.11219504530767561,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009730302020940229,
      "loss": 2.3174,
      "step": 16513
    },
    {
      "epoch": 0.11220183965426968,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000973026716307646,
      "loss": 2.333,
      "step": 16514
    },
    {
      "epoch": 0.11220863400086373,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009730232303022632,
      "loss": 2.1985,
      "step": 16515
    },
    {
      "epoch": 0.11221542834745779,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009730197440778763,
      "loss": 2.0496,
      "step": 16516
    },
    {
      "epoch": 0.11222222269405184,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009730162576344864,
      "loss": 2.2854,
      "step": 16517
    },
    {
      "epoch": 0.1122290170406459,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009730127709720955,
      "loss": 2.3134,
      "step": 16518
    },
    {
      "epoch": 0.11223581138723997,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009730092840907052,
      "loss": 2.3122,
      "step": 16519
    },
    {
      "epoch": 0.11224260573383402,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009730057969903169,
      "loss": 2.209,
      "step": 16520
    },
    {
      "epoch": 0.11224940008042808,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009730023096709324,
      "loss": 2.1227,
      "step": 16521
    },
    {
      "epoch": 0.11225619442702213,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009729988221325532,
      "loss": 2.1268,
      "step": 16522
    },
    {
      "epoch": 0.11226298877361619,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009729953343751809,
      "loss": 2.2794,
      "step": 16523
    },
    {
      "epoch": 0.11226978312021026,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009729918463988173,
      "loss": 2.2331,
      "step": 16524
    },
    {
      "epoch": 0.1122765774668043,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009729883582034639,
      "loss": 2.2774,
      "step": 16525
    },
    {
      "epoch": 0.11228337181339837,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009729848697891222,
      "loss": 2.1008,
      "step": 16526
    },
    {
      "epoch": 0.11229016615999242,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000972981381155794,
      "loss": 2.1183,
      "step": 16527
    },
    {
      "epoch": 0.11229696050658648,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009729778923034809,
      "loss": 2.112,
      "step": 16528
    },
    {
      "epoch": 0.11230375485318055,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009729744032321844,
      "loss": 2.0853,
      "step": 16529
    },
    {
      "epoch": 0.1123105491997746,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009729709139419063,
      "loss": 2.0283,
      "step": 16530
    },
    {
      "epoch": 0.11231734354636866,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009729674244326478,
      "loss": 2.2443,
      "step": 16531
    },
    {
      "epoch": 0.11232413789296271,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009729639347044111,
      "loss": 1.9834,
      "step": 16532
    },
    {
      "epoch": 0.11233093223955677,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009729604447571974,
      "loss": 2.3276,
      "step": 16533
    },
    {
      "epoch": 0.11233772658615084,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009729569545910084,
      "loss": 2.2636,
      "step": 16534
    },
    {
      "epoch": 0.11234452093274488,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009729534642058458,
      "loss": 2.0861,
      "step": 16535
    },
    {
      "epoch": 0.11235131527933895,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009729499736017112,
      "loss": 2.0997,
      "step": 16536
    },
    {
      "epoch": 0.112358109625933,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009729464827786061,
      "loss": 2.1445,
      "step": 16537
    },
    {
      "epoch": 0.11236490397252706,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009729429917365323,
      "loss": 2.1544,
      "step": 16538
    },
    {
      "epoch": 0.11237169831912111,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009729395004754912,
      "loss": 2.2422,
      "step": 16539
    },
    {
      "epoch": 0.11237849266571517,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009729360089954846,
      "loss": 2.1529,
      "step": 16540
    },
    {
      "epoch": 0.11238528701230924,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000972932517296514,
      "loss": 2.1685,
      "step": 16541
    },
    {
      "epoch": 0.11239208135890329,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009729290253785812,
      "loss": 2.3788,
      "step": 16542
    },
    {
      "epoch": 0.11239887570549735,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009729255332416875,
      "loss": 2.1393,
      "step": 16543
    },
    {
      "epoch": 0.1124056700520914,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009729220408858348,
      "loss": 2.2785,
      "step": 16544
    },
    {
      "epoch": 0.11241246439868546,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009729185483110245,
      "loss": 2.1012,
      "step": 16545
    },
    {
      "epoch": 0.11241925874527953,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009729150555172585,
      "loss": 2.2567,
      "step": 16546
    },
    {
      "epoch": 0.11242605309187358,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009729115625045382,
      "loss": 2.2969,
      "step": 16547
    },
    {
      "epoch": 0.11243284743846764,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009729080692728653,
      "loss": 2.2543,
      "step": 16548
    },
    {
      "epoch": 0.11243964178506169,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009729045758222414,
      "loss": 2.2734,
      "step": 16549
    },
    {
      "epoch": 0.11244643613165575,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000972901082152668,
      "loss": 2.2231,
      "step": 16550
    },
    {
      "epoch": 0.11245323047824982,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009728975882641468,
      "loss": 2.1928,
      "step": 16551
    },
    {
      "epoch": 0.11246002482484387,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009728940941566795,
      "loss": 2.3009,
      "step": 16552
    },
    {
      "epoch": 0.11246681917143793,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009728905998302677,
      "loss": 2.3484,
      "step": 16553
    },
    {
      "epoch": 0.11247361351803198,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000972887105284913,
      "loss": 2.0969,
      "step": 16554
    },
    {
      "epoch": 0.11248040786462604,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009728836105206169,
      "loss": 2.2867,
      "step": 16555
    },
    {
      "epoch": 0.1124872022112201,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009728801155373812,
      "loss": 2.2383,
      "step": 16556
    },
    {
      "epoch": 0.11249399655781415,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009728766203352073,
      "loss": 2.2849,
      "step": 16557
    },
    {
      "epoch": 0.11250079090440822,
      "grad_norm": 0.125,
      "learning_rate": 0.0009728731249140969,
      "loss": 2.0212,
      "step": 16558
    },
    {
      "epoch": 0.11250758525100227,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009728696292740518,
      "loss": 2.2387,
      "step": 16559
    },
    {
      "epoch": 0.11251437959759633,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009728661334150734,
      "loss": 2.201,
      "step": 16560
    },
    {
      "epoch": 0.1125211739441904,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009728626373371636,
      "loss": 2.2332,
      "step": 16561
    },
    {
      "epoch": 0.11252796829078444,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009728591410403237,
      "loss": 2.2533,
      "step": 16562
    },
    {
      "epoch": 0.11253476263737851,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009728556445245554,
      "loss": 2.1218,
      "step": 16563
    },
    {
      "epoch": 0.11254155698397256,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009728521477898604,
      "loss": 2.2332,
      "step": 16564
    },
    {
      "epoch": 0.11254835133056662,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009728486508362402,
      "loss": 2.2298,
      "step": 16565
    },
    {
      "epoch": 0.11255514567716067,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009728451536636966,
      "loss": 2.2021,
      "step": 16566
    },
    {
      "epoch": 0.11256194002375473,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009728416562722311,
      "loss": 2.1852,
      "step": 16567
    },
    {
      "epoch": 0.1125687343703488,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009728381586618453,
      "loss": 2.2321,
      "step": 16568
    },
    {
      "epoch": 0.11257552871694285,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009728346608325408,
      "loss": 2.2688,
      "step": 16569
    },
    {
      "epoch": 0.11258232306353691,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009728311627843194,
      "loss": 2.306,
      "step": 16570
    },
    {
      "epoch": 0.11258911741013096,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009728276645171825,
      "loss": 2.2703,
      "step": 16571
    },
    {
      "epoch": 0.11259591175672502,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009728241660311317,
      "loss": 2.2374,
      "step": 16572
    },
    {
      "epoch": 0.11260270610331909,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009728206673261689,
      "loss": 2.2872,
      "step": 16573
    },
    {
      "epoch": 0.11260950044991314,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009728171684022955,
      "loss": 2.2123,
      "step": 16574
    },
    {
      "epoch": 0.1126162947965072,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009728136692595131,
      "loss": 2.13,
      "step": 16575
    },
    {
      "epoch": 0.11262308914310125,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009728101698978235,
      "loss": 2.2695,
      "step": 16576
    },
    {
      "epoch": 0.11262988348969531,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009728066703172283,
      "loss": 2.2167,
      "step": 16577
    },
    {
      "epoch": 0.11263667783628938,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009728031705177289,
      "loss": 2.1891,
      "step": 16578
    },
    {
      "epoch": 0.11264347218288343,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009727996704993272,
      "loss": 2.114,
      "step": 16579
    },
    {
      "epoch": 0.11265026652947749,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009727961702620245,
      "loss": 2.2386,
      "step": 16580
    },
    {
      "epoch": 0.11265706087607154,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009727926698058226,
      "loss": 2.2685,
      "step": 16581
    },
    {
      "epoch": 0.1126638552226656,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009727891691307232,
      "loss": 2.2145,
      "step": 16582
    },
    {
      "epoch": 0.11267064956925966,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009727856682367278,
      "loss": 2.1742,
      "step": 16583
    },
    {
      "epoch": 0.11267744391585371,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009727821671238379,
      "loss": 2.2228,
      "step": 16584
    },
    {
      "epoch": 0.11268423826244778,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009727786657920554,
      "loss": 2.1198,
      "step": 16585
    },
    {
      "epoch": 0.11269103260904183,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009727751642413819,
      "loss": 2.1464,
      "step": 16586
    },
    {
      "epoch": 0.11269782695563589,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009727716624718188,
      "loss": 2.1853,
      "step": 16587
    },
    {
      "epoch": 0.11270462130222994,
      "grad_norm": 0.146484375,
      "learning_rate": 0.000972768160483368,
      "loss": 2.2548,
      "step": 16588
    },
    {
      "epoch": 0.112711415648824,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009727646582760308,
      "loss": 2.0608,
      "step": 16589
    },
    {
      "epoch": 0.11271820999541807,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009727611558498089,
      "loss": 2.1242,
      "step": 16590
    },
    {
      "epoch": 0.11272500434201212,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009727576532047042,
      "loss": 2.2594,
      "step": 16591
    },
    {
      "epoch": 0.11273179868860618,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000972754150340718,
      "loss": 2.2411,
      "step": 16592
    },
    {
      "epoch": 0.11273859303520023,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009727506472578522,
      "loss": 2.126,
      "step": 16593
    },
    {
      "epoch": 0.1127453873817943,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009727471439561081,
      "loss": 2.1522,
      "step": 16594
    },
    {
      "epoch": 0.11275218172838836,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009727436404354876,
      "loss": 2.3197,
      "step": 16595
    },
    {
      "epoch": 0.1127589760749824,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009727401366959921,
      "loss": 2.2456,
      "step": 16596
    },
    {
      "epoch": 0.11276577042157647,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009727366327376235,
      "loss": 2.174,
      "step": 16597
    },
    {
      "epoch": 0.11277256476817052,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009727331285603831,
      "loss": 2.2544,
      "step": 16598
    },
    {
      "epoch": 0.11277935911476458,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009727296241642728,
      "loss": 2.1727,
      "step": 16599
    },
    {
      "epoch": 0.11278615346135865,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000972726119549294,
      "loss": 2.1084,
      "step": 16600
    },
    {
      "epoch": 0.1127929478079527,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009727226147154485,
      "loss": 2.2724,
      "step": 16601
    },
    {
      "epoch": 0.11279974215454676,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009727191096627379,
      "loss": 2.1685,
      "step": 16602
    },
    {
      "epoch": 0.11280653650114081,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009727156043911637,
      "loss": 2.4093,
      "step": 16603
    },
    {
      "epoch": 0.11281333084773487,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009727120989007277,
      "loss": 2.1498,
      "step": 16604
    },
    {
      "epoch": 0.11282012519432894,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009727085931914313,
      "loss": 2.2227,
      "step": 16605
    },
    {
      "epoch": 0.11282691954092298,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009727050872632761,
      "loss": 2.2702,
      "step": 16606
    },
    {
      "epoch": 0.11283371388751705,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000972701581116264,
      "loss": 2.1466,
      "step": 16607
    },
    {
      "epoch": 0.1128405082341111,
      "grad_norm": 0.125,
      "learning_rate": 0.0009726980747503967,
      "loss": 1.9391,
      "step": 16608
    },
    {
      "epoch": 0.11284730258070516,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009726945681656754,
      "loss": 2.1405,
      "step": 16609
    },
    {
      "epoch": 0.11285409692729921,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009726910613621018,
      "loss": 2.2713,
      "step": 16610
    },
    {
      "epoch": 0.11286089127389327,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000972687554339678,
      "loss": 2.1267,
      "step": 16611
    },
    {
      "epoch": 0.11286768562048734,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000972684047098405,
      "loss": 2.0883,
      "step": 16612
    },
    {
      "epoch": 0.11287447996708139,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009726805396382848,
      "loss": 2.1363,
      "step": 16613
    },
    {
      "epoch": 0.11288127431367545,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009726770319593189,
      "loss": 2.2065,
      "step": 16614
    },
    {
      "epoch": 0.1128880686602695,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000972673524061509,
      "loss": 2.1128,
      "step": 16615
    },
    {
      "epoch": 0.11289486300686356,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009726700159448567,
      "loss": 2.1758,
      "step": 16616
    },
    {
      "epoch": 0.11290165735345763,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009726665076093635,
      "loss": 2.1433,
      "step": 16617
    },
    {
      "epoch": 0.11290845170005168,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009726629990550312,
      "loss": 2.1876,
      "step": 16618
    },
    {
      "epoch": 0.11291524604664574,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009726594902818613,
      "loss": 2.222,
      "step": 16619
    },
    {
      "epoch": 0.11292204039323979,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009726559812898556,
      "loss": 2.1631,
      "step": 16620
    },
    {
      "epoch": 0.11292883473983385,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009726524720790155,
      "loss": 2.197,
      "step": 16621
    },
    {
      "epoch": 0.11293562908642792,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009726489626493426,
      "loss": 2.0831,
      "step": 16622
    },
    {
      "epoch": 0.11294242343302197,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009726454530008388,
      "loss": 2.33,
      "step": 16623
    },
    {
      "epoch": 0.11294921777961603,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009726419431335056,
      "loss": 2.2362,
      "step": 16624
    },
    {
      "epoch": 0.11295601212621008,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009726384330473446,
      "loss": 2.3986,
      "step": 16625
    },
    {
      "epoch": 0.11296280647280414,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009726349227423575,
      "loss": 2.0431,
      "step": 16626
    },
    {
      "epoch": 0.1129696008193982,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009726314122185456,
      "loss": 2.17,
      "step": 16627
    },
    {
      "epoch": 0.11297639516599225,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000972627901475911,
      "loss": 2.1394,
      "step": 16628
    },
    {
      "epoch": 0.11298318951258632,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009726243905144549,
      "loss": 2.1167,
      "step": 16629
    },
    {
      "epoch": 0.11298998385918037,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009726208793341794,
      "loss": 2.2381,
      "step": 16630
    },
    {
      "epoch": 0.11299677820577443,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009726173679350856,
      "loss": 2.1827,
      "step": 16631
    },
    {
      "epoch": 0.1130035725523685,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009726138563171754,
      "loss": 2.1232,
      "step": 16632
    },
    {
      "epoch": 0.11301036689896254,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009726103444804505,
      "loss": 2.2846,
      "step": 16633
    },
    {
      "epoch": 0.11301716124555661,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009726068324249123,
      "loss": 2.0138,
      "step": 16634
    },
    {
      "epoch": 0.11302395559215066,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009726033201505626,
      "loss": 2.2601,
      "step": 16635
    },
    {
      "epoch": 0.11303074993874472,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009725998076574032,
      "loss": 2.1865,
      "step": 16636
    },
    {
      "epoch": 0.11303754428533877,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009725962949454352,
      "loss": 2.1339,
      "step": 16637
    },
    {
      "epoch": 0.11304433863193283,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009725927820146607,
      "loss": 2.1638,
      "step": 16638
    },
    {
      "epoch": 0.1130511329785269,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009725892688650812,
      "loss": 2.213,
      "step": 16639
    },
    {
      "epoch": 0.11305792732512095,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009725857554966982,
      "loss": 2.311,
      "step": 16640
    },
    {
      "epoch": 0.11306472167171501,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009725822419095134,
      "loss": 2.2292,
      "step": 16641
    },
    {
      "epoch": 0.11307151601830906,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009725787281035284,
      "loss": 2.2137,
      "step": 16642
    },
    {
      "epoch": 0.11307831036490312,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009725752140787449,
      "loss": 2.0238,
      "step": 16643
    },
    {
      "epoch": 0.11308510471149719,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009725716998351646,
      "loss": 2.2139,
      "step": 16644
    },
    {
      "epoch": 0.11309189905809124,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000972568185372789,
      "loss": 2.1709,
      "step": 16645
    },
    {
      "epoch": 0.1130986934046853,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009725646706916197,
      "loss": 2.1149,
      "step": 16646
    },
    {
      "epoch": 0.11310548775127935,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009725611557916583,
      "loss": 2.0629,
      "step": 16647
    },
    {
      "epoch": 0.11311228209787341,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009725576406729066,
      "loss": 2.1961,
      "step": 16648
    },
    {
      "epoch": 0.11311907644446748,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000972554125335366,
      "loss": 2.3596,
      "step": 16649
    },
    {
      "epoch": 0.11312587079106153,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009725506097790384,
      "loss": 2.1996,
      "step": 16650
    },
    {
      "epoch": 0.11313266513765559,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009725470940039253,
      "loss": 2.1733,
      "step": 16651
    },
    {
      "epoch": 0.11313945948424964,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009725435780100283,
      "loss": 2.1373,
      "step": 16652
    },
    {
      "epoch": 0.1131462538308437,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000972540061797349,
      "loss": 2.3136,
      "step": 16653
    },
    {
      "epoch": 0.11315304817743776,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.000972536545365889,
      "loss": 2.2218,
      "step": 16654
    },
    {
      "epoch": 0.11315984252403181,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009725330287156501,
      "loss": 2.1903,
      "step": 16655
    },
    {
      "epoch": 0.11316663687062588,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009725295118466337,
      "loss": 2.3081,
      "step": 16656
    },
    {
      "epoch": 0.11317343121721993,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009725259947588417,
      "loss": 2.1047,
      "step": 16657
    },
    {
      "epoch": 0.11318022556381399,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009725224774522756,
      "loss": 2.3208,
      "step": 16658
    },
    {
      "epoch": 0.11318701991040804,
      "grad_norm": 0.1181640625,
      "learning_rate": 0.0009725189599269369,
      "loss": 2.2024,
      "step": 16659
    },
    {
      "epoch": 0.1131938142570021,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009725154421828274,
      "loss": 2.2558,
      "step": 16660
    },
    {
      "epoch": 0.11320060860359617,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009725119242199486,
      "loss": 2.0133,
      "step": 16661
    },
    {
      "epoch": 0.11320740295019022,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009725084060383022,
      "loss": 2.3711,
      "step": 16662
    },
    {
      "epoch": 0.11321419729678428,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.00097250488763789,
      "loss": 2.0545,
      "step": 16663
    },
    {
      "epoch": 0.11322099164337833,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009725013690187132,
      "loss": 2.1352,
      "step": 16664
    },
    {
      "epoch": 0.1132277859899724,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009724978501807739,
      "loss": 2.3176,
      "step": 16665
    },
    {
      "epoch": 0.11323458033656646,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009724943311240733,
      "loss": 2.0682,
      "step": 16666
    },
    {
      "epoch": 0.1132413746831605,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009724908118486134,
      "loss": 2.1646,
      "step": 16667
    },
    {
      "epoch": 0.11324816902975457,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009724872923543958,
      "loss": 2.1176,
      "step": 16668
    },
    {
      "epoch": 0.11325496337634862,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009724837726414218,
      "loss": 2.2143,
      "step": 16669
    },
    {
      "epoch": 0.11326175772294268,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009724802527096931,
      "loss": 2.1513,
      "step": 16670
    },
    {
      "epoch": 0.11326855206953675,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009724767325592116,
      "loss": 2.1377,
      "step": 16671
    },
    {
      "epoch": 0.1132753464161308,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009724732121899789,
      "loss": 2.204,
      "step": 16672
    },
    {
      "epoch": 0.11328214076272486,
      "grad_norm": 0.125,
      "learning_rate": 0.0009724696916019964,
      "loss": 2.1103,
      "step": 16673
    },
    {
      "epoch": 0.11328893510931891,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009724661707952658,
      "loss": 2.1667,
      "step": 16674
    },
    {
      "epoch": 0.11329572945591297,
      "grad_norm": 0.125,
      "learning_rate": 0.0009724626497697889,
      "loss": 2.0562,
      "step": 16675
    },
    {
      "epoch": 0.11330252380250704,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009724591285255672,
      "loss": 2.1318,
      "step": 16676
    },
    {
      "epoch": 0.11330931814910108,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009724556070626022,
      "loss": 2.163,
      "step": 16677
    },
    {
      "epoch": 0.11331611249569515,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009724520853808958,
      "loss": 2.2479,
      "step": 16678
    },
    {
      "epoch": 0.1133229068422892,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009724485634804495,
      "loss": 2.068,
      "step": 16679
    },
    {
      "epoch": 0.11332970118888326,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009724450413612648,
      "loss": 2.0899,
      "step": 16680
    },
    {
      "epoch": 0.11333649553547732,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009724415190233436,
      "loss": 2.0391,
      "step": 16681
    },
    {
      "epoch": 0.11334328988207137,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009724379964666874,
      "loss": 2.3289,
      "step": 16682
    },
    {
      "epoch": 0.11335008422866544,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009724344736912978,
      "loss": 2.1383,
      "step": 16683
    },
    {
      "epoch": 0.11335687857525949,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009724309506971765,
      "loss": 2.052,
      "step": 16684
    },
    {
      "epoch": 0.11336367292185355,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.000972427427484325,
      "loss": 2.4132,
      "step": 16685
    },
    {
      "epoch": 0.1133704672684476,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000972423904052745,
      "loss": 2.2246,
      "step": 16686
    },
    {
      "epoch": 0.11337726161504166,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009724203804024383,
      "loss": 2.2208,
      "step": 16687
    },
    {
      "epoch": 0.11338405596163573,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009724168565334062,
      "loss": 2.0253,
      "step": 16688
    },
    {
      "epoch": 0.11339085030822978,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009724133324456505,
      "loss": 2.226,
      "step": 16689
    },
    {
      "epoch": 0.11339764465482384,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000972409808139173,
      "loss": 2.1288,
      "step": 16690
    },
    {
      "epoch": 0.11340443900141789,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009724062836139751,
      "loss": 2.1616,
      "step": 16691
    },
    {
      "epoch": 0.11341123334801195,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009724027588700586,
      "loss": 2.1191,
      "step": 16692
    },
    {
      "epoch": 0.11341802769460602,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009723992339074248,
      "loss": 2.1212,
      "step": 16693
    },
    {
      "epoch": 0.11342482204120007,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009723957087260758,
      "loss": 2.3036,
      "step": 16694
    },
    {
      "epoch": 0.11343161638779413,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000972392183326013,
      "loss": 2.0913,
      "step": 16695
    },
    {
      "epoch": 0.11343841073438818,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009723886577072379,
      "loss": 2.1965,
      "step": 16696
    },
    {
      "epoch": 0.11344520508098224,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009723851318697523,
      "loss": 2.054,
      "step": 16697
    },
    {
      "epoch": 0.1134519994275763,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009723816058135578,
      "loss": 2.239,
      "step": 16698
    },
    {
      "epoch": 0.11345879377417036,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009723780795386561,
      "loss": 2.1335,
      "step": 16699
    },
    {
      "epoch": 0.11346558812076442,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009723745530450487,
      "loss": 2.2187,
      "step": 16700
    },
    {
      "epoch": 0.11347238246735847,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009723710263327372,
      "loss": 2.1571,
      "step": 16701
    },
    {
      "epoch": 0.11347917681395253,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009723674994017236,
      "loss": 2.2294,
      "step": 16702
    },
    {
      "epoch": 0.1134859711605466,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000972363972252009,
      "loss": 2.2345,
      "step": 16703
    },
    {
      "epoch": 0.11349276550714064,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009723604448835955,
      "loss": 2.2812,
      "step": 16704
    },
    {
      "epoch": 0.11349955985373471,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009723569172964844,
      "loss": 2.2229,
      "step": 16705
    },
    {
      "epoch": 0.11350635420032876,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009723533894906776,
      "loss": 2.1223,
      "step": 16706
    },
    {
      "epoch": 0.11351314854692282,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009723498614661763,
      "loss": 2.1035,
      "step": 16707
    },
    {
      "epoch": 0.11351994289351687,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009723463332229828,
      "loss": 2.1504,
      "step": 16708
    },
    {
      "epoch": 0.11352673724011093,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000972342804761098,
      "loss": 2.2181,
      "step": 16709
    },
    {
      "epoch": 0.113533531586705,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009723392760805241,
      "loss": 2.1804,
      "step": 16710
    },
    {
      "epoch": 0.11354032593329905,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009723357471812626,
      "loss": 2.2557,
      "step": 16711
    },
    {
      "epoch": 0.11354712027989311,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009723322180633149,
      "loss": 2.0949,
      "step": 16712
    },
    {
      "epoch": 0.11355391462648716,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000972328688726683,
      "loss": 2.1413,
      "step": 16713
    },
    {
      "epoch": 0.11356070897308122,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009723251591713683,
      "loss": 2.1753,
      "step": 16714
    },
    {
      "epoch": 0.11356750331967529,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009723216293973723,
      "loss": 2.1918,
      "step": 16715
    },
    {
      "epoch": 0.11357429766626934,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009723180994046967,
      "loss": 2.224,
      "step": 16716
    },
    {
      "epoch": 0.1135810920128634,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009723145691933434,
      "loss": 2.2727,
      "step": 16717
    },
    {
      "epoch": 0.11358788635945745,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000972311038763314,
      "loss": 2.1063,
      "step": 16718
    },
    {
      "epoch": 0.11359468070605151,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009723075081146098,
      "loss": 2.2397,
      "step": 16719
    },
    {
      "epoch": 0.11360147505264558,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009723039772472328,
      "loss": 2.1993,
      "step": 16720
    },
    {
      "epoch": 0.11360826939923963,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009723004461611844,
      "loss": 2.0067,
      "step": 16721
    },
    {
      "epoch": 0.11361506374583369,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009722969148564662,
      "loss": 2.1869,
      "step": 16722
    },
    {
      "epoch": 0.11362185809242774,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009722933833330801,
      "loss": 2.1827,
      "step": 16723
    },
    {
      "epoch": 0.1136286524390218,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009722898515910275,
      "loss": 2.1935,
      "step": 16724
    },
    {
      "epoch": 0.11363544678561587,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009722863196303102,
      "loss": 2.1976,
      "step": 16725
    },
    {
      "epoch": 0.11364224113220991,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009722827874509296,
      "loss": 2.2523,
      "step": 16726
    },
    {
      "epoch": 0.11364903547880398,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009722792550528876,
      "loss": 2.1376,
      "step": 16727
    },
    {
      "epoch": 0.11365582982539803,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009722757224361856,
      "loss": 2.0117,
      "step": 16728
    },
    {
      "epoch": 0.11366262417199209,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009722721896008255,
      "loss": 2.2878,
      "step": 16729
    },
    {
      "epoch": 0.11366941851858614,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009722686565468089,
      "loss": 2.1169,
      "step": 16730
    },
    {
      "epoch": 0.1136762128651802,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000972265123274137,
      "loss": 2.0585,
      "step": 16731
    },
    {
      "epoch": 0.11368300721177427,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000972261589782812,
      "loss": 2.18,
      "step": 16732
    },
    {
      "epoch": 0.11368980155836832,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009722580560728351,
      "loss": 2.2161,
      "step": 16733
    },
    {
      "epoch": 0.11369659590496238,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009722545221442083,
      "loss": 2.1055,
      "step": 16734
    },
    {
      "epoch": 0.11370339025155643,
      "grad_norm": 0.1171875,
      "learning_rate": 0.000972250987996933,
      "loss": 2.0232,
      "step": 16735
    },
    {
      "epoch": 0.1137101845981505,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000972247453631011,
      "loss": 2.0965,
      "step": 16736
    },
    {
      "epoch": 0.11371697894474456,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009722439190464438,
      "loss": 2.0807,
      "step": 16737
    },
    {
      "epoch": 0.1137237732913386,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000972240384243233,
      "loss": 2.1183,
      "step": 16738
    },
    {
      "epoch": 0.11373056763793267,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009722368492213804,
      "loss": 2.1104,
      "step": 16739
    },
    {
      "epoch": 0.11373736198452672,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009722333139808875,
      "loss": 2.0793,
      "step": 16740
    },
    {
      "epoch": 0.11374415633112078,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000972229778521756,
      "loss": 2.1814,
      "step": 16741
    },
    {
      "epoch": 0.11375095067771485,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009722262428439876,
      "loss": 2.2576,
      "step": 16742
    },
    {
      "epoch": 0.1137577450243089,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009722227069475837,
      "loss": 2.1406,
      "step": 16743
    },
    {
      "epoch": 0.11376453937090296,
      "grad_norm": 0.125,
      "learning_rate": 0.0009722191708325463,
      "loss": 2.1757,
      "step": 16744
    },
    {
      "epoch": 0.11377133371749701,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009722156344988767,
      "loss": 2.1137,
      "step": 16745
    },
    {
      "epoch": 0.11377812806409107,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009722120979465767,
      "loss": 2.0714,
      "step": 16746
    },
    {
      "epoch": 0.11378492241068514,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000972208561175648,
      "loss": 2.1643,
      "step": 16747
    },
    {
      "epoch": 0.11379171675727918,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000972205024186092,
      "loss": 2.1152,
      "step": 16748
    },
    {
      "epoch": 0.11379851110387325,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009722014869779106,
      "loss": 2.1864,
      "step": 16749
    },
    {
      "epoch": 0.1138053054504673,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009721979495511053,
      "loss": 2.2439,
      "step": 16750
    },
    {
      "epoch": 0.11381209979706136,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009721944119056778,
      "loss": 2.2053,
      "step": 16751
    },
    {
      "epoch": 0.11381889414365542,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009721908740416296,
      "loss": 2.218,
      "step": 16752
    },
    {
      "epoch": 0.11382568849024947,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009721873359589625,
      "loss": 2.2416,
      "step": 16753
    },
    {
      "epoch": 0.11383248283684354,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000972183797657678,
      "loss": 2.0721,
      "step": 16754
    },
    {
      "epoch": 0.11383927718343759,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000972180259137778,
      "loss": 2.2152,
      "step": 16755
    },
    {
      "epoch": 0.11384607153003165,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009721767203992637,
      "loss": 2.2191,
      "step": 16756
    },
    {
      "epoch": 0.1138528658766257,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009721731814421371,
      "loss": 2.1678,
      "step": 16757
    },
    {
      "epoch": 0.11385966022321976,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009721696422663999,
      "loss": 2.0797,
      "step": 16758
    },
    {
      "epoch": 0.11386645456981383,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009721661028720534,
      "loss": 2.1367,
      "step": 16759
    },
    {
      "epoch": 0.11387324891640788,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009721625632590993,
      "loss": 2.1201,
      "step": 16760
    },
    {
      "epoch": 0.11388004326300194,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009721590234275396,
      "loss": 2.256,
      "step": 16761
    },
    {
      "epoch": 0.11388683760959599,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009721554833773756,
      "loss": 2.2202,
      "step": 16762
    },
    {
      "epoch": 0.11389363195619005,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000972151943108609,
      "loss": 2.2146,
      "step": 16763
    },
    {
      "epoch": 0.11390042630278412,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009721484026212415,
      "loss": 2.4155,
      "step": 16764
    },
    {
      "epoch": 0.11390722064937817,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009721448619152746,
      "loss": 2.2472,
      "step": 16765
    },
    {
      "epoch": 0.11391401499597223,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009721413209907101,
      "loss": 2.3482,
      "step": 16766
    },
    {
      "epoch": 0.11392080934256628,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009721377798475495,
      "loss": 2.1944,
      "step": 16767
    },
    {
      "epoch": 0.11392760368916034,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009721342384857947,
      "loss": 2.1175,
      "step": 16768
    },
    {
      "epoch": 0.1139343980357544,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009721306969054471,
      "loss": 2.1529,
      "step": 16769
    },
    {
      "epoch": 0.11394119238234846,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009721271551065083,
      "loss": 2.205,
      "step": 16770
    },
    {
      "epoch": 0.11394798672894252,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009721236130889802,
      "loss": 2.246,
      "step": 16771
    },
    {
      "epoch": 0.11395478107553657,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009721200708528641,
      "loss": 2.0659,
      "step": 16772
    },
    {
      "epoch": 0.11396157542213063,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009721165283981619,
      "loss": 2.3108,
      "step": 16773
    },
    {
      "epoch": 0.1139683697687247,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009721129857248751,
      "loss": 2.242,
      "step": 16774
    },
    {
      "epoch": 0.11397516411531874,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009721094428330055,
      "loss": 2.2211,
      "step": 16775
    },
    {
      "epoch": 0.11398195846191281,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009721058997225546,
      "loss": 2.2612,
      "step": 16776
    },
    {
      "epoch": 0.11398875280850686,
      "grad_norm": 0.140625,
      "learning_rate": 0.000972102356393524,
      "loss": 2.1644,
      "step": 16777
    },
    {
      "epoch": 0.11399554715510092,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009720988128459156,
      "loss": 2.3166,
      "step": 16778
    },
    {
      "epoch": 0.11400234150169497,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009720952690797307,
      "loss": 2.3402,
      "step": 16779
    },
    {
      "epoch": 0.11400913584828903,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009720917250949713,
      "loss": 2.1844,
      "step": 16780
    },
    {
      "epoch": 0.1140159301948831,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009720881808916387,
      "loss": 2.1504,
      "step": 16781
    },
    {
      "epoch": 0.11402272454147715,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009720846364697347,
      "loss": 2.1292,
      "step": 16782
    },
    {
      "epoch": 0.11402951888807121,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009720810918292608,
      "loss": 2.1674,
      "step": 16783
    },
    {
      "epoch": 0.11403631323466526,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000972077546970219,
      "loss": 2.3646,
      "step": 16784
    },
    {
      "epoch": 0.11404310758125932,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009720740018926105,
      "loss": 2.2392,
      "step": 16785
    },
    {
      "epoch": 0.11404990192785339,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009720704565964373,
      "loss": 2.2413,
      "step": 16786
    },
    {
      "epoch": 0.11405669627444744,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009720669110817009,
      "loss": 2.3265,
      "step": 16787
    },
    {
      "epoch": 0.1140634906210415,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009720633653484028,
      "loss": 2.1456,
      "step": 16788
    },
    {
      "epoch": 0.11407028496763555,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009720598193965449,
      "loss": 2.2475,
      "step": 16789
    },
    {
      "epoch": 0.11407707931422961,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009720562732261287,
      "loss": 2.1787,
      "step": 16790
    },
    {
      "epoch": 0.11408387366082368,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009720527268371557,
      "loss": 2.1711,
      "step": 16791
    },
    {
      "epoch": 0.11409066800741773,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009720491802296279,
      "loss": 2.2696,
      "step": 16792
    },
    {
      "epoch": 0.11409746235401179,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009720456334035466,
      "loss": 2.2031,
      "step": 16793
    },
    {
      "epoch": 0.11410425670060584,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009720420863589136,
      "loss": 2.169,
      "step": 16794
    },
    {
      "epoch": 0.1141110510471999,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009720385390957306,
      "loss": 2.1425,
      "step": 16795
    },
    {
      "epoch": 0.11411784539379397,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009720349916139991,
      "loss": 2.137,
      "step": 16796
    },
    {
      "epoch": 0.11412463974038801,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009720314439137208,
      "loss": 2.1945,
      "step": 16797
    },
    {
      "epoch": 0.11413143408698208,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009720278959948974,
      "loss": 2.1213,
      "step": 16798
    },
    {
      "epoch": 0.11413822843357613,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009720243478575306,
      "loss": 2.0759,
      "step": 16799
    },
    {
      "epoch": 0.11414502278017019,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009720207995016217,
      "loss": 2.1893,
      "step": 16800
    },
    {
      "epoch": 0.11415181712676425,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009720172509271727,
      "loss": 2.219,
      "step": 16801
    },
    {
      "epoch": 0.1141586114733583,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009720137021341853,
      "loss": 2.0737,
      "step": 16802
    },
    {
      "epoch": 0.11416540581995237,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009720101531226607,
      "loss": 2.1668,
      "step": 16803
    },
    {
      "epoch": 0.11417220016654642,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009720066038926009,
      "loss": 2.1578,
      "step": 16804
    },
    {
      "epoch": 0.11417899451314048,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009720030544440075,
      "loss": 2.2308,
      "step": 16805
    },
    {
      "epoch": 0.11418578885973453,
      "grad_norm": 0.125,
      "learning_rate": 0.0009719995047768822,
      "loss": 2.0784,
      "step": 16806
    },
    {
      "epoch": 0.1141925832063286,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009719959548912262,
      "loss": 2.2159,
      "step": 16807
    },
    {
      "epoch": 0.11419937755292266,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009719924047870417,
      "loss": 2.3013,
      "step": 16808
    },
    {
      "epoch": 0.1142061718995167,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009719888544643303,
      "loss": 2.0622,
      "step": 16809
    },
    {
      "epoch": 0.11421296624611077,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009719853039230933,
      "loss": 2.0924,
      "step": 16810
    },
    {
      "epoch": 0.11421976059270482,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009719817531633324,
      "loss": 2.146,
      "step": 16811
    },
    {
      "epoch": 0.11422655493929888,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009719782021850494,
      "loss": 2.1611,
      "step": 16812
    },
    {
      "epoch": 0.11423334928589295,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000971974650988246,
      "loss": 2.2591,
      "step": 16813
    },
    {
      "epoch": 0.114240143632487,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0009719710995729238,
      "loss": 2.0434,
      "step": 16814
    },
    {
      "epoch": 0.11424693797908106,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009719675479390843,
      "loss": 2.1306,
      "step": 16815
    },
    {
      "epoch": 0.11425373232567511,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009719639960867292,
      "loss": 2.0721,
      "step": 16816
    },
    {
      "epoch": 0.11426052667226917,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009719604440158602,
      "loss": 2.2254,
      "step": 16817
    },
    {
      "epoch": 0.11426732101886324,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000971956891726479,
      "loss": 2.3134,
      "step": 16818
    },
    {
      "epoch": 0.11427411536545728,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000971953339218587,
      "loss": 2.1422,
      "step": 16819
    },
    {
      "epoch": 0.11428090971205135,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009719497864921862,
      "loss": 2.1776,
      "step": 16820
    },
    {
      "epoch": 0.1142877040586454,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000971946233547278,
      "loss": 2.1925,
      "step": 16821
    },
    {
      "epoch": 0.11429449840523946,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009719426803838641,
      "loss": 2.1717,
      "step": 16822
    },
    {
      "epoch": 0.11430129275183352,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009719391270019462,
      "loss": 2.3501,
      "step": 16823
    },
    {
      "epoch": 0.11430808709842757,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009719355734015259,
      "loss": 2.0125,
      "step": 16824
    },
    {
      "epoch": 0.11431488144502164,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009719320195826046,
      "loss": 2.1291,
      "step": 16825
    },
    {
      "epoch": 0.11432167579161569,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009719284655451845,
      "loss": 2.1405,
      "step": 16826
    },
    {
      "epoch": 0.11432847013820975,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009719249112892667,
      "loss": 2.1412,
      "step": 16827
    },
    {
      "epoch": 0.1143352644848038,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009719213568148533,
      "loss": 2.2744,
      "step": 16828
    },
    {
      "epoch": 0.11434205883139786,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009719178021219455,
      "loss": 2.118,
      "step": 16829
    },
    {
      "epoch": 0.11434885317799193,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009719142472105454,
      "loss": 2.1601,
      "step": 16830
    },
    {
      "epoch": 0.11435564752458598,
      "grad_norm": 0.125,
      "learning_rate": 0.0009719106920806542,
      "loss": 2.0887,
      "step": 16831
    },
    {
      "epoch": 0.11436244187118004,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009719071367322738,
      "loss": 2.2059,
      "step": 16832
    },
    {
      "epoch": 0.11436923621777409,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009719035811654058,
      "loss": 2.2378,
      "step": 16833
    },
    {
      "epoch": 0.11437603056436815,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009719000253800519,
      "loss": 2.1475,
      "step": 16834
    },
    {
      "epoch": 0.11438282491096222,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009718964693762138,
      "loss": 2.2748,
      "step": 16835
    },
    {
      "epoch": 0.11438961925755627,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009718929131538929,
      "loss": 2.1799,
      "step": 16836
    },
    {
      "epoch": 0.11439641360415033,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000971889356713091,
      "loss": 2.3323,
      "step": 16837
    },
    {
      "epoch": 0.11440320795074438,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009718858000538098,
      "loss": 2.1003,
      "step": 16838
    },
    {
      "epoch": 0.11441000229733844,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009718822431760508,
      "loss": 2.2795,
      "step": 16839
    },
    {
      "epoch": 0.1144167966439325,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009718786860798158,
      "loss": 2.1673,
      "step": 16840
    },
    {
      "epoch": 0.11442359099052656,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009718751287651064,
      "loss": 2.2883,
      "step": 16841
    },
    {
      "epoch": 0.11443038533712062,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009718715712319242,
      "loss": 2.1959,
      "step": 16842
    },
    {
      "epoch": 0.11443717968371467,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009718680134802709,
      "loss": 2.1162,
      "step": 16843
    },
    {
      "epoch": 0.11444397403030873,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009718644555101481,
      "loss": 2.2591,
      "step": 16844
    },
    {
      "epoch": 0.1144507683769028,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009718608973215574,
      "loss": 2.129,
      "step": 16845
    },
    {
      "epoch": 0.11445756272349684,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009718573389145004,
      "loss": 2.3657,
      "step": 16846
    },
    {
      "epoch": 0.11446435707009091,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009718537802889791,
      "loss": 2.121,
      "step": 16847
    },
    {
      "epoch": 0.11447115141668496,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000971850221444995,
      "loss": 2.2399,
      "step": 16848
    },
    {
      "epoch": 0.11447794576327902,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009718466623825493,
      "loss": 2.0696,
      "step": 16849
    },
    {
      "epoch": 0.11448474010987307,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009718431031016441,
      "loss": 2.2624,
      "step": 16850
    },
    {
      "epoch": 0.11449153445646713,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009718395436022811,
      "loss": 2.1761,
      "step": 16851
    },
    {
      "epoch": 0.1144983288030612,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009718359838844619,
      "loss": 2.152,
      "step": 16852
    },
    {
      "epoch": 0.11450512314965525,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009718324239481878,
      "loss": 2.2791,
      "step": 16853
    },
    {
      "epoch": 0.11451191749624931,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009718288637934608,
      "loss": 2.2151,
      "step": 16854
    },
    {
      "epoch": 0.11451871184284336,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009718253034202824,
      "loss": 2.2718,
      "step": 16855
    },
    {
      "epoch": 0.11452550618943742,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009718217428286545,
      "loss": 2.2432,
      "step": 16856
    },
    {
      "epoch": 0.11453230053603149,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009718181820185783,
      "loss": 2.1904,
      "step": 16857
    },
    {
      "epoch": 0.11453909488262554,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009718146209900558,
      "loss": 2.1931,
      "step": 16858
    },
    {
      "epoch": 0.1145458892292196,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009718110597430885,
      "loss": 2.2201,
      "step": 16859
    },
    {
      "epoch": 0.11455268357581365,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000971807498277678,
      "loss": 2.1325,
      "step": 16860
    },
    {
      "epoch": 0.11455947792240771,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009718039365938262,
      "loss": 2.2363,
      "step": 16861
    },
    {
      "epoch": 0.11456627226900178,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009718003746915345,
      "loss": 2.1586,
      "step": 16862
    },
    {
      "epoch": 0.11457306661559583,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009717968125708045,
      "loss": 2.2865,
      "step": 16863
    },
    {
      "epoch": 0.11457986096218989,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009717932502316382,
      "loss": 2.1926,
      "step": 16864
    },
    {
      "epoch": 0.11458665530878394,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009717896876740371,
      "loss": 2.1781,
      "step": 16865
    },
    {
      "epoch": 0.114593449655378,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009717861248980025,
      "loss": 2.1818,
      "step": 16866
    },
    {
      "epoch": 0.11460024400197207,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009717825619035365,
      "loss": 2.2141,
      "step": 16867
    },
    {
      "epoch": 0.11460703834856611,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009717789986906407,
      "loss": 2.0775,
      "step": 16868
    },
    {
      "epoch": 0.11461383269516018,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009717754352593165,
      "loss": 2.2673,
      "step": 16869
    },
    {
      "epoch": 0.11462062704175423,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009717718716095656,
      "loss": 2.2081,
      "step": 16870
    },
    {
      "epoch": 0.11462742138834829,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009717683077413899,
      "loss": 2.0469,
      "step": 16871
    },
    {
      "epoch": 0.11463421573494235,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009717647436547907,
      "loss": 2.27,
      "step": 16872
    },
    {
      "epoch": 0.1146410100815364,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009717611793497701,
      "loss": 2.2197,
      "step": 16873
    },
    {
      "epoch": 0.11464780442813047,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009717576148263291,
      "loss": 2.1994,
      "step": 16874
    },
    {
      "epoch": 0.11465459877472452,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00097175405008447,
      "loss": 2.155,
      "step": 16875
    },
    {
      "epoch": 0.11466139312131858,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009717504851241941,
      "loss": 2.1698,
      "step": 16876
    },
    {
      "epoch": 0.11466818746791263,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009717469199455032,
      "loss": 2.3531,
      "step": 16877
    },
    {
      "epoch": 0.1146749818145067,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000971743354548399,
      "loss": 2.1232,
      "step": 16878
    },
    {
      "epoch": 0.11468177616110076,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009717397889328828,
      "loss": 2.2438,
      "step": 16879
    },
    {
      "epoch": 0.1146885705076948,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009717362230989565,
      "loss": 2.0809,
      "step": 16880
    },
    {
      "epoch": 0.11469536485428887,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009717326570466219,
      "loss": 2.2077,
      "step": 16881
    },
    {
      "epoch": 0.11470215920088292,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009717290907758803,
      "loss": 2.1704,
      "step": 16882
    },
    {
      "epoch": 0.11470895354747698,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009717255242867336,
      "loss": 2.1169,
      "step": 16883
    },
    {
      "epoch": 0.11471574789407105,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009717219575791834,
      "loss": 2.2582,
      "step": 16884
    },
    {
      "epoch": 0.1147225422406651,
      "grad_norm": 0.125,
      "learning_rate": 0.0009717183906532314,
      "loss": 2.1475,
      "step": 16885
    },
    {
      "epoch": 0.11472933658725916,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009717148235088792,
      "loss": 2.177,
      "step": 16886
    },
    {
      "epoch": 0.11473613093385321,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009717112561461283,
      "loss": 2.1816,
      "step": 16887
    },
    {
      "epoch": 0.11474292528044727,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009717076885649807,
      "loss": 2.2769,
      "step": 16888
    },
    {
      "epoch": 0.11474971962704134,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009717041207654378,
      "loss": 2.3875,
      "step": 16889
    },
    {
      "epoch": 0.11475651397363539,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009717005527475011,
      "loss": 2.1003,
      "step": 16890
    },
    {
      "epoch": 0.11476330832022945,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009716969845111728,
      "loss": 2.2493,
      "step": 16891
    },
    {
      "epoch": 0.1147701026668235,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009716934160564539,
      "loss": 2.2021,
      "step": 16892
    },
    {
      "epoch": 0.11477689701341756,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009716898473833464,
      "loss": 2.2071,
      "step": 16893
    },
    {
      "epoch": 0.11478369136001162,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000971686278491852,
      "loss": 2.1171,
      "step": 16894
    },
    {
      "epoch": 0.11479048570660567,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009716827093819722,
      "loss": 2.2069,
      "step": 16895
    },
    {
      "epoch": 0.11479728005319974,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009716791400537088,
      "loss": 2.2246,
      "step": 16896
    },
    {
      "epoch": 0.11480407439979379,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009716755705070632,
      "loss": 2.2768,
      "step": 16897
    },
    {
      "epoch": 0.11481086874638785,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009716720007420374,
      "loss": 2.1514,
      "step": 16898
    },
    {
      "epoch": 0.1148176630929819,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009716684307586327,
      "loss": 2.3525,
      "step": 16899
    },
    {
      "epoch": 0.11482445743957596,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000971664860556851,
      "loss": 2.1616,
      "step": 16900
    },
    {
      "epoch": 0.11483125178617003,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000971661290136694,
      "loss": 2.1382,
      "step": 16901
    },
    {
      "epoch": 0.11483804613276408,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009716577194981631,
      "loss": 1.9964,
      "step": 16902
    },
    {
      "epoch": 0.11484484047935814,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009716541486412602,
      "loss": 2.2339,
      "step": 16903
    },
    {
      "epoch": 0.11485163482595219,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009716505775659867,
      "loss": 1.9777,
      "step": 16904
    },
    {
      "epoch": 0.11485842917254625,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009716470062723444,
      "loss": 2.2666,
      "step": 16905
    },
    {
      "epoch": 0.11486522351914032,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000971643434760335,
      "loss": 2.247,
      "step": 16906
    },
    {
      "epoch": 0.11487201786573437,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.00097163986302996,
      "loss": 2.2275,
      "step": 16907
    },
    {
      "epoch": 0.11487881221232843,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009716362910812213,
      "loss": 2.1714,
      "step": 16908
    },
    {
      "epoch": 0.11488560655892248,
      "grad_norm": 0.21484375,
      "learning_rate": 0.0009716327189141203,
      "loss": 2.2682,
      "step": 16909
    },
    {
      "epoch": 0.11489240090551654,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009716291465286589,
      "loss": 2.1487,
      "step": 16910
    },
    {
      "epoch": 0.1148991952521106,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009716255739248384,
      "loss": 2.1296,
      "step": 16911
    },
    {
      "epoch": 0.11490598959870466,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009716220011026608,
      "loss": 2.2204,
      "step": 16912
    },
    {
      "epoch": 0.11491278394529872,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009716184280621277,
      "loss": 1.9788,
      "step": 16913
    },
    {
      "epoch": 0.11491957829189277,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009716148548032404,
      "loss": 2.1922,
      "step": 16914
    },
    {
      "epoch": 0.11492637263848683,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009716112813260011,
      "loss": 2.2724,
      "step": 16915
    },
    {
      "epoch": 0.1149331669850809,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009716077076304111,
      "loss": 2.2011,
      "step": 16916
    },
    {
      "epoch": 0.11493996133167494,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009716041337164721,
      "loss": 2.2347,
      "step": 16917
    },
    {
      "epoch": 0.11494675567826901,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009716005595841857,
      "loss": 2.2252,
      "step": 16918
    },
    {
      "epoch": 0.11495355002486306,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009715969852335537,
      "loss": 2.2736,
      "step": 16919
    },
    {
      "epoch": 0.11496034437145712,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009715934106645778,
      "loss": 2.1116,
      "step": 16920
    },
    {
      "epoch": 0.11496713871805118,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009715898358772596,
      "loss": 2.2397,
      "step": 16921
    },
    {
      "epoch": 0.11497393306464523,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009715862608716007,
      "loss": 2.2779,
      "step": 16922
    },
    {
      "epoch": 0.1149807274112393,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009715826856476026,
      "loss": 2.1454,
      "step": 16923
    },
    {
      "epoch": 0.11498752175783335,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009715791102052672,
      "loss": 2.1941,
      "step": 16924
    },
    {
      "epoch": 0.11499431610442741,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009715755345445961,
      "loss": 2.4069,
      "step": 16925
    },
    {
      "epoch": 0.11500111045102146,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009715719586655909,
      "loss": 2.1848,
      "step": 16926
    },
    {
      "epoch": 0.11500790479761552,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009715683825682534,
      "loss": 2.3497,
      "step": 16927
    },
    {
      "epoch": 0.11501469914420959,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000971564806252585,
      "loss": 2.1727,
      "step": 16928
    },
    {
      "epoch": 0.11502149349080364,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009715612297185876,
      "loss": 2.1554,
      "step": 16929
    },
    {
      "epoch": 0.1150282878373977,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009715576529662626,
      "loss": 2.3333,
      "step": 16930
    },
    {
      "epoch": 0.11503508218399175,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009715540759956121,
      "loss": 2.1416,
      "step": 16931
    },
    {
      "epoch": 0.11504187653058581,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009715504988066372,
      "loss": 2.2286,
      "step": 16932
    },
    {
      "epoch": 0.11504867087717988,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00097154692139934,
      "loss": 2.1465,
      "step": 16933
    },
    {
      "epoch": 0.11505546522377393,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009715433437737219,
      "loss": 2.1462,
      "step": 16934
    },
    {
      "epoch": 0.11506225957036799,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009715397659297847,
      "loss": 2.0619,
      "step": 16935
    },
    {
      "epoch": 0.11506905391696204,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00097153618786753,
      "loss": 2.3624,
      "step": 16936
    },
    {
      "epoch": 0.1150758482635561,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009715326095869592,
      "loss": 2.1589,
      "step": 16937
    },
    {
      "epoch": 0.11508264261015017,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009715290310880744,
      "loss": 2.2818,
      "step": 16938
    },
    {
      "epoch": 0.11508943695674421,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009715254523708772,
      "loss": 2.1404,
      "step": 16939
    },
    {
      "epoch": 0.11509623130333828,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009715218734353689,
      "loss": 2.2099,
      "step": 16940
    },
    {
      "epoch": 0.11510302564993233,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009715182942815515,
      "loss": 2.1756,
      "step": 16941
    },
    {
      "epoch": 0.11510981999652639,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009715147149094265,
      "loss": 2.1848,
      "step": 16942
    },
    {
      "epoch": 0.11511661434312045,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009715111353189956,
      "loss": 2.1458,
      "step": 16943
    },
    {
      "epoch": 0.1151234086897145,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009715075555102603,
      "loss": 2.1162,
      "step": 16944
    },
    {
      "epoch": 0.11513020303630857,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009715039754832227,
      "loss": 2.2011,
      "step": 16945
    },
    {
      "epoch": 0.11513699738290262,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000971500395237884,
      "loss": 2.103,
      "step": 16946
    },
    {
      "epoch": 0.11514379172949668,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000971496814774246,
      "loss": 2.2001,
      "step": 16947
    },
    {
      "epoch": 0.11515058607609073,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009714932340923103,
      "loss": 2.1232,
      "step": 16948
    },
    {
      "epoch": 0.1151573804226848,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000971489653192079,
      "loss": 2.1681,
      "step": 16949
    },
    {
      "epoch": 0.11516417476927886,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000971486072073553,
      "loss": 2.169,
      "step": 16950
    },
    {
      "epoch": 0.1151709691158729,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009714824907367345,
      "loss": 2.1834,
      "step": 16951
    },
    {
      "epoch": 0.11517776346246697,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009714789091816251,
      "loss": 2.2543,
      "step": 16952
    },
    {
      "epoch": 0.11518455780906102,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009714753274082261,
      "loss": 2.3065,
      "step": 16953
    },
    {
      "epoch": 0.11519135215565508,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009714717454165398,
      "loss": 2.1072,
      "step": 16954
    },
    {
      "epoch": 0.11519814650224915,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009714681632065673,
      "loss": 2.1396,
      "step": 16955
    },
    {
      "epoch": 0.1152049408488432,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009714645807783104,
      "loss": 2.0974,
      "step": 16956
    },
    {
      "epoch": 0.11521173519543726,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009714609981317709,
      "loss": 2.2104,
      "step": 16957
    },
    {
      "epoch": 0.11521852954203131,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009714574152669503,
      "loss": 2.1962,
      "step": 16958
    },
    {
      "epoch": 0.11522532388862537,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009714538321838504,
      "loss": 2.2344,
      "step": 16959
    },
    {
      "epoch": 0.11523211823521944,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009714502488824727,
      "loss": 2.2563,
      "step": 16960
    },
    {
      "epoch": 0.11523891258181349,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009714466653628189,
      "loss": 2.29,
      "step": 16961
    },
    {
      "epoch": 0.11524570692840755,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009714430816248908,
      "loss": 2.0929,
      "step": 16962
    },
    {
      "epoch": 0.1152525012750016,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009714394976686898,
      "loss": 2.1516,
      "step": 16963
    },
    {
      "epoch": 0.11525929562159566,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009714359134942179,
      "loss": 2.2003,
      "step": 16964
    },
    {
      "epoch": 0.11526608996818972,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009714323291014766,
      "loss": 2.1079,
      "step": 16965
    },
    {
      "epoch": 0.11527288431478377,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009714287444904675,
      "loss": 2.2329,
      "step": 16966
    },
    {
      "epoch": 0.11527967866137784,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009714251596611922,
      "loss": 2.2218,
      "step": 16967
    },
    {
      "epoch": 0.11528647300797189,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009714215746136524,
      "loss": 2.011,
      "step": 16968
    },
    {
      "epoch": 0.11529326735456595,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.00097141798934785,
      "loss": 2.1437,
      "step": 16969
    },
    {
      "epoch": 0.11530006170116,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009714144038637864,
      "loss": 1.9439,
      "step": 16970
    },
    {
      "epoch": 0.11530685604775406,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009714108181614633,
      "loss": 2.2865,
      "step": 16971
    },
    {
      "epoch": 0.11531365039434813,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009714072322408824,
      "loss": 2.1602,
      "step": 16972
    },
    {
      "epoch": 0.11532044474094218,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009714036461020454,
      "loss": 2.1339,
      "step": 16973
    },
    {
      "epoch": 0.11532723908753624,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000971400059744954,
      "loss": 2.1729,
      "step": 16974
    },
    {
      "epoch": 0.11533403343413029,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009713964731696095,
      "loss": 2.2359,
      "step": 16975
    },
    {
      "epoch": 0.11534082778072435,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000971392886376014,
      "loss": 2.1698,
      "step": 16976
    },
    {
      "epoch": 0.11534762212731842,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009713892993641691,
      "loss": 2.2924,
      "step": 16977
    },
    {
      "epoch": 0.11535441647391247,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009713857121340762,
      "loss": 2.2348,
      "step": 16978
    },
    {
      "epoch": 0.11536121082050653,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009713821246857371,
      "loss": 2.2531,
      "step": 16979
    },
    {
      "epoch": 0.11536800516710058,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009713785370191537,
      "loss": 2.2752,
      "step": 16980
    },
    {
      "epoch": 0.11537479951369464,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009713749491343273,
      "loss": 2.1106,
      "step": 16981
    },
    {
      "epoch": 0.1153815938602887,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009713713610312596,
      "loss": 2.2314,
      "step": 16982
    },
    {
      "epoch": 0.11538838820688276,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009713677727099526,
      "loss": 2.2492,
      "step": 16983
    },
    {
      "epoch": 0.11539518255347682,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009713641841704075,
      "loss": 2.2483,
      "step": 16984
    },
    {
      "epoch": 0.11540197690007087,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009713605954126263,
      "loss": 2.2624,
      "step": 16985
    },
    {
      "epoch": 0.11540877124666493,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009713570064366106,
      "loss": 2.2853,
      "step": 16986
    },
    {
      "epoch": 0.115415565593259,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000971353417242362,
      "loss": 2.1885,
      "step": 16987
    },
    {
      "epoch": 0.11542235993985304,
      "grad_norm": 0.140625,
      "learning_rate": 0.000971349827829882,
      "loss": 2.2385,
      "step": 16988
    },
    {
      "epoch": 0.11542915428644711,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009713462381991725,
      "loss": 2.2049,
      "step": 16989
    },
    {
      "epoch": 0.11543594863304116,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009713426483502352,
      "loss": 2.1913,
      "step": 16990
    },
    {
      "epoch": 0.11544274297963522,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009713390582830716,
      "loss": 2.2372,
      "step": 16991
    },
    {
      "epoch": 0.11544953732622928,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009713354679976834,
      "loss": 2.2976,
      "step": 16992
    },
    {
      "epoch": 0.11545633167282333,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009713318774940723,
      "loss": 2.177,
      "step": 16993
    },
    {
      "epoch": 0.1154631260194174,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.00097132828677224,
      "loss": 2.2325,
      "step": 16994
    },
    {
      "epoch": 0.11546992036601145,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000971324695832188,
      "loss": 2.2169,
      "step": 16995
    },
    {
      "epoch": 0.11547671471260551,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009713211046739181,
      "loss": 2.1652,
      "step": 16996
    },
    {
      "epoch": 0.11548350905919956,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000971317513297432,
      "loss": 2.2473,
      "step": 16997
    },
    {
      "epoch": 0.11549030340579362,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009713139217027313,
      "loss": 2.2228,
      "step": 16998
    },
    {
      "epoch": 0.11549709775238769,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009713103298898176,
      "loss": 2.2564,
      "step": 16999
    },
    {
      "epoch": 0.11550389209898174,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009713067378586926,
      "loss": 2.083,
      "step": 17000
    },
    {
      "epoch": 0.1155106864455758,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000971303145609358,
      "loss": 2.1686,
      "step": 17001
    },
    {
      "epoch": 0.11551748079216985,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009712995531418153,
      "loss": 2.0428,
      "step": 17002
    },
    {
      "epoch": 0.11552427513876391,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009712959604560666,
      "loss": 2.1158,
      "step": 17003
    },
    {
      "epoch": 0.11553106948535798,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009712923675521131,
      "loss": 2.1899,
      "step": 17004
    },
    {
      "epoch": 0.11553786383195203,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009712887744299567,
      "loss": 2.1547,
      "step": 17005
    },
    {
      "epoch": 0.11554465817854609,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000971285181089599,
      "loss": 2.1916,
      "step": 17006
    },
    {
      "epoch": 0.11555145252514014,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009712815875310417,
      "loss": 2.0779,
      "step": 17007
    },
    {
      "epoch": 0.1155582468717342,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009712779937542864,
      "loss": 2.2331,
      "step": 17008
    },
    {
      "epoch": 0.11556504121832827,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009712743997593347,
      "loss": 2.293,
      "step": 17009
    },
    {
      "epoch": 0.11557183556492232,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009712708055461884,
      "loss": 2.2749,
      "step": 17010
    },
    {
      "epoch": 0.11557862991151638,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009712672111148491,
      "loss": 2.2228,
      "step": 17011
    },
    {
      "epoch": 0.11558542425811043,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009712636164653186,
      "loss": 2.1973,
      "step": 17012
    },
    {
      "epoch": 0.11559221860470449,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009712600215975983,
      "loss": 2.0954,
      "step": 17013
    },
    {
      "epoch": 0.11559901295129855,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009712564265116901,
      "loss": 2.2576,
      "step": 17014
    },
    {
      "epoch": 0.1156058072978926,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009712528312075955,
      "loss": 2.1637,
      "step": 17015
    },
    {
      "epoch": 0.11561260164448667,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009712492356853164,
      "loss": 1.9933,
      "step": 17016
    },
    {
      "epoch": 0.11561939599108072,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009712456399448542,
      "loss": 2.1295,
      "step": 17017
    },
    {
      "epoch": 0.11562619033767478,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009712420439862106,
      "loss": 2.1748,
      "step": 17018
    },
    {
      "epoch": 0.11563298468426883,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009712384478093874,
      "loss": 2.1331,
      "step": 17019
    },
    {
      "epoch": 0.1156397790308629,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009712348514143862,
      "loss": 2.2045,
      "step": 17020
    },
    {
      "epoch": 0.11564657337745696,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009712312548012088,
      "loss": 2.0432,
      "step": 17021
    },
    {
      "epoch": 0.115653367724051,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009712276579698566,
      "loss": 2.0987,
      "step": 17022
    },
    {
      "epoch": 0.11566016207064507,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009712240609203313,
      "loss": 2.3463,
      "step": 17023
    },
    {
      "epoch": 0.11566695641723912,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009712204636526349,
      "loss": 2.0858,
      "step": 17024
    },
    {
      "epoch": 0.11567375076383318,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009712168661667686,
      "loss": 2.1681,
      "step": 17025
    },
    {
      "epoch": 0.11568054511042725,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009712132684627345,
      "loss": 2.2117,
      "step": 17026
    },
    {
      "epoch": 0.1156873394570213,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000971209670540534,
      "loss": 2.2078,
      "step": 17027
    },
    {
      "epoch": 0.11569413380361536,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009712060724001688,
      "loss": 2.1039,
      "step": 17028
    },
    {
      "epoch": 0.11570092815020941,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009712024740416406,
      "loss": 2.0963,
      "step": 17029
    },
    {
      "epoch": 0.11570772249680347,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009711988754649509,
      "loss": 2.121,
      "step": 17030
    },
    {
      "epoch": 0.11571451684339754,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009711952766701017,
      "loss": 2.149,
      "step": 17031
    },
    {
      "epoch": 0.11572131118999159,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009711916776570943,
      "loss": 2.1803,
      "step": 17032
    },
    {
      "epoch": 0.11572810553658565,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009711880784259308,
      "loss": 2.1804,
      "step": 17033
    },
    {
      "epoch": 0.1157348998831797,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009711844789766124,
      "loss": 2.1983,
      "step": 17034
    },
    {
      "epoch": 0.11574169422977376,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009711808793091412,
      "loss": 2.1947,
      "step": 17035
    },
    {
      "epoch": 0.11574848857636783,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009711772794235185,
      "loss": 2.169,
      "step": 17036
    },
    {
      "epoch": 0.11575528292296187,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009711736793197462,
      "loss": 1.9578,
      "step": 17037
    },
    {
      "epoch": 0.11576207726955594,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009711700789978259,
      "loss": 2.2363,
      "step": 17038
    },
    {
      "epoch": 0.11576887161614999,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009711664784577592,
      "loss": 2.0953,
      "step": 17039
    },
    {
      "epoch": 0.11577566596274405,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009711628776995478,
      "loss": 2.1993,
      "step": 17040
    },
    {
      "epoch": 0.11578246030933811,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009711592767231935,
      "loss": 2.0487,
      "step": 17041
    },
    {
      "epoch": 0.11578925465593216,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009711556755286978,
      "loss": 2.2548,
      "step": 17042
    },
    {
      "epoch": 0.11579604900252623,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009711520741160624,
      "loss": 2.1795,
      "step": 17043
    },
    {
      "epoch": 0.11580284334912028,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009711484724852891,
      "loss": 2.2345,
      "step": 17044
    },
    {
      "epoch": 0.11580963769571434,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009711448706363792,
      "loss": 2.0676,
      "step": 17045
    },
    {
      "epoch": 0.11581643204230839,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009711412685693349,
      "loss": 2.0557,
      "step": 17046
    },
    {
      "epoch": 0.11582322638890245,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009711376662841575,
      "loss": 2.1969,
      "step": 17047
    },
    {
      "epoch": 0.11583002073549652,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009711340637808487,
      "loss": 2.3033,
      "step": 17048
    },
    {
      "epoch": 0.11583681508209057,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009711304610594102,
      "loss": 2.199,
      "step": 17049
    },
    {
      "epoch": 0.11584360942868463,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009711268581198438,
      "loss": 2.2398,
      "step": 17050
    },
    {
      "epoch": 0.11585040377527868,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009711232549621512,
      "loss": 2.2757,
      "step": 17051
    },
    {
      "epoch": 0.11585719812187274,
      "grad_norm": 0.1650390625,
      "learning_rate": 0.0009711196515863336,
      "loss": 2.337,
      "step": 17052
    },
    {
      "epoch": 0.1158639924684668,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009711160479923933,
      "loss": 2.1689,
      "step": 17053
    },
    {
      "epoch": 0.11587078681506086,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009711124441803314,
      "loss": 2.2152,
      "step": 17054
    },
    {
      "epoch": 0.11587758116165492,
      "grad_norm": 0.125,
      "learning_rate": 0.00097110884015015,
      "loss": 2.1689,
      "step": 17055
    },
    {
      "epoch": 0.11588437550824897,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009711052359018507,
      "loss": 2.271,
      "step": 17056
    },
    {
      "epoch": 0.11589116985484303,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009711016314354349,
      "loss": 2.2577,
      "step": 17057
    },
    {
      "epoch": 0.1158979642014371,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009710980267509045,
      "loss": 2.0789,
      "step": 17058
    },
    {
      "epoch": 0.11590475854803114,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009710944218482611,
      "loss": 2.0946,
      "step": 17059
    },
    {
      "epoch": 0.11591155289462521,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009710908167275066,
      "loss": 2.1196,
      "step": 17060
    },
    {
      "epoch": 0.11591834724121926,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009710872113886422,
      "loss": 2.1558,
      "step": 17061
    },
    {
      "epoch": 0.11592514158781332,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009710836058316699,
      "loss": 2.1336,
      "step": 17062
    },
    {
      "epoch": 0.11593193593440738,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009710800000565913,
      "loss": 2.1435,
      "step": 17063
    },
    {
      "epoch": 0.11593873028100143,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000971076394063408,
      "loss": 2.1929,
      "step": 17064
    },
    {
      "epoch": 0.1159455246275955,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009710727878521217,
      "loss": 2.0447,
      "step": 17065
    },
    {
      "epoch": 0.11595231897418955,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009710691814227341,
      "loss": 2.1027,
      "step": 17066
    },
    {
      "epoch": 0.11595911332078361,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000971065574775247,
      "loss": 2.0461,
      "step": 17067
    },
    {
      "epoch": 0.11596590766737766,
      "grad_norm": 0.236328125,
      "learning_rate": 0.000971061967909662,
      "loss": 2.0702,
      "step": 17068
    },
    {
      "epoch": 0.11597270201397172,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009710583608259805,
      "loss": 2.2728,
      "step": 17069
    },
    {
      "epoch": 0.11597949636056579,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009710547535242045,
      "loss": 2.1639,
      "step": 17070
    },
    {
      "epoch": 0.11598629070715984,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009710511460043355,
      "loss": 2.1145,
      "step": 17071
    },
    {
      "epoch": 0.1159930850537539,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009710475382663752,
      "loss": 2.1919,
      "step": 17072
    },
    {
      "epoch": 0.11599987940034795,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009710439303103253,
      "loss": 2.2254,
      "step": 17073
    },
    {
      "epoch": 0.11600667374694201,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009710403221361877,
      "loss": 2.2372,
      "step": 17074
    },
    {
      "epoch": 0.11601346809353608,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009710367137439636,
      "loss": 2.2073,
      "step": 17075
    },
    {
      "epoch": 0.11602026244013013,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009710331051336549,
      "loss": 2.112,
      "step": 17076
    },
    {
      "epoch": 0.11602705678672419,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009710294963052633,
      "loss": 2.1752,
      "step": 17077
    },
    {
      "epoch": 0.11603385113331824,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009710258872587902,
      "loss": 2.2401,
      "step": 17078
    },
    {
      "epoch": 0.1160406454799123,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000971022277994238,
      "loss": 2.2613,
      "step": 17079
    },
    {
      "epoch": 0.11604743982650637,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009710186685116075,
      "loss": 2.2058,
      "step": 17080
    },
    {
      "epoch": 0.11605423417310042,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009710150588109009,
      "loss": 2.1989,
      "step": 17081
    },
    {
      "epoch": 0.11606102851969448,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009710114488921198,
      "loss": 2.213,
      "step": 17082
    },
    {
      "epoch": 0.11606782286628853,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009710078387552656,
      "loss": 2.2106,
      "step": 17083
    },
    {
      "epoch": 0.11607461721288259,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009710042284003404,
      "loss": 2.1585,
      "step": 17084
    },
    {
      "epoch": 0.11608141155947665,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009710006178273455,
      "loss": 2.2076,
      "step": 17085
    },
    {
      "epoch": 0.1160882059060707,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009709970070362828,
      "loss": 2.1459,
      "step": 17086
    },
    {
      "epoch": 0.11609500025266477,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009709933960271537,
      "loss": 2.0495,
      "step": 17087
    },
    {
      "epoch": 0.11610179459925882,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009709897847999603,
      "loss": 2.2028,
      "step": 17088
    },
    {
      "epoch": 0.11610858894585288,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009709861733547038,
      "loss": 2.2911,
      "step": 17089
    },
    {
      "epoch": 0.11611538329244693,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009709825616913863,
      "loss": 2.2038,
      "step": 17090
    },
    {
      "epoch": 0.116122177639041,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009709789498100092,
      "loss": 2.0955,
      "step": 17091
    },
    {
      "epoch": 0.11612897198563506,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009709753377105742,
      "loss": 2.1939,
      "step": 17092
    },
    {
      "epoch": 0.1161357663322291,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009709717253930831,
      "loss": 2.3747,
      "step": 17093
    },
    {
      "epoch": 0.11614256067882317,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009709681128575373,
      "loss": 2.1351,
      "step": 17094
    },
    {
      "epoch": 0.11614935502541722,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009709645001039389,
      "loss": 2.1005,
      "step": 17095
    },
    {
      "epoch": 0.11615614937201128,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009709608871322893,
      "loss": 2.0394,
      "step": 17096
    },
    {
      "epoch": 0.11616294371860535,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009709572739425901,
      "loss": 2.1617,
      "step": 17097
    },
    {
      "epoch": 0.1161697380651994,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009709536605348433,
      "loss": 2.2563,
      "step": 17098
    },
    {
      "epoch": 0.11617653241179346,
      "grad_norm": 0.140625,
      "learning_rate": 0.00097095004690905,
      "loss": 2.2577,
      "step": 17099
    },
    {
      "epoch": 0.11618332675838751,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009709464330652125,
      "loss": 2.1713,
      "step": 17100
    },
    {
      "epoch": 0.11619012110498157,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009709428190033322,
      "loss": 2.3741,
      "step": 17101
    },
    {
      "epoch": 0.11619691545157564,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009709392047234107,
      "loss": 2.2671,
      "step": 17102
    },
    {
      "epoch": 0.11620370979816969,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009709355902254497,
      "loss": 2.2368,
      "step": 17103
    },
    {
      "epoch": 0.11621050414476375,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000970931975509451,
      "loss": 2.2714,
      "step": 17104
    },
    {
      "epoch": 0.1162172984913578,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009709283605754161,
      "loss": 2.2121,
      "step": 17105
    },
    {
      "epoch": 0.11622409283795186,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009709247454233467,
      "loss": 2.1068,
      "step": 17106
    },
    {
      "epoch": 0.11623088718454593,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009709211300532448,
      "loss": 2.2132,
      "step": 17107
    },
    {
      "epoch": 0.11623768153113997,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009709175144651116,
      "loss": 2.119,
      "step": 17108
    },
    {
      "epoch": 0.11624447587773404,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009709138986589492,
      "loss": 2.1022,
      "step": 17109
    },
    {
      "epoch": 0.11625127022432809,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009709102826347589,
      "loss": 2.0175,
      "step": 17110
    },
    {
      "epoch": 0.11625806457092215,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009709066663925425,
      "loss": 2.1583,
      "step": 17111
    },
    {
      "epoch": 0.11626485891751621,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009709030499323018,
      "loss": 2.2168,
      "step": 17112
    },
    {
      "epoch": 0.11627165326411026,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009708994332540384,
      "loss": 2.2206,
      "step": 17113
    },
    {
      "epoch": 0.11627844761070433,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009708958163577539,
      "loss": 2.0836,
      "step": 17114
    },
    {
      "epoch": 0.11628524195729838,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009708921992434501,
      "loss": 2.3549,
      "step": 17115
    },
    {
      "epoch": 0.11629203630389244,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009708885819111286,
      "loss": 2.213,
      "step": 17116
    },
    {
      "epoch": 0.11629883065048649,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009708849643607911,
      "loss": 2.2756,
      "step": 17117
    },
    {
      "epoch": 0.11630562499708055,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009708813465924391,
      "loss": 2.1126,
      "step": 17118
    },
    {
      "epoch": 0.11631241934367462,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009708777286060745,
      "loss": 2.1058,
      "step": 17119
    },
    {
      "epoch": 0.11631921369026867,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009708741104016991,
      "loss": 2.2089,
      "step": 17120
    },
    {
      "epoch": 0.11632600803686273,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009708704919793144,
      "loss": 2.271,
      "step": 17121
    },
    {
      "epoch": 0.11633280238345678,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009708668733389217,
      "loss": 2.1331,
      "step": 17122
    },
    {
      "epoch": 0.11633959673005084,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009708632544805233,
      "loss": 2.1829,
      "step": 17123
    },
    {
      "epoch": 0.1163463910766449,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009708596354041206,
      "loss": 2.4447,
      "step": 17124
    },
    {
      "epoch": 0.11635318542323896,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009708560161097153,
      "loss": 2.0349,
      "step": 17125
    },
    {
      "epoch": 0.11635997976983302,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009708523965973089,
      "loss": 2.3145,
      "step": 17126
    },
    {
      "epoch": 0.11636677411642707,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009708487768669034,
      "loss": 2.2992,
      "step": 17127
    },
    {
      "epoch": 0.11637356846302113,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009708451569185001,
      "loss": 2.092,
      "step": 17128
    },
    {
      "epoch": 0.1163803628096152,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009708415367521012,
      "loss": 2.0776,
      "step": 17129
    },
    {
      "epoch": 0.11638715715620924,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000970837916367708,
      "loss": 2.0635,
      "step": 17130
    },
    {
      "epoch": 0.11639395150280331,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000970834295765322,
      "loss": 2.1705,
      "step": 17131
    },
    {
      "epoch": 0.11640074584939736,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009708306749449454,
      "loss": 2.3297,
      "step": 17132
    },
    {
      "epoch": 0.11640754019599142,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009708270539065795,
      "loss": 2.0915,
      "step": 17133
    },
    {
      "epoch": 0.11641433454258548,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000970823432650226,
      "loss": 2.1101,
      "step": 17134
    },
    {
      "epoch": 0.11642112888917953,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009708198111758867,
      "loss": 2.2083,
      "step": 17135
    },
    {
      "epoch": 0.1164279232357736,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009708161894835632,
      "loss": 2.11,
      "step": 17136
    },
    {
      "epoch": 0.11643471758236765,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009708125675732573,
      "loss": 2.3183,
      "step": 17137
    },
    {
      "epoch": 0.11644151192896171,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009708089454449704,
      "loss": 2.1516,
      "step": 17138
    },
    {
      "epoch": 0.11644830627555576,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009708053230987045,
      "loss": 2.2673,
      "step": 17139
    },
    {
      "epoch": 0.11645510062214982,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009708017005344611,
      "loss": 2.1731,
      "step": 17140
    },
    {
      "epoch": 0.11646189496874389,
      "grad_norm": 0.125,
      "learning_rate": 0.0009707980777522419,
      "loss": 2.2233,
      "step": 17141
    },
    {
      "epoch": 0.11646868931533794,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009707944547520486,
      "loss": 2.1345,
      "step": 17142
    },
    {
      "epoch": 0.116475483661932,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009707908315338829,
      "loss": 2.1926,
      "step": 17143
    },
    {
      "epoch": 0.11648227800852605,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009707872080977464,
      "loss": 2.1364,
      "step": 17144
    },
    {
      "epoch": 0.11648907235512011,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009707835844436407,
      "loss": 2.1666,
      "step": 17145
    },
    {
      "epoch": 0.11649586670171418,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009707799605715678,
      "loss": 2.1594,
      "step": 17146
    },
    {
      "epoch": 0.11650266104830823,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009707763364815291,
      "loss": 2.3447,
      "step": 17147
    },
    {
      "epoch": 0.11650945539490229,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009707727121735264,
      "loss": 2.0455,
      "step": 17148
    },
    {
      "epoch": 0.11651624974149634,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009707690876475612,
      "loss": 2.0289,
      "step": 17149
    },
    {
      "epoch": 0.1165230440880904,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009707654629036353,
      "loss": 2.0692,
      "step": 17150
    },
    {
      "epoch": 0.11652983843468447,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.0009707618379417506,
      "loss": 2.2002,
      "step": 17151
    },
    {
      "epoch": 0.11653663278127852,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009707582127619083,
      "loss": 2.1778,
      "step": 17152
    },
    {
      "epoch": 0.11654342712787258,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009707545873641106,
      "loss": 2.3384,
      "step": 17153
    },
    {
      "epoch": 0.11655022147446663,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009707509617483587,
      "loss": 2.2839,
      "step": 17154
    },
    {
      "epoch": 0.11655701582106069,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009707473359146547,
      "loss": 2.2238,
      "step": 17155
    },
    {
      "epoch": 0.11656381016765476,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000970743709863,
      "loss": 2.1916,
      "step": 17156
    },
    {
      "epoch": 0.1165706045142488,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009707400835933964,
      "loss": 2.3057,
      "step": 17157
    },
    {
      "epoch": 0.11657739886084287,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009707364571058455,
      "loss": 2.1249,
      "step": 17158
    },
    {
      "epoch": 0.11658419320743692,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009707328304003488,
      "loss": 2.1532,
      "step": 17159
    },
    {
      "epoch": 0.11659098755403098,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009707292034769085,
      "loss": 2.1723,
      "step": 17160
    },
    {
      "epoch": 0.11659778190062504,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009707255763355258,
      "loss": 2.1855,
      "step": 17161
    },
    {
      "epoch": 0.1166045762472191,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009707219489762027,
      "loss": 2.2511,
      "step": 17162
    },
    {
      "epoch": 0.11661137059381316,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009707183213989406,
      "loss": 2.2192,
      "step": 17163
    },
    {
      "epoch": 0.1166181649404072,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009707146936037414,
      "loss": 2.2639,
      "step": 17164
    },
    {
      "epoch": 0.11662495928700127,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009707110655906067,
      "loss": 2.103,
      "step": 17165
    },
    {
      "epoch": 0.11663175363359532,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009707074373595381,
      "loss": 2.1964,
      "step": 17166
    },
    {
      "epoch": 0.11663854798018938,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009707038089105376,
      "loss": 2.368,
      "step": 17167
    },
    {
      "epoch": 0.11664534232678345,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009707001802436063,
      "loss": 2.1618,
      "step": 17168
    },
    {
      "epoch": 0.1166521366733775,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009706965513587464,
      "loss": 2.189,
      "step": 17169
    },
    {
      "epoch": 0.11665893101997156,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009706929222559593,
      "loss": 2.1598,
      "step": 17170
    },
    {
      "epoch": 0.11666572536656561,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009706892929352468,
      "loss": 2.2314,
      "step": 17171
    },
    {
      "epoch": 0.11667251971315967,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009706856633966105,
      "loss": 2.1972,
      "step": 17172
    },
    {
      "epoch": 0.11667931405975374,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009706820336400523,
      "loss": 2.09,
      "step": 17173
    },
    {
      "epoch": 0.11668610840634779,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009706784036655736,
      "loss": 2.1308,
      "step": 17174
    },
    {
      "epoch": 0.11669290275294185,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009706747734731763,
      "loss": 2.1999,
      "step": 17175
    },
    {
      "epoch": 0.1166996970995359,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009706711430628618,
      "loss": 2.2091,
      "step": 17176
    },
    {
      "epoch": 0.11670649144612996,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000970667512434632,
      "loss": 2.2212,
      "step": 17177
    },
    {
      "epoch": 0.11671328579272403,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009706638815884885,
      "loss": 2.1078,
      "step": 17178
    },
    {
      "epoch": 0.11672008013931807,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009706602505244332,
      "loss": 2.1486,
      "step": 17179
    },
    {
      "epoch": 0.11672687448591214,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009706566192424675,
      "loss": 2.282,
      "step": 17180
    },
    {
      "epoch": 0.11673366883250619,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009706529877425931,
      "loss": 2.2507,
      "step": 17181
    },
    {
      "epoch": 0.11674046317910025,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009706493560248118,
      "loss": 2.1292,
      "step": 17182
    },
    {
      "epoch": 0.11674725752569431,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009706457240891254,
      "loss": 2.2624,
      "step": 17183
    },
    {
      "epoch": 0.11675405187228836,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009706420919355352,
      "loss": 2.1542,
      "step": 17184
    },
    {
      "epoch": 0.11676084621888243,
      "grad_norm": 0.17578125,
      "learning_rate": 0.0009706384595640433,
      "loss": 2.1037,
      "step": 17185
    },
    {
      "epoch": 0.11676764056547648,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009706348269746511,
      "loss": 2.2748,
      "step": 17186
    },
    {
      "epoch": 0.11677443491207054,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009706311941673604,
      "loss": 2.199,
      "step": 17187
    },
    {
      "epoch": 0.11678122925866459,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009706275611421728,
      "loss": 2.2378,
      "step": 17188
    },
    {
      "epoch": 0.11678802360525865,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00097062392789909,
      "loss": 2.1601,
      "step": 17189
    },
    {
      "epoch": 0.11679481795185272,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009706202944381138,
      "loss": 2.251,
      "step": 17190
    },
    {
      "epoch": 0.11680161229844677,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009706166607592458,
      "loss": 2.1211,
      "step": 17191
    },
    {
      "epoch": 0.11680840664504083,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009706130268624878,
      "loss": 2.2663,
      "step": 17192
    },
    {
      "epoch": 0.11681520099163488,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009706093927478411,
      "loss": 2.3372,
      "step": 17193
    },
    {
      "epoch": 0.11682199533822894,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009706057584153078,
      "loss": 2.18,
      "step": 17194
    },
    {
      "epoch": 0.116828789684823,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009706021238648895,
      "loss": 2.2064,
      "step": 17195
    },
    {
      "epoch": 0.11683558403141706,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009705984890965876,
      "loss": 2.1009,
      "step": 17196
    },
    {
      "epoch": 0.11684237837801112,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009705948541104041,
      "loss": 2.2251,
      "step": 17197
    },
    {
      "epoch": 0.11684917272460517,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009705912189063406,
      "loss": 2.2447,
      "step": 17198
    },
    {
      "epoch": 0.11685596707119923,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009705875834843987,
      "loss": 2.2562,
      "step": 17199
    },
    {
      "epoch": 0.1168627614177933,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009705839478445802,
      "loss": 2.1314,
      "step": 17200
    },
    {
      "epoch": 0.11686955576438735,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009705803119868868,
      "loss": 2.2749,
      "step": 17201
    },
    {
      "epoch": 0.11687635011098141,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00097057667591132,
      "loss": 2.2139,
      "step": 17202
    },
    {
      "epoch": 0.11688314445757546,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009705730396178815,
      "loss": 2.1986,
      "step": 17203
    },
    {
      "epoch": 0.11688993880416952,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009705694031065733,
      "loss": 2.16,
      "step": 17204
    },
    {
      "epoch": 0.11689673315076358,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009705657663773968,
      "loss": 2.2685,
      "step": 17205
    },
    {
      "epoch": 0.11690352749735763,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009705621294303537,
      "loss": 2.1825,
      "step": 17206
    },
    {
      "epoch": 0.1169103218439517,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009705584922654457,
      "loss": 2.3736,
      "step": 17207
    },
    {
      "epoch": 0.11691711619054575,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009705548548826746,
      "loss": 2.1649,
      "step": 17208
    },
    {
      "epoch": 0.11692391053713981,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009705512172820419,
      "loss": 2.2396,
      "step": 17209
    },
    {
      "epoch": 0.11693070488373386,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009705475794635494,
      "loss": 2.2717,
      "step": 17210
    },
    {
      "epoch": 0.11693749923032792,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009705439414271988,
      "loss": 2.0774,
      "step": 17211
    },
    {
      "epoch": 0.11694429357692199,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009705403031729917,
      "loss": 2.1679,
      "step": 17212
    },
    {
      "epoch": 0.11695108792351604,
      "grad_norm": 0.146484375,
      "learning_rate": 0.00097053666470093,
      "loss": 2.2878,
      "step": 17213
    },
    {
      "epoch": 0.1169578822701101,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000970533026011015,
      "loss": 2.4118,
      "step": 17214
    },
    {
      "epoch": 0.11696467661670415,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009705293871032487,
      "loss": 2.1422,
      "step": 17215
    },
    {
      "epoch": 0.11697147096329821,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009705257479776326,
      "loss": 1.9911,
      "step": 17216
    },
    {
      "epoch": 0.11697826530989228,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009705221086341686,
      "loss": 2.1177,
      "step": 17217
    },
    {
      "epoch": 0.11698505965648633,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009705184690728582,
      "loss": 2.1255,
      "step": 17218
    },
    {
      "epoch": 0.11699185400308039,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009705148292937032,
      "loss": 2.1465,
      "step": 17219
    },
    {
      "epoch": 0.11699864834967444,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009705111892967051,
      "loss": 2.2441,
      "step": 17220
    },
    {
      "epoch": 0.1170054426962685,
      "grad_norm": 0.125,
      "learning_rate": 0.0009705075490818658,
      "loss": 2.1614,
      "step": 17221
    },
    {
      "epoch": 0.11701223704286257,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009705039086491869,
      "loss": 2.1497,
      "step": 17222
    },
    {
      "epoch": 0.11701903138945662,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.00097050026799867,
      "loss": 2.1775,
      "step": 17223
    },
    {
      "epoch": 0.11702582573605068,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000970496627130317,
      "loss": 2.2582,
      "step": 17224
    },
    {
      "epoch": 0.11703262008264473,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009704929860441295,
      "loss": 2.1893,
      "step": 17225
    },
    {
      "epoch": 0.11703941442923879,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009704893447401091,
      "loss": 2.2315,
      "step": 17226
    },
    {
      "epoch": 0.11704620877583286,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009704857032182575,
      "loss": 2.1782,
      "step": 17227
    },
    {
      "epoch": 0.1170530031224269,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009704820614785763,
      "loss": 2.2079,
      "step": 17228
    },
    {
      "epoch": 0.11705979746902097,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009704784195210674,
      "loss": 2.2404,
      "step": 17229
    },
    {
      "epoch": 0.11706659181561502,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009704747773457323,
      "loss": 2.1661,
      "step": 17230
    },
    {
      "epoch": 0.11707338616220908,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009704711349525729,
      "loss": 2.2743,
      "step": 17231
    },
    {
      "epoch": 0.11708018050880314,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009704674923415907,
      "loss": 2.2214,
      "step": 17232
    },
    {
      "epoch": 0.1170869748553972,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009704638495127876,
      "loss": 2.0684,
      "step": 17233
    },
    {
      "epoch": 0.11709376920199126,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009704602064661648,
      "loss": 2.0771,
      "step": 17234
    },
    {
      "epoch": 0.1171005635485853,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009704565632017246,
      "loss": 2.229,
      "step": 17235
    },
    {
      "epoch": 0.11710735789517937,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009704529197194682,
      "loss": 2.2069,
      "step": 17236
    },
    {
      "epoch": 0.11711415224177342,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009704492760193977,
      "loss": 2.2215,
      "step": 17237
    },
    {
      "epoch": 0.11712094658836748,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009704456321015145,
      "loss": 2.2438,
      "step": 17238
    },
    {
      "epoch": 0.11712774093496155,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009704419879658203,
      "loss": 2.1141,
      "step": 17239
    },
    {
      "epoch": 0.1171345352815556,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009704383436123169,
      "loss": 2.3233,
      "step": 17240
    },
    {
      "epoch": 0.11714132962814966,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009704346990410059,
      "loss": 2.076,
      "step": 17241
    },
    {
      "epoch": 0.11714812397474371,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000970431054251889,
      "loss": 2.1329,
      "step": 17242
    },
    {
      "epoch": 0.11715491832133777,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000970427409244968,
      "loss": 2.1778,
      "step": 17243
    },
    {
      "epoch": 0.11716171266793184,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009704237640202444,
      "loss": 2.2798,
      "step": 17244
    },
    {
      "epoch": 0.11716850701452589,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009704201185777202,
      "loss": 2.1459,
      "step": 17245
    },
    {
      "epoch": 0.11717530136111995,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009704164729173967,
      "loss": 2.188,
      "step": 17246
    },
    {
      "epoch": 0.117182095707714,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009704128270392759,
      "loss": 2.234,
      "step": 17247
    },
    {
      "epoch": 0.11718889005430806,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009704091809433591,
      "loss": 2.2464,
      "step": 17248
    },
    {
      "epoch": 0.11719568440090213,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009704055346296485,
      "loss": 2.1608,
      "step": 17249
    },
    {
      "epoch": 0.11720247874749617,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009704018880981454,
      "loss": 2.2307,
      "step": 17250
    },
    {
      "epoch": 0.11720927309409024,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009703982413488516,
      "loss": 2.1336,
      "step": 17251
    },
    {
      "epoch": 0.11721606744068429,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009703945943817689,
      "loss": 2.3108,
      "step": 17252
    },
    {
      "epoch": 0.11722286178727835,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009703909471968989,
      "loss": 2.1617,
      "step": 17253
    },
    {
      "epoch": 0.11722965613387241,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009703872997942433,
      "loss": 2.1491,
      "step": 17254
    },
    {
      "epoch": 0.11723645048046646,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009703836521738037,
      "loss": 2.1654,
      "step": 17255
    },
    {
      "epoch": 0.11724324482706053,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000970380004335582,
      "loss": 2.1488,
      "step": 17256
    },
    {
      "epoch": 0.11725003917365458,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009703763562795795,
      "loss": 2.1852,
      "step": 17257
    },
    {
      "epoch": 0.11725683352024864,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009703727080057982,
      "loss": 2.3003,
      "step": 17258
    },
    {
      "epoch": 0.11726362786684269,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009703690595142398,
      "loss": 2.2637,
      "step": 17259
    },
    {
      "epoch": 0.11727042221343675,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000970365410804906,
      "loss": 2.2478,
      "step": 17260
    },
    {
      "epoch": 0.11727721656003082,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009703617618777983,
      "loss": 2.1353,
      "step": 17261
    },
    {
      "epoch": 0.11728401090662487,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009703581127329185,
      "loss": 2.0625,
      "step": 17262
    },
    {
      "epoch": 0.11729080525321893,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009703544633702684,
      "loss": 2.1676,
      "step": 17263
    },
    {
      "epoch": 0.11729759959981298,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009703508137898493,
      "loss": 2.2675,
      "step": 17264
    },
    {
      "epoch": 0.11730439394640704,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009703471639916634,
      "loss": 2.088,
      "step": 17265
    },
    {
      "epoch": 0.1173111882930011,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009703435139757122,
      "loss": 2.189,
      "step": 17266
    },
    {
      "epoch": 0.11731798263959516,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009703398637419973,
      "loss": 2.0362,
      "step": 17267
    },
    {
      "epoch": 0.11732477698618922,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009703362132905203,
      "loss": 2.1087,
      "step": 17268
    },
    {
      "epoch": 0.11733157133278327,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009703325626212831,
      "loss": 2.2662,
      "step": 17269
    },
    {
      "epoch": 0.11733836567937733,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009703289117342873,
      "loss": 2.1454,
      "step": 17270
    },
    {
      "epoch": 0.1173451600259714,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009703252606295347,
      "loss": 2.3226,
      "step": 17271
    },
    {
      "epoch": 0.11735195437256545,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009703216093070268,
      "loss": 2.378,
      "step": 17272
    },
    {
      "epoch": 0.11735874871915951,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009703179577667654,
      "loss": 2.3077,
      "step": 17273
    },
    {
      "epoch": 0.11736554306575356,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009703143060087523,
      "loss": 2.1847,
      "step": 17274
    },
    {
      "epoch": 0.11737233741234762,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000970310654032989,
      "loss": 2.1176,
      "step": 17275
    },
    {
      "epoch": 0.11737913175894168,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000970307001839477,
      "loss": 2.1535,
      "step": 17276
    },
    {
      "epoch": 0.11738592610553573,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009703033494282185,
      "loss": 2.1498,
      "step": 17277
    },
    {
      "epoch": 0.1173927204521298,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000970299696799215,
      "loss": 2.3496,
      "step": 17278
    },
    {
      "epoch": 0.11739951479872385,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009702960439524679,
      "loss": 2.0797,
      "step": 17279
    },
    {
      "epoch": 0.11740630914531791,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009702923908879794,
      "loss": 2.1446,
      "step": 17280
    },
    {
      "epoch": 0.11741310349191197,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009702887376057507,
      "loss": 2.1281,
      "step": 17281
    },
    {
      "epoch": 0.11741989783850602,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009702850841057838,
      "loss": 1.9583,
      "step": 17282
    },
    {
      "epoch": 0.11742669218510009,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009702814303880803,
      "loss": 2.289,
      "step": 17283
    },
    {
      "epoch": 0.11743348653169414,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000970277776452642,
      "loss": 2.2291,
      "step": 17284
    },
    {
      "epoch": 0.1174402808782882,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009702741222994702,
      "loss": 2.0289,
      "step": 17285
    },
    {
      "epoch": 0.11744707522488225,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009702704679285671,
      "loss": 2.1669,
      "step": 17286
    },
    {
      "epoch": 0.11745386957147631,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009702668133399341,
      "loss": 2.2736,
      "step": 17287
    },
    {
      "epoch": 0.11746066391807038,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000970263158533573,
      "loss": 2.3335,
      "step": 17288
    },
    {
      "epoch": 0.11746745826466443,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009702595035094853,
      "loss": 2.2973,
      "step": 17289
    },
    {
      "epoch": 0.11747425261125849,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000970255848267673,
      "loss": 2.1014,
      "step": 17290
    },
    {
      "epoch": 0.11748104695785254,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009702521928081376,
      "loss": 2.179,
      "step": 17291
    },
    {
      "epoch": 0.1174878413044466,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009702485371308807,
      "loss": 2.181,
      "step": 17292
    },
    {
      "epoch": 0.11749463565104067,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009702448812359042,
      "loss": 2.2747,
      "step": 17293
    },
    {
      "epoch": 0.11750142999763472,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009702412251232097,
      "loss": 2.0623,
      "step": 17294
    },
    {
      "epoch": 0.11750822434422878,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009702375687927991,
      "loss": 2.0882,
      "step": 17295
    },
    {
      "epoch": 0.11751501869082283,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009702339122446737,
      "loss": 2.2715,
      "step": 17296
    },
    {
      "epoch": 0.11752181303741689,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009702302554788353,
      "loss": 2.1809,
      "step": 17297
    },
    {
      "epoch": 0.11752860738401096,
      "grad_norm": 0.125,
      "learning_rate": 0.0009702265984952858,
      "loss": 2.1635,
      "step": 17298
    },
    {
      "epoch": 0.117535401730605,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009702229412940268,
      "loss": 2.143,
      "step": 17299
    },
    {
      "epoch": 0.11754219607719907,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009702192838750601,
      "loss": 2.1338,
      "step": 17300
    },
    {
      "epoch": 0.11754899042379312,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000970215626238387,
      "loss": 2.1004,
      "step": 17301
    },
    {
      "epoch": 0.11755578477038718,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009702119683840096,
      "loss": 2.3535,
      "step": 17302
    },
    {
      "epoch": 0.11756257911698124,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009702083103119294,
      "loss": 2.0511,
      "step": 17303
    },
    {
      "epoch": 0.1175693734635753,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009702046520221481,
      "loss": 2.1523,
      "step": 17304
    },
    {
      "epoch": 0.11757616781016936,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009702009935146676,
      "loss": 2.1568,
      "step": 17305
    },
    {
      "epoch": 0.11758296215676341,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009701973347894892,
      "loss": 2.2391,
      "step": 17306
    },
    {
      "epoch": 0.11758975650335747,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000970193675846615,
      "loss": 2.1377,
      "step": 17307
    },
    {
      "epoch": 0.11759655084995152,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009701900166860464,
      "loss": 2.0228,
      "step": 17308
    },
    {
      "epoch": 0.11760334519654558,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009701863573077854,
      "loss": 2.1406,
      "step": 17309
    },
    {
      "epoch": 0.11761013954313965,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009701826977118333,
      "loss": 2.1634,
      "step": 17310
    },
    {
      "epoch": 0.1176169338897337,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009701790378981921,
      "loss": 2.2213,
      "step": 17311
    },
    {
      "epoch": 0.11762372823632776,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009701753778668634,
      "loss": 2.0217,
      "step": 17312
    },
    {
      "epoch": 0.11763052258292181,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009701717176178489,
      "loss": 2.1892,
      "step": 17313
    },
    {
      "epoch": 0.11763731692951587,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009701680571511502,
      "loss": 2.1677,
      "step": 17314
    },
    {
      "epoch": 0.11764411127610994,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009701643964667693,
      "loss": 2.1924,
      "step": 17315
    },
    {
      "epoch": 0.11765090562270399,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009701607355647074,
      "loss": 2.2055,
      "step": 17316
    },
    {
      "epoch": 0.11765769996929805,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009701570744449666,
      "loss": 2.2316,
      "step": 17317
    },
    {
      "epoch": 0.1176644943158921,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009701534131075485,
      "loss": 2.1704,
      "step": 17318
    },
    {
      "epoch": 0.11767128866248616,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009701497515524546,
      "loss": 2.1769,
      "step": 17319
    },
    {
      "epoch": 0.11767808300908023,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000970146089779687,
      "loss": 2.2331,
      "step": 17320
    },
    {
      "epoch": 0.11768487735567427,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000970142427789247,
      "loss": 2.2734,
      "step": 17321
    },
    {
      "epoch": 0.11769167170226834,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009701387655811364,
      "loss": 2.1143,
      "step": 17322
    },
    {
      "epoch": 0.11769846604886239,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000970135103155357,
      "loss": 2.2607,
      "step": 17323
    },
    {
      "epoch": 0.11770526039545645,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009701314405119104,
      "loss": 2.0867,
      "step": 17324
    },
    {
      "epoch": 0.11771205474205051,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009701277776507985,
      "loss": 2.2802,
      "step": 17325
    },
    {
      "epoch": 0.11771884908864456,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009701241145720225,
      "loss": 2.2236,
      "step": 17326
    },
    {
      "epoch": 0.11772564343523863,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009701204512755848,
      "loss": 2.2423,
      "step": 17327
    },
    {
      "epoch": 0.11773243778183268,
      "grad_norm": 0.125,
      "learning_rate": 0.0009701167877614865,
      "loss": 2.056,
      "step": 17328
    },
    {
      "epoch": 0.11773923212842674,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009701131240297296,
      "loss": 2.1259,
      "step": 17329
    },
    {
      "epoch": 0.11774602647502079,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009701094600803157,
      "loss": 2.1162,
      "step": 17330
    },
    {
      "epoch": 0.11775282082161485,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009701057959132464,
      "loss": 2.2804,
      "step": 17331
    },
    {
      "epoch": 0.11775961516820892,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009701021315285237,
      "loss": 2.3532,
      "step": 17332
    },
    {
      "epoch": 0.11776640951480297,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009700984669261489,
      "loss": 2.0757,
      "step": 17333
    },
    {
      "epoch": 0.11777320386139703,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000970094802106124,
      "loss": 2.0384,
      "step": 17334
    },
    {
      "epoch": 0.11777999820799108,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009700911370684506,
      "loss": 2.1685,
      "step": 17335
    },
    {
      "epoch": 0.11778679255458514,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009700874718131302,
      "loss": 2.1347,
      "step": 17336
    },
    {
      "epoch": 0.1177935869011792,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000970083806340165,
      "loss": 2.3048,
      "step": 17337
    },
    {
      "epoch": 0.11780038124777326,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009700801406495562,
      "loss": 2.232,
      "step": 17338
    },
    {
      "epoch": 0.11780717559436732,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009700764747413058,
      "loss": 2.1239,
      "step": 17339
    },
    {
      "epoch": 0.11781396994096137,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009700728086154154,
      "loss": 2.1521,
      "step": 17340
    },
    {
      "epoch": 0.11782076428755543,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009700691422718866,
      "loss": 2.2051,
      "step": 17341
    },
    {
      "epoch": 0.1178275586341495,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009700654757107211,
      "loss": 2.1365,
      "step": 17342
    },
    {
      "epoch": 0.11783435298074355,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009700618089319208,
      "loss": 2.3181,
      "step": 17343
    },
    {
      "epoch": 0.11784114732733761,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009700581419354872,
      "loss": 2.0294,
      "step": 17344
    },
    {
      "epoch": 0.11784794167393166,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009700544747214221,
      "loss": 2.1587,
      "step": 17345
    },
    {
      "epoch": 0.11785473602052572,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009700508072897271,
      "loss": 2.1572,
      "step": 17346
    },
    {
      "epoch": 0.11786153036711979,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000970047139640404,
      "loss": 2.2828,
      "step": 17347
    },
    {
      "epoch": 0.11786832471371383,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009700434717734546,
      "loss": 2.0369,
      "step": 17348
    },
    {
      "epoch": 0.1178751190603079,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009700398036888803,
      "loss": 2.1494,
      "step": 17349
    },
    {
      "epoch": 0.11788191340690195,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009700361353866832,
      "loss": 2.2212,
      "step": 17350
    },
    {
      "epoch": 0.11788870775349601,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009700324668668644,
      "loss": 2.0112,
      "step": 17351
    },
    {
      "epoch": 0.11789550210009007,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009700287981294261,
      "loss": 2.1846,
      "step": 17352
    },
    {
      "epoch": 0.11790229644668412,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009700251291743699,
      "loss": 2.1601,
      "step": 17353
    },
    {
      "epoch": 0.11790909079327819,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009700214600016975,
      "loss": 2.124,
      "step": 17354
    },
    {
      "epoch": 0.11791588513987224,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009700177906114105,
      "loss": 2.1043,
      "step": 17355
    },
    {
      "epoch": 0.1179226794864663,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009700141210035105,
      "loss": 2.1512,
      "step": 17356
    },
    {
      "epoch": 0.11792947383306035,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009700104511779996,
      "loss": 2.1486,
      "step": 17357
    },
    {
      "epoch": 0.11793626817965441,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009700067811348791,
      "loss": 2.1792,
      "step": 17358
    },
    {
      "epoch": 0.11794306252624848,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009700031108741509,
      "loss": 2.1454,
      "step": 17359
    },
    {
      "epoch": 0.11794985687284253,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009699994403958165,
      "loss": 2.2555,
      "step": 17360
    },
    {
      "epoch": 0.11795665121943659,
      "grad_norm": 0.125,
      "learning_rate": 0.000969995769699878,
      "loss": 2.0796,
      "step": 17361
    },
    {
      "epoch": 0.11796344556603064,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009699920987863367,
      "loss": 2.2054,
      "step": 17362
    },
    {
      "epoch": 0.1179702399126247,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009699884276551944,
      "loss": 2.1487,
      "step": 17363
    },
    {
      "epoch": 0.11797703425921877,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000969984756306453,
      "loss": 2.2477,
      "step": 17364
    },
    {
      "epoch": 0.11798382860581282,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009699810847401141,
      "loss": 2.1763,
      "step": 17365
    },
    {
      "epoch": 0.11799062295240688,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009699774129561791,
      "loss": 2.0701,
      "step": 17366
    },
    {
      "epoch": 0.11799741729900093,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009699737409546501,
      "loss": 2.3938,
      "step": 17367
    },
    {
      "epoch": 0.11800421164559499,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009699700687355287,
      "loss": 2.2955,
      "step": 17368
    },
    {
      "epoch": 0.11801100599218906,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009699663962988163,
      "loss": 2.3398,
      "step": 17369
    },
    {
      "epoch": 0.1180178003387831,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000969962723644515,
      "loss": 2.1073,
      "step": 17370
    },
    {
      "epoch": 0.11802459468537717,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009699590507726264,
      "loss": 2.1828,
      "step": 17371
    },
    {
      "epoch": 0.11803138903197122,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000969955377683152,
      "loss": 2.159,
      "step": 17372
    },
    {
      "epoch": 0.11803818337856528,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009699517043760938,
      "loss": 2.0756,
      "step": 17373
    },
    {
      "epoch": 0.11804497772515934,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009699480308514532,
      "loss": 2.166,
      "step": 17374
    },
    {
      "epoch": 0.1180517720717534,
      "grad_norm": 0.177734375,
      "learning_rate": 0.0009699443571092322,
      "loss": 2.0772,
      "step": 17375
    },
    {
      "epoch": 0.11805856641834746,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009699406831494323,
      "loss": 2.0603,
      "step": 17376
    },
    {
      "epoch": 0.11806536076494151,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009699370089720552,
      "loss": 2.246,
      "step": 17377
    },
    {
      "epoch": 0.11807215511153557,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009699333345771028,
      "loss": 2.2272,
      "step": 17378
    },
    {
      "epoch": 0.11807894945812962,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009699296599645765,
      "loss": 2.0972,
      "step": 17379
    },
    {
      "epoch": 0.11808574380472368,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009699259851344782,
      "loss": 2.1083,
      "step": 17380
    },
    {
      "epoch": 0.11809253815131775,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009699223100868095,
      "loss": 2.2408,
      "step": 17381
    },
    {
      "epoch": 0.1180993324979118,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009699186348215723,
      "loss": 2.2185,
      "step": 17382
    },
    {
      "epoch": 0.11810612684450586,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000969914959338768,
      "loss": 2.0873,
      "step": 17383
    },
    {
      "epoch": 0.11811292119109991,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009699112836383985,
      "loss": 2.0997,
      "step": 17384
    },
    {
      "epoch": 0.11811971553769397,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009699076077204656,
      "loss": 2.2042,
      "step": 17385
    },
    {
      "epoch": 0.11812650988428804,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009699039315849707,
      "loss": 2.3396,
      "step": 17386
    },
    {
      "epoch": 0.11813330423088209,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009699002552319158,
      "loss": 2.1406,
      "step": 17387
    },
    {
      "epoch": 0.11814009857747615,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009698965786613023,
      "loss": 2.1566,
      "step": 17388
    },
    {
      "epoch": 0.1181468929240702,
      "grad_norm": 0.234375,
      "learning_rate": 0.0009698929018731321,
      "loss": 2.19,
      "step": 17389
    },
    {
      "epoch": 0.11815368727066426,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009698892248674072,
      "loss": 2.241,
      "step": 17390
    },
    {
      "epoch": 0.11816048161725833,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009698855476441286,
      "loss": 2.2626,
      "step": 17391
    },
    {
      "epoch": 0.11816727596385238,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009698818702032985,
      "loss": 2.2387,
      "step": 17392
    },
    {
      "epoch": 0.11817407031044644,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009698781925449185,
      "loss": 2.2688,
      "step": 17393
    },
    {
      "epoch": 0.11818086465704049,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009698745146689903,
      "loss": 2.1775,
      "step": 17394
    },
    {
      "epoch": 0.11818765900363455,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009698708365755155,
      "loss": 2.2624,
      "step": 17395
    },
    {
      "epoch": 0.11819445335022861,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000969867158264496,
      "loss": 2.2003,
      "step": 17396
    },
    {
      "epoch": 0.11820124769682266,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009698634797359332,
      "loss": 2.1473,
      "step": 17397
    },
    {
      "epoch": 0.11820804204341673,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009698598009898292,
      "loss": 2.2724,
      "step": 17398
    },
    {
      "epoch": 0.11821483639001078,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009698561220261854,
      "loss": 2.1475,
      "step": 17399
    },
    {
      "epoch": 0.11822163073660484,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009698524428450036,
      "loss": 2.1215,
      "step": 17400
    },
    {
      "epoch": 0.1182284250831989,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009698487634462856,
      "loss": 2.1946,
      "step": 17401
    },
    {
      "epoch": 0.11823521942979295,
      "grad_norm": 0.140625,
      "learning_rate": 0.000969845083830033,
      "loss": 2.2014,
      "step": 17402
    },
    {
      "epoch": 0.11824201377638702,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009698414039962473,
      "loss": 2.1614,
      "step": 17403
    },
    {
      "epoch": 0.11824880812298107,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009698377239449307,
      "loss": 2.2694,
      "step": 17404
    },
    {
      "epoch": 0.11825560246957513,
      "grad_norm": 0.125,
      "learning_rate": 0.0009698340436760844,
      "loss": 2.1591,
      "step": 17405
    },
    {
      "epoch": 0.11826239681616918,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009698303631897105,
      "loss": 2.1681,
      "step": 17406
    },
    {
      "epoch": 0.11826919116276324,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009698266824858104,
      "loss": 2.2466,
      "step": 17407
    },
    {
      "epoch": 0.1182759855093573,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000969823001564386,
      "loss": 2.1987,
      "step": 17408
    },
    {
      "epoch": 0.11828277985595136,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009698193204254388,
      "loss": 2.0423,
      "step": 17409
    },
    {
      "epoch": 0.11828957420254542,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009698156390689709,
      "loss": 2.163,
      "step": 17410
    },
    {
      "epoch": 0.11829636854913947,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009698119574949835,
      "loss": 2.1272,
      "step": 17411
    },
    {
      "epoch": 0.11830316289573353,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009698082757034788,
      "loss": 2.1656,
      "step": 17412
    },
    {
      "epoch": 0.1183099572423276,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000969804593694458,
      "loss": 2.2056,
      "step": 17413
    },
    {
      "epoch": 0.11831675158892165,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009698009114679231,
      "loss": 2.1403,
      "step": 17414
    },
    {
      "epoch": 0.11832354593551571,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009697972290238759,
      "loss": 2.1043,
      "step": 17415
    },
    {
      "epoch": 0.11833034028210976,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009697935463623178,
      "loss": 2.3715,
      "step": 17416
    },
    {
      "epoch": 0.11833713462870382,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009697898634832509,
      "loss": 2.1561,
      "step": 17417
    },
    {
      "epoch": 0.11834392897529789,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009697861803866766,
      "loss": 2.2618,
      "step": 17418
    },
    {
      "epoch": 0.11835072332189193,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009697824970725966,
      "loss": 2.0805,
      "step": 17419
    },
    {
      "epoch": 0.118357517668486,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009697788135410126,
      "loss": 2.2954,
      "step": 17420
    },
    {
      "epoch": 0.11836431201508005,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009697751297919266,
      "loss": 2.1352,
      "step": 17421
    },
    {
      "epoch": 0.11837110636167411,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009697714458253399,
      "loss": 2.1786,
      "step": 17422
    },
    {
      "epoch": 0.11837790070826817,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009697677616412546,
      "loss": 2.147,
      "step": 17423
    },
    {
      "epoch": 0.11838469505486222,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009697640772396721,
      "loss": 2.213,
      "step": 17424
    },
    {
      "epoch": 0.11839148940145629,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009697603926205943,
      "loss": 2.1932,
      "step": 17425
    },
    {
      "epoch": 0.11839828374805034,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009697567077840228,
      "loss": 2.1312,
      "step": 17426
    },
    {
      "epoch": 0.1184050780946444,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009697530227299594,
      "loss": 2.0527,
      "step": 17427
    },
    {
      "epoch": 0.11841187244123845,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009697493374584055,
      "loss": 2.1467,
      "step": 17428
    },
    {
      "epoch": 0.11841866678783251,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009697456519693633,
      "loss": 1.9947,
      "step": 17429
    },
    {
      "epoch": 0.11842546113442658,
      "grad_norm": 0.140625,
      "learning_rate": 0.000969741966262834,
      "loss": 2.1568,
      "step": 17430
    },
    {
      "epoch": 0.11843225548102063,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009697382803388197,
      "loss": 2.0742,
      "step": 17431
    },
    {
      "epoch": 0.11843904982761469,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000969734594197322,
      "loss": 2.2301,
      "step": 17432
    },
    {
      "epoch": 0.11844584417420874,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009697309078383425,
      "loss": 2.1751,
      "step": 17433
    },
    {
      "epoch": 0.1184526385208028,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000969727221261883,
      "loss": 2.2505,
      "step": 17434
    },
    {
      "epoch": 0.11845943286739687,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009697235344679452,
      "loss": 2.2609,
      "step": 17435
    },
    {
      "epoch": 0.11846622721399092,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009697198474565307,
      "loss": 2.17,
      "step": 17436
    },
    {
      "epoch": 0.11847302156058498,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009697161602276414,
      "loss": 2.1334,
      "step": 17437
    },
    {
      "epoch": 0.11847981590717903,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009697124727812789,
      "loss": 2.2406,
      "step": 17438
    },
    {
      "epoch": 0.11848661025377309,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009697087851174449,
      "loss": 2.1924,
      "step": 17439
    },
    {
      "epoch": 0.11849340460036716,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009697050972361411,
      "loss": 2.3884,
      "step": 17440
    },
    {
      "epoch": 0.1185001989469612,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009697014091373693,
      "loss": 2.167,
      "step": 17441
    },
    {
      "epoch": 0.11850699329355527,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009696977208211309,
      "loss": 2.1825,
      "step": 17442
    },
    {
      "epoch": 0.11851378764014932,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000969694032287428,
      "loss": 2.2688,
      "step": 17443
    },
    {
      "epoch": 0.11852058198674338,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009696903435362622,
      "loss": 2.0979,
      "step": 17444
    },
    {
      "epoch": 0.11852737633333744,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000969686654567635,
      "loss": 2.2854,
      "step": 17445
    },
    {
      "epoch": 0.1185341706799315,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009696829653815484,
      "loss": 2.0832,
      "step": 17446
    },
    {
      "epoch": 0.11854096502652556,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000969679275978004,
      "loss": 2.2298,
      "step": 17447
    },
    {
      "epoch": 0.11854775937311961,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009696755863570033,
      "loss": 2.1458,
      "step": 17448
    },
    {
      "epoch": 0.11855455371971367,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009696718965185484,
      "loss": 2.2475,
      "step": 17449
    },
    {
      "epoch": 0.11856134806630772,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009696682064626406,
      "loss": 2.1663,
      "step": 17450
    },
    {
      "epoch": 0.11856814241290178,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009696645161892818,
      "loss": 2.2776,
      "step": 17451
    },
    {
      "epoch": 0.11857493675949585,
      "grad_norm": 0.140625,
      "learning_rate": 0.000969660825698474,
      "loss": 2.1687,
      "step": 17452
    },
    {
      "epoch": 0.1185817311060899,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009696571349902184,
      "loss": 2.1683,
      "step": 17453
    },
    {
      "epoch": 0.11858852545268396,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009696534440645169,
      "loss": 2.1189,
      "step": 17454
    },
    {
      "epoch": 0.11859531979927801,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009696497529213714,
      "loss": 2.1598,
      "step": 17455
    },
    {
      "epoch": 0.11860211414587207,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009696460615607834,
      "loss": 2.1909,
      "step": 17456
    },
    {
      "epoch": 0.11860890849246614,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009696423699827546,
      "loss": 2.2408,
      "step": 17457
    },
    {
      "epoch": 0.11861570283906019,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009696386781872867,
      "loss": 2.4219,
      "step": 17458
    },
    {
      "epoch": 0.11862249718565425,
      "grad_norm": 0.125,
      "learning_rate": 0.0009696349861743817,
      "loss": 2.1812,
      "step": 17459
    },
    {
      "epoch": 0.1186292915322483,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009696312939440409,
      "loss": 2.1779,
      "step": 17460
    },
    {
      "epoch": 0.11863608587884236,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009696276014962661,
      "loss": 2.1136,
      "step": 17461
    },
    {
      "epoch": 0.11864288022543643,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009696239088310592,
      "loss": 2.2622,
      "step": 17462
    },
    {
      "epoch": 0.11864967457203048,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009696202159484219,
      "loss": 2.2576,
      "step": 17463
    },
    {
      "epoch": 0.11865646891862454,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009696165228483557,
      "loss": 2.0764,
      "step": 17464
    },
    {
      "epoch": 0.11866326326521859,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009696128295308626,
      "loss": 2.2718,
      "step": 17465
    },
    {
      "epoch": 0.11867005761181265,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009696091359959441,
      "loss": 2.1097,
      "step": 17466
    },
    {
      "epoch": 0.11867685195840671,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009696054422436018,
      "loss": 2.2658,
      "step": 17467
    },
    {
      "epoch": 0.11868364630500076,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009696017482738377,
      "loss": 2.2583,
      "step": 17468
    },
    {
      "epoch": 0.11869044065159483,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009695980540866534,
      "loss": 2.139,
      "step": 17469
    },
    {
      "epoch": 0.11869723499818888,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009695943596820505,
      "loss": 2.0345,
      "step": 17470
    },
    {
      "epoch": 0.11870402934478294,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009695906650600308,
      "loss": 2.2153,
      "step": 17471
    },
    {
      "epoch": 0.118710823691377,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009695869702205962,
      "loss": 2.1604,
      "step": 17472
    },
    {
      "epoch": 0.11871761803797105,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009695832751637479,
      "loss": 2.2154,
      "step": 17473
    },
    {
      "epoch": 0.11872441238456512,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009695795798894881,
      "loss": 2.2872,
      "step": 17474
    },
    {
      "epoch": 0.11873120673115917,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009695758843978183,
      "loss": 2.2621,
      "step": 17475
    },
    {
      "epoch": 0.11873800107775323,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009695721886887403,
      "loss": 2.2718,
      "step": 17476
    },
    {
      "epoch": 0.11874479542434728,
      "grad_norm": 0.125,
      "learning_rate": 0.0009695684927622557,
      "loss": 2.0827,
      "step": 17477
    },
    {
      "epoch": 0.11875158977094134,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009695647966183662,
      "loss": 2.1772,
      "step": 17478
    },
    {
      "epoch": 0.1187583841175354,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009695611002570738,
      "loss": 2.2157,
      "step": 17479
    },
    {
      "epoch": 0.11876517846412946,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009695574036783799,
      "loss": 2.2213,
      "step": 17480
    },
    {
      "epoch": 0.11877197281072352,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009695537068822862,
      "loss": 2.2921,
      "step": 17481
    },
    {
      "epoch": 0.11877876715731757,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009695500098687946,
      "loss": 2.1353,
      "step": 17482
    },
    {
      "epoch": 0.11878556150391163,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009695463126379067,
      "loss": 2.0693,
      "step": 17483
    },
    {
      "epoch": 0.1187923558505057,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009695426151896243,
      "loss": 2.2816,
      "step": 17484
    },
    {
      "epoch": 0.11879915019709975,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009695389175239491,
      "loss": 2.222,
      "step": 17485
    },
    {
      "epoch": 0.11880594454369381,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009695352196408826,
      "loss": 2.1407,
      "step": 17486
    },
    {
      "epoch": 0.11881273889028786,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009695315215404268,
      "loss": 2.1337,
      "step": 17487
    },
    {
      "epoch": 0.11881953323688192,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009695278232225833,
      "loss": 2.1543,
      "step": 17488
    },
    {
      "epoch": 0.11882632758347599,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009695241246873536,
      "loss": 2.0602,
      "step": 17489
    },
    {
      "epoch": 0.11883312193007003,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009695204259347398,
      "loss": 2.212,
      "step": 17490
    },
    {
      "epoch": 0.1188399162766641,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009695167269647433,
      "loss": 2.1058,
      "step": 17491
    },
    {
      "epoch": 0.11884671062325815,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009695130277773661,
      "loss": 2.0466,
      "step": 17492
    },
    {
      "epoch": 0.11885350496985221,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009695093283726097,
      "loss": 2.1368,
      "step": 17493
    },
    {
      "epoch": 0.11886029931644627,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009695056287504758,
      "loss": 2.2659,
      "step": 17494
    },
    {
      "epoch": 0.11886709366304032,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009695019289109663,
      "loss": 2.2798,
      "step": 17495
    },
    {
      "epoch": 0.11887388800963439,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009694982288540826,
      "loss": 2.1894,
      "step": 17496
    },
    {
      "epoch": 0.11888068235622844,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009694945285798267,
      "loss": 2.1639,
      "step": 17497
    },
    {
      "epoch": 0.1188874767028225,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009694908280882003,
      "loss": 2.1696,
      "step": 17498
    },
    {
      "epoch": 0.11889427104941655,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009694871273792048,
      "loss": 2.1894,
      "step": 17499
    },
    {
      "epoch": 0.11890106539601061,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009694834264528424,
      "loss": 2.1788,
      "step": 17500
    },
    {
      "epoch": 0.11890785974260468,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009694797253091144,
      "loss": 2.1554,
      "step": 17501
    },
    {
      "epoch": 0.11891465408919873,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009694760239480227,
      "loss": 2.2961,
      "step": 17502
    },
    {
      "epoch": 0.11892144843579279,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009694723223695691,
      "loss": 2.2981,
      "step": 17503
    },
    {
      "epoch": 0.11892824278238684,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009694686205737551,
      "loss": 2.2295,
      "step": 17504
    },
    {
      "epoch": 0.1189350371289809,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009694649185605824,
      "loss": 2.2704,
      "step": 17505
    },
    {
      "epoch": 0.11894183147557497,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000969461216330053,
      "loss": 2.3158,
      "step": 17506
    },
    {
      "epoch": 0.11894862582216902,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009694575138821684,
      "loss": 2.1795,
      "step": 17507
    },
    {
      "epoch": 0.11895542016876308,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009694538112169302,
      "loss": 2.3151,
      "step": 17508
    },
    {
      "epoch": 0.11896221451535713,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009694501083343405,
      "loss": 2.2903,
      "step": 17509
    },
    {
      "epoch": 0.11896900886195119,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009694464052344006,
      "loss": 2.1757,
      "step": 17510
    },
    {
      "epoch": 0.11897580320854526,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009694427019171125,
      "loss": 2.1127,
      "step": 17511
    },
    {
      "epoch": 0.1189825975551393,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009694389983824778,
      "loss": 2.3435,
      "step": 17512
    },
    {
      "epoch": 0.11898939190173337,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009694352946304982,
      "loss": 2.0504,
      "step": 17513
    },
    {
      "epoch": 0.11899618624832742,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009694315906611754,
      "loss": 2.21,
      "step": 17514
    },
    {
      "epoch": 0.11900298059492148,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009694278864745112,
      "loss": 2.1163,
      "step": 17515
    },
    {
      "epoch": 0.11900977494151554,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009694241820705073,
      "loss": 2.1622,
      "step": 17516
    },
    {
      "epoch": 0.1190165692881096,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009694204774491654,
      "loss": 2.17,
      "step": 17517
    },
    {
      "epoch": 0.11902336363470366,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009694167726104872,
      "loss": 2.1462,
      "step": 17518
    },
    {
      "epoch": 0.11903015798129771,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009694130675544743,
      "loss": 2.1778,
      "step": 17519
    },
    {
      "epoch": 0.11903695232789177,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009694093622811287,
      "loss": 2.1878,
      "step": 17520
    },
    {
      "epoch": 0.11904374667448583,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009694056567904518,
      "loss": 2.1936,
      "step": 17521
    },
    {
      "epoch": 0.11905054102107988,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009694019510824455,
      "loss": 2.1994,
      "step": 17522
    },
    {
      "epoch": 0.11905733536767395,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009693982451571115,
      "loss": 2.1799,
      "step": 17523
    },
    {
      "epoch": 0.119064129714268,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009693945390144514,
      "loss": 2.1597,
      "step": 17524
    },
    {
      "epoch": 0.11907092406086206,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009693908326544672,
      "loss": 2.1393,
      "step": 17525
    },
    {
      "epoch": 0.11907771840745611,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009693871260771603,
      "loss": 2.1549,
      "step": 17526
    },
    {
      "epoch": 0.11908451275405017,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009693834192825326,
      "loss": 2.2276,
      "step": 17527
    },
    {
      "epoch": 0.11909130710064424,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009693797122705856,
      "loss": 2.1765,
      "step": 17528
    },
    {
      "epoch": 0.11909810144723829,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009693760050413214,
      "loss": 2.2509,
      "step": 17529
    },
    {
      "epoch": 0.11910489579383235,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009693722975947414,
      "loss": 2.2079,
      "step": 17530
    },
    {
      "epoch": 0.1191116901404264,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009693685899308473,
      "loss": 2.3005,
      "step": 17531
    },
    {
      "epoch": 0.11911848448702046,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000969364882049641,
      "loss": 2.2174,
      "step": 17532
    },
    {
      "epoch": 0.11912527883361453,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009693611739511242,
      "loss": 2.0351,
      "step": 17533
    },
    {
      "epoch": 0.11913207318020858,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009693574656352986,
      "loss": 2.3116,
      "step": 17534
    },
    {
      "epoch": 0.11913886752680264,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009693537571021657,
      "loss": 2.2692,
      "step": 17535
    },
    {
      "epoch": 0.11914566187339669,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009693500483517275,
      "loss": 2.2558,
      "step": 17536
    },
    {
      "epoch": 0.11915245621999075,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009693463393839856,
      "loss": 2.1528,
      "step": 17537
    },
    {
      "epoch": 0.11915925056658482,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009693426301989417,
      "loss": 2.1112,
      "step": 17538
    },
    {
      "epoch": 0.11916604491317886,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009693389207965975,
      "loss": 2.2204,
      "step": 17539
    },
    {
      "epoch": 0.11917283925977293,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009693352111769549,
      "loss": 2.2404,
      "step": 17540
    },
    {
      "epoch": 0.11917963360636698,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009693315013400154,
      "loss": 1.9861,
      "step": 17541
    },
    {
      "epoch": 0.11918642795296104,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009693277912857808,
      "loss": 2.2571,
      "step": 17542
    },
    {
      "epoch": 0.1191932222995551,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009693240810142527,
      "loss": 2.2609,
      "step": 17543
    },
    {
      "epoch": 0.11920001664614915,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009693203705254332,
      "loss": 2.2083,
      "step": 17544
    },
    {
      "epoch": 0.11920681099274322,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009693166598193234,
      "loss": 2.1952,
      "step": 17545
    },
    {
      "epoch": 0.11921360533933727,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009693129488959257,
      "loss": 2.1217,
      "step": 17546
    },
    {
      "epoch": 0.11922039968593133,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009693092377552413,
      "loss": 2.1153,
      "step": 17547
    },
    {
      "epoch": 0.11922719403252538,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009693055263972722,
      "loss": 2.2224,
      "step": 17548
    },
    {
      "epoch": 0.11923398837911944,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00096930181482202,
      "loss": 2.0732,
      "step": 17549
    },
    {
      "epoch": 0.1192407827257135,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009692981030294863,
      "loss": 2.1392,
      "step": 17550
    },
    {
      "epoch": 0.11924757707230756,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009692943910196731,
      "loss": 2.1103,
      "step": 17551
    },
    {
      "epoch": 0.11925437141890162,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000969290678792582,
      "loss": 2.2302,
      "step": 17552
    },
    {
      "epoch": 0.11926116576549567,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009692869663482145,
      "loss": 2.0576,
      "step": 17553
    },
    {
      "epoch": 0.11926796011208973,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009692832536865728,
      "loss": 2.1961,
      "step": 17554
    },
    {
      "epoch": 0.1192747544586838,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009692795408076581,
      "loss": 1.9864,
      "step": 17555
    },
    {
      "epoch": 0.11928154880527785,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009692758277114725,
      "loss": 2.2892,
      "step": 17556
    },
    {
      "epoch": 0.11928834315187191,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009692721143980174,
      "loss": 2.3015,
      "step": 17557
    },
    {
      "epoch": 0.11929513749846596,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000969268400867295,
      "loss": 2.2091,
      "step": 17558
    },
    {
      "epoch": 0.11930193184506002,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009692646871193064,
      "loss": 2.4154,
      "step": 17559
    },
    {
      "epoch": 0.11930872619165409,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009692609731540537,
      "loss": 2.2117,
      "step": 17560
    },
    {
      "epoch": 0.11931552053824813,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009692572589715386,
      "loss": 2.2278,
      "step": 17561
    },
    {
      "epoch": 0.1193223148848422,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009692535445717627,
      "loss": 2.1583,
      "step": 17562
    },
    {
      "epoch": 0.11932910923143625,
      "grad_norm": 0.125,
      "learning_rate": 0.0009692498299547279,
      "loss": 2.1246,
      "step": 17563
    },
    {
      "epoch": 0.11933590357803031,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009692461151204358,
      "loss": 2.2993,
      "step": 17564
    },
    {
      "epoch": 0.11934269792462437,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009692424000688881,
      "loss": 2.2012,
      "step": 17565
    },
    {
      "epoch": 0.11934949227121842,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009692386848000865,
      "loss": 2.2142,
      "step": 17566
    },
    {
      "epoch": 0.11935628661781249,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009692349693140328,
      "loss": 2.2007,
      "step": 17567
    },
    {
      "epoch": 0.11936308096440654,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009692312536107286,
      "loss": 2.0625,
      "step": 17568
    },
    {
      "epoch": 0.1193698753110006,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009692275376901758,
      "loss": 2.0906,
      "step": 17569
    },
    {
      "epoch": 0.11937666965759465,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009692238215523761,
      "loss": 2.2554,
      "step": 17570
    },
    {
      "epoch": 0.11938346400418871,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000969220105197331,
      "loss": 2.125,
      "step": 17571
    },
    {
      "epoch": 0.11939025835078278,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009692163886250425,
      "loss": 2.3196,
      "step": 17572
    },
    {
      "epoch": 0.11939705269737683,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009692126718355121,
      "loss": 2.0802,
      "step": 17573
    },
    {
      "epoch": 0.11940384704397089,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009692089548287417,
      "loss": 2.1345,
      "step": 17574
    },
    {
      "epoch": 0.11941064139056494,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009692052376047329,
      "loss": 2.1414,
      "step": 17575
    },
    {
      "epoch": 0.119417435737159,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009692015201634874,
      "loss": 2.0681,
      "step": 17576
    },
    {
      "epoch": 0.11942423008375307,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000969197802505007,
      "loss": 2.2108,
      "step": 17577
    },
    {
      "epoch": 0.11943102443034712,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009691940846292935,
      "loss": 2.1911,
      "step": 17578
    },
    {
      "epoch": 0.11943781877694118,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009691903665363483,
      "loss": 2.1814,
      "step": 17579
    },
    {
      "epoch": 0.11944461312353523,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009691866482261735,
      "loss": 2.3545,
      "step": 17580
    },
    {
      "epoch": 0.11945140747012929,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009691829296987706,
      "loss": 2.1444,
      "step": 17581
    },
    {
      "epoch": 0.11945820181672336,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009691792109541416,
      "loss": 2.244,
      "step": 17582
    },
    {
      "epoch": 0.1194649961633174,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009691754919922877,
      "loss": 2.1887,
      "step": 17583
    },
    {
      "epoch": 0.11947179050991147,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000969171772813211,
      "loss": 2.1541,
      "step": 17584
    },
    {
      "epoch": 0.11947858485650552,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009691680534169133,
      "loss": 2.2286,
      "step": 17585
    },
    {
      "epoch": 0.11948537920309958,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000969164333803396,
      "loss": 2.3298,
      "step": 17586
    },
    {
      "epoch": 0.11949217354969364,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009691606139726611,
      "loss": 2.2915,
      "step": 17587
    },
    {
      "epoch": 0.1194989678962877,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009691568939247102,
      "loss": 2.0874,
      "step": 17588
    },
    {
      "epoch": 0.11950576224288176,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009691531736595449,
      "loss": 2.1833,
      "step": 17589
    },
    {
      "epoch": 0.11951255658947581,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009691494531771672,
      "loss": 2.1813,
      "step": 17590
    },
    {
      "epoch": 0.11951935093606987,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009691457324775786,
      "loss": 2.2556,
      "step": 17591
    },
    {
      "epoch": 0.11952614528266393,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000969142011560781,
      "loss": 2.1866,
      "step": 17592
    },
    {
      "epoch": 0.11953293962925798,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009691382904267761,
      "loss": 2.1534,
      "step": 17593
    },
    {
      "epoch": 0.11953973397585205,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009691345690755653,
      "loss": 2.1872,
      "step": 17594
    },
    {
      "epoch": 0.1195465283224461,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009691308475071507,
      "loss": 2.2219,
      "step": 17595
    },
    {
      "epoch": 0.11955332266904016,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000969127125721534,
      "loss": 2.2632,
      "step": 17596
    },
    {
      "epoch": 0.11956011701563421,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009691234037187167,
      "loss": 2.1969,
      "step": 17597
    },
    {
      "epoch": 0.11956691136222827,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009691196814987006,
      "loss": 2.1542,
      "step": 17598
    },
    {
      "epoch": 0.11957370570882234,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009691159590614876,
      "loss": 2.1783,
      "step": 17599
    },
    {
      "epoch": 0.11958050005541639,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009691122364070791,
      "loss": 2.2148,
      "step": 17600
    },
    {
      "epoch": 0.11958729440201045,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009691085135354772,
      "loss": 2.2332,
      "step": 17601
    },
    {
      "epoch": 0.1195940887486045,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009691047904466833,
      "loss": 2.2318,
      "step": 17602
    },
    {
      "epoch": 0.11960088309519856,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009691010671406993,
      "loss": 2.0354,
      "step": 17603
    },
    {
      "epoch": 0.11960767744179263,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000969097343617527,
      "loss": 2.084,
      "step": 17604
    },
    {
      "epoch": 0.11961447178838668,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009690936198771678,
      "loss": 2.2948,
      "step": 17605
    },
    {
      "epoch": 0.11962126613498074,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009690898959196237,
      "loss": 2.2458,
      "step": 17606
    },
    {
      "epoch": 0.11962806048157479,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009690861717448965,
      "loss": 2.0631,
      "step": 17607
    },
    {
      "epoch": 0.11963485482816885,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009690824473529877,
      "loss": 1.9294,
      "step": 17608
    },
    {
      "epoch": 0.11964164917476292,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009690787227438991,
      "loss": 2.3351,
      "step": 17609
    },
    {
      "epoch": 0.11964844352135696,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009690749979176323,
      "loss": 2.3634,
      "step": 17610
    },
    {
      "epoch": 0.11965523786795103,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009690712728741892,
      "loss": 2.1792,
      "step": 17611
    },
    {
      "epoch": 0.11966203221454508,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009690675476135718,
      "loss": 2.1714,
      "step": 17612
    },
    {
      "epoch": 0.11966882656113914,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009690638221357812,
      "loss": 2.104,
      "step": 17613
    },
    {
      "epoch": 0.1196756209077332,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009690600964408194,
      "loss": 2.1957,
      "step": 17614
    },
    {
      "epoch": 0.11968241525432725,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009690563705286882,
      "loss": 2.1343,
      "step": 17615
    },
    {
      "epoch": 0.11968920960092132,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009690526443993893,
      "loss": 2.0464,
      "step": 17616
    },
    {
      "epoch": 0.11969600394751537,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009690489180529245,
      "loss": 2.2671,
      "step": 17617
    },
    {
      "epoch": 0.11970279829410943,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009690451914892953,
      "loss": 2.2582,
      "step": 17618
    },
    {
      "epoch": 0.11970959264070348,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009690414647085035,
      "loss": 2.2946,
      "step": 17619
    },
    {
      "epoch": 0.11971638698729754,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.000969037737710551,
      "loss": 2.0888,
      "step": 17620
    },
    {
      "epoch": 0.1197231813338916,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009690340104954394,
      "loss": 2.2221,
      "step": 17621
    },
    {
      "epoch": 0.11972997568048566,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009690302830631705,
      "loss": 2.2533,
      "step": 17622
    },
    {
      "epoch": 0.11973677002707972,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009690265554137458,
      "loss": 2.2455,
      "step": 17623
    },
    {
      "epoch": 0.11974356437367377,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009690228275471673,
      "loss": 2.1299,
      "step": 17624
    },
    {
      "epoch": 0.11975035872026783,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009690190994634365,
      "loss": 2.13,
      "step": 17625
    },
    {
      "epoch": 0.1197571530668619,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009690153711625554,
      "loss": 2.1519,
      "step": 17626
    },
    {
      "epoch": 0.11976394741345595,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009690116426445254,
      "loss": 2.1171,
      "step": 17627
    },
    {
      "epoch": 0.11977074176005001,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009690079139093484,
      "loss": 2.1173,
      "step": 17628
    },
    {
      "epoch": 0.11977753610664406,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009690041849570261,
      "loss": 2.3351,
      "step": 17629
    },
    {
      "epoch": 0.11978433045323812,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009690004557875603,
      "loss": 2.2021,
      "step": 17630
    },
    {
      "epoch": 0.11979112479983219,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009689967264009527,
      "loss": 2.1345,
      "step": 17631
    },
    {
      "epoch": 0.11979791914642623,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009689929967972049,
      "loss": 2.179,
      "step": 17632
    },
    {
      "epoch": 0.1198047134930203,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009689892669763187,
      "loss": 2.231,
      "step": 17633
    },
    {
      "epoch": 0.11981150783961435,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000968985536938296,
      "loss": 2.2612,
      "step": 17634
    },
    {
      "epoch": 0.11981830218620841,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009689818066831383,
      "loss": 2.1228,
      "step": 17635
    },
    {
      "epoch": 0.11982509653280247,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009689780762108473,
      "loss": 2.2181,
      "step": 17636
    },
    {
      "epoch": 0.11983189087939652,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000968974345521425,
      "loss": 2.2953,
      "step": 17637
    },
    {
      "epoch": 0.11983868522599059,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009689706146148729,
      "loss": 2.159,
      "step": 17638
    },
    {
      "epoch": 0.11984547957258464,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009689668834911927,
      "loss": 2.206,
      "step": 17639
    },
    {
      "epoch": 0.1198522739191787,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009689631521503862,
      "loss": 2.2157,
      "step": 17640
    },
    {
      "epoch": 0.11985906826577275,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009689594205924552,
      "loss": 2.2292,
      "step": 17641
    },
    {
      "epoch": 0.11986586261236681,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009689556888174014,
      "loss": 2.1574,
      "step": 17642
    },
    {
      "epoch": 0.11987265695896088,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009689519568252265,
      "loss": 2.2396,
      "step": 17643
    },
    {
      "epoch": 0.11987945130555493,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009689482246159322,
      "loss": 2.1703,
      "step": 17644
    },
    {
      "epoch": 0.11988624565214899,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009689444921895201,
      "loss": 2.2412,
      "step": 17645
    },
    {
      "epoch": 0.11989303999874304,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009689407595459922,
      "loss": 2.2013,
      "step": 17646
    },
    {
      "epoch": 0.1198998343453371,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009689370266853501,
      "loss": 2.2534,
      "step": 17647
    },
    {
      "epoch": 0.11990662869193117,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009689332936075956,
      "loss": 2.3562,
      "step": 17648
    },
    {
      "epoch": 0.11991342303852522,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009689295603127302,
      "loss": 2.1806,
      "step": 17649
    },
    {
      "epoch": 0.11992021738511928,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009689258268007559,
      "loss": 2.0034,
      "step": 17650
    },
    {
      "epoch": 0.11992701173171333,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009689220930716744,
      "loss": 2.1633,
      "step": 17651
    },
    {
      "epoch": 0.11993380607830739,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009689183591254872,
      "loss": 2.1546,
      "step": 17652
    },
    {
      "epoch": 0.11994060042490146,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009689146249621963,
      "loss": 2.1582,
      "step": 17653
    },
    {
      "epoch": 0.1199473947714955,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009689108905818032,
      "loss": 2.2071,
      "step": 17654
    },
    {
      "epoch": 0.11995418911808957,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009689071559843096,
      "loss": 2.0606,
      "step": 17655
    },
    {
      "epoch": 0.11996098346468362,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009689034211697176,
      "loss": 2.2082,
      "step": 17656
    },
    {
      "epoch": 0.11996777781127768,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009688996861380286,
      "loss": 2.2266,
      "step": 17657
    },
    {
      "epoch": 0.11997457215787175,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009688959508892445,
      "loss": 2.2112,
      "step": 17658
    },
    {
      "epoch": 0.1199813665044658,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009688922154233668,
      "loss": 2.1547,
      "step": 17659
    },
    {
      "epoch": 0.11998816085105986,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009688884797403976,
      "loss": 2.1419,
      "step": 17660
    },
    {
      "epoch": 0.11999495519765391,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009688847438403382,
      "loss": 2.0901,
      "step": 17661
    },
    {
      "epoch": 0.12000174954424797,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009688810077231907,
      "loss": 2.2699,
      "step": 17662
    },
    {
      "epoch": 0.12000854389084203,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009688772713889565,
      "loss": 2.0934,
      "step": 17663
    },
    {
      "epoch": 0.12001533823743608,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009688735348376376,
      "loss": 2.2312,
      "step": 17664
    },
    {
      "epoch": 0.12002213258403015,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009688697980692356,
      "loss": 2.2124,
      "step": 17665
    },
    {
      "epoch": 0.1200289269306242,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009688660610837523,
      "loss": 2.2048,
      "step": 17666
    },
    {
      "epoch": 0.12003572127721826,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009688623238811894,
      "loss": 2.2777,
      "step": 17667
    },
    {
      "epoch": 0.12004251562381231,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009688585864615486,
      "loss": 2.1018,
      "step": 17668
    },
    {
      "epoch": 0.12004930997040637,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009688548488248315,
      "loss": 2.1332,
      "step": 17669
    },
    {
      "epoch": 0.12005610431700044,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009688511109710401,
      "loss": 2.1477,
      "step": 17670
    },
    {
      "epoch": 0.12006289866359449,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009688473729001761,
      "loss": 2.0727,
      "step": 17671
    },
    {
      "epoch": 0.12006969301018855,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009688436346122412,
      "loss": 2.0397,
      "step": 17672
    },
    {
      "epoch": 0.1200764873567826,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009688398961072368,
      "loss": 2.2672,
      "step": 17673
    },
    {
      "epoch": 0.12008328170337666,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009688361573851651,
      "loss": 2.2168,
      "step": 17674
    },
    {
      "epoch": 0.12009007604997073,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009688324184460276,
      "loss": 2.0305,
      "step": 17675
    },
    {
      "epoch": 0.12009687039656478,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000968828679289826,
      "loss": 2.1655,
      "step": 17676
    },
    {
      "epoch": 0.12010366474315884,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009688249399165621,
      "loss": 2.092,
      "step": 17677
    },
    {
      "epoch": 0.12011045908975289,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009688212003262376,
      "loss": 2.2768,
      "step": 17678
    },
    {
      "epoch": 0.12011725343634695,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009688174605188544,
      "loss": 2.4543,
      "step": 17679
    },
    {
      "epoch": 0.12012404778294102,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009688137204944139,
      "loss": 2.1989,
      "step": 17680
    },
    {
      "epoch": 0.12013084212953506,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009688099802529182,
      "loss": 2.1242,
      "step": 17681
    },
    {
      "epoch": 0.12013763647612913,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009688062397943687,
      "loss": 2.1756,
      "step": 17682
    },
    {
      "epoch": 0.12014443082272318,
      "grad_norm": 0.125,
      "learning_rate": 0.0009688024991187674,
      "loss": 2.0779,
      "step": 17683
    },
    {
      "epoch": 0.12015122516931724,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000968798758226116,
      "loss": 2.1123,
      "step": 17684
    },
    {
      "epoch": 0.1201580195159113,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000968795017116416,
      "loss": 2.1938,
      "step": 17685
    },
    {
      "epoch": 0.12016481386250535,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009687912757896693,
      "loss": 2.2861,
      "step": 17686
    },
    {
      "epoch": 0.12017160820909942,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009687875342458776,
      "loss": 2.2776,
      "step": 17687
    },
    {
      "epoch": 0.12017840255569347,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009687837924850426,
      "loss": 2.1445,
      "step": 17688
    },
    {
      "epoch": 0.12018519690228753,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009687800505071662,
      "loss": 2.1646,
      "step": 17689
    },
    {
      "epoch": 0.12019199124888158,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00096877630831225,
      "loss": 2.191,
      "step": 17690
    },
    {
      "epoch": 0.12019878559547564,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009687725659002957,
      "loss": 2.0503,
      "step": 17691
    },
    {
      "epoch": 0.1202055799420697,
      "grad_norm": 0.16796875,
      "learning_rate": 0.0009687688232713052,
      "loss": 2.3128,
      "step": 17692
    },
    {
      "epoch": 0.12021237428866376,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009687650804252798,
      "loss": 2.1605,
      "step": 17693
    },
    {
      "epoch": 0.12021916863525782,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000968761337362222,
      "loss": 2.4175,
      "step": 17694
    },
    {
      "epoch": 0.12022596298185187,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009687575940821327,
      "loss": 2.2178,
      "step": 17695
    },
    {
      "epoch": 0.12023275732844593,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009687538505850142,
      "loss": 2.2149,
      "step": 17696
    },
    {
      "epoch": 0.12023955167504,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009687501068708679,
      "loss": 2.2118,
      "step": 17697
    },
    {
      "epoch": 0.12024634602163405,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009687463629396958,
      "loss": 2.1303,
      "step": 17698
    },
    {
      "epoch": 0.12025314036822811,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009687426187914994,
      "loss": 2.2325,
      "step": 17699
    },
    {
      "epoch": 0.12025993471482216,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009687388744262806,
      "loss": 2.1654,
      "step": 17700
    },
    {
      "epoch": 0.12026672906141622,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000968735129844041,
      "loss": 2.3261,
      "step": 17701
    },
    {
      "epoch": 0.12027352340801029,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009687313850447825,
      "loss": 2.2163,
      "step": 17702
    },
    {
      "epoch": 0.12028031775460434,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009687276400285068,
      "loss": 2.1786,
      "step": 17703
    },
    {
      "epoch": 0.1202871121011984,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009687238947952155,
      "loss": 2.2287,
      "step": 17704
    },
    {
      "epoch": 0.12029390644779245,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009687201493449103,
      "loss": 2.1489,
      "step": 17705
    },
    {
      "epoch": 0.12030070079438651,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009687164036775932,
      "loss": 1.9983,
      "step": 17706
    },
    {
      "epoch": 0.12030749514098057,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009687126577932656,
      "loss": 2.2693,
      "step": 17707
    },
    {
      "epoch": 0.12031428948757462,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009687089116919296,
      "loss": 2.0989,
      "step": 17708
    },
    {
      "epoch": 0.12032108383416869,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009687051653735867,
      "loss": 2.3178,
      "step": 17709
    },
    {
      "epoch": 0.12032787818076274,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009687014188382387,
      "loss": 2.1641,
      "step": 17710
    },
    {
      "epoch": 0.1203346725273568,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009686976720858873,
      "loss": 2.1001,
      "step": 17711
    },
    {
      "epoch": 0.12034146687395086,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009686939251165342,
      "loss": 2.2214,
      "step": 17712
    },
    {
      "epoch": 0.12034826122054491,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009686901779301812,
      "loss": 2.1303,
      "step": 17713
    },
    {
      "epoch": 0.12035505556713898,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009686864305268302,
      "loss": 2.2167,
      "step": 17714
    },
    {
      "epoch": 0.12036184991373303,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009686826829064826,
      "loss": 1.9625,
      "step": 17715
    },
    {
      "epoch": 0.12036864426032709,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009686789350691402,
      "loss": 2.1647,
      "step": 17716
    },
    {
      "epoch": 0.12037543860692114,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009686751870148049,
      "loss": 2.1,
      "step": 17717
    },
    {
      "epoch": 0.1203822329535152,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009686714387434783,
      "loss": 2.1489,
      "step": 17718
    },
    {
      "epoch": 0.12038902730010927,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009686676902551624,
      "loss": 2.1184,
      "step": 17719
    },
    {
      "epoch": 0.12039582164670332,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009686639415498586,
      "loss": 2.0062,
      "step": 17720
    },
    {
      "epoch": 0.12040261599329738,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009686601926275686,
      "loss": 2.2599,
      "step": 17721
    },
    {
      "epoch": 0.12040941033989143,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009686564434882945,
      "loss": 2.2984,
      "step": 17722
    },
    {
      "epoch": 0.12041620468648549,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000968652694132038,
      "loss": 2.282,
      "step": 17723
    },
    {
      "epoch": 0.12042299903307956,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009686489445588003,
      "loss": 2.1079,
      "step": 17724
    },
    {
      "epoch": 0.1204297933796736,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009686451947685837,
      "loss": 2.1641,
      "step": 17725
    },
    {
      "epoch": 0.12043658772626767,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009686414447613898,
      "loss": 2.0267,
      "step": 17726
    },
    {
      "epoch": 0.12044338207286172,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009686376945372203,
      "loss": 2.1151,
      "step": 17727
    },
    {
      "epoch": 0.12045017641945578,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009686339440960768,
      "loss": 2.2839,
      "step": 17728
    },
    {
      "epoch": 0.12045697076604985,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009686301934379613,
      "loss": 2.0913,
      "step": 17729
    },
    {
      "epoch": 0.1204637651126439,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009686264425628752,
      "loss": 2.123,
      "step": 17730
    },
    {
      "epoch": 0.12047055945923796,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009686226914708206,
      "loss": 2.1752,
      "step": 17731
    },
    {
      "epoch": 0.12047735380583201,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000968618940161799,
      "loss": 2.2088,
      "step": 17732
    },
    {
      "epoch": 0.12048414815242607,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009686151886358122,
      "loss": 2.2232,
      "step": 17733
    },
    {
      "epoch": 0.12049094249902013,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000968611436892862,
      "loss": 2.2027,
      "step": 17734
    },
    {
      "epoch": 0.12049773684561418,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009686076849329499,
      "loss": 2.1539,
      "step": 17735
    },
    {
      "epoch": 0.12050453119220825,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009686039327560781,
      "loss": 2.1616,
      "step": 17736
    },
    {
      "epoch": 0.1205113255388023,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009686001803622477,
      "loss": 2.202,
      "step": 17737
    },
    {
      "epoch": 0.12051811988539636,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000968596427751461,
      "loss": 2.1551,
      "step": 17738
    },
    {
      "epoch": 0.12052491423199041,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009685926749237195,
      "loss": 2.106,
      "step": 17739
    },
    {
      "epoch": 0.12053170857858447,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000968588921879025,
      "loss": 2.1263,
      "step": 17740
    },
    {
      "epoch": 0.12053850292517854,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009685851686173792,
      "loss": 2.3548,
      "step": 17741
    },
    {
      "epoch": 0.12054529727177259,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009685814151387837,
      "loss": 2.1068,
      "step": 17742
    },
    {
      "epoch": 0.12055209161836665,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009685776614432407,
      "loss": 2.1435,
      "step": 17743
    },
    {
      "epoch": 0.1205588859649607,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009685739075307513,
      "loss": 2.1814,
      "step": 17744
    },
    {
      "epoch": 0.12056568031155476,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009685701534013176,
      "loss": 2.2702,
      "step": 17745
    },
    {
      "epoch": 0.12057247465814883,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009685663990549413,
      "loss": 2.348,
      "step": 17746
    },
    {
      "epoch": 0.12057926900474288,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009685626444916242,
      "loss": 2.1747,
      "step": 17747
    },
    {
      "epoch": 0.12058606335133694,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000968558889711368,
      "loss": 2.0752,
      "step": 17748
    },
    {
      "epoch": 0.12059285769793099,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009685551347141743,
      "loss": 2.2085,
      "step": 17749
    },
    {
      "epoch": 0.12059965204452505,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000968551379500045,
      "loss": 2.1572,
      "step": 17750
    },
    {
      "epoch": 0.12060644639111912,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009685476240689818,
      "loss": 2.1535,
      "step": 17751
    },
    {
      "epoch": 0.12061324073771316,
      "grad_norm": 0.125,
      "learning_rate": 0.0009685438684209864,
      "loss": 2.1311,
      "step": 17752
    },
    {
      "epoch": 0.12062003508430723,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009685401125560605,
      "loss": 2.3123,
      "step": 17753
    },
    {
      "epoch": 0.12062682943090128,
      "grad_norm": 0.140625,
      "learning_rate": 0.000968536356474206,
      "loss": 2.0649,
      "step": 17754
    },
    {
      "epoch": 0.12063362377749534,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009685326001754245,
      "loss": 2.3755,
      "step": 17755
    },
    {
      "epoch": 0.1206404181240894,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009685288436597177,
      "loss": 2.2329,
      "step": 17756
    },
    {
      "epoch": 0.12064721247068345,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009685250869270876,
      "loss": 2.0464,
      "step": 17757
    },
    {
      "epoch": 0.12065400681727752,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009685213299775355,
      "loss": 2.2766,
      "step": 17758
    },
    {
      "epoch": 0.12066080116387157,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009685175728110636,
      "loss": 2.1983,
      "step": 17759
    },
    {
      "epoch": 0.12066759551046563,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009685138154276734,
      "loss": 2.29,
      "step": 17760
    },
    {
      "epoch": 0.12067438985705968,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009685100578273667,
      "loss": 2.1558,
      "step": 17761
    },
    {
      "epoch": 0.12068118420365374,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000968506300010145,
      "loss": 2.1763,
      "step": 17762
    },
    {
      "epoch": 0.12068797855024781,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009685025419760105,
      "loss": 2.2607,
      "step": 17763
    },
    {
      "epoch": 0.12069477289684186,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009684987837249646,
      "loss": 2.0771,
      "step": 17764
    },
    {
      "epoch": 0.12070156724343592,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009684950252570091,
      "loss": 2.4454,
      "step": 17765
    },
    {
      "epoch": 0.12070836159002997,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009684912665721459,
      "loss": 2.2714,
      "step": 17766
    },
    {
      "epoch": 0.12071515593662403,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009684875076703765,
      "loss": 2.1772,
      "step": 17767
    },
    {
      "epoch": 0.1207219502832181,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009684837485517028,
      "loss": 2.2174,
      "step": 17768
    },
    {
      "epoch": 0.12072874462981215,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009684799892161265,
      "loss": 2.2468,
      "step": 17769
    },
    {
      "epoch": 0.12073553897640621,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009684762296636493,
      "loss": 2.2574,
      "step": 17770
    },
    {
      "epoch": 0.12074233332300026,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009684724698942731,
      "loss": 2.4273,
      "step": 17771
    },
    {
      "epoch": 0.12074912766959432,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009684687099079994,
      "loss": 2.1837,
      "step": 17772
    },
    {
      "epoch": 0.12075592201618839,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.00096846494970483,
      "loss": 2.1746,
      "step": 17773
    },
    {
      "epoch": 0.12076271636278244,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009684611892847668,
      "loss": 2.097,
      "step": 17774
    },
    {
      "epoch": 0.1207695107093765,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009684574286478114,
      "loss": 2.1453,
      "step": 17775
    },
    {
      "epoch": 0.12077630505597055,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009684536677939656,
      "loss": 2.1729,
      "step": 17776
    },
    {
      "epoch": 0.12078309940256461,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000968449906723231,
      "loss": 2.1288,
      "step": 17777
    },
    {
      "epoch": 0.12078989374915867,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009684461454356097,
      "loss": 2.2735,
      "step": 17778
    },
    {
      "epoch": 0.12079668809575272,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009684423839311031,
      "loss": 2.0089,
      "step": 17779
    },
    {
      "epoch": 0.12080348244234679,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000968438622209713,
      "loss": 2.2193,
      "step": 17780
    },
    {
      "epoch": 0.12081027678894084,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009684348602714412,
      "loss": 2.1124,
      "step": 17781
    },
    {
      "epoch": 0.1208170711355349,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009684310981162894,
      "loss": 2.1516,
      "step": 17782
    },
    {
      "epoch": 0.12082386548212896,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009684273357442594,
      "loss": 2.1376,
      "step": 17783
    },
    {
      "epoch": 0.12083065982872301,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000968423573155353,
      "loss": 2.1208,
      "step": 17784
    },
    {
      "epoch": 0.12083745417531708,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009684198103495718,
      "loss": 2.1847,
      "step": 17785
    },
    {
      "epoch": 0.12084424852191113,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009684160473269176,
      "loss": 2.014,
      "step": 17786
    },
    {
      "epoch": 0.12085104286850519,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009684122840873922,
      "loss": 2.2014,
      "step": 17787
    },
    {
      "epoch": 0.12085783721509924,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009684085206309971,
      "loss": 2.2847,
      "step": 17788
    },
    {
      "epoch": 0.1208646315616933,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009684047569577343,
      "loss": 2.3657,
      "step": 17789
    },
    {
      "epoch": 0.12087142590828737,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009684009930676057,
      "loss": 2.2883,
      "step": 17790
    },
    {
      "epoch": 0.12087822025488142,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009683972289606125,
      "loss": 2.2577,
      "step": 17791
    },
    {
      "epoch": 0.12088501460147548,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009683934646367569,
      "loss": 2.1203,
      "step": 17792
    },
    {
      "epoch": 0.12089180894806953,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009683897000960404,
      "loss": 2.1179,
      "step": 17793
    },
    {
      "epoch": 0.12089860329466359,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000968385935338465,
      "loss": 2.2453,
      "step": 17794
    },
    {
      "epoch": 0.12090539764125766,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009683821703640321,
      "loss": 2.2976,
      "step": 17795
    },
    {
      "epoch": 0.1209121919878517,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009683784051727438,
      "loss": 2.2574,
      "step": 17796
    },
    {
      "epoch": 0.12091898633444577,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009683746397646015,
      "loss": 2.27,
      "step": 17797
    },
    {
      "epoch": 0.12092578068103982,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009683708741396071,
      "loss": 2.2608,
      "step": 17798
    },
    {
      "epoch": 0.12093257502763388,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009683671082977625,
      "loss": 2.2099,
      "step": 17799
    },
    {
      "epoch": 0.12093936937422795,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009683633422390692,
      "loss": 2.2338,
      "step": 17800
    },
    {
      "epoch": 0.120946163720822,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009683595759635291,
      "loss": 2.2404,
      "step": 17801
    },
    {
      "epoch": 0.12095295806741606,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000968355809471144,
      "loss": 2.1858,
      "step": 17802
    },
    {
      "epoch": 0.12095975241401011,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009683520427619152,
      "loss": 2.2373,
      "step": 17803
    },
    {
      "epoch": 0.12096654676060417,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009683482758358451,
      "loss": 2.1007,
      "step": 17804
    },
    {
      "epoch": 0.12097334110719823,
      "grad_norm": 0.140625,
      "learning_rate": 0.000968344508692935,
      "loss": 2.1989,
      "step": 17805
    },
    {
      "epoch": 0.12098013545379228,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009683407413331867,
      "loss": 2.1795,
      "step": 17806
    },
    {
      "epoch": 0.12098692980038635,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009683369737566021,
      "loss": 2.1606,
      "step": 17807
    },
    {
      "epoch": 0.1209937241469804,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009683332059631827,
      "loss": 2.1212,
      "step": 17808
    },
    {
      "epoch": 0.12100051849357446,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009683294379529306,
      "loss": 2.2778,
      "step": 17809
    },
    {
      "epoch": 0.12100731284016851,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009683256697258473,
      "loss": 2.1387,
      "step": 17810
    },
    {
      "epoch": 0.12101410718676257,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009683219012819345,
      "loss": 2.2821,
      "step": 17811
    },
    {
      "epoch": 0.12102090153335664,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009683181326211941,
      "loss": 2.1175,
      "step": 17812
    },
    {
      "epoch": 0.12102769587995069,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009683143637436276,
      "loss": 2.294,
      "step": 17813
    },
    {
      "epoch": 0.12103449022654475,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009683105946492371,
      "loss": 2.1835,
      "step": 17814
    },
    {
      "epoch": 0.1210412845731388,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009683068253380241,
      "loss": 2.1181,
      "step": 17815
    },
    {
      "epoch": 0.12104807891973286,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009683030558099903,
      "loss": 2.1225,
      "step": 17816
    },
    {
      "epoch": 0.12105487326632693,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009682992860651379,
      "loss": 2.3138,
      "step": 17817
    },
    {
      "epoch": 0.12106166761292098,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009682955161034679,
      "loss": 2.1453,
      "step": 17818
    },
    {
      "epoch": 0.12106846195951504,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009682917459249827,
      "loss": 2.1545,
      "step": 17819
    },
    {
      "epoch": 0.12107525630610909,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009682879755296836,
      "loss": 2.3346,
      "step": 17820
    },
    {
      "epoch": 0.12108205065270315,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009682842049175727,
      "loss": 2.2028,
      "step": 17821
    },
    {
      "epoch": 0.12108884499929722,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009682804340886515,
      "loss": 2.3526,
      "step": 17822
    },
    {
      "epoch": 0.12109563934589126,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0009682766630429218,
      "loss": 2.1067,
      "step": 17823
    },
    {
      "epoch": 0.12110243369248533,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009682728917803856,
      "loss": 2.0938,
      "step": 17824
    },
    {
      "epoch": 0.12110922803907938,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009682691203010442,
      "loss": 2.1105,
      "step": 17825
    },
    {
      "epoch": 0.12111602238567344,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009682653486048996,
      "loss": 2.1713,
      "step": 17826
    },
    {
      "epoch": 0.1211228167322675,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009682615766919536,
      "loss": 2.2535,
      "step": 17827
    },
    {
      "epoch": 0.12112961107886155,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009682578045622077,
      "loss": 2.1057,
      "step": 17828
    },
    {
      "epoch": 0.12113640542545562,
      "grad_norm": 0.1630859375,
      "learning_rate": 0.0009682540322156638,
      "loss": 2.2018,
      "step": 17829
    },
    {
      "epoch": 0.12114319977204967,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009682502596523239,
      "loss": 2.322,
      "step": 17830
    },
    {
      "epoch": 0.12114999411864373,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009682464868721892,
      "loss": 2.2037,
      "step": 17831
    },
    {
      "epoch": 0.1211567884652378,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000968242713875262,
      "loss": 2.131,
      "step": 17832
    },
    {
      "epoch": 0.12116358281183184,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009682389406615436,
      "loss": 2.1124,
      "step": 17833
    },
    {
      "epoch": 0.12117037715842591,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009682351672310361,
      "loss": 2.2231,
      "step": 17834
    },
    {
      "epoch": 0.12117717150501996,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000968231393583741,
      "loss": 2.1453,
      "step": 17835
    },
    {
      "epoch": 0.12118396585161402,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00096822761971966,
      "loss": 2.3022,
      "step": 17836
    },
    {
      "epoch": 0.12119076019820807,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009682238456387951,
      "loss": 2.2423,
      "step": 17837
    },
    {
      "epoch": 0.12119755454480213,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009682200713411479,
      "loss": 2.1403,
      "step": 17838
    },
    {
      "epoch": 0.1212043488913962,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009682162968267203,
      "loss": 2.2186,
      "step": 17839
    },
    {
      "epoch": 0.12121114323799025,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009682125220955138,
      "loss": 2.2584,
      "step": 17840
    },
    {
      "epoch": 0.12121793758458431,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009682087471475304,
      "loss": 2.2065,
      "step": 17841
    },
    {
      "epoch": 0.12122473193117836,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009682049719827716,
      "loss": 2.2335,
      "step": 17842
    },
    {
      "epoch": 0.12123152627777242,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009682011966012393,
      "loss": 2.1106,
      "step": 17843
    },
    {
      "epoch": 0.12123832062436649,
      "grad_norm": 0.125,
      "learning_rate": 0.0009681974210029352,
      "loss": 2.1187,
      "step": 17844
    },
    {
      "epoch": 0.12124511497096054,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000968193645187861,
      "loss": 2.2779,
      "step": 17845
    },
    {
      "epoch": 0.1212519093175546,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009681898691560187,
      "loss": 2.1301,
      "step": 17846
    },
    {
      "epoch": 0.12125870366414865,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009681860929074096,
      "loss": 2.1673,
      "step": 17847
    },
    {
      "epoch": 0.12126549801074271,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009681823164420359,
      "loss": 2.2137,
      "step": 17848
    },
    {
      "epoch": 0.12127229235733678,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009681785397598991,
      "loss": 2.2091,
      "step": 17849
    },
    {
      "epoch": 0.12127908670393082,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000968174762861001,
      "loss": 2.1066,
      "step": 17850
    },
    {
      "epoch": 0.12128588105052489,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009681709857453434,
      "loss": 2.3268,
      "step": 17851
    },
    {
      "epoch": 0.12129267539711894,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009681672084129278,
      "loss": 2.3502,
      "step": 17852
    },
    {
      "epoch": 0.121299469743713,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009681634308637563,
      "loss": 2.2227,
      "step": 17853
    },
    {
      "epoch": 0.12130626409030706,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009681596530978306,
      "loss": 2.0776,
      "step": 17854
    },
    {
      "epoch": 0.12131305843690111,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009681558751151523,
      "loss": 2.1117,
      "step": 17855
    },
    {
      "epoch": 0.12131985278349518,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009681520969157232,
      "loss": 2.263,
      "step": 17856
    },
    {
      "epoch": 0.12132664713008923,
      "grad_norm": 0.125,
      "learning_rate": 0.0009681483184995448,
      "loss": 2.2583,
      "step": 17857
    },
    {
      "epoch": 0.12133344147668329,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009681445398666193,
      "loss": 2.2077,
      "step": 17858
    },
    {
      "epoch": 0.12134023582327734,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009681407610169484,
      "loss": 2.1798,
      "step": 17859
    },
    {
      "epoch": 0.1213470301698714,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009681369819505334,
      "loss": 2.068,
      "step": 17860
    },
    {
      "epoch": 0.12135382451646547,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009681332026673765,
      "loss": 2.3002,
      "step": 17861
    },
    {
      "epoch": 0.12136061886305952,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009681294231674793,
      "loss": 2.1905,
      "step": 17862
    },
    {
      "epoch": 0.12136741320965358,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009681256434508434,
      "loss": 2.1572,
      "step": 17863
    },
    {
      "epoch": 0.12137420755624763,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009681218635174709,
      "loss": 2.2314,
      "step": 17864
    },
    {
      "epoch": 0.12138100190284169,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009681180833673633,
      "loss": 2.2128,
      "step": 17865
    },
    {
      "epoch": 0.12138779624943576,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009681143030005223,
      "loss": 2.1366,
      "step": 17866
    },
    {
      "epoch": 0.1213945905960298,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009681105224169498,
      "loss": 2.1894,
      "step": 17867
    },
    {
      "epoch": 0.12140138494262387,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009681067416166474,
      "loss": 2.2531,
      "step": 17868
    },
    {
      "epoch": 0.12140817928921792,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000968102960599617,
      "loss": 2.0958,
      "step": 17869
    },
    {
      "epoch": 0.12141497363581198,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009680991793658603,
      "loss": 2.0366,
      "step": 17870
    },
    {
      "epoch": 0.12142176798240605,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000968095397915379,
      "loss": 2.2756,
      "step": 17871
    },
    {
      "epoch": 0.1214285623290001,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009680916162481751,
      "loss": 2.3463,
      "step": 17872
    },
    {
      "epoch": 0.12143535667559416,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009680878343642501,
      "loss": 2.1497,
      "step": 17873
    },
    {
      "epoch": 0.12144215102218821,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009680840522636056,
      "loss": 2.0442,
      "step": 17874
    },
    {
      "epoch": 0.12144894536878227,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009680802699462437,
      "loss": 2.1587,
      "step": 17875
    },
    {
      "epoch": 0.12145573971537633,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009680764874121659,
      "loss": 2.1617,
      "step": 17876
    },
    {
      "epoch": 0.12146253406197038,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009680727046613742,
      "loss": 2.0085,
      "step": 17877
    },
    {
      "epoch": 0.12146932840856445,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009680689216938701,
      "loss": 2.0774,
      "step": 17878
    },
    {
      "epoch": 0.1214761227551585,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009680651385096555,
      "loss": 2.3095,
      "step": 17879
    },
    {
      "epoch": 0.12148291710175256,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000968061355108732,
      "loss": 2.1341,
      "step": 17880
    },
    {
      "epoch": 0.12148971144834661,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009680575714911015,
      "loss": 2.1447,
      "step": 17881
    },
    {
      "epoch": 0.12149650579494067,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009680537876567658,
      "loss": 2.1841,
      "step": 17882
    },
    {
      "epoch": 0.12150330014153474,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009680500036057265,
      "loss": 2.1271,
      "step": 17883
    },
    {
      "epoch": 0.12151009448812879,
      "grad_norm": 0.1640625,
      "learning_rate": 0.0009680462193379856,
      "loss": 2.4455,
      "step": 17884
    },
    {
      "epoch": 0.12151688883472285,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009680424348535444,
      "loss": 2.0438,
      "step": 17885
    },
    {
      "epoch": 0.1215236831813169,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000968038650152405,
      "loss": 2.1485,
      "step": 17886
    },
    {
      "epoch": 0.12153047752791096,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009680348652345691,
      "loss": 2.2103,
      "step": 17887
    },
    {
      "epoch": 0.12153727187450503,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009680310801000384,
      "loss": 2.0316,
      "step": 17888
    },
    {
      "epoch": 0.12154406622109908,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009680272947488147,
      "loss": 2.2349,
      "step": 17889
    },
    {
      "epoch": 0.12155086056769314,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009680235091808996,
      "loss": 2.2159,
      "step": 17890
    },
    {
      "epoch": 0.12155765491428719,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009680197233962951,
      "loss": 2.1484,
      "step": 17891
    },
    {
      "epoch": 0.12156444926088125,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009680159373950028,
      "loss": 2.3526,
      "step": 17892
    },
    {
      "epoch": 0.12157124360747532,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009680121511770245,
      "loss": 2.241,
      "step": 17893
    },
    {
      "epoch": 0.12157803795406937,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009680083647423621,
      "loss": 2.2179,
      "step": 17894
    },
    {
      "epoch": 0.12158483230066343,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000968004578091017,
      "loss": 2.1659,
      "step": 17895
    },
    {
      "epoch": 0.12159162664725748,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009680007912229911,
      "loss": 2.3098,
      "step": 17896
    },
    {
      "epoch": 0.12159842099385154,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009679970041382863,
      "loss": 1.9749,
      "step": 17897
    },
    {
      "epoch": 0.1216052153404456,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009679932168369043,
      "loss": 2.268,
      "step": 17898
    },
    {
      "epoch": 0.12161200968703965,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009679894293188467,
      "loss": 2.1987,
      "step": 17899
    },
    {
      "epoch": 0.12161880403363372,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009679856415841155,
      "loss": 2.1631,
      "step": 17900
    },
    {
      "epoch": 0.12162559838022777,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009679818536327122,
      "loss": 2.0839,
      "step": 17901
    },
    {
      "epoch": 0.12163239272682183,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009679780654646386,
      "loss": 2.2031,
      "step": 17902
    },
    {
      "epoch": 0.1216391870734159,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009679742770798967,
      "loss": 2.2029,
      "step": 17903
    },
    {
      "epoch": 0.12164598142000994,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009679704884784879,
      "loss": 2.2198,
      "step": 17904
    },
    {
      "epoch": 0.12165277576660401,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009679666996604141,
      "loss": 2.2362,
      "step": 17905
    },
    {
      "epoch": 0.12165957011319806,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009679629106256773,
      "loss": 2.0371,
      "step": 17906
    },
    {
      "epoch": 0.12166636445979212,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000967959121374279,
      "loss": 2.2057,
      "step": 17907
    },
    {
      "epoch": 0.12167315880638617,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009679553319062209,
      "loss": 2.14,
      "step": 17908
    },
    {
      "epoch": 0.12167995315298023,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009679515422215049,
      "loss": 2.1,
      "step": 17909
    },
    {
      "epoch": 0.1216867474995743,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009679477523201327,
      "loss": 2.06,
      "step": 17910
    },
    {
      "epoch": 0.12169354184616835,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000967943962202106,
      "loss": 2.1026,
      "step": 17911
    },
    {
      "epoch": 0.12170033619276241,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009679401718674266,
      "loss": 2.1151,
      "step": 17912
    },
    {
      "epoch": 0.12170713053935646,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009679363813160963,
      "loss": 2.1308,
      "step": 17913
    },
    {
      "epoch": 0.12171392488595052,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009679325905481169,
      "loss": 2.3384,
      "step": 17914
    },
    {
      "epoch": 0.12172071923254459,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009679287995634899,
      "loss": 2.1803,
      "step": 17915
    },
    {
      "epoch": 0.12172751357913864,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009679250083622173,
      "loss": 2.2792,
      "step": 17916
    },
    {
      "epoch": 0.1217343079257327,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009679212169443008,
      "loss": 2.1266,
      "step": 17917
    },
    {
      "epoch": 0.12174110227232675,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009679174253097423,
      "loss": 2.1989,
      "step": 17918
    },
    {
      "epoch": 0.12174789661892081,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009679136334585433,
      "loss": 2.1323,
      "step": 17919
    },
    {
      "epoch": 0.12175469096551488,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009679098413907055,
      "loss": 2.3082,
      "step": 17920
    },
    {
      "epoch": 0.12176148531210892,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009679060491062308,
      "loss": 2.1391,
      "step": 17921
    },
    {
      "epoch": 0.12176827965870299,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009679022566051212,
      "loss": 2.2418,
      "step": 17922
    },
    {
      "epoch": 0.12177507400529704,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000967898463887378,
      "loss": 2.3107,
      "step": 17923
    },
    {
      "epoch": 0.1217818683518911,
      "grad_norm": 0.314453125,
      "learning_rate": 0.0009678946709530033,
      "loss": 2.2915,
      "step": 17924
    },
    {
      "epoch": 0.12178866269848516,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009678908778019987,
      "loss": 2.0823,
      "step": 17925
    },
    {
      "epoch": 0.12179545704507921,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000967887084434366,
      "loss": 2.272,
      "step": 17926
    },
    {
      "epoch": 0.12180225139167328,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009678832908501069,
      "loss": 2.3167,
      "step": 17927
    },
    {
      "epoch": 0.12180904573826733,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009678794970492233,
      "loss": 2.1872,
      "step": 17928
    },
    {
      "epoch": 0.12181584008486139,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009678757030317167,
      "loss": 1.9937,
      "step": 17929
    },
    {
      "epoch": 0.12182263443145544,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009678719087975891,
      "loss": 1.9996,
      "step": 17930
    },
    {
      "epoch": 0.1218294287780495,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009678681143468422,
      "loss": 2.0754,
      "step": 17931
    },
    {
      "epoch": 0.12183622312464357,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009678643196794776,
      "loss": 2.1242,
      "step": 17932
    },
    {
      "epoch": 0.12184301747123762,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009678605247954975,
      "loss": 2.2603,
      "step": 17933
    },
    {
      "epoch": 0.12184981181783168,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000967856729694903,
      "loss": 2.2394,
      "step": 17934
    },
    {
      "epoch": 0.12185660616442573,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009678529343776963,
      "loss": 1.9916,
      "step": 17935
    },
    {
      "epoch": 0.12186340051101979,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009678491388438791,
      "loss": 2.2493,
      "step": 17936
    },
    {
      "epoch": 0.12187019485761386,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009678453430934531,
      "loss": 2.1271,
      "step": 17937
    },
    {
      "epoch": 0.1218769892042079,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009678415471264201,
      "loss": 2.131,
      "step": 17938
    },
    {
      "epoch": 0.12188378355080197,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009678377509427818,
      "loss": 2.2291,
      "step": 17939
    },
    {
      "epoch": 0.12189057789739602,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009678339545425398,
      "loss": 2.2626,
      "step": 17940
    },
    {
      "epoch": 0.12189737224399008,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009678301579256963,
      "loss": 2.0959,
      "step": 17941
    },
    {
      "epoch": 0.12190416659058415,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009678263610922528,
      "loss": 2.0636,
      "step": 17942
    },
    {
      "epoch": 0.1219109609371782,
      "grad_norm": 0.1484375,
      "learning_rate": 0.000967822564042211,
      "loss": 2.0776,
      "step": 17943
    },
    {
      "epoch": 0.12191775528377226,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009678187667755727,
      "loss": 2.1794,
      "step": 17944
    },
    {
      "epoch": 0.12192454963036631,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009678149692923395,
      "loss": 2.1968,
      "step": 17945
    },
    {
      "epoch": 0.12193134397696037,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009678111715925135,
      "loss": 2.1295,
      "step": 17946
    },
    {
      "epoch": 0.12193813832355443,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009678073736760964,
      "loss": 2.1332,
      "step": 17947
    },
    {
      "epoch": 0.12194493267014848,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009678035755430896,
      "loss": 2.1589,
      "step": 17948
    },
    {
      "epoch": 0.12195172701674255,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009677997771934953,
      "loss": 2.2665,
      "step": 17949
    },
    {
      "epoch": 0.1219585213633366,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009677959786273151,
      "loss": 2.0866,
      "step": 17950
    },
    {
      "epoch": 0.12196531570993066,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009677921798445506,
      "loss": 2.2638,
      "step": 17951
    },
    {
      "epoch": 0.12197211005652472,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009677883808452037,
      "loss": 1.962,
      "step": 17952
    },
    {
      "epoch": 0.12197890440311877,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009677845816292762,
      "loss": 2.2142,
      "step": 17953
    },
    {
      "epoch": 0.12198569874971284,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009677807821967697,
      "loss": 2.1305,
      "step": 17954
    },
    {
      "epoch": 0.12199249309630689,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009677769825476861,
      "loss": 2.1052,
      "step": 17955
    },
    {
      "epoch": 0.12199928744290095,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009677731826820272,
      "loss": 2.1786,
      "step": 17956
    },
    {
      "epoch": 0.122006081789495,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009677693825997947,
      "loss": 2.1172,
      "step": 17957
    },
    {
      "epoch": 0.12201287613608906,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009677655823009901,
      "loss": 2.0545,
      "step": 17958
    },
    {
      "epoch": 0.12201967048268313,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009677617817856157,
      "loss": 2.1153,
      "step": 17959
    },
    {
      "epoch": 0.12202646482927718,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009677579810536727,
      "loss": 2.0602,
      "step": 17960
    },
    {
      "epoch": 0.12203325917587124,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009677541801051632,
      "loss": 2.3074,
      "step": 17961
    },
    {
      "epoch": 0.12204005352246529,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000967750378940089,
      "loss": 2.0705,
      "step": 17962
    },
    {
      "epoch": 0.12204684786905935,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009677465775584516,
      "loss": 2.2235,
      "step": 17963
    },
    {
      "epoch": 0.12205364221565342,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009677427759602529,
      "loss": 2.1289,
      "step": 17964
    },
    {
      "epoch": 0.12206043656224747,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009677389741454947,
      "loss": 2.3485,
      "step": 17965
    },
    {
      "epoch": 0.12206723090884153,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009677351721141788,
      "loss": 2.2791,
      "step": 17966
    },
    {
      "epoch": 0.12207402525543558,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009677313698663068,
      "loss": 2.3521,
      "step": 17967
    },
    {
      "epoch": 0.12208081960202964,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009677275674018806,
      "loss": 2.1094,
      "step": 17968
    },
    {
      "epoch": 0.1220876139486237,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009677237647209017,
      "loss": 2.1672,
      "step": 17969
    },
    {
      "epoch": 0.12209440829521775,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009677199618233723,
      "loss": 2.2466,
      "step": 17970
    },
    {
      "epoch": 0.12210120264181182,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009677161587092937,
      "loss": 2.2386,
      "step": 17971
    },
    {
      "epoch": 0.12210799698840587,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009677123553786681,
      "loss": 2.1047,
      "step": 17972
    },
    {
      "epoch": 0.12211479133499993,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000967708551831497,
      "loss": 2.1868,
      "step": 17973
    },
    {
      "epoch": 0.122121585681594,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009677047480677821,
      "loss": 2.2268,
      "step": 17974
    },
    {
      "epoch": 0.12212838002818804,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009677009440875253,
      "loss": 2.0763,
      "step": 17975
    },
    {
      "epoch": 0.12213517437478211,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009676971398907282,
      "loss": 2.3798,
      "step": 17976
    },
    {
      "epoch": 0.12214196872137616,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009676933354773929,
      "loss": 2.3139,
      "step": 17977
    },
    {
      "epoch": 0.12214876306797022,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009676895308475209,
      "loss": 2.1546,
      "step": 17978
    },
    {
      "epoch": 0.12215555741456427,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009676857260011139,
      "loss": 2.2321,
      "step": 17979
    },
    {
      "epoch": 0.12216235176115833,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009676819209381738,
      "loss": 2.2057,
      "step": 17980
    },
    {
      "epoch": 0.1221691461077524,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009676781156587025,
      "loss": 2.1403,
      "step": 17981
    },
    {
      "epoch": 0.12217594045434645,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009676743101627014,
      "loss": 2.1525,
      "step": 17982
    },
    {
      "epoch": 0.12218273480094051,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009676705044501726,
      "loss": 2.1772,
      "step": 17983
    },
    {
      "epoch": 0.12218952914753456,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009676666985211175,
      "loss": 2.0756,
      "step": 17984
    },
    {
      "epoch": 0.12219632349412862,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009676628923755383,
      "loss": 2.1935,
      "step": 17985
    },
    {
      "epoch": 0.12220311784072269,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009676590860134364,
      "loss": 2.0586,
      "step": 17986
    },
    {
      "epoch": 0.12220991218731674,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009676552794348138,
      "loss": 2.1609,
      "step": 17987
    },
    {
      "epoch": 0.1222167065339108,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009676514726396721,
      "loss": 2.2024,
      "step": 17988
    },
    {
      "epoch": 0.12222350088050485,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009676476656280132,
      "loss": 2.3133,
      "step": 17989
    },
    {
      "epoch": 0.12223029522709891,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009676438583998389,
      "loss": 2.288,
      "step": 17990
    },
    {
      "epoch": 0.12223708957369298,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009676400509551507,
      "loss": 2.2259,
      "step": 17991
    },
    {
      "epoch": 0.12224388392028702,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009676362432939504,
      "loss": 2.2068,
      "step": 17992
    },
    {
      "epoch": 0.12225067826688109,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.00096763243541624,
      "loss": 2.3106,
      "step": 17993
    },
    {
      "epoch": 0.12225747261347514,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009676286273220212,
      "loss": 2.3486,
      "step": 17994
    },
    {
      "epoch": 0.1222642669600692,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009676248190112956,
      "loss": 2.2607,
      "step": 17995
    },
    {
      "epoch": 0.12227106130666326,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009676210104840651,
      "loss": 2.1124,
      "step": 17996
    },
    {
      "epoch": 0.12227785565325731,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009676172017403315,
      "loss": 2.1186,
      "step": 17997
    },
    {
      "epoch": 0.12228464999985138,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009676133927800965,
      "loss": 2.2668,
      "step": 17998
    },
    {
      "epoch": 0.12229144434644543,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009676095836033618,
      "loss": 2.2344,
      "step": 17999
    },
    {
      "epoch": 0.12229823869303949,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009676057742101292,
      "loss": 2.3138,
      "step": 18000
    },
    {
      "epoch": 0.12230503303963354,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009676019646004006,
      "loss": 2.0454,
      "step": 18001
    },
    {
      "epoch": 0.1223118273862276,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009675981547741775,
      "loss": 2.2533,
      "step": 18002
    },
    {
      "epoch": 0.12231862173282167,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009675943447314619,
      "loss": 2.2706,
      "step": 18003
    },
    {
      "epoch": 0.12232541607941572,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009675905344722556,
      "loss": 2.1901,
      "step": 18004
    },
    {
      "epoch": 0.12233221042600978,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00096758672399656,
      "loss": 2.3786,
      "step": 18005
    },
    {
      "epoch": 0.12233900477260383,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009675829133043771,
      "loss": 2.2977,
      "step": 18006
    },
    {
      "epoch": 0.12234579911919789,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000967579102395709,
      "loss": 2.2975,
      "step": 18007
    },
    {
      "epoch": 0.12235259346579196,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009675752912705569,
      "loss": 2.2159,
      "step": 18008
    },
    {
      "epoch": 0.122359387812386,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009675714799289229,
      "loss": 2.1388,
      "step": 18009
    },
    {
      "epoch": 0.12236618215898007,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009675676683708086,
      "loss": 2.2785,
      "step": 18010
    },
    {
      "epoch": 0.12237297650557412,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009675638565962157,
      "loss": 2.1073,
      "step": 18011
    },
    {
      "epoch": 0.12237977085216818,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009675600446051462,
      "loss": 2.2956,
      "step": 18012
    },
    {
      "epoch": 0.12238656519876225,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009675562323976019,
      "loss": 2.1257,
      "step": 18013
    },
    {
      "epoch": 0.1223933595453563,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009675524199735842,
      "loss": 2.0438,
      "step": 18014
    },
    {
      "epoch": 0.12240015389195036,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009675486073330952,
      "loss": 2.2349,
      "step": 18015
    },
    {
      "epoch": 0.12240694823854441,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009675447944761367,
      "loss": 2.1913,
      "step": 18016
    },
    {
      "epoch": 0.12241374258513847,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009675409814027101,
      "loss": 2.4226,
      "step": 18017
    },
    {
      "epoch": 0.12242053693173253,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009675371681128175,
      "loss": 2.1559,
      "step": 18018
    },
    {
      "epoch": 0.12242733127832658,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009675333546064604,
      "loss": 2.1231,
      "step": 18019
    },
    {
      "epoch": 0.12243412562492065,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009675295408836409,
      "loss": 2.2467,
      "step": 18020
    },
    {
      "epoch": 0.1224409199715147,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009675257269443605,
      "loss": 2.3083,
      "step": 18021
    },
    {
      "epoch": 0.12244771431810876,
      "grad_norm": 0.125,
      "learning_rate": 0.000967521912788621,
      "loss": 2.1366,
      "step": 18022
    },
    {
      "epoch": 0.12245450866470282,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009675180984164242,
      "loss": 2.2458,
      "step": 18023
    },
    {
      "epoch": 0.12246130301129687,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009675142838277721,
      "loss": 2.1901,
      "step": 18024
    },
    {
      "epoch": 0.12246809735789094,
      "grad_norm": 0.125,
      "learning_rate": 0.0009675104690226662,
      "loss": 2.3054,
      "step": 18025
    },
    {
      "epoch": 0.12247489170448499,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009675066540011082,
      "loss": 2.2765,
      "step": 18026
    },
    {
      "epoch": 0.12248168605107905,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009675028387630998,
      "loss": 2.115,
      "step": 18027
    },
    {
      "epoch": 0.1224884803976731,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009674990233086431,
      "loss": 2.1949,
      "step": 18028
    },
    {
      "epoch": 0.12249527474426716,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009674952076377399,
      "loss": 2.1876,
      "step": 18029
    },
    {
      "epoch": 0.12250206909086123,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009674913917503916,
      "loss": 2.3406,
      "step": 18030
    },
    {
      "epoch": 0.12250886343745528,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009674875756466002,
      "loss": 2.2262,
      "step": 18031
    },
    {
      "epoch": 0.12251565778404934,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009674837593263673,
      "loss": 2.1768,
      "step": 18032
    },
    {
      "epoch": 0.12252245213064339,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000967479942789695,
      "loss": 2.1479,
      "step": 18033
    },
    {
      "epoch": 0.12252924647723745,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009674761260365846,
      "loss": 2.1354,
      "step": 18034
    },
    {
      "epoch": 0.12253604082383152,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009674723090670383,
      "loss": 2.2059,
      "step": 18035
    },
    {
      "epoch": 0.12254283517042557,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009674684918810576,
      "loss": 2.1576,
      "step": 18036
    },
    {
      "epoch": 0.12254962951701963,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009674646744786443,
      "loss": 2.2874,
      "step": 18037
    },
    {
      "epoch": 0.12255642386361368,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009674608568598002,
      "loss": 2.2534,
      "step": 18038
    },
    {
      "epoch": 0.12256321821020774,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009674570390245272,
      "loss": 2.2148,
      "step": 18039
    },
    {
      "epoch": 0.1225700125568018,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009674532209728268,
      "loss": 2.1647,
      "step": 18040
    },
    {
      "epoch": 0.12257680690339585,
      "grad_norm": 0.125,
      "learning_rate": 0.0009674494027047012,
      "loss": 2.0885,
      "step": 18041
    },
    {
      "epoch": 0.12258360124998992,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009674455842201516,
      "loss": 2.0628,
      "step": 18042
    },
    {
      "epoch": 0.12259039559658397,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009674417655191801,
      "loss": 2.2035,
      "step": 18043
    },
    {
      "epoch": 0.12259718994317803,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009674379466017886,
      "loss": 2.1173,
      "step": 18044
    },
    {
      "epoch": 0.1226039842897721,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009674341274679785,
      "loss": 2.1865,
      "step": 18045
    },
    {
      "epoch": 0.12261077863636614,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009674303081177518,
      "loss": 2.3269,
      "step": 18046
    },
    {
      "epoch": 0.12261757298296021,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009674264885511103,
      "loss": 2.1531,
      "step": 18047
    },
    {
      "epoch": 0.12262436732955426,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009674226687680556,
      "loss": 2.2985,
      "step": 18048
    },
    {
      "epoch": 0.12263116167614832,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009674188487685896,
      "loss": 2.1455,
      "step": 18049
    },
    {
      "epoch": 0.12263795602274237,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009674150285527141,
      "loss": 2.287,
      "step": 18050
    },
    {
      "epoch": 0.12264475036933643,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009674112081204307,
      "loss": 2.1592,
      "step": 18051
    },
    {
      "epoch": 0.1226515447159305,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009674073874717413,
      "loss": 2.2387,
      "step": 18052
    },
    {
      "epoch": 0.12265833906252455,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009674035666066476,
      "loss": 2.2482,
      "step": 18053
    },
    {
      "epoch": 0.12266513340911861,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009673997455251516,
      "loss": 2.1684,
      "step": 18054
    },
    {
      "epoch": 0.12267192775571266,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009673959242272546,
      "loss": 2.2065,
      "step": 18055
    },
    {
      "epoch": 0.12267872210230672,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009673921027129589,
      "loss": 2.2729,
      "step": 18056
    },
    {
      "epoch": 0.12268551644890079,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009673882809822658,
      "loss": 2.2562,
      "step": 18057
    },
    {
      "epoch": 0.12269231079549484,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009673844590351774,
      "loss": 2.1349,
      "step": 18058
    },
    {
      "epoch": 0.1226991051420889,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009673806368716954,
      "loss": 2.2941,
      "step": 18059
    },
    {
      "epoch": 0.12270589948868295,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009673768144918214,
      "loss": 2.3023,
      "step": 18060
    },
    {
      "epoch": 0.12271269383527701,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009673729918955574,
      "loss": 2.1518,
      "step": 18061
    },
    {
      "epoch": 0.12271948818187108,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009673691690829049,
      "loss": 2.2952,
      "step": 18062
    },
    {
      "epoch": 0.12272628252846512,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009673653460538661,
      "loss": 2.0546,
      "step": 18063
    },
    {
      "epoch": 0.12273307687505919,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009673615228084423,
      "loss": 2.1977,
      "step": 18064
    },
    {
      "epoch": 0.12273987122165324,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009673576993466355,
      "loss": 2.1932,
      "step": 18065
    },
    {
      "epoch": 0.1227466655682473,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009673538756684475,
      "loss": 2.2061,
      "step": 18066
    },
    {
      "epoch": 0.12275345991484136,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009673500517738798,
      "loss": 2.1948,
      "step": 18067
    },
    {
      "epoch": 0.12276025426143541,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009673462276629345,
      "loss": 2.1942,
      "step": 18068
    },
    {
      "epoch": 0.12276704860802948,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009673424033356135,
      "loss": 2.274,
      "step": 18069
    },
    {
      "epoch": 0.12277384295462353,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009673385787919181,
      "loss": 2.2983,
      "step": 18070
    },
    {
      "epoch": 0.12278063730121759,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009673347540318503,
      "loss": 2.3533,
      "step": 18071
    },
    {
      "epoch": 0.12278743164781165,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000967330929055412,
      "loss": 2.0434,
      "step": 18072
    },
    {
      "epoch": 0.1227942259944057,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009673271038626046,
      "loss": 2.1345,
      "step": 18073
    },
    {
      "epoch": 0.12280102034099977,
      "grad_norm": 0.16015625,
      "learning_rate": 0.0009673232784534303,
      "loss": 2.2254,
      "step": 18074
    },
    {
      "epoch": 0.12280781468759382,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009673194528278905,
      "loss": 2.1924,
      "step": 18075
    },
    {
      "epoch": 0.12281460903418788,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009673156269859874,
      "loss": 2.1706,
      "step": 18076
    },
    {
      "epoch": 0.12282140338078193,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009673118009277225,
      "loss": 2.1701,
      "step": 18077
    },
    {
      "epoch": 0.12282819772737599,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009673079746530974,
      "loss": 2.1951,
      "step": 18078
    },
    {
      "epoch": 0.12283499207397006,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009673041481621141,
      "loss": 2.2444,
      "step": 18079
    },
    {
      "epoch": 0.1228417864205641,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009673003214547744,
      "loss": 2.1042,
      "step": 18080
    },
    {
      "epoch": 0.12284858076715817,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00096729649453108,
      "loss": 2.1569,
      "step": 18081
    },
    {
      "epoch": 0.12285537511375222,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009672926673910328,
      "loss": 2.1692,
      "step": 18082
    },
    {
      "epoch": 0.12286216946034628,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009672888400346343,
      "loss": 2.3051,
      "step": 18083
    },
    {
      "epoch": 0.12286896380694035,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.0009672850124618864,
      "loss": 2.1856,
      "step": 18084
    },
    {
      "epoch": 0.1228757581535344,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009672811846727909,
      "loss": 2.0924,
      "step": 18085
    },
    {
      "epoch": 0.12288255250012846,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009672773566673497,
      "loss": 2.0928,
      "step": 18086
    },
    {
      "epoch": 0.12288934684672251,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009672735284455644,
      "loss": 2.241,
      "step": 18087
    },
    {
      "epoch": 0.12289614119331657,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009672697000074368,
      "loss": 2.1636,
      "step": 18088
    },
    {
      "epoch": 0.12290293553991063,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009672658713529687,
      "loss": 2.1108,
      "step": 18089
    },
    {
      "epoch": 0.12290972988650468,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009672620424821618,
      "loss": 2.1726,
      "step": 18090
    },
    {
      "epoch": 0.12291652423309875,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009672582133950178,
      "loss": 2.2404,
      "step": 18091
    },
    {
      "epoch": 0.1229233185796928,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009672543840915389,
      "loss": 2.0853,
      "step": 18092
    },
    {
      "epoch": 0.12293011292628686,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009672505545717264,
      "loss": 2.1646,
      "step": 18093
    },
    {
      "epoch": 0.12293690727288092,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009672467248355823,
      "loss": 2.2267,
      "step": 18094
    },
    {
      "epoch": 0.12294370161947497,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009672428948831083,
      "loss": 2.2712,
      "step": 18095
    },
    {
      "epoch": 0.12295049596606904,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009672390647143062,
      "loss": 2.2214,
      "step": 18096
    },
    {
      "epoch": 0.12295729031266309,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009672352343291777,
      "loss": 2.2589,
      "step": 18097
    },
    {
      "epoch": 0.12296408465925715,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009672314037277248,
      "loss": 2.1254,
      "step": 18098
    },
    {
      "epoch": 0.1229708790058512,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009672275729099492,
      "loss": 2.1823,
      "step": 18099
    },
    {
      "epoch": 0.12297767335244526,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009672237418758522,
      "loss": 2.166,
      "step": 18100
    },
    {
      "epoch": 0.12298446769903933,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009672199106254362,
      "loss": 2.2131,
      "step": 18101
    },
    {
      "epoch": 0.12299126204563338,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009672160791587028,
      "loss": 1.9612,
      "step": 18102
    },
    {
      "epoch": 0.12299805639222744,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009672122474756536,
      "loss": 2.1352,
      "step": 18103
    },
    {
      "epoch": 0.12300485073882149,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009672084155762906,
      "loss": 2.1116,
      "step": 18104
    },
    {
      "epoch": 0.12301164508541555,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009672045834606154,
      "loss": 2.1266,
      "step": 18105
    },
    {
      "epoch": 0.12301843943200962,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009672007511286299,
      "loss": 2.2202,
      "step": 18106
    },
    {
      "epoch": 0.12302523377860367,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009671969185803355,
      "loss": 2.0856,
      "step": 18107
    },
    {
      "epoch": 0.12303202812519773,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009671930858157345,
      "loss": 2.3789,
      "step": 18108
    },
    {
      "epoch": 0.12303882247179178,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009671892528348285,
      "loss": 2.2549,
      "step": 18109
    },
    {
      "epoch": 0.12304561681838584,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009671854196376193,
      "loss": 2.0922,
      "step": 18110
    },
    {
      "epoch": 0.1230524111649799,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009671815862241084,
      "loss": 2.1041,
      "step": 18111
    },
    {
      "epoch": 0.12305920551157395,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009671777525942979,
      "loss": 2.2246,
      "step": 18112
    },
    {
      "epoch": 0.12306599985816802,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009671739187481895,
      "loss": 2.2257,
      "step": 18113
    },
    {
      "epoch": 0.12307279420476207,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009671700846857849,
      "loss": 2.2693,
      "step": 18114
    },
    {
      "epoch": 0.12307958855135613,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009671662504070858,
      "loss": 2.3512,
      "step": 18115
    },
    {
      "epoch": 0.1230863828979502,
      "grad_norm": 0.12060546875,
      "learning_rate": 0.0009671624159120942,
      "loss": 2.0309,
      "step": 18116
    },
    {
      "epoch": 0.12309317724454424,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009671585812008117,
      "loss": 2.319,
      "step": 18117
    },
    {
      "epoch": 0.12309997159113831,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009671547462732402,
      "loss": 2.2703,
      "step": 18118
    },
    {
      "epoch": 0.12310676593773236,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009671509111293813,
      "loss": 2.3435,
      "step": 18119
    },
    {
      "epoch": 0.12311356028432642,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000967147075769237,
      "loss": 2.1603,
      "step": 18120
    },
    {
      "epoch": 0.12312035463092047,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009671432401928088,
      "loss": 2.1348,
      "step": 18121
    },
    {
      "epoch": 0.12312714897751453,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009671394044000988,
      "loss": 2.2999,
      "step": 18122
    },
    {
      "epoch": 0.1231339433241086,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009671355683911086,
      "loss": 2.2697,
      "step": 18123
    },
    {
      "epoch": 0.12314073767070265,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00096713173216584,
      "loss": 2.1401,
      "step": 18124
    },
    {
      "epoch": 0.12314753201729671,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009671278957242946,
      "loss": 2.1634,
      "step": 18125
    },
    {
      "epoch": 0.12315432636389076,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009671240590664745,
      "loss": 2.1716,
      "step": 18126
    },
    {
      "epoch": 0.12316112071048482,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009671202221923813,
      "loss": 2.1117,
      "step": 18127
    },
    {
      "epoch": 0.12316791505707889,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009671163851020168,
      "loss": 2.1439,
      "step": 18128
    },
    {
      "epoch": 0.12317470940367294,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009671125477953826,
      "loss": 2.1113,
      "step": 18129
    },
    {
      "epoch": 0.123181503750267,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009671087102724809,
      "loss": 2.1832,
      "step": 18130
    },
    {
      "epoch": 0.12318829809686105,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000967104872533313,
      "loss": 2.2041,
      "step": 18131
    },
    {
      "epoch": 0.12319509244345511,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000967101034577881,
      "loss": 2.211,
      "step": 18132
    },
    {
      "epoch": 0.12320188679004918,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009670971964061865,
      "loss": 2.1168,
      "step": 18133
    },
    {
      "epoch": 0.12320868113664322,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009670933580182315,
      "loss": 2.1789,
      "step": 18134
    },
    {
      "epoch": 0.12321547548323729,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009670895194140174,
      "loss": 2.214,
      "step": 18135
    },
    {
      "epoch": 0.12322226982983134,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009670856805935464,
      "loss": 2.3486,
      "step": 18136
    },
    {
      "epoch": 0.1232290641764254,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009670818415568201,
      "loss": 2.1346,
      "step": 18137
    },
    {
      "epoch": 0.12323585852301946,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009670780023038402,
      "loss": 2.2474,
      "step": 18138
    },
    {
      "epoch": 0.12324265286961351,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009670741628346085,
      "loss": 2.2168,
      "step": 18139
    },
    {
      "epoch": 0.12324944721620758,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009670703231491269,
      "loss": 2.2508,
      "step": 18140
    },
    {
      "epoch": 0.12325624156280163,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000967066483247397,
      "loss": 2.0773,
      "step": 18141
    },
    {
      "epoch": 0.12326303590939569,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009670626431294207,
      "loss": 2.353,
      "step": 18142
    },
    {
      "epoch": 0.12326983025598975,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009670588027951999,
      "loss": 2.2145,
      "step": 18143
    },
    {
      "epoch": 0.1232766246025838,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000967054962244736,
      "loss": 2.217,
      "step": 18144
    },
    {
      "epoch": 0.12328341894917787,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009670511214780311,
      "loss": 2.1583,
      "step": 18145
    },
    {
      "epoch": 0.12329021329577192,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000967047280495087,
      "loss": 2.2006,
      "step": 18146
    },
    {
      "epoch": 0.12329700764236598,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009670434392959052,
      "loss": 2.2722,
      "step": 18147
    },
    {
      "epoch": 0.12330380198896003,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009670395978804878,
      "loss": 2.2512,
      "step": 18148
    },
    {
      "epoch": 0.12331059633555409,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009670357562488363,
      "loss": 2.3097,
      "step": 18149
    },
    {
      "epoch": 0.12331739068214816,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009670319144009526,
      "loss": 2.1567,
      "step": 18150
    },
    {
      "epoch": 0.1233241850287422,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009670280723368385,
      "loss": 2.1767,
      "step": 18151
    },
    {
      "epoch": 0.12333097937533627,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009670242300564958,
      "loss": 2.2869,
      "step": 18152
    },
    {
      "epoch": 0.12333777372193032,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009670203875599262,
      "loss": 2.2234,
      "step": 18153
    },
    {
      "epoch": 0.12334456806852438,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009670165448471315,
      "loss": 2.1653,
      "step": 18154
    },
    {
      "epoch": 0.12335136241511845,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009670127019181134,
      "loss": 2.2002,
      "step": 18155
    },
    {
      "epoch": 0.1233581567617125,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009670088587728739,
      "loss": 2.2118,
      "step": 18156
    },
    {
      "epoch": 0.12336495110830656,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009670050154114147,
      "loss": 2.4535,
      "step": 18157
    },
    {
      "epoch": 0.12337174545490061,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009670011718337374,
      "loss": 2.1957,
      "step": 18158
    },
    {
      "epoch": 0.12337853980149467,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009669973280398439,
      "loss": 2.1224,
      "step": 18159
    },
    {
      "epoch": 0.12338533414808874,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009669934840297361,
      "loss": 2.1681,
      "step": 18160
    },
    {
      "epoch": 0.12339212849468278,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009669896398034157,
      "loss": 2.2258,
      "step": 18161
    },
    {
      "epoch": 0.12339892284127685,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009669857953608844,
      "loss": 2.1804,
      "step": 18162
    },
    {
      "epoch": 0.1234057171878709,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000966981950702144,
      "loss": 2.0672,
      "step": 18163
    },
    {
      "epoch": 0.12341251153446496,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0009669781058271962,
      "loss": 2.1513,
      "step": 18164
    },
    {
      "epoch": 0.12341930588105902,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009669742607360431,
      "loss": 2.0881,
      "step": 18165
    },
    {
      "epoch": 0.12342610022765307,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009669704154286863,
      "loss": 2.1848,
      "step": 18166
    },
    {
      "epoch": 0.12343289457424714,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009669665699051273,
      "loss": 2.0021,
      "step": 18167
    },
    {
      "epoch": 0.12343968892084119,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009669627241653682,
      "loss": 2.3598,
      "step": 18168
    },
    {
      "epoch": 0.12344648326743525,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000966958878209411,
      "loss": 2.2066,
      "step": 18169
    },
    {
      "epoch": 0.1234532776140293,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009669550320372569,
      "loss": 2.3275,
      "step": 18170
    },
    {
      "epoch": 0.12346007196062336,
      "grad_norm": 0.125,
      "learning_rate": 0.000966951185648908,
      "loss": 1.8733,
      "step": 18171
    },
    {
      "epoch": 0.12346686630721743,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009669473390443662,
      "loss": 2.205,
      "step": 18172
    },
    {
      "epoch": 0.12347366065381148,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000966943492223633,
      "loss": 2.2387,
      "step": 18173
    },
    {
      "epoch": 0.12348045500040554,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009669396451867103,
      "loss": 2.3653,
      "step": 18174
    },
    {
      "epoch": 0.12348724934699959,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009669357979336,
      "loss": 2.2532,
      "step": 18175
    },
    {
      "epoch": 0.12349404369359365,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009669319504643037,
      "loss": 2.2343,
      "step": 18176
    },
    {
      "epoch": 0.12350083804018772,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009669281027788232,
      "loss": 2.1952,
      "step": 18177
    },
    {
      "epoch": 0.12350763238678177,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009669242548771605,
      "loss": 2.1753,
      "step": 18178
    },
    {
      "epoch": 0.12351442673337583,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009669204067593171,
      "loss": 2.281,
      "step": 18179
    },
    {
      "epoch": 0.12352122107996988,
      "grad_norm": 0.125,
      "learning_rate": 0.000966916558425295,
      "loss": 2.1444,
      "step": 18180
    },
    {
      "epoch": 0.12352801542656394,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009669127098750959,
      "loss": 2.2787,
      "step": 18181
    },
    {
      "epoch": 0.123534809773158,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009669088611087216,
      "loss": 2.0964,
      "step": 18182
    },
    {
      "epoch": 0.12354160411975205,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009669050121261737,
      "loss": 2.2947,
      "step": 18183
    },
    {
      "epoch": 0.12354839846634612,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009669011629274542,
      "loss": 2.1869,
      "step": 18184
    },
    {
      "epoch": 0.12355519281294017,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009668973135125647,
      "loss": 2.2148,
      "step": 18185
    },
    {
      "epoch": 0.12356198715953423,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009668934638815072,
      "loss": 2.2121,
      "step": 18186
    },
    {
      "epoch": 0.1235687815061283,
      "grad_norm": 0.125,
      "learning_rate": 0.0009668896140342836,
      "loss": 2.0307,
      "step": 18187
    },
    {
      "epoch": 0.12357557585272234,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009668857639708951,
      "loss": 2.1831,
      "step": 18188
    },
    {
      "epoch": 0.12358237019931641,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000966881913691344,
      "loss": 2.0828,
      "step": 18189
    },
    {
      "epoch": 0.12358916454591046,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000966878063195632,
      "loss": 2.1734,
      "step": 18190
    },
    {
      "epoch": 0.12359595889250452,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009668742124837606,
      "loss": 2.3332,
      "step": 18191
    },
    {
      "epoch": 0.12360275323909858,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000966870361555732,
      "loss": 2.1965,
      "step": 18192
    },
    {
      "epoch": 0.12360954758569263,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009668665104115477,
      "loss": 2.0501,
      "step": 18193
    },
    {
      "epoch": 0.1236163419322867,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009668626590512095,
      "loss": 2.1252,
      "step": 18194
    },
    {
      "epoch": 0.12362313627888075,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009668588074747194,
      "loss": 2.2259,
      "step": 18195
    },
    {
      "epoch": 0.12362993062547481,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009668549556820788,
      "loss": 2.2033,
      "step": 18196
    },
    {
      "epoch": 0.12363672497206886,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009668511036732899,
      "loss": 2.287,
      "step": 18197
    },
    {
      "epoch": 0.12364351931866292,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009668472514483542,
      "loss": 2.1252,
      "step": 18198
    },
    {
      "epoch": 0.12365031366525699,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009668433990072736,
      "loss": 2.0111,
      "step": 18199
    },
    {
      "epoch": 0.12365710801185104,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009668395463500499,
      "loss": 2.2361,
      "step": 18200
    },
    {
      "epoch": 0.1236639023584451,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009668356934766847,
      "loss": 2.2432,
      "step": 18201
    },
    {
      "epoch": 0.12367069670503915,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009668318403871801,
      "loss": 2.1966,
      "step": 18202
    },
    {
      "epoch": 0.12367749105163321,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009668279870815376,
      "loss": 2.2466,
      "step": 18203
    },
    {
      "epoch": 0.12368428539822728,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009668241335597592,
      "loss": 2.0333,
      "step": 18204
    },
    {
      "epoch": 0.12369107974482133,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009668202798218464,
      "loss": 2.221,
      "step": 18205
    },
    {
      "epoch": 0.12369787409141539,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009668164258678013,
      "loss": 2.2005,
      "step": 18206
    },
    {
      "epoch": 0.12370466843800944,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009668125716976255,
      "loss": 2.2297,
      "step": 18207
    },
    {
      "epoch": 0.1237114627846035,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009668087173113208,
      "loss": 2.239,
      "step": 18208
    },
    {
      "epoch": 0.12371825713119756,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009668048627088891,
      "loss": 2.2837,
      "step": 18209
    },
    {
      "epoch": 0.12372505147779161,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009668010078903319,
      "loss": 2.3514,
      "step": 18210
    },
    {
      "epoch": 0.12373184582438568,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009667971528556513,
      "loss": 2.1264,
      "step": 18211
    },
    {
      "epoch": 0.12373864017097973,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009667932976048493,
      "loss": 2.0233,
      "step": 18212
    },
    {
      "epoch": 0.12374543451757379,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009667894421379269,
      "loss": 2.1639,
      "step": 18213
    },
    {
      "epoch": 0.12375222886416785,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009667855864548866,
      "loss": 2.1767,
      "step": 18214
    },
    {
      "epoch": 0.1237590232107619,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009667817305557298,
      "loss": 2.1423,
      "step": 18215
    },
    {
      "epoch": 0.12376581755735597,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009667778744404583,
      "loss": 2.2114,
      "step": 18216
    },
    {
      "epoch": 0.12377261190395002,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009667740181090742,
      "loss": 2.0862,
      "step": 18217
    },
    {
      "epoch": 0.12377940625054408,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000966770161561579,
      "loss": 2.0888,
      "step": 18218
    },
    {
      "epoch": 0.12378620059713813,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009667663047979747,
      "loss": 2.1325,
      "step": 18219
    },
    {
      "epoch": 0.12379299494373219,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009667624478182629,
      "loss": 2.1996,
      "step": 18220
    },
    {
      "epoch": 0.12379978929032626,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009667585906224452,
      "loss": 2.2018,
      "step": 18221
    },
    {
      "epoch": 0.1238065836369203,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009667547332105239,
      "loss": 2.0632,
      "step": 18222
    },
    {
      "epoch": 0.12381337798351437,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009667508755825003,
      "loss": 2.1291,
      "step": 18223
    },
    {
      "epoch": 0.12382017233010842,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009667470177383766,
      "loss": 2.0957,
      "step": 18224
    },
    {
      "epoch": 0.12382696667670248,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009667431596781542,
      "loss": 2.029,
      "step": 18225
    },
    {
      "epoch": 0.12383376102329655,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009667393014018352,
      "loss": 2.1719,
      "step": 18226
    },
    {
      "epoch": 0.1238405553698906,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009667354429094213,
      "loss": 2.1359,
      "step": 18227
    },
    {
      "epoch": 0.12384734971648466,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009667315842009141,
      "loss": 2.3326,
      "step": 18228
    },
    {
      "epoch": 0.12385414406307871,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009667277252763156,
      "loss": 2.174,
      "step": 18229
    },
    {
      "epoch": 0.12386093840967277,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009667238661356275,
      "loss": 2.1016,
      "step": 18230
    },
    {
      "epoch": 0.12386773275626684,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009667200067788515,
      "loss": 2.2207,
      "step": 18231
    },
    {
      "epoch": 0.12387452710286088,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009667161472059895,
      "loss": 2.2045,
      "step": 18232
    },
    {
      "epoch": 0.12388132144945495,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009667122874170435,
      "loss": 2.2254,
      "step": 18233
    },
    {
      "epoch": 0.123888115796049,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009667084274120148,
      "loss": 2.3092,
      "step": 18234
    },
    {
      "epoch": 0.12389491014264306,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009667045671909055,
      "loss": 2.1629,
      "step": 18235
    },
    {
      "epoch": 0.12390170448923712,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009667007067537174,
      "loss": 2.0602,
      "step": 18236
    },
    {
      "epoch": 0.12390849883583117,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009666968461004524,
      "loss": 2.2073,
      "step": 18237
    },
    {
      "epoch": 0.12391529318242524,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000966692985231112,
      "loss": 2.2738,
      "step": 18238
    },
    {
      "epoch": 0.12392208752901929,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009666891241456979,
      "loss": 2.0762,
      "step": 18239
    },
    {
      "epoch": 0.12392888187561335,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009666852628442122,
      "loss": 2.3056,
      "step": 18240
    },
    {
      "epoch": 0.1239356762222074,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009666814013266566,
      "loss": 2.2749,
      "step": 18241
    },
    {
      "epoch": 0.12394247056880146,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000966677539593033,
      "loss": 2.2471,
      "step": 18242
    },
    {
      "epoch": 0.12394926491539553,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009666736776433428,
      "loss": 2.1106,
      "step": 18243
    },
    {
      "epoch": 0.12395605926198958,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009666698154775881,
      "loss": 2.2832,
      "step": 18244
    },
    {
      "epoch": 0.12396285360858364,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009666659530957706,
      "loss": 2.3277,
      "step": 18245
    },
    {
      "epoch": 0.12396964795517769,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009666620904978922,
      "loss": 2.1368,
      "step": 18246
    },
    {
      "epoch": 0.12397644230177175,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009666582276839546,
      "loss": 2.2007,
      "step": 18247
    },
    {
      "epoch": 0.12398323664836582,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009666543646539596,
      "loss": 2.2669,
      "step": 18248
    },
    {
      "epoch": 0.12399003099495987,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.000966650501407909,
      "loss": 2.014,
      "step": 18249
    },
    {
      "epoch": 0.12399682534155393,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009666466379458045,
      "loss": 2.2303,
      "step": 18250
    },
    {
      "epoch": 0.12400361968814798,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000966642774267648,
      "loss": 2.2793,
      "step": 18251
    },
    {
      "epoch": 0.12401041403474204,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009666389103734412,
      "loss": 2.2987,
      "step": 18252
    },
    {
      "epoch": 0.1240172083813361,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009666350462631859,
      "loss": 2.3332,
      "step": 18253
    },
    {
      "epoch": 0.12402400272793015,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009666311819368841,
      "loss": 2.1725,
      "step": 18254
    },
    {
      "epoch": 0.12403079707452422,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009666273173945373,
      "loss": 2.0796,
      "step": 18255
    },
    {
      "epoch": 0.12403759142111827,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009666234526361472,
      "loss": 2.1897,
      "step": 18256
    },
    {
      "epoch": 0.12404438576771233,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009666195876617161,
      "loss": 2.1692,
      "step": 18257
    },
    {
      "epoch": 0.1240511801143064,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009666157224712453,
      "loss": 2.2213,
      "step": 18258
    },
    {
      "epoch": 0.12405797446090044,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009666118570647368,
      "loss": 2.1774,
      "step": 18259
    },
    {
      "epoch": 0.12406476880749451,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009666079914421924,
      "loss": 2.2189,
      "step": 18260
    },
    {
      "epoch": 0.12407156315408856,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009666041256036139,
      "loss": 2.1084,
      "step": 18261
    },
    {
      "epoch": 0.12407835750068262,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000966600259549003,
      "loss": 2.1717,
      "step": 18262
    },
    {
      "epoch": 0.12408515184727668,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009665963932783615,
      "loss": 2.2093,
      "step": 18263
    },
    {
      "epoch": 0.12409194619387073,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009665925267916912,
      "loss": 2.1579,
      "step": 18264
    },
    {
      "epoch": 0.1240987405404648,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000966588660088994,
      "loss": 2.2558,
      "step": 18265
    },
    {
      "epoch": 0.12410553488705885,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009665847931702715,
      "loss": 2.183,
      "step": 18266
    },
    {
      "epoch": 0.12411232923365291,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009665809260355256,
      "loss": 2.0937,
      "step": 18267
    },
    {
      "epoch": 0.12411912358024696,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009665770586847582,
      "loss": 2.2265,
      "step": 18268
    },
    {
      "epoch": 0.12412591792684102,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009665731911179708,
      "loss": 2.1299,
      "step": 18269
    },
    {
      "epoch": 0.12413271227343509,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009665693233351654,
      "loss": 2.0719,
      "step": 18270
    },
    {
      "epoch": 0.12413950662002914,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009665654553363439,
      "loss": 2.1831,
      "step": 18271
    },
    {
      "epoch": 0.1241463009666232,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009665615871215078,
      "loss": 2.337,
      "step": 18272
    },
    {
      "epoch": 0.12415309531321725,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009665577186906591,
      "loss": 2.1886,
      "step": 18273
    },
    {
      "epoch": 0.12415988965981131,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009665538500437993,
      "loss": 2.1796,
      "step": 18274
    },
    {
      "epoch": 0.12416668400640538,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009665499811809307,
      "loss": 2.1983,
      "step": 18275
    },
    {
      "epoch": 0.12417347835299943,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009665461121020548,
      "loss": 2.0931,
      "step": 18276
    },
    {
      "epoch": 0.12418027269959349,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009665422428071733,
      "loss": 2.1305,
      "step": 18277
    },
    {
      "epoch": 0.12418706704618754,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000966538373296288,
      "loss": 2.2685,
      "step": 18278
    },
    {
      "epoch": 0.1241938613927816,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009665345035694009,
      "loss": 2.097,
      "step": 18279
    },
    {
      "epoch": 0.12420065573937566,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009665306336265138,
      "loss": 2.1374,
      "step": 18280
    },
    {
      "epoch": 0.12420745008596971,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009665267634676282,
      "loss": 2.2027,
      "step": 18281
    },
    {
      "epoch": 0.12421424443256378,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009665228930927461,
      "loss": 2.0731,
      "step": 18282
    },
    {
      "epoch": 0.12422103877915783,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009665190225018692,
      "loss": 2.0917,
      "step": 18283
    },
    {
      "epoch": 0.12422783312575189,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009665151516949994,
      "loss": 2.1622,
      "step": 18284
    },
    {
      "epoch": 0.12423462747234595,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009665112806721383,
      "loss": 2.0412,
      "step": 18285
    },
    {
      "epoch": 0.12424142181894,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000966507409433288,
      "loss": 2.0545,
      "step": 18286
    },
    {
      "epoch": 0.12424821616553407,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009665035379784502,
      "loss": 2.129,
      "step": 18287
    },
    {
      "epoch": 0.12425501051212812,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009664996663076264,
      "loss": 2.242,
      "step": 18288
    },
    {
      "epoch": 0.12426180485872218,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009664957944208186,
      "loss": 2.1889,
      "step": 18289
    },
    {
      "epoch": 0.12426859920531623,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009664919223180286,
      "loss": 2.1589,
      "step": 18290
    },
    {
      "epoch": 0.1242753935519103,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009664880499992584,
      "loss": 2.0947,
      "step": 18291
    },
    {
      "epoch": 0.12428218789850436,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009664841774645093,
      "loss": 2.0565,
      "step": 18292
    },
    {
      "epoch": 0.1242889822450984,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009664803047137835,
      "loss": 2.1066,
      "step": 18293
    },
    {
      "epoch": 0.12429577659169247,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009664764317470827,
      "loss": 2.1009,
      "step": 18294
    },
    {
      "epoch": 0.12430257093828652,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009664725585644086,
      "loss": 2.1155,
      "step": 18295
    },
    {
      "epoch": 0.12430936528488058,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009664686851657632,
      "loss": 2.301,
      "step": 18296
    },
    {
      "epoch": 0.12431615963147465,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000966464811551148,
      "loss": 2.166,
      "step": 18297
    },
    {
      "epoch": 0.1243229539780687,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009664609377205649,
      "loss": 2.2163,
      "step": 18298
    },
    {
      "epoch": 0.12432974832466276,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009664570636740157,
      "loss": 2.2915,
      "step": 18299
    },
    {
      "epoch": 0.12433654267125681,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009664531894115025,
      "loss": 2.2172,
      "step": 18300
    },
    {
      "epoch": 0.12434333701785087,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009664493149330267,
      "loss": 2.2177,
      "step": 18301
    },
    {
      "epoch": 0.12435013136444494,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009664454402385901,
      "loss": 2.077,
      "step": 18302
    },
    {
      "epoch": 0.12435692571103898,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009664415653281946,
      "loss": 2.148,
      "step": 18303
    },
    {
      "epoch": 0.12436372005763305,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009664376902018421,
      "loss": 2.0691,
      "step": 18304
    },
    {
      "epoch": 0.1243705144042271,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009664338148595343,
      "loss": 2.1737,
      "step": 18305
    },
    {
      "epoch": 0.12437730875082116,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000966429939301273,
      "loss": 2.1104,
      "step": 18306
    },
    {
      "epoch": 0.12438410309741522,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009664260635270599,
      "loss": 2.2726,
      "step": 18307
    },
    {
      "epoch": 0.12439089744400927,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000966422187536897,
      "loss": 2.1624,
      "step": 18308
    },
    {
      "epoch": 0.12439769179060334,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009664183113307858,
      "loss": 2.1125,
      "step": 18309
    },
    {
      "epoch": 0.12440448613719739,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009664144349087285,
      "loss": 2.1709,
      "step": 18310
    },
    {
      "epoch": 0.12441128048379145,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009664105582707265,
      "loss": 2.0493,
      "step": 18311
    },
    {
      "epoch": 0.12441807483038551,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009664066814167818,
      "loss": 2.2383,
      "step": 18312
    },
    {
      "epoch": 0.12442486917697956,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009664028043468962,
      "loss": 2.1436,
      "step": 18313
    },
    {
      "epoch": 0.12443166352357363,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009663989270610714,
      "loss": 2.2496,
      "step": 18314
    },
    {
      "epoch": 0.12443845787016768,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009663950495593092,
      "loss": 2.1967,
      "step": 18315
    },
    {
      "epoch": 0.12444525221676174,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009663911718416114,
      "loss": 2.1507,
      "step": 18316
    },
    {
      "epoch": 0.12445204656335579,
      "grad_norm": 0.125,
      "learning_rate": 0.00096638729390798,
      "loss": 2.0713,
      "step": 18317
    },
    {
      "epoch": 0.12445884090994985,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009663834157584164,
      "loss": 2.4111,
      "step": 18318
    },
    {
      "epoch": 0.12446563525654392,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.000966379537392923,
      "loss": 2.1398,
      "step": 18319
    },
    {
      "epoch": 0.12447242960313797,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009663756588115009,
      "loss": 2.1897,
      "step": 18320
    },
    {
      "epoch": 0.12447922394973203,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009663717800141523,
      "loss": 2.3077,
      "step": 18321
    },
    {
      "epoch": 0.12448601829632608,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009663679010008789,
      "loss": 2.305,
      "step": 18322
    },
    {
      "epoch": 0.12449281264292014,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009663640217716825,
      "loss": 2.2837,
      "step": 18323
    },
    {
      "epoch": 0.1244996069895142,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000966360142326565,
      "loss": 2.0445,
      "step": 18324
    },
    {
      "epoch": 0.12450640133610825,
      "grad_norm": 0.140625,
      "learning_rate": 0.000966356262665528,
      "loss": 2.1802,
      "step": 18325
    },
    {
      "epoch": 0.12451319568270232,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009663523827885734,
      "loss": 2.1234,
      "step": 18326
    },
    {
      "epoch": 0.12451999002929637,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009663485026957031,
      "loss": 2.1574,
      "step": 18327
    },
    {
      "epoch": 0.12452678437589043,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009663446223869186,
      "loss": 2.2408,
      "step": 18328
    },
    {
      "epoch": 0.1245335787224845,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009663407418622221,
      "loss": 2.094,
      "step": 18329
    },
    {
      "epoch": 0.12454037306907854,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000966336861121615,
      "loss": 2.1872,
      "step": 18330
    },
    {
      "epoch": 0.12454716741567261,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009663329801650996,
      "loss": 2.2421,
      "step": 18331
    },
    {
      "epoch": 0.12455396176226666,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009663290989926769,
      "loss": 2.2878,
      "step": 18332
    },
    {
      "epoch": 0.12456075610886072,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009663252176043496,
      "loss": 2.1823,
      "step": 18333
    },
    {
      "epoch": 0.12456755045545478,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009663213360001188,
      "loss": 2.3021,
      "step": 18334
    },
    {
      "epoch": 0.12457434480204883,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009663174541799867,
      "loss": 2.2423,
      "step": 18335
    },
    {
      "epoch": 0.1245811391486429,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009663135721439548,
      "loss": 2.2815,
      "step": 18336
    },
    {
      "epoch": 0.12458793349523695,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009663096898920252,
      "loss": 2.1748,
      "step": 18337
    },
    {
      "epoch": 0.12459472784183101,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009663058074241995,
      "loss": 2.1592,
      "step": 18338
    },
    {
      "epoch": 0.12460152218842506,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009663019247404796,
      "loss": 2.2657,
      "step": 18339
    },
    {
      "epoch": 0.12460831653501912,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009662980418408673,
      "loss": 2.2276,
      "step": 18340
    },
    {
      "epoch": 0.12461511088161319,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009662941587253643,
      "loss": 2.0287,
      "step": 18341
    },
    {
      "epoch": 0.12462190522820724,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009662902753939725,
      "loss": 2.2004,
      "step": 18342
    },
    {
      "epoch": 0.1246286995748013,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009662863918466936,
      "loss": 2.2086,
      "step": 18343
    },
    {
      "epoch": 0.12463549392139535,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009662825080835294,
      "loss": 2.1758,
      "step": 18344
    },
    {
      "epoch": 0.12464228826798941,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009662786241044818,
      "loss": 2.2643,
      "step": 18345
    },
    {
      "epoch": 0.12464908261458348,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009662747399095526,
      "loss": 2.2625,
      "step": 18346
    },
    {
      "epoch": 0.12465587696117753,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009662708554987434,
      "loss": 2.1226,
      "step": 18347
    },
    {
      "epoch": 0.12466267130777159,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009662669708720562,
      "loss": 2.1379,
      "step": 18348
    },
    {
      "epoch": 0.12466946565436564,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009662630860294927,
      "loss": 2.1752,
      "step": 18349
    },
    {
      "epoch": 0.1246762600009597,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009662592009710549,
      "loss": 2.1094,
      "step": 18350
    },
    {
      "epoch": 0.12468305434755377,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009662553156967443,
      "loss": 2.3233,
      "step": 18351
    },
    {
      "epoch": 0.12468984869414781,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009662514302065628,
      "loss": 2.3048,
      "step": 18352
    },
    {
      "epoch": 0.12469664304074188,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009662475445005123,
      "loss": 2.1897,
      "step": 18353
    },
    {
      "epoch": 0.12470343738733593,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009662436585785945,
      "loss": 2.0777,
      "step": 18354
    },
    {
      "epoch": 0.12471023173392999,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009662397724408114,
      "loss": 2.3245,
      "step": 18355
    },
    {
      "epoch": 0.12471702608052405,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009662358860871645,
      "loss": 2.1911,
      "step": 18356
    },
    {
      "epoch": 0.1247238204271181,
      "grad_norm": 0.125,
      "learning_rate": 0.0009662319995176556,
      "loss": 2.1671,
      "step": 18357
    },
    {
      "epoch": 0.12473061477371217,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009662281127322867,
      "loss": 2.1171,
      "step": 18358
    },
    {
      "epoch": 0.12473740912030622,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009662242257310596,
      "loss": 2.1927,
      "step": 18359
    },
    {
      "epoch": 0.12474420346690028,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000966220338513976,
      "loss": 2.1856,
      "step": 18360
    },
    {
      "epoch": 0.12475099781349433,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009662164510810379,
      "loss": 2.2594,
      "step": 18361
    },
    {
      "epoch": 0.1247577921600884,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009662125634322468,
      "loss": 2.1992,
      "step": 18362
    },
    {
      "epoch": 0.12476458650668246,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009662086755676045,
      "loss": 2.1325,
      "step": 18363
    },
    {
      "epoch": 0.1247713808532765,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000966204787487113,
      "loss": 2.1233,
      "step": 18364
    },
    {
      "epoch": 0.12477817519987057,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000966200899190774,
      "loss": 2.0046,
      "step": 18365
    },
    {
      "epoch": 0.12478496954646462,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009661970106785893,
      "loss": 2.1718,
      "step": 18366
    },
    {
      "epoch": 0.12479176389305868,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009661931219505608,
      "loss": 2.1689,
      "step": 18367
    },
    {
      "epoch": 0.12479855823965275,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009661892330066903,
      "loss": 2.1958,
      "step": 18368
    },
    {
      "epoch": 0.1248053525862468,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009661853438469795,
      "loss": 2.1905,
      "step": 18369
    },
    {
      "epoch": 0.12481214693284086,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009661814544714302,
      "loss": 2.1842,
      "step": 18370
    },
    {
      "epoch": 0.12481894127943491,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009661775648800441,
      "loss": 2.1929,
      "step": 18371
    },
    {
      "epoch": 0.12482573562602897,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009661736750728233,
      "loss": 2.1283,
      "step": 18372
    },
    {
      "epoch": 0.12483252997262304,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009661697850497695,
      "loss": 2.2203,
      "step": 18373
    },
    {
      "epoch": 0.12483932431921708,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009661658948108842,
      "loss": 2.3819,
      "step": 18374
    },
    {
      "epoch": 0.12484611866581115,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009661620043561696,
      "loss": 1.9576,
      "step": 18375
    },
    {
      "epoch": 0.1248529130124052,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009661581136856274,
      "loss": 2.1239,
      "step": 18376
    },
    {
      "epoch": 0.12485970735899926,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009661542227992592,
      "loss": 2.3039,
      "step": 18377
    },
    {
      "epoch": 0.12486650170559332,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009661503316970669,
      "loss": 2.106,
      "step": 18378
    },
    {
      "epoch": 0.12487329605218737,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009661464403790524,
      "loss": 2.0553,
      "step": 18379
    },
    {
      "epoch": 0.12488009039878144,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009661425488452175,
      "loss": 2.1885,
      "step": 18380
    },
    {
      "epoch": 0.12488688474537549,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009661386570955639,
      "loss": 2.1743,
      "step": 18381
    },
    {
      "epoch": 0.12489367909196955,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009661347651300934,
      "loss": 2.3514,
      "step": 18382
    },
    {
      "epoch": 0.12490047343856361,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009661308729488079,
      "loss": 2.0438,
      "step": 18383
    },
    {
      "epoch": 0.12490726778515766,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009661269805517093,
      "loss": 2.1741,
      "step": 18384
    },
    {
      "epoch": 0.12491406213175173,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000966123087938799,
      "loss": 2.1916,
      "step": 18385
    },
    {
      "epoch": 0.12492085647834578,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000966119195110079,
      "loss": 2.2222,
      "step": 18386
    },
    {
      "epoch": 0.12492765082493984,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009661153020655513,
      "loss": 2.1313,
      "step": 18387
    },
    {
      "epoch": 0.12493444517153389,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009661114088052176,
      "loss": 2.0979,
      "step": 18388
    },
    {
      "epoch": 0.12494123951812795,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009661075153290797,
      "loss": 2.2191,
      "step": 18389
    },
    {
      "epoch": 0.12494803386472202,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009661036216371393,
      "loss": 2.2132,
      "step": 18390
    },
    {
      "epoch": 0.12495482821131607,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009660997277293981,
      "loss": 2.2309,
      "step": 18391
    },
    {
      "epoch": 0.12496162255791013,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009660958336058583,
      "loss": 2.1728,
      "step": 18392
    },
    {
      "epoch": 0.12496841690450418,
      "grad_norm": 0.125,
      "learning_rate": 0.0009660919392665215,
      "loss": 2.0611,
      "step": 18393
    },
    {
      "epoch": 0.12497521125109824,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009660880447113894,
      "loss": 2.1836,
      "step": 18394
    },
    {
      "epoch": 0.1249820055976923,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009660841499404638,
      "loss": 2.2653,
      "step": 18395
    },
    {
      "epoch": 0.12498879994428636,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009660802549537466,
      "loss": 2.1037,
      "step": 18396
    },
    {
      "epoch": 0.12499559429088042,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009660763597512396,
      "loss": 2.189,
      "step": 18397
    },
    {
      "epoch": 0.12500238863747448,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009660724643329446,
      "loss": 2.1918,
      "step": 18398
    },
    {
      "epoch": 0.12500918298406852,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009660685686988634,
      "loss": 2.1102,
      "step": 18399
    },
    {
      "epoch": 0.12501597733066258,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009660646728489977,
      "loss": 2.1748,
      "step": 18400
    },
    {
      "epoch": 0.12502277167725664,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009660607767833496,
      "loss": 1.9034,
      "step": 18401
    },
    {
      "epoch": 0.1250295660238507,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009660568805019206,
      "loss": 2.0725,
      "step": 18402
    },
    {
      "epoch": 0.12503636037044477,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009660529840047126,
      "loss": 2.0703,
      "step": 18403
    },
    {
      "epoch": 0.1250431547170388,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009660490872917273,
      "loss": 2.1163,
      "step": 18404
    },
    {
      "epoch": 0.12504994906363287,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009660451903629668,
      "loss": 2.169,
      "step": 18405
    },
    {
      "epoch": 0.12505674341022693,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009660412932184327,
      "loss": 2.1184,
      "step": 18406
    },
    {
      "epoch": 0.125063537756821,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009660373958581268,
      "loss": 2.2048,
      "step": 18407
    },
    {
      "epoch": 0.12507033210341506,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009660334982820509,
      "loss": 2.2202,
      "step": 18408
    },
    {
      "epoch": 0.1250771264500091,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009660296004902068,
      "loss": 2.2198,
      "step": 18409
    },
    {
      "epoch": 0.12508392079660316,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009660257024825964,
      "loss": 2.2478,
      "step": 18410
    },
    {
      "epoch": 0.12509071514319722,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009660218042592215,
      "loss": 2.2683,
      "step": 18411
    },
    {
      "epoch": 0.1250975094897913,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009660179058200838,
      "loss": 2.1414,
      "step": 18412
    },
    {
      "epoch": 0.12510430383638535,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009660140071651851,
      "loss": 2.2173,
      "step": 18413
    },
    {
      "epoch": 0.12511109818297939,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009660101082945272,
      "loss": 2.2371,
      "step": 18414
    },
    {
      "epoch": 0.12511789252957345,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000966006209208112,
      "loss": 2.0359,
      "step": 18415
    },
    {
      "epoch": 0.1251246868761675,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009660023099059413,
      "loss": 2.0807,
      "step": 18416
    },
    {
      "epoch": 0.12513148122276158,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009659984103880168,
      "loss": 2.1597,
      "step": 18417
    },
    {
      "epoch": 0.12513827556935564,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009659945106543405,
      "loss": 2.2424,
      "step": 18418
    },
    {
      "epoch": 0.12514506991594967,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009659906107049142,
      "loss": 2.1994,
      "step": 18419
    },
    {
      "epoch": 0.12515186426254374,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009659867105397393,
      "loss": 2.0495,
      "step": 18420
    },
    {
      "epoch": 0.1251586586091378,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009659828101588181,
      "loss": 2.3189,
      "step": 18421
    },
    {
      "epoch": 0.12516545295573187,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009659789095621521,
      "loss": 2.0773,
      "step": 18422
    },
    {
      "epoch": 0.12517224730232593,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009659750087497432,
      "loss": 2.1491,
      "step": 18423
    },
    {
      "epoch": 0.12517904164891996,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009659711077215933,
      "loss": 1.9717,
      "step": 18424
    },
    {
      "epoch": 0.12518583599551403,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000965967206477704,
      "loss": 2.1333,
      "step": 18425
    },
    {
      "epoch": 0.1251926303421081,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009659633050180774,
      "loss": 2.1605,
      "step": 18426
    },
    {
      "epoch": 0.12519942468870215,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000965959403342715,
      "loss": 2.2812,
      "step": 18427
    },
    {
      "epoch": 0.12520621903529622,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009659555014516187,
      "loss": 2.2034,
      "step": 18428
    },
    {
      "epoch": 0.12521301338189025,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009659515993447904,
      "loss": 2.2479,
      "step": 18429
    },
    {
      "epoch": 0.12521980772848432,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009659476970222318,
      "loss": 1.992,
      "step": 18430
    },
    {
      "epoch": 0.12522660207507838,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009659437944839449,
      "loss": 2.0788,
      "step": 18431
    },
    {
      "epoch": 0.12523339642167244,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009659398917299313,
      "loss": 2.2324,
      "step": 18432
    },
    {
      "epoch": 0.1252401907682665,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009659359887601929,
      "loss": 2.2097,
      "step": 18433
    },
    {
      "epoch": 0.12524698511486054,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009659320855747313,
      "loss": 2.1759,
      "step": 18434
    },
    {
      "epoch": 0.1252537794614546,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009659281821735486,
      "loss": 2.1291,
      "step": 18435
    },
    {
      "epoch": 0.12526057380804867,
      "grad_norm": 0.1591796875,
      "learning_rate": 0.0009659242785566466,
      "loss": 2.1736,
      "step": 18436
    },
    {
      "epoch": 0.12526736815464273,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009659203747240269,
      "loss": 2.203,
      "step": 18437
    },
    {
      "epoch": 0.1252741625012368,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009659164706756913,
      "loss": 2.0878,
      "step": 18438
    },
    {
      "epoch": 0.12528095684783083,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000965912566411642,
      "loss": 2.3497,
      "step": 18439
    },
    {
      "epoch": 0.1252877511944249,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009659086619318803,
      "loss": 2.1958,
      "step": 18440
    },
    {
      "epoch": 0.12529454554101896,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009659047572364084,
      "loss": 2.1094,
      "step": 18441
    },
    {
      "epoch": 0.12530133988761302,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009659008523252278,
      "loss": 2.1166,
      "step": 18442
    },
    {
      "epoch": 0.12530813423420709,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009658969471983403,
      "loss": 2.1972,
      "step": 18443
    },
    {
      "epoch": 0.12531492858080112,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009658930418557481,
      "loss": 2.1495,
      "step": 18444
    },
    {
      "epoch": 0.12532172292739518,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009658891362974526,
      "loss": 2.132,
      "step": 18445
    },
    {
      "epoch": 0.12532851727398925,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000965885230523456,
      "loss": 2.313,
      "step": 18446
    },
    {
      "epoch": 0.1253353116205833,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009658813245337597,
      "loss": 2.2601,
      "step": 18447
    },
    {
      "epoch": 0.12534210596717735,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009658774183283657,
      "loss": 2.2791,
      "step": 18448
    },
    {
      "epoch": 0.1253489003137714,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009658735119072759,
      "loss": 2.2435,
      "step": 18449
    },
    {
      "epoch": 0.12535569466036547,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009658696052704919,
      "loss": 2.1754,
      "step": 18450
    },
    {
      "epoch": 0.12536248900695954,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009658656984180156,
      "loss": 2.0319,
      "step": 18451
    },
    {
      "epoch": 0.1253692833535536,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009658617913498489,
      "loss": 2.0887,
      "step": 18452
    },
    {
      "epoch": 0.12537607770014764,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009658578840659935,
      "loss": 2.354,
      "step": 18453
    },
    {
      "epoch": 0.1253828720467417,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009658539765664512,
      "loss": 2.3193,
      "step": 18454
    },
    {
      "epoch": 0.12538966639333576,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000965850068851224,
      "loss": 2.2268,
      "step": 18455
    },
    {
      "epoch": 0.12539646073992983,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009658461609203134,
      "loss": 2.1422,
      "step": 18456
    },
    {
      "epoch": 0.1254032550865239,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009658422527737214,
      "loss": 2.0583,
      "step": 18457
    },
    {
      "epoch": 0.12541004943311793,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009658383444114498,
      "loss": 2.0387,
      "step": 18458
    },
    {
      "epoch": 0.125416843779712,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009658344358335004,
      "loss": 2.048,
      "step": 18459
    },
    {
      "epoch": 0.12542363812630605,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009658305270398749,
      "loss": 2.3025,
      "step": 18460
    },
    {
      "epoch": 0.12543043247290012,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009658266180305753,
      "loss": 2.2539,
      "step": 18461
    },
    {
      "epoch": 0.12543722681949418,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009658227088056033,
      "loss": 2.2259,
      "step": 18462
    },
    {
      "epoch": 0.12544402116608822,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009658187993649607,
      "loss": 2.2309,
      "step": 18463
    },
    {
      "epoch": 0.12545081551268228,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009658148897086494,
      "loss": 2.0558,
      "step": 18464
    },
    {
      "epoch": 0.12545760985927634,
      "grad_norm": 0.125,
      "learning_rate": 0.0009658109798366712,
      "loss": 2.0743,
      "step": 18465
    },
    {
      "epoch": 0.1254644042058704,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009658070697490278,
      "loss": 2.0929,
      "step": 18466
    },
    {
      "epoch": 0.12547119855246447,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009658031594457209,
      "loss": 2.2184,
      "step": 18467
    },
    {
      "epoch": 0.1254779928990585,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009657992489267526,
      "loss": 2.1084,
      "step": 18468
    },
    {
      "epoch": 0.12548478724565257,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009657953381921248,
      "loss": 2.1627,
      "step": 18469
    },
    {
      "epoch": 0.12549158159224663,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009657914272418387,
      "loss": 2.2833,
      "step": 18470
    },
    {
      "epoch": 0.1254983759388407,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009657875160758967,
      "loss": 2.1734,
      "step": 18471
    },
    {
      "epoch": 0.12550517028543476,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009657836046943005,
      "loss": 2.1993,
      "step": 18472
    },
    {
      "epoch": 0.1255119646320288,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009657796930970517,
      "loss": 2.1628,
      "step": 18473
    },
    {
      "epoch": 0.12551875897862286,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009657757812841524,
      "loss": 2.1779,
      "step": 18474
    },
    {
      "epoch": 0.12552555332521692,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009657718692556042,
      "loss": 2.1622,
      "step": 18475
    },
    {
      "epoch": 0.12553234767181098,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009657679570114089,
      "loss": 2.1971,
      "step": 18476
    },
    {
      "epoch": 0.12553914201840505,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009657640445515683,
      "loss": 2.1788,
      "step": 18477
    },
    {
      "epoch": 0.12554593636499908,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009657601318760845,
      "loss": 2.3585,
      "step": 18478
    },
    {
      "epoch": 0.12555273071159315,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009657562189849589,
      "loss": 2.0749,
      "step": 18479
    },
    {
      "epoch": 0.1255595250581872,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009657523058781936,
      "loss": 2.0584,
      "step": 18480
    },
    {
      "epoch": 0.12556631940478127,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009657483925557903,
      "loss": 2.3315,
      "step": 18481
    },
    {
      "epoch": 0.12557311375137534,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009657444790177507,
      "loss": 2.1685,
      "step": 18482
    },
    {
      "epoch": 0.12557990809796937,
      "grad_norm": 0.125,
      "learning_rate": 0.000965740565264077,
      "loss": 2.072,
      "step": 18483
    },
    {
      "epoch": 0.12558670244456344,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009657366512947707,
      "loss": 2.2955,
      "step": 18484
    },
    {
      "epoch": 0.1255934967911575,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009657327371098335,
      "loss": 2.2123,
      "step": 18485
    },
    {
      "epoch": 0.12560029113775156,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009657288227092676,
      "loss": 2.1393,
      "step": 18486
    },
    {
      "epoch": 0.12560708548434563,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009657249080930745,
      "loss": 2.1406,
      "step": 18487
    },
    {
      "epoch": 0.12561387983093966,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009657209932612561,
      "loss": 2.2203,
      "step": 18488
    },
    {
      "epoch": 0.12562067417753373,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009657170782138143,
      "loss": 1.9756,
      "step": 18489
    },
    {
      "epoch": 0.1256274685241278,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009657131629507508,
      "loss": 2.0885,
      "step": 18490
    },
    {
      "epoch": 0.12563426287072185,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009657092474720672,
      "loss": 2.2857,
      "step": 18491
    },
    {
      "epoch": 0.12564105721731592,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009657053317777657,
      "loss": 2.2533,
      "step": 18492
    },
    {
      "epoch": 0.12564785156390995,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009657014158678481,
      "loss": 2.3065,
      "step": 18493
    },
    {
      "epoch": 0.12565464591050401,
      "grad_norm": 0.125,
      "learning_rate": 0.0009656974997423161,
      "loss": 2.0562,
      "step": 18494
    },
    {
      "epoch": 0.12566144025709808,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009656935834011713,
      "loss": 2.2437,
      "step": 18495
    },
    {
      "epoch": 0.12566823460369214,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000965689666844416,
      "loss": 2.0856,
      "step": 18496
    },
    {
      "epoch": 0.12567502895028618,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009656857500720514,
      "loss": 2.1933,
      "step": 18497
    },
    {
      "epoch": 0.12568182329688024,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009656818330840797,
      "loss": 2.144,
      "step": 18498
    },
    {
      "epoch": 0.1256886176434743,
      "grad_norm": 0.125,
      "learning_rate": 0.0009656779158805028,
      "loss": 2.0766,
      "step": 18499
    },
    {
      "epoch": 0.12569541199006837,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009656739984613223,
      "loss": 2.2618,
      "step": 18500
    },
    {
      "epoch": 0.12570220633666243,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009656700808265401,
      "loss": 2.2049,
      "step": 18501
    },
    {
      "epoch": 0.12570900068325647,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.000965666162976158,
      "loss": 2.2719,
      "step": 18502
    },
    {
      "epoch": 0.12571579502985053,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009656622449101777,
      "loss": 2.0058,
      "step": 18503
    },
    {
      "epoch": 0.1257225893764446,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009656583266286012,
      "loss": 2.1753,
      "step": 18504
    },
    {
      "epoch": 0.12572938372303866,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009656544081314303,
      "loss": 2.192,
      "step": 18505
    },
    {
      "epoch": 0.12573617806963272,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009656504894186666,
      "loss": 2.3204,
      "step": 18506
    },
    {
      "epoch": 0.12574297241622676,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009656465704903121,
      "loss": 2.2193,
      "step": 18507
    },
    {
      "epoch": 0.12574976676282082,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009656426513463685,
      "loss": 2.3021,
      "step": 18508
    },
    {
      "epoch": 0.12575656110941488,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009656387319868378,
      "loss": 2.1938,
      "step": 18509
    },
    {
      "epoch": 0.12576335545600895,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009656348124117218,
      "loss": 2.1363,
      "step": 18510
    },
    {
      "epoch": 0.125770149802603,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009656308926210221,
      "loss": 2.1472,
      "step": 18511
    },
    {
      "epoch": 0.12577694414919705,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009656269726147406,
      "loss": 2.3103,
      "step": 18512
    },
    {
      "epoch": 0.1257837384957911,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009656230523928793,
      "loss": 2.1692,
      "step": 18513
    },
    {
      "epoch": 0.12579053284238517,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009656191319554398,
      "loss": 2.1221,
      "step": 18514
    },
    {
      "epoch": 0.12579732718897924,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009656152113024239,
      "loss": 2.1338,
      "step": 18515
    },
    {
      "epoch": 0.1258041215355733,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009656112904338335,
      "loss": 2.2958,
      "step": 18516
    },
    {
      "epoch": 0.12581091588216733,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009656073693496705,
      "loss": 2.1914,
      "step": 18517
    },
    {
      "epoch": 0.1258177102287614,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009656034480499364,
      "loss": 2.1614,
      "step": 18518
    },
    {
      "epoch": 0.12582450457535546,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009655995265346335,
      "loss": 2.156,
      "step": 18519
    },
    {
      "epoch": 0.12583129892194952,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009655956048037633,
      "loss": 2.2998,
      "step": 18520
    },
    {
      "epoch": 0.1258380932685436,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009655916828573277,
      "loss": 2.0796,
      "step": 18521
    },
    {
      "epoch": 0.12584488761513762,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009655877606953285,
      "loss": 2.251,
      "step": 18522
    },
    {
      "epoch": 0.1258516819617317,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009655838383177673,
      "loss": 2.1405,
      "step": 18523
    },
    {
      "epoch": 0.12585847630832575,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009655799157246463,
      "loss": 2.2179,
      "step": 18524
    },
    {
      "epoch": 0.12586527065491981,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009655759929159671,
      "loss": 2.037,
      "step": 18525
    },
    {
      "epoch": 0.12587206500151388,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009655720698917317,
      "loss": 2.0982,
      "step": 18526
    },
    {
      "epoch": 0.1258788593481079,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009655681466519414,
      "loss": 2.2677,
      "step": 18527
    },
    {
      "epoch": 0.12588565369470198,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009655642231965986,
      "loss": 2.1463,
      "step": 18528
    },
    {
      "epoch": 0.12589244804129604,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000965560299525705,
      "loss": 2.1904,
      "step": 18529
    },
    {
      "epoch": 0.1258992423878901,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009655563756392622,
      "loss": 2.2457,
      "step": 18530
    },
    {
      "epoch": 0.12590603673448417,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009655524515372723,
      "loss": 2.1225,
      "step": 18531
    },
    {
      "epoch": 0.1259128310810782,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009655485272197367,
      "loss": 2.2509,
      "step": 18532
    },
    {
      "epoch": 0.12591962542767227,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009655446026866577,
      "loss": 2.249,
      "step": 18533
    },
    {
      "epoch": 0.12592641977426633,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009655406779380366,
      "loss": 2.2629,
      "step": 18534
    },
    {
      "epoch": 0.1259332141208604,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009655367529738758,
      "loss": 2.1962,
      "step": 18535
    },
    {
      "epoch": 0.12594000846745446,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009655328277941766,
      "loss": 2.1247,
      "step": 18536
    },
    {
      "epoch": 0.1259468028140485,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009655289023989412,
      "loss": 2.1072,
      "step": 18537
    },
    {
      "epoch": 0.12595359716064256,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009655249767881712,
      "loss": 2.1296,
      "step": 18538
    },
    {
      "epoch": 0.12596039150723662,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009655210509618684,
      "loss": 2.127,
      "step": 18539
    },
    {
      "epoch": 0.12596718585383068,
      "grad_norm": 0.1689453125,
      "learning_rate": 0.0009655171249200348,
      "loss": 2.1705,
      "step": 18540
    },
    {
      "epoch": 0.12597398020042472,
      "grad_norm": 0.125,
      "learning_rate": 0.000965513198662672,
      "loss": 2.0875,
      "step": 18541
    },
    {
      "epoch": 0.12598077454701878,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000965509272189782,
      "loss": 2.2677,
      "step": 18542
    },
    {
      "epoch": 0.12598756889361284,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009655053455013666,
      "loss": 2.0996,
      "step": 18543
    },
    {
      "epoch": 0.1259943632402069,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009655014185974275,
      "loss": 2.1073,
      "step": 18544
    },
    {
      "epoch": 0.12600115758680097,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009654974914779666,
      "loss": 2.1701,
      "step": 18545
    },
    {
      "epoch": 0.126007951933395,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009654935641429856,
      "loss": 2.2168,
      "step": 18546
    },
    {
      "epoch": 0.12601474627998907,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009654896365924866,
      "loss": 2.1512,
      "step": 18547
    },
    {
      "epoch": 0.12602154062658313,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000965485708826471,
      "loss": 2.1198,
      "step": 18548
    },
    {
      "epoch": 0.1260283349731772,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000965481780844941,
      "loss": 2.1118,
      "step": 18549
    },
    {
      "epoch": 0.12603512931977126,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009654778526478982,
      "loss": 2.1594,
      "step": 18550
    },
    {
      "epoch": 0.1260419236663653,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009654739242353445,
      "loss": 2.0561,
      "step": 18551
    },
    {
      "epoch": 0.12604871801295936,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009654699956072819,
      "loss": 2.1892,
      "step": 18552
    },
    {
      "epoch": 0.12605551235955342,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009654660667637119,
      "loss": 2.3316,
      "step": 18553
    },
    {
      "epoch": 0.1260623067061475,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009654621377046363,
      "loss": 2.1851,
      "step": 18554
    },
    {
      "epoch": 0.12606910105274155,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009654582084300572,
      "loss": 2.0305,
      "step": 18555
    },
    {
      "epoch": 0.12607589539933559,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009654542789399763,
      "loss": 2.1612,
      "step": 18556
    },
    {
      "epoch": 0.12608268974592965,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009654503492343954,
      "loss": 2.1545,
      "step": 18557
    },
    {
      "epoch": 0.1260894840925237,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009654464193133163,
      "loss": 2.1993,
      "step": 18558
    },
    {
      "epoch": 0.12609627843911778,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009654424891767408,
      "loss": 2.2703,
      "step": 18559
    },
    {
      "epoch": 0.12610307278571184,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009654385588246708,
      "loss": 2.2088,
      "step": 18560
    },
    {
      "epoch": 0.12610986713230588,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009654346282571082,
      "loss": 2.248,
      "step": 18561
    },
    {
      "epoch": 0.12611666147889994,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009654306974740545,
      "loss": 2.0619,
      "step": 18562
    },
    {
      "epoch": 0.126123455825494,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009654267664755119,
      "loss": 2.2203,
      "step": 18563
    },
    {
      "epoch": 0.12613025017208807,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009654228352614819,
      "loss": 2.1935,
      "step": 18564
    },
    {
      "epoch": 0.12613704451868213,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009654189038319665,
      "loss": 2.0939,
      "step": 18565
    },
    {
      "epoch": 0.12614383886527616,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009654149721869676,
      "loss": 2.2158,
      "step": 18566
    },
    {
      "epoch": 0.12615063321187023,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009654110403264868,
      "loss": 2.232,
      "step": 18567
    },
    {
      "epoch": 0.1261574275584643,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000965407108250526,
      "loss": 2.107,
      "step": 18568
    },
    {
      "epoch": 0.12616422190505835,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009654031759590871,
      "loss": 2.2837,
      "step": 18569
    },
    {
      "epoch": 0.12617101625165242,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009653992434521719,
      "loss": 2.1824,
      "step": 18570
    },
    {
      "epoch": 0.12617781059824645,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009653953107297822,
      "loss": 2.0263,
      "step": 18571
    },
    {
      "epoch": 0.12618460494484052,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009653913777919196,
      "loss": 2.1833,
      "step": 18572
    },
    {
      "epoch": 0.12619139929143458,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009653874446385863,
      "loss": 2.178,
      "step": 18573
    },
    {
      "epoch": 0.12619819363802864,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000965383511269784,
      "loss": 2.0412,
      "step": 18574
    },
    {
      "epoch": 0.1262049879846227,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009653795776855143,
      "loss": 2.0817,
      "step": 18575
    },
    {
      "epoch": 0.12621178233121674,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009653756438857792,
      "loss": 2.2683,
      "step": 18576
    },
    {
      "epoch": 0.1262185766778108,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009653717098705806,
      "loss": 2.1651,
      "step": 18577
    },
    {
      "epoch": 0.12622537102440487,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009653677756399203,
      "loss": 2.1653,
      "step": 18578
    },
    {
      "epoch": 0.12623216537099893,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009653638411938,
      "loss": 2.1312,
      "step": 18579
    },
    {
      "epoch": 0.126238959717593,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009653599065322215,
      "loss": 2.2915,
      "step": 18580
    },
    {
      "epoch": 0.12624575406418703,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009653559716551867,
      "loss": 2.2394,
      "step": 18581
    },
    {
      "epoch": 0.1262525484107811,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009653520365626976,
      "loss": 2.2097,
      "step": 18582
    },
    {
      "epoch": 0.12625934275737516,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009653481012547555,
      "loss": 2.1052,
      "step": 18583
    },
    {
      "epoch": 0.12626613710396922,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009653441657313628,
      "loss": 2.1389,
      "step": 18584
    },
    {
      "epoch": 0.12627293145056329,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009653402299925211,
      "loss": 2.1591,
      "step": 18585
    },
    {
      "epoch": 0.12627972579715732,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009653362940382321,
      "loss": 2.279,
      "step": 18586
    },
    {
      "epoch": 0.12628652014375139,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000965332357868498,
      "loss": 2.263,
      "step": 18587
    },
    {
      "epoch": 0.12629331449034545,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009653284214833201,
      "loss": 2.1954,
      "step": 18588
    },
    {
      "epoch": 0.1263001088369395,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009653244848827005,
      "loss": 2.1064,
      "step": 18589
    },
    {
      "epoch": 0.12630690318353355,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009653205480666412,
      "loss": 2.2839,
      "step": 18590
    },
    {
      "epoch": 0.1263136975301276,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009653166110351435,
      "loss": 2.099,
      "step": 18591
    },
    {
      "epoch": 0.12632049187672167,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009653126737882098,
      "loss": 2.224,
      "step": 18592
    },
    {
      "epoch": 0.12632728622331574,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009653087363258415,
      "loss": 2.2083,
      "step": 18593
    },
    {
      "epoch": 0.1263340805699098,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009653047986480407,
      "loss": 2.1871,
      "step": 18594
    },
    {
      "epoch": 0.12634087491650384,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000965300860754809,
      "loss": 2.2161,
      "step": 18595
    },
    {
      "epoch": 0.1263476692630979,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009652969226461484,
      "loss": 2.2089,
      "step": 18596
    },
    {
      "epoch": 0.12635446360969196,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009652929843220607,
      "loss": 2.2309,
      "step": 18597
    },
    {
      "epoch": 0.12636125795628603,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009652890457825476,
      "loss": 2.1871,
      "step": 18598
    },
    {
      "epoch": 0.1263680523028801,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000965285107027611,
      "loss": 2.1309,
      "step": 18599
    },
    {
      "epoch": 0.12637484664947413,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009652811680572528,
      "loss": 2.1165,
      "step": 18600
    },
    {
      "epoch": 0.1263816409960682,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009652772288714749,
      "loss": 2.1152,
      "step": 18601
    },
    {
      "epoch": 0.12638843534266225,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009652732894702787,
      "loss": 2.1625,
      "step": 18602
    },
    {
      "epoch": 0.12639522968925632,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009652693498536665,
      "loss": 2.1393,
      "step": 18603
    },
    {
      "epoch": 0.12640202403585038,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009652654100216398,
      "loss": 2.1731,
      "step": 18604
    },
    {
      "epoch": 0.12640881838244442,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009652614699742006,
      "loss": 2.0985,
      "step": 18605
    },
    {
      "epoch": 0.12641561272903848,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009652575297113506,
      "loss": 2.3273,
      "step": 18606
    },
    {
      "epoch": 0.12642240707563254,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009652535892330918,
      "loss": 2.1733,
      "step": 18607
    },
    {
      "epoch": 0.1264292014222266,
      "grad_norm": 0.125,
      "learning_rate": 0.0009652496485394258,
      "loss": 2.1347,
      "step": 18608
    },
    {
      "epoch": 0.12643599576882067,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009652457076303547,
      "loss": 2.1741,
      "step": 18609
    },
    {
      "epoch": 0.1264427901154147,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009652417665058802,
      "loss": 2.09,
      "step": 18610
    },
    {
      "epoch": 0.12644958446200877,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009652378251660039,
      "loss": 2.2099,
      "step": 18611
    },
    {
      "epoch": 0.12645637880860283,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009652338836107279,
      "loss": 2.189,
      "step": 18612
    },
    {
      "epoch": 0.1264631731551969,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009652299418400538,
      "loss": 2.1562,
      "step": 18613
    },
    {
      "epoch": 0.12646996750179096,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009652259998539839,
      "loss": 2.0425,
      "step": 18614
    },
    {
      "epoch": 0.126476761848385,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009652220576525194,
      "loss": 2.3575,
      "step": 18615
    },
    {
      "epoch": 0.12648355619497906,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009652181152356626,
      "loss": 2.2322,
      "step": 18616
    },
    {
      "epoch": 0.12649035054157312,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009652141726034151,
      "loss": 2.2291,
      "step": 18617
    },
    {
      "epoch": 0.12649714488816718,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009652102297557787,
      "loss": 2.1233,
      "step": 18618
    },
    {
      "epoch": 0.12650393923476125,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009652062866927555,
      "loss": 2.2171,
      "step": 18619
    },
    {
      "epoch": 0.12651073358135528,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009652023434143468,
      "loss": 2.1362,
      "step": 18620
    },
    {
      "epoch": 0.12651752792794935,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009651983999205551,
      "loss": 2.2282,
      "step": 18621
    },
    {
      "epoch": 0.1265243222745434,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009651944562113817,
      "loss": 2.2118,
      "step": 18622
    },
    {
      "epoch": 0.12653111662113747,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009651905122868285,
      "loss": 2.2903,
      "step": 18623
    },
    {
      "epoch": 0.12653791096773154,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009651865681468976,
      "loss": 2.2454,
      "step": 18624
    },
    {
      "epoch": 0.12654470531432557,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009651826237915906,
      "loss": 2.2122,
      "step": 18625
    },
    {
      "epoch": 0.12655149966091964,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009651786792209094,
      "loss": 2.112,
      "step": 18626
    },
    {
      "epoch": 0.1265582940075137,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009651747344348557,
      "loss": 2.155,
      "step": 18627
    },
    {
      "epoch": 0.12656508835410776,
      "grad_norm": 0.125,
      "learning_rate": 0.0009651707894334315,
      "loss": 2.0054,
      "step": 18628
    },
    {
      "epoch": 0.12657188270070183,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009651668442166386,
      "loss": 2.2257,
      "step": 18629
    },
    {
      "epoch": 0.12657867704729586,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009651628987844787,
      "loss": 2.2643,
      "step": 18630
    },
    {
      "epoch": 0.12658547139388993,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009651589531369538,
      "loss": 2.0887,
      "step": 18631
    },
    {
      "epoch": 0.126592265740484,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009651550072740656,
      "loss": 2.1207,
      "step": 18632
    },
    {
      "epoch": 0.12659906008707805,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009651510611958159,
      "loss": 2.1769,
      "step": 18633
    },
    {
      "epoch": 0.12660585443367212,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009651471149022066,
      "loss": 2.0929,
      "step": 18634
    },
    {
      "epoch": 0.12661264878026615,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009651431683932397,
      "loss": 2.2125,
      "step": 18635
    },
    {
      "epoch": 0.12661944312686021,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009651392216689166,
      "loss": 2.005,
      "step": 18636
    },
    {
      "epoch": 0.12662623747345428,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009651352747292395,
      "loss": 2.2485,
      "step": 18637
    },
    {
      "epoch": 0.12663303182004834,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009651313275742101,
      "loss": 2.2051,
      "step": 18638
    },
    {
      "epoch": 0.12663982616664238,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009651273802038302,
      "loss": 2.0354,
      "step": 18639
    },
    {
      "epoch": 0.12664662051323644,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009651234326181016,
      "loss": 2.0546,
      "step": 18640
    },
    {
      "epoch": 0.1266534148598305,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009651194848170262,
      "loss": 2.253,
      "step": 18641
    },
    {
      "epoch": 0.12666020920642457,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009651155368006059,
      "loss": 2.1385,
      "step": 18642
    },
    {
      "epoch": 0.12666700355301863,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009651115885688424,
      "loss": 2.2083,
      "step": 18643
    },
    {
      "epoch": 0.12667379789961267,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009651076401217376,
      "loss": 2.3616,
      "step": 18644
    },
    {
      "epoch": 0.12668059224620673,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009651036914592932,
      "loss": 2.1527,
      "step": 18645
    },
    {
      "epoch": 0.1266873865928008,
      "grad_norm": 0.34375,
      "learning_rate": 0.0009650997425815112,
      "loss": 2.1462,
      "step": 18646
    },
    {
      "epoch": 0.12669418093939486,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009650957934883934,
      "loss": 2.2535,
      "step": 18647
    },
    {
      "epoch": 0.12670097528598892,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009650918441799414,
      "loss": 2.1401,
      "step": 18648
    },
    {
      "epoch": 0.12670776963258296,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009650878946561574,
      "loss": 2.207,
      "step": 18649
    },
    {
      "epoch": 0.12671456397917702,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000965083944917043,
      "loss": 2.1871,
      "step": 18650
    },
    {
      "epoch": 0.12672135832577108,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009650799949625999,
      "loss": 2.2742,
      "step": 18651
    },
    {
      "epoch": 0.12672815267236515,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009650760447928303,
      "loss": 2.2526,
      "step": 18652
    },
    {
      "epoch": 0.1267349470189592,
      "grad_norm": 0.1865234375,
      "learning_rate": 0.0009650720944077358,
      "loss": 2.2381,
      "step": 18653
    },
    {
      "epoch": 0.12674174136555325,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009650681438073181,
      "loss": 2.2803,
      "step": 18654
    },
    {
      "epoch": 0.1267485357121473,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009650641929915793,
      "loss": 2.1472,
      "step": 18655
    },
    {
      "epoch": 0.12675533005874137,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000965060241960521,
      "loss": 2.0902,
      "step": 18656
    },
    {
      "epoch": 0.12676212440533544,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009650562907141454,
      "loss": 2.2127,
      "step": 18657
    },
    {
      "epoch": 0.1267689187519295,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009650523392524537,
      "loss": 2.1349,
      "step": 18658
    },
    {
      "epoch": 0.12677571309852353,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009650483875754483,
      "loss": 2.2245,
      "step": 18659
    },
    {
      "epoch": 0.1267825074451176,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009650444356831309,
      "loss": 2.1085,
      "step": 18660
    },
    {
      "epoch": 0.12678930179171166,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009650404835755031,
      "loss": 2.2768,
      "step": 18661
    },
    {
      "epoch": 0.12679609613830573,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009650365312525671,
      "loss": 2.1468,
      "step": 18662
    },
    {
      "epoch": 0.1268028904848998,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009650325787143244,
      "loss": 2.2302,
      "step": 18663
    },
    {
      "epoch": 0.12680968483149382,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009650286259607767,
      "loss": 2.2893,
      "step": 18664
    },
    {
      "epoch": 0.1268164791780879,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009650246729919265,
      "loss": 2.2425,
      "step": 18665
    },
    {
      "epoch": 0.12682327352468195,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009650207198077749,
      "loss": 2.2694,
      "step": 18666
    },
    {
      "epoch": 0.12683006787127601,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009650167664083242,
      "loss": 2.094,
      "step": 18667
    },
    {
      "epoch": 0.12683686221787008,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000965012812793576,
      "loss": 2.1623,
      "step": 18668
    },
    {
      "epoch": 0.1268436565644641,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009650088589635322,
      "loss": 2.103,
      "step": 18669
    },
    {
      "epoch": 0.12685045091105818,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009650049049181947,
      "loss": 2.1328,
      "step": 18670
    },
    {
      "epoch": 0.12685724525765224,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009650009506575652,
      "loss": 2.1289,
      "step": 18671
    },
    {
      "epoch": 0.1268640396042463,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009649969961816456,
      "loss": 2.1543,
      "step": 18672
    },
    {
      "epoch": 0.12687083395084037,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009649930414904377,
      "loss": 2.147,
      "step": 18673
    },
    {
      "epoch": 0.1268776282974344,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009649890865839434,
      "loss": 2.1694,
      "step": 18674
    },
    {
      "epoch": 0.12688442264402847,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009649851314621645,
      "loss": 2.1153,
      "step": 18675
    },
    {
      "epoch": 0.12689121699062253,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009649811761251027,
      "loss": 2.0638,
      "step": 18676
    },
    {
      "epoch": 0.1268980113372166,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.00096497722057276,
      "loss": 2.0858,
      "step": 18677
    },
    {
      "epoch": 0.12690480568381066,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009649732648051383,
      "loss": 2.3242,
      "step": 18678
    },
    {
      "epoch": 0.1269116000304047,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009649693088222391,
      "loss": 2.1861,
      "step": 18679
    },
    {
      "epoch": 0.12691839437699876,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009649653526240646,
      "loss": 2.3021,
      "step": 18680
    },
    {
      "epoch": 0.12692518872359282,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009649613962106164,
      "loss": 2.1009,
      "step": 18681
    },
    {
      "epoch": 0.12693198307018688,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009649574395818965,
      "loss": 2.0965,
      "step": 18682
    },
    {
      "epoch": 0.12693877741678095,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009649534827379065,
      "loss": 2.1868,
      "step": 18683
    },
    {
      "epoch": 0.12694557176337498,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009649495256786485,
      "loss": 2.15,
      "step": 18684
    },
    {
      "epoch": 0.12695236610996904,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000964945568404124,
      "loss": 2.3115,
      "step": 18685
    },
    {
      "epoch": 0.1269591604565631,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009649416109143352,
      "loss": 2.01,
      "step": 18686
    },
    {
      "epoch": 0.12696595480315717,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009649376532092838,
      "loss": 2.1942,
      "step": 18687
    },
    {
      "epoch": 0.1269727491497512,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009649336952889714,
      "loss": 2.2187,
      "step": 18688
    },
    {
      "epoch": 0.12697954349634527,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009649297371534003,
      "loss": 2.1688,
      "step": 18689
    },
    {
      "epoch": 0.12698633784293933,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009649257788025719,
      "loss": 2.0987,
      "step": 18690
    },
    {
      "epoch": 0.1269931321895334,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009649218202364881,
      "loss": 2.1501,
      "step": 18691
    },
    {
      "epoch": 0.12699992653612746,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009649178614551511,
      "loss": 2.2187,
      "step": 18692
    },
    {
      "epoch": 0.1270067208827215,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009649139024585622,
      "loss": 2.1479,
      "step": 18693
    },
    {
      "epoch": 0.12701351522931556,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009649099432467236,
      "loss": 2.171,
      "step": 18694
    },
    {
      "epoch": 0.12702030957590962,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000964905983819637,
      "loss": 2.1243,
      "step": 18695
    },
    {
      "epoch": 0.1270271039225037,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009649020241773043,
      "loss": 2.1979,
      "step": 18696
    },
    {
      "epoch": 0.12703389826909775,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009648980643197271,
      "loss": 2.1613,
      "step": 18697
    },
    {
      "epoch": 0.12704069261569179,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009648941042469078,
      "loss": 2.1103,
      "step": 18698
    },
    {
      "epoch": 0.12704748696228585,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009648901439588476,
      "loss": 2.0997,
      "step": 18699
    },
    {
      "epoch": 0.1270542813088799,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009648861834555486,
      "loss": 1.9976,
      "step": 18700
    },
    {
      "epoch": 0.12706107565547398,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009648822227370127,
      "loss": 2.115,
      "step": 18701
    },
    {
      "epoch": 0.12706787000206804,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009648782618032416,
      "loss": 2.1752,
      "step": 18702
    },
    {
      "epoch": 0.12707466434866208,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009648743006542372,
      "loss": 2.2573,
      "step": 18703
    },
    {
      "epoch": 0.12708145869525614,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009648703392900013,
      "loss": 2.1863,
      "step": 18704
    },
    {
      "epoch": 0.1270882530418502,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009648663777105359,
      "loss": 2.2038,
      "step": 18705
    },
    {
      "epoch": 0.12709504738844427,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009648624159158426,
      "loss": 2.126,
      "step": 18706
    },
    {
      "epoch": 0.12710184173503833,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009648584539059233,
      "loss": 2.1148,
      "step": 18707
    },
    {
      "epoch": 0.12710863608163236,
      "grad_norm": 0.142578125,
      "learning_rate": 0.00096485449168078,
      "loss": 2.0777,
      "step": 18708
    },
    {
      "epoch": 0.12711543042822643,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009648505292404141,
      "loss": 2.3125,
      "step": 18709
    },
    {
      "epoch": 0.1271222247748205,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009648465665848281,
      "loss": 2.1897,
      "step": 18710
    },
    {
      "epoch": 0.12712901912141455,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009648426037140234,
      "loss": 2.0828,
      "step": 18711
    },
    {
      "epoch": 0.12713581346800862,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009648386406280017,
      "loss": 2.1496,
      "step": 18712
    },
    {
      "epoch": 0.12714260781460265,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009648346773267651,
      "loss": 2.1719,
      "step": 18713
    },
    {
      "epoch": 0.12714940216119672,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009648307138103155,
      "loss": 2.2715,
      "step": 18714
    },
    {
      "epoch": 0.12715619650779078,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009648267500786545,
      "loss": 2.1848,
      "step": 18715
    },
    {
      "epoch": 0.12716299085438484,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009648227861317841,
      "loss": 2.1848,
      "step": 18716
    },
    {
      "epoch": 0.1271697852009789,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000964818821969706,
      "loss": 2.3646,
      "step": 18717
    },
    {
      "epoch": 0.12717657954757294,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009648148575924222,
      "loss": 2.2354,
      "step": 18718
    },
    {
      "epoch": 0.127183373894167,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009648108929999344,
      "loss": 2.1465,
      "step": 18719
    },
    {
      "epoch": 0.12719016824076107,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009648069281922445,
      "loss": 2.1284,
      "step": 18720
    },
    {
      "epoch": 0.12719696258735513,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009648029631693545,
      "loss": 2.2984,
      "step": 18721
    },
    {
      "epoch": 0.1272037569339492,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009647989979312658,
      "loss": 2.1562,
      "step": 18722
    },
    {
      "epoch": 0.12721055128054323,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009647950324779807,
      "loss": 2.1925,
      "step": 18723
    },
    {
      "epoch": 0.1272173456271373,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009647910668095005,
      "loss": 2.1564,
      "step": 18724
    },
    {
      "epoch": 0.12722413997373136,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009647871009258275,
      "loss": 2.3116,
      "step": 18725
    },
    {
      "epoch": 0.12723093432032542,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009647831348269636,
      "loss": 2.0025,
      "step": 18726
    },
    {
      "epoch": 0.1272377286669195,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009647791685129104,
      "loss": 2.2207,
      "step": 18727
    },
    {
      "epoch": 0.12724452301351352,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009647752019836696,
      "loss": 2.459,
      "step": 18728
    },
    {
      "epoch": 0.12725131736010759,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009647712352392432,
      "loss": 2.2394,
      "step": 18729
    },
    {
      "epoch": 0.12725811170670165,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009647672682796333,
      "loss": 2.2373,
      "step": 18730
    },
    {
      "epoch": 0.1272649060532957,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009647633011048414,
      "loss": 2.1901,
      "step": 18731
    },
    {
      "epoch": 0.12727170039988978,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009647593337148693,
      "loss": 2.1623,
      "step": 18732
    },
    {
      "epoch": 0.1272784947464838,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009647553661097189,
      "loss": 2.1821,
      "step": 18733
    },
    {
      "epoch": 0.12728528909307787,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009647513982893922,
      "loss": 2.195,
      "step": 18734
    },
    {
      "epoch": 0.12729208343967194,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000964747430253891,
      "loss": 2.1523,
      "step": 18735
    },
    {
      "epoch": 0.127298877786266,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009647434620032169,
      "loss": 2.4541,
      "step": 18736
    },
    {
      "epoch": 0.12730567213286004,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000964739493537372,
      "loss": 2.229,
      "step": 18737
    },
    {
      "epoch": 0.1273124664794541,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009647355248563581,
      "loss": 2.2913,
      "step": 18738
    },
    {
      "epoch": 0.12731926082604816,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009647315559601768,
      "loss": 2.1562,
      "step": 18739
    },
    {
      "epoch": 0.12732605517264223,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009647275868488303,
      "loss": 2.1285,
      "step": 18740
    },
    {
      "epoch": 0.1273328495192363,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009647236175223202,
      "loss": 2.2191,
      "step": 18741
    },
    {
      "epoch": 0.12733964386583033,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009647196479806484,
      "loss": 2.3999,
      "step": 18742
    },
    {
      "epoch": 0.1273464382124244,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009647156782238166,
      "loss": 2.1787,
      "step": 18743
    },
    {
      "epoch": 0.12735323255901845,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000964711708251827,
      "loss": 2.114,
      "step": 18744
    },
    {
      "epoch": 0.12736002690561252,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009647077380646811,
      "loss": 2.0022,
      "step": 18745
    },
    {
      "epoch": 0.12736682125220658,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009647037676623807,
      "loss": 2.2055,
      "step": 18746
    },
    {
      "epoch": 0.12737361559880062,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009646997970449281,
      "loss": 2.3367,
      "step": 18747
    },
    {
      "epoch": 0.12738040994539468,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009646958262123246,
      "loss": 2.3146,
      "step": 18748
    },
    {
      "epoch": 0.12738720429198874,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009646918551645722,
      "loss": 2.3032,
      "step": 18749
    },
    {
      "epoch": 0.1273939986385828,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000964687883901673,
      "loss": 2.1683,
      "step": 18750
    },
    {
      "epoch": 0.12740079298517687,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009646839124236284,
      "loss": 2.135,
      "step": 18751
    },
    {
      "epoch": 0.1274075873317709,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009646799407304408,
      "loss": 2.2023,
      "step": 18752
    },
    {
      "epoch": 0.12741438167836497,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009646759688221114,
      "loss": 2.0538,
      "step": 18753
    },
    {
      "epoch": 0.12742117602495903,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009646719966986426,
      "loss": 2.171,
      "step": 18754
    },
    {
      "epoch": 0.1274279703715531,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009646680243600359,
      "loss": 2.3197,
      "step": 18755
    },
    {
      "epoch": 0.12743476471814716,
      "grad_norm": 0.123046875,
      "learning_rate": 0.000964664051806293,
      "loss": 2.069,
      "step": 18756
    },
    {
      "epoch": 0.1274415590647412,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009646600790374163,
      "loss": 2.0743,
      "step": 18757
    },
    {
      "epoch": 0.12744835341133526,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009646561060534071,
      "loss": 2.2063,
      "step": 18758
    },
    {
      "epoch": 0.12745514775792932,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0009646521328542676,
      "loss": 2.0993,
      "step": 18759
    },
    {
      "epoch": 0.12746194210452338,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009646481594399994,
      "loss": 2.1968,
      "step": 18760
    },
    {
      "epoch": 0.12746873645111745,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009646441858106045,
      "loss": 2.032,
      "step": 18761
    },
    {
      "epoch": 0.12747553079771148,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009646402119660847,
      "loss": 2.2074,
      "step": 18762
    },
    {
      "epoch": 0.12748232514430555,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009646362379064417,
      "loss": 2.4596,
      "step": 18763
    },
    {
      "epoch": 0.1274891194908996,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009646322636316774,
      "loss": 2.2252,
      "step": 18764
    },
    {
      "epoch": 0.12749591383749367,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009646282891417938,
      "loss": 2.093,
      "step": 18765
    },
    {
      "epoch": 0.12750270818408774,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009646243144367925,
      "loss": 2.189,
      "step": 18766
    },
    {
      "epoch": 0.12750950253068177,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009646203395166758,
      "loss": 2.2457,
      "step": 18767
    },
    {
      "epoch": 0.12751629687727584,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009646163643814449,
      "loss": 2.031,
      "step": 18768
    },
    {
      "epoch": 0.1275230912238699,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009646123890311019,
      "loss": 2.261,
      "step": 18769
    },
    {
      "epoch": 0.12752988557046396,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009646084134656489,
      "loss": 2.2686,
      "step": 18770
    },
    {
      "epoch": 0.12753667991705803,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009646044376850875,
      "loss": 2.1372,
      "step": 18771
    },
    {
      "epoch": 0.12754347426365206,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009646004616894195,
      "loss": 2.1913,
      "step": 18772
    },
    {
      "epoch": 0.12755026861024613,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009645964854786467,
      "loss": 2.2584,
      "step": 18773
    },
    {
      "epoch": 0.1275570629568402,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009645925090527712,
      "loss": 2.16,
      "step": 18774
    },
    {
      "epoch": 0.12756385730343425,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009645885324117947,
      "loss": 2.2081,
      "step": 18775
    },
    {
      "epoch": 0.12757065165002832,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009645845555557192,
      "loss": 2.1952,
      "step": 18776
    },
    {
      "epoch": 0.12757744599662235,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000964580578484546,
      "loss": 2.1806,
      "step": 18777
    },
    {
      "epoch": 0.12758424034321642,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009645766011982776,
      "loss": 2.0986,
      "step": 18778
    },
    {
      "epoch": 0.12759103468981048,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009645726236969154,
      "loss": 2.1764,
      "step": 18779
    },
    {
      "epoch": 0.12759782903640454,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009645686459804615,
      "loss": 2.2514,
      "step": 18780
    },
    {
      "epoch": 0.12760462338299858,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009645646680489176,
      "loss": 2.332,
      "step": 18781
    },
    {
      "epoch": 0.12761141772959264,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009645606899022856,
      "loss": 2.1253,
      "step": 18782
    },
    {
      "epoch": 0.1276182120761867,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009645567115405673,
      "loss": 2.2752,
      "step": 18783
    },
    {
      "epoch": 0.12762500642278077,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009645527329637647,
      "loss": 2.276,
      "step": 18784
    },
    {
      "epoch": 0.12763180076937483,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009645487541718793,
      "loss": 2.1838,
      "step": 18785
    },
    {
      "epoch": 0.12763859511596887,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009645447751649133,
      "loss": 2.1877,
      "step": 18786
    },
    {
      "epoch": 0.12764538946256293,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009645407959428683,
      "loss": 2.1958,
      "step": 18787
    },
    {
      "epoch": 0.127652183809157,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009645368165057465,
      "loss": 2.214,
      "step": 18788
    },
    {
      "epoch": 0.12765897815575106,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009645328368535491,
      "loss": 2.235,
      "step": 18789
    },
    {
      "epoch": 0.12766577250234512,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009645288569862785,
      "loss": 2.3607,
      "step": 18790
    },
    {
      "epoch": 0.12767256684893916,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009645248769039363,
      "loss": 2.1991,
      "step": 18791
    },
    {
      "epoch": 0.12767936119553322,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009645208966065246,
      "loss": 2.265,
      "step": 18792
    },
    {
      "epoch": 0.12768615554212728,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009645169160940448,
      "loss": 2.1224,
      "step": 18793
    },
    {
      "epoch": 0.12769294988872135,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009645129353664992,
      "loss": 2.0951,
      "step": 18794
    },
    {
      "epoch": 0.1276997442353154,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009645089544238893,
      "loss": 2.352,
      "step": 18795
    },
    {
      "epoch": 0.12770653858190945,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009645049732662172,
      "loss": 2.2667,
      "step": 18796
    },
    {
      "epoch": 0.1277133329285035,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009645009918934845,
      "loss": 2.1394,
      "step": 18797
    },
    {
      "epoch": 0.12772012727509757,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009644970103056932,
      "loss": 2.1242,
      "step": 18798
    },
    {
      "epoch": 0.12772692162169164,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009644930285028451,
      "loss": 2.2727,
      "step": 18799
    },
    {
      "epoch": 0.1277337159682857,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009644890464849421,
      "loss": 2.1778,
      "step": 18800
    },
    {
      "epoch": 0.12774051031487973,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009644850642519861,
      "loss": 2.4043,
      "step": 18801
    },
    {
      "epoch": 0.1277473046614738,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009644810818039786,
      "loss": 2.2916,
      "step": 18802
    },
    {
      "epoch": 0.12775409900806786,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009644770991409218,
      "loss": 2.2165,
      "step": 18803
    },
    {
      "epoch": 0.12776089335466193,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009644731162628174,
      "loss": 2.2509,
      "step": 18804
    },
    {
      "epoch": 0.127767687701256,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009644691331696674,
      "loss": 2.2181,
      "step": 18805
    },
    {
      "epoch": 0.12777448204785002,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009644651498614733,
      "loss": 2.1177,
      "step": 18806
    },
    {
      "epoch": 0.1277812763944441,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009644611663382374,
      "loss": 2.2833,
      "step": 18807
    },
    {
      "epoch": 0.12778807074103815,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000964457182599961,
      "loss": 2.2325,
      "step": 18808
    },
    {
      "epoch": 0.12779486508763221,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009644531986466465,
      "loss": 2.2618,
      "step": 18809
    },
    {
      "epoch": 0.12780165943422628,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009644492144782955,
      "loss": 2.2464,
      "step": 18810
    },
    {
      "epoch": 0.1278084537808203,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009644452300949097,
      "loss": 2.1686,
      "step": 18811
    },
    {
      "epoch": 0.12781524812741438,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009644412454964911,
      "loss": 2.2085,
      "step": 18812
    },
    {
      "epoch": 0.12782204247400844,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009644372606830416,
      "loss": 2.1947,
      "step": 18813
    },
    {
      "epoch": 0.1278288368206025,
      "grad_norm": 0.140625,
      "learning_rate": 0.000964433275654563,
      "loss": 2.1186,
      "step": 18814
    },
    {
      "epoch": 0.12783563116719657,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000964429290411057,
      "loss": 2.2579,
      "step": 18815
    },
    {
      "epoch": 0.1278424255137906,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009644253049525257,
      "loss": 2.2947,
      "step": 18816
    },
    {
      "epoch": 0.12784921986038467,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009644213192789707,
      "loss": 2.2797,
      "step": 18817
    },
    {
      "epoch": 0.12785601420697873,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009644173333903939,
      "loss": 2.1226,
      "step": 18818
    },
    {
      "epoch": 0.1278628085535728,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009644133472867973,
      "loss": 2.1547,
      "step": 18819
    },
    {
      "epoch": 0.12786960290016686,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009644093609681828,
      "loss": 2.2077,
      "step": 18820
    },
    {
      "epoch": 0.1278763972467609,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009644053744345518,
      "loss": 2.1085,
      "step": 18821
    },
    {
      "epoch": 0.12788319159335496,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009644013876859065,
      "loss": 2.1359,
      "step": 18822
    },
    {
      "epoch": 0.12788998593994902,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009643974007222488,
      "loss": 2.0954,
      "step": 18823
    },
    {
      "epoch": 0.12789678028654308,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009643934135435804,
      "loss": 2.1334,
      "step": 18824
    },
    {
      "epoch": 0.12790357463313715,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009643894261499031,
      "loss": 2.3473,
      "step": 18825
    },
    {
      "epoch": 0.12791036897973118,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009643854385412189,
      "loss": 2.2812,
      "step": 18826
    },
    {
      "epoch": 0.12791716332632524,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009643814507175294,
      "loss": 2.147,
      "step": 18827
    },
    {
      "epoch": 0.1279239576729193,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009643774626788369,
      "loss": 2.1411,
      "step": 18828
    },
    {
      "epoch": 0.12793075201951337,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009643734744251426,
      "loss": 2.2041,
      "step": 18829
    },
    {
      "epoch": 0.1279375463661074,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000964369485956449,
      "loss": 2.2263,
      "step": 18830
    },
    {
      "epoch": 0.12794434071270147,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009643654972727576,
      "loss": 2.0868,
      "step": 18831
    },
    {
      "epoch": 0.12795113505929553,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009643615083740702,
      "loss": 2.1545,
      "step": 18832
    },
    {
      "epoch": 0.1279579294058896,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009643575192603888,
      "loss": 2.2371,
      "step": 18833
    },
    {
      "epoch": 0.12796472375248366,
      "grad_norm": 0.19921875,
      "learning_rate": 0.0009643535299317152,
      "loss": 2.3246,
      "step": 18834
    },
    {
      "epoch": 0.1279715180990777,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009643495403880512,
      "loss": 2.1597,
      "step": 18835
    },
    {
      "epoch": 0.12797831244567176,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009643455506293988,
      "loss": 2.0853,
      "step": 18836
    },
    {
      "epoch": 0.12798510679226582,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009643415606557596,
      "loss": 2.2702,
      "step": 18837
    },
    {
      "epoch": 0.1279919011388599,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009643375704671357,
      "loss": 2.137,
      "step": 18838
    },
    {
      "epoch": 0.12799869548545395,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009643335800635286,
      "loss": 2.1379,
      "step": 18839
    },
    {
      "epoch": 0.128005489832048,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009643295894449407,
      "loss": 2.1,
      "step": 18840
    },
    {
      "epoch": 0.12801228417864205,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009643255986113733,
      "loss": 2.1931,
      "step": 18841
    },
    {
      "epoch": 0.1280190785252361,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009643216075628285,
      "loss": 2.2467,
      "step": 18842
    },
    {
      "epoch": 0.12802587287183018,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009643176162993082,
      "loss": 2.0599,
      "step": 18843
    },
    {
      "epoch": 0.12803266721842424,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009643136248208141,
      "loss": 2.1377,
      "step": 18844
    },
    {
      "epoch": 0.12803946156501828,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009643096331273481,
      "loss": 2.2761,
      "step": 18845
    },
    {
      "epoch": 0.12804625591161234,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009643056412189121,
      "loss": 2.161,
      "step": 18846
    },
    {
      "epoch": 0.1280530502582064,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009643016490955079,
      "loss": 2.2125,
      "step": 18847
    },
    {
      "epoch": 0.12805984460480047,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009642976567571373,
      "loss": 2.1705,
      "step": 18848
    },
    {
      "epoch": 0.12806663895139453,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009642936642038023,
      "loss": 2.1573,
      "step": 18849
    },
    {
      "epoch": 0.12807343329798856,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009642896714355048,
      "loss": 2.2509,
      "step": 18850
    },
    {
      "epoch": 0.12808022764458263,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009642856784522463,
      "loss": 2.2304,
      "step": 18851
    },
    {
      "epoch": 0.1280870219911767,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009642816852540289,
      "loss": 2.2058,
      "step": 18852
    },
    {
      "epoch": 0.12809381633777076,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009642776918408543,
      "loss": 2.2372,
      "step": 18853
    },
    {
      "epoch": 0.12810061068436482,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009642736982127245,
      "loss": 2.312,
      "step": 18854
    },
    {
      "epoch": 0.12810740503095885,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009642697043696413,
      "loss": 2.0082,
      "step": 18855
    },
    {
      "epoch": 0.12811419937755292,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009642657103116066,
      "loss": 2.1072,
      "step": 18856
    },
    {
      "epoch": 0.12812099372414698,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009642617160386223,
      "loss": 2.3308,
      "step": 18857
    },
    {
      "epoch": 0.12812778807074104,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009642577215506899,
      "loss": 2.0827,
      "step": 18858
    },
    {
      "epoch": 0.1281345824173351,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009642537268478116,
      "loss": 2.1037,
      "step": 18859
    },
    {
      "epoch": 0.12814137676392914,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009642497319299892,
      "loss": 2.1206,
      "step": 18860
    },
    {
      "epoch": 0.1281481711105232,
      "grad_norm": 0.125,
      "learning_rate": 0.0009642457367972245,
      "loss": 2.1147,
      "step": 18861
    },
    {
      "epoch": 0.12815496545711727,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009642417414495193,
      "loss": 2.0809,
      "step": 18862
    },
    {
      "epoch": 0.12816175980371133,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009642377458868757,
      "loss": 2.2037,
      "step": 18863
    },
    {
      "epoch": 0.1281685541503054,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000964233750109295,
      "loss": 2.0758,
      "step": 18864
    },
    {
      "epoch": 0.12817534849689943,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009642297541167795,
      "loss": 2.1152,
      "step": 18865
    },
    {
      "epoch": 0.1281821428434935,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009642257579093311,
      "loss": 2.1549,
      "step": 18866
    },
    {
      "epoch": 0.12818893719008756,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009642217614869514,
      "loss": 2.1411,
      "step": 18867
    },
    {
      "epoch": 0.12819573153668162,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009642177648496425,
      "loss": 2.252,
      "step": 18868
    },
    {
      "epoch": 0.1282025258832757,
      "grad_norm": 0.1201171875,
      "learning_rate": 0.0009642137679974058,
      "loss": 2.01,
      "step": 18869
    },
    {
      "epoch": 0.12820932022986972,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009642097709302435,
      "loss": 2.1166,
      "step": 18870
    },
    {
      "epoch": 0.12821611457646379,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009642057736481576,
      "loss": 1.9605,
      "step": 18871
    },
    {
      "epoch": 0.12822290892305785,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009642017761511496,
      "loss": 2.2927,
      "step": 18872
    },
    {
      "epoch": 0.1282297032696519,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009641977784392215,
      "loss": 2.1458,
      "step": 18873
    },
    {
      "epoch": 0.12823649761624598,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009641937805123752,
      "loss": 2.0697,
      "step": 18874
    },
    {
      "epoch": 0.12824329196284,
      "grad_norm": 0.125,
      "learning_rate": 0.0009641897823706125,
      "loss": 2.0806,
      "step": 18875
    },
    {
      "epoch": 0.12825008630943407,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009641857840139352,
      "loss": 2.0919,
      "step": 18876
    },
    {
      "epoch": 0.12825688065602814,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009641817854423453,
      "loss": 2.3342,
      "step": 18877
    },
    {
      "epoch": 0.1282636750026222,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009641777866558445,
      "loss": 2.1053,
      "step": 18878
    },
    {
      "epoch": 0.12827046934921624,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009641737876544348,
      "loss": 2.1195,
      "step": 18879
    },
    {
      "epoch": 0.1282772636958103,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009641697884381177,
      "loss": 2.1741,
      "step": 18880
    },
    {
      "epoch": 0.12828405804240436,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009641657890068955,
      "loss": 2.1706,
      "step": 18881
    },
    {
      "epoch": 0.12829085238899843,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009641617893607699,
      "loss": 2.0413,
      "step": 18882
    },
    {
      "epoch": 0.1282976467355925,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009641577894997426,
      "loss": 2.2183,
      "step": 18883
    },
    {
      "epoch": 0.12830444108218653,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009641537894238157,
      "loss": 2.2524,
      "step": 18884
    },
    {
      "epoch": 0.1283112354287806,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009641497891329908,
      "loss": 2.1099,
      "step": 18885
    },
    {
      "epoch": 0.12831802977537465,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009641457886272698,
      "loss": 2.2279,
      "step": 18886
    },
    {
      "epoch": 0.12832482412196872,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009641417879066547,
      "loss": 2.2607,
      "step": 18887
    },
    {
      "epoch": 0.12833161846856278,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009641377869711474,
      "loss": 2.2192,
      "step": 18888
    },
    {
      "epoch": 0.12833841281515682,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009641337858207496,
      "loss": 2.2746,
      "step": 18889
    },
    {
      "epoch": 0.12834520716175088,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000964129784455463,
      "loss": 2.118,
      "step": 18890
    },
    {
      "epoch": 0.12835200150834494,
      "grad_norm": 0.125,
      "learning_rate": 0.0009641257828752897,
      "loss": 2.1337,
      "step": 18891
    },
    {
      "epoch": 0.128358795854939,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009641217810802316,
      "loss": 2.1241,
      "step": 18892
    },
    {
      "epoch": 0.12836559020153307,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009641177790702903,
      "loss": 2.1834,
      "step": 18893
    },
    {
      "epoch": 0.1283723845481271,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009641137768454679,
      "loss": 2.2161,
      "step": 18894
    },
    {
      "epoch": 0.12837917889472117,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009641097744057661,
      "loss": 2.3038,
      "step": 18895
    },
    {
      "epoch": 0.12838597324131523,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009641057717511868,
      "loss": 1.9882,
      "step": 18896
    },
    {
      "epoch": 0.1283927675879093,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009641017688817317,
      "loss": 2.2999,
      "step": 18897
    },
    {
      "epoch": 0.12839956193450336,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009640977657974029,
      "loss": 2.0781,
      "step": 18898
    },
    {
      "epoch": 0.1284063562810974,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009640937624982022,
      "loss": 2.0905,
      "step": 18899
    },
    {
      "epoch": 0.12841315062769146,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009640897589841314,
      "loss": 2.1446,
      "step": 18900
    },
    {
      "epoch": 0.12841994497428552,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009640857552551924,
      "loss": 2.1856,
      "step": 18901
    },
    {
      "epoch": 0.12842673932087958,
      "grad_norm": 0.125,
      "learning_rate": 0.0009640817513113869,
      "loss": 2.0896,
      "step": 18902
    },
    {
      "epoch": 0.12843353366747365,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009640777471527169,
      "loss": 2.2515,
      "step": 18903
    },
    {
      "epoch": 0.12844032801406768,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009640737427791844,
      "loss": 2.1562,
      "step": 18904
    },
    {
      "epoch": 0.12844712236066175,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009640697381907908,
      "loss": 2.3148,
      "step": 18905
    },
    {
      "epoch": 0.1284539167072558,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009640657333875385,
      "loss": 2.3088,
      "step": 18906
    },
    {
      "epoch": 0.12846071105384987,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009640617283694289,
      "loss": 2.0921,
      "step": 18907
    },
    {
      "epoch": 0.12846750540044394,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009640577231364641,
      "loss": 2.242,
      "step": 18908
    },
    {
      "epoch": 0.12847429974703797,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009640537176886459,
      "loss": 2.2451,
      "step": 18909
    },
    {
      "epoch": 0.12848109409363204,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009640497120259762,
      "loss": 2.1064,
      "step": 18910
    },
    {
      "epoch": 0.1284878884402261,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009640457061484566,
      "loss": 2.1965,
      "step": 18911
    },
    {
      "epoch": 0.12849468278682016,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009640417000560894,
      "loss": 2.1204,
      "step": 18912
    },
    {
      "epoch": 0.12850147713341423,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009640376937488761,
      "loss": 2.2883,
      "step": 18913
    },
    {
      "epoch": 0.12850827148000826,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009640336872268187,
      "loss": 2.1496,
      "step": 18914
    },
    {
      "epoch": 0.12851506582660233,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009640296804899189,
      "loss": 2.1395,
      "step": 18915
    },
    {
      "epoch": 0.1285218601731964,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009640256735381789,
      "loss": 2.2243,
      "step": 18916
    },
    {
      "epoch": 0.12852865451979045,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009640216663716002,
      "loss": 2.0888,
      "step": 18917
    },
    {
      "epoch": 0.12853544886638452,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009640176589901848,
      "loss": 2.1593,
      "step": 18918
    },
    {
      "epoch": 0.12854224321297855,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009640136513939345,
      "loss": 2.2759,
      "step": 18919
    },
    {
      "epoch": 0.12854903755957262,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009640096435828512,
      "loss": 2.1079,
      "step": 18920
    },
    {
      "epoch": 0.12855583190616668,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009640056355569369,
      "loss": 2.1607,
      "step": 18921
    },
    {
      "epoch": 0.12856262625276074,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009640016273161932,
      "loss": 2.0896,
      "step": 18922
    },
    {
      "epoch": 0.1285694205993548,
      "grad_norm": 0.125,
      "learning_rate": 0.0009639976188606221,
      "loss": 2.1779,
      "step": 18923
    },
    {
      "epoch": 0.12857621494594884,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009639936101902254,
      "loss": 2.1189,
      "step": 18924
    },
    {
      "epoch": 0.1285830092925429,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009639896013050049,
      "loss": 2.1761,
      "step": 18925
    },
    {
      "epoch": 0.12858980363913697,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009639855922049626,
      "loss": 2.1742,
      "step": 18926
    },
    {
      "epoch": 0.12859659798573103,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009639815828901003,
      "loss": 2.3082,
      "step": 18927
    },
    {
      "epoch": 0.12860339233232507,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009639775733604199,
      "loss": 2.2823,
      "step": 18928
    },
    {
      "epoch": 0.12861018667891913,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000963973563615923,
      "loss": 2.2018,
      "step": 18929
    },
    {
      "epoch": 0.1286169810255132,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009639695536566119,
      "loss": 2.1409,
      "step": 18930
    },
    {
      "epoch": 0.12862377537210726,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009639655434824881,
      "loss": 2.1907,
      "step": 18931
    },
    {
      "epoch": 0.12863056971870132,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009639615330935536,
      "loss": 2.2348,
      "step": 18932
    },
    {
      "epoch": 0.12863736406529536,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009639575224898103,
      "loss": 2.2036,
      "step": 18933
    },
    {
      "epoch": 0.12864415841188942,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009639535116712599,
      "loss": 1.9131,
      "step": 18934
    },
    {
      "epoch": 0.12865095275848348,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009639495006379044,
      "loss": 2.0589,
      "step": 18935
    },
    {
      "epoch": 0.12865774710507755,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009639454893897454,
      "loss": 2.2,
      "step": 18936
    },
    {
      "epoch": 0.1286645414516716,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009639414779267852,
      "loss": 2.3067,
      "step": 18937
    },
    {
      "epoch": 0.12867133579826565,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009639374662490254,
      "loss": 2.1749,
      "step": 18938
    },
    {
      "epoch": 0.1286781301448597,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009639334543564677,
      "loss": 2.131,
      "step": 18939
    },
    {
      "epoch": 0.12868492449145377,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009639294422491143,
      "loss": 2.0919,
      "step": 18940
    },
    {
      "epoch": 0.12869171883804784,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009639254299269668,
      "loss": 2.0361,
      "step": 18941
    },
    {
      "epoch": 0.1286985131846419,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000963921417390027,
      "loss": 2.1971,
      "step": 18942
    },
    {
      "epoch": 0.12870530753123594,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009639174046382973,
      "loss": 2.0713,
      "step": 18943
    },
    {
      "epoch": 0.12871210187783,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009639133916717788,
      "loss": 2.1816,
      "step": 18944
    },
    {
      "epoch": 0.12871889622442406,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009639093784904738,
      "loss": 2.2294,
      "step": 18945
    },
    {
      "epoch": 0.12872569057101813,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009639053650943841,
      "loss": 2.2224,
      "step": 18946
    },
    {
      "epoch": 0.1287324849176122,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009639013514835115,
      "loss": 2.2377,
      "step": 18947
    },
    {
      "epoch": 0.12873927926420622,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009638973376578581,
      "loss": 2.1531,
      "step": 18948
    },
    {
      "epoch": 0.1287460736108003,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009638933236174255,
      "loss": 2.1341,
      "step": 18949
    },
    {
      "epoch": 0.12875286795739435,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009638893093622154,
      "loss": 2.1033,
      "step": 18950
    },
    {
      "epoch": 0.12875966230398841,
      "grad_norm": 0.125,
      "learning_rate": 0.0009638852948922298,
      "loss": 2.1008,
      "step": 18951
    },
    {
      "epoch": 0.12876645665058248,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.000963881280207471,
      "loss": 2.1138,
      "step": 18952
    },
    {
      "epoch": 0.1287732509971765,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009638772653079401,
      "loss": 2.158,
      "step": 18953
    },
    {
      "epoch": 0.12878004534377058,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009638732501936395,
      "loss": 2.2062,
      "step": 18954
    },
    {
      "epoch": 0.12878683969036464,
      "grad_norm": 0.12109375,
      "learning_rate": 0.000963869234864571,
      "loss": 2.0131,
      "step": 18955
    },
    {
      "epoch": 0.1287936340369587,
      "grad_norm": 0.125,
      "learning_rate": 0.0009638652193207364,
      "loss": 2.1401,
      "step": 18956
    },
    {
      "epoch": 0.12880042838355277,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009638612035621375,
      "loss": 2.3051,
      "step": 18957
    },
    {
      "epoch": 0.1288072227301468,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000963857187588776,
      "loss": 2.236,
      "step": 18958
    },
    {
      "epoch": 0.12881401707674087,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000963853171400654,
      "loss": 2.3674,
      "step": 18959
    },
    {
      "epoch": 0.12882081142333493,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009638491549977733,
      "loss": 2.1609,
      "step": 18960
    },
    {
      "epoch": 0.128827605769929,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009638451383801359,
      "loss": 2.1571,
      "step": 18961
    },
    {
      "epoch": 0.12883440011652306,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009638411215477434,
      "loss": 2.2769,
      "step": 18962
    },
    {
      "epoch": 0.1288411944631171,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009638371045005978,
      "loss": 2.2354,
      "step": 18963
    },
    {
      "epoch": 0.12884798880971116,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009638330872387008,
      "loss": 2.1349,
      "step": 18964
    },
    {
      "epoch": 0.12885478315630522,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009638290697620547,
      "loss": 2.2613,
      "step": 18965
    },
    {
      "epoch": 0.12886157750289928,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009638250520706608,
      "loss": 2.3683,
      "step": 18966
    },
    {
      "epoch": 0.12886837184949335,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009638210341645213,
      "loss": 2.2056,
      "step": 18967
    },
    {
      "epoch": 0.12887516619608738,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009638170160436379,
      "loss": 2.1347,
      "step": 18968
    },
    {
      "epoch": 0.12888196054268145,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009638129977080128,
      "loss": 2.2085,
      "step": 18969
    },
    {
      "epoch": 0.1288887548892755,
      "grad_norm": 0.171875,
      "learning_rate": 0.0009638089791576473,
      "loss": 2.4324,
      "step": 18970
    },
    {
      "epoch": 0.12889554923586957,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009638049603925438,
      "loss": 2.1492,
      "step": 18971
    },
    {
      "epoch": 0.12890234358246364,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009638009414127038,
      "loss": 2.115,
      "step": 18972
    },
    {
      "epoch": 0.12890913792905767,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009637969222181292,
      "loss": 2.193,
      "step": 18973
    },
    {
      "epoch": 0.12891593227565173,
      "grad_norm": 0.14453125,
      "learning_rate": 0.000963792902808822,
      "loss": 2.1941,
      "step": 18974
    },
    {
      "epoch": 0.1289227266222458,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009637888831847841,
      "loss": 2.1545,
      "step": 18975
    },
    {
      "epoch": 0.12892952096883986,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009637848633460172,
      "loss": 2.2042,
      "step": 18976
    },
    {
      "epoch": 0.1289363153154339,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009637808432925233,
      "loss": 2.1804,
      "step": 18977
    },
    {
      "epoch": 0.12894310966202796,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000963776823024304,
      "loss": 2.2736,
      "step": 18978
    },
    {
      "epoch": 0.12894990400862202,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009637728025413615,
      "loss": 2.2783,
      "step": 18979
    },
    {
      "epoch": 0.1289566983552161,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009637687818436975,
      "loss": 2.2518,
      "step": 18980
    },
    {
      "epoch": 0.12896349270181015,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009637647609313137,
      "loss": 2.2724,
      "step": 18981
    },
    {
      "epoch": 0.1289702870484042,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009637607398042123,
      "loss": 2.0812,
      "step": 18982
    },
    {
      "epoch": 0.12897708139499825,
      "grad_norm": 0.15234375,
      "learning_rate": 0.000963756718462395,
      "loss": 2.3171,
      "step": 18983
    },
    {
      "epoch": 0.1289838757415923,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009637526969058636,
      "loss": 2.1289,
      "step": 18984
    },
    {
      "epoch": 0.12899067008818638,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.00096374867513462,
      "loss": 2.2146,
      "step": 18985
    },
    {
      "epoch": 0.12899746443478044,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009637446531486662,
      "loss": 2.232,
      "step": 18986
    },
    {
      "epoch": 0.12900425878137448,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009637406309480038,
      "loss": 2.3619,
      "step": 18987
    },
    {
      "epoch": 0.12901105312796854,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009637366085326348,
      "loss": 2.1156,
      "step": 18988
    },
    {
      "epoch": 0.1290178474745626,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009637325859025612,
      "loss": 2.1824,
      "step": 18989
    },
    {
      "epoch": 0.12902464182115667,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009637285630577847,
      "loss": 2.1924,
      "step": 18990
    },
    {
      "epoch": 0.12903143616775073,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000963724539998307,
      "loss": 2.1166,
      "step": 18991
    },
    {
      "epoch": 0.12903823051434476,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009637205167241304,
      "loss": 2.2464,
      "step": 18992
    },
    {
      "epoch": 0.12904502486093883,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009637164932352564,
      "loss": 2.2067,
      "step": 18993
    },
    {
      "epoch": 0.1290518192075329,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.000963712469531687,
      "loss": 2.2088,
      "step": 18994
    },
    {
      "epoch": 0.12905861355412696,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009637084456134241,
      "loss": 2.1433,
      "step": 18995
    },
    {
      "epoch": 0.12906540790072102,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009637044214804694,
      "loss": 2.201,
      "step": 18996
    },
    {
      "epoch": 0.12907220224731505,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009637003971328248,
      "loss": 2.3583,
      "step": 18997
    },
    {
      "epoch": 0.12907899659390912,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009636963725704925,
      "loss": 2.1571,
      "step": 18998
    },
    {
      "epoch": 0.12908579094050318,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000963692347793474,
      "loss": 2.2389,
      "step": 18999
    },
    {
      "epoch": 0.12909258528709724,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009636883228017712,
      "loss": 2.0557,
      "step": 19000
    },
    {
      "epoch": 0.1290993796336913,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009636842975953859,
      "loss": 2.2083,
      "step": 19001
    },
    {
      "epoch": 0.12910617398028534,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009636802721743203,
      "loss": 2.0957,
      "step": 19002
    },
    {
      "epoch": 0.1291129683268794,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000963676246538576,
      "loss": 2.2742,
      "step": 19003
    },
    {
      "epoch": 0.12911976267347347,
      "grad_norm": 0.1748046875,
      "learning_rate": 0.0009636722206881548,
      "loss": 2.4042,
      "step": 19004
    },
    {
      "epoch": 0.12912655702006753,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009636681946230588,
      "loss": 2.2585,
      "step": 19005
    },
    {
      "epoch": 0.1291333513666616,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009636641683432896,
      "loss": 2.1856,
      "step": 19006
    },
    {
      "epoch": 0.12914014571325563,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009636601418488493,
      "loss": 2.1768,
      "step": 19007
    },
    {
      "epoch": 0.1291469400598497,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009636561151397397,
      "loss": 2.1852,
      "step": 19008
    },
    {
      "epoch": 0.12915373440644376,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009636520882159627,
      "loss": 2.3507,
      "step": 19009
    },
    {
      "epoch": 0.12916052875303782,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00096364806107752,
      "loss": 2.1209,
      "step": 19010
    },
    {
      "epoch": 0.1291673230996319,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009636440337244135,
      "loss": 2.3412,
      "step": 19011
    },
    {
      "epoch": 0.12917411744622592,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009636400061566453,
      "loss": 2.1,
      "step": 19012
    },
    {
      "epoch": 0.12918091179281999,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009636359783742169,
      "loss": 2.2075,
      "step": 19013
    },
    {
      "epoch": 0.12918770613941405,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009636319503771305,
      "loss": 2.1036,
      "step": 19014
    },
    {
      "epoch": 0.1291945004860081,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009636279221653878,
      "loss": 2.1448,
      "step": 19015
    },
    {
      "epoch": 0.12920129483260218,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009636238937389905,
      "loss": 2.1555,
      "step": 19016
    },
    {
      "epoch": 0.1292080891791962,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009636198650979409,
      "loss": 1.9943,
      "step": 19017
    },
    {
      "epoch": 0.12921488352579028,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009636158362422405,
      "loss": 2.2907,
      "step": 19018
    },
    {
      "epoch": 0.12922167787238434,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009636118071718914,
      "loss": 2.226,
      "step": 19019
    },
    {
      "epoch": 0.1292284722189784,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009636077778868952,
      "loss": 2.0522,
      "step": 19020
    },
    {
      "epoch": 0.12923526656557244,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009636037483872539,
      "loss": 2.0622,
      "step": 19021
    },
    {
      "epoch": 0.1292420609121665,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009635997186729694,
      "loss": 2.0881,
      "step": 19022
    },
    {
      "epoch": 0.12924885525876056,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009635956887440437,
      "loss": 2.3223,
      "step": 19023
    },
    {
      "epoch": 0.12925564960535463,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009635916586004784,
      "loss": 2.1572,
      "step": 19024
    },
    {
      "epoch": 0.1292624439519487,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009635876282422755,
      "loss": 2.1489,
      "step": 19025
    },
    {
      "epoch": 0.12926923829854273,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009635835976694368,
      "loss": 2.1376,
      "step": 19026
    },
    {
      "epoch": 0.1292760326451368,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009635795668819642,
      "loss": 2.1181,
      "step": 19027
    },
    {
      "epoch": 0.12928282699173085,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009635755358798597,
      "loss": 2.1373,
      "step": 19028
    },
    {
      "epoch": 0.12928962133832492,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000963571504663125,
      "loss": 2.279,
      "step": 19029
    },
    {
      "epoch": 0.12929641568491898,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009635674732317619,
      "loss": 2.2678,
      "step": 19030
    },
    {
      "epoch": 0.12930321003151302,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009635634415857724,
      "loss": 2.0384,
      "step": 19031
    },
    {
      "epoch": 0.12931000437810708,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009635594097251583,
      "loss": 2.2586,
      "step": 19032
    },
    {
      "epoch": 0.12931679872470114,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009635553776499217,
      "loss": 2.0963,
      "step": 19033
    },
    {
      "epoch": 0.1293235930712952,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009635513453600642,
      "loss": 2.1584,
      "step": 19034
    },
    {
      "epoch": 0.12933038741788927,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009635473128555876,
      "loss": 2.167,
      "step": 19035
    },
    {
      "epoch": 0.1293371817644833,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.000963543280136494,
      "loss": 2.1535,
      "step": 19036
    },
    {
      "epoch": 0.12934397611107737,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009635392472027852,
      "loss": 2.1657,
      "step": 19037
    },
    {
      "epoch": 0.12935077045767143,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009635352140544631,
      "loss": 2.0727,
      "step": 19038
    },
    {
      "epoch": 0.1293575648042655,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009635311806915294,
      "loss": 2.0876,
      "step": 19039
    },
    {
      "epoch": 0.12936435915085956,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009635271471139862,
      "loss": 2.1671,
      "step": 19040
    },
    {
      "epoch": 0.1293711534974536,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009635231133218351,
      "loss": 2.2316,
      "step": 19041
    },
    {
      "epoch": 0.12937794784404766,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009635190793150781,
      "loss": 2.2275,
      "step": 19042
    },
    {
      "epoch": 0.12938474219064172,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009635150450937171,
      "loss": 2.2369,
      "step": 19043
    },
    {
      "epoch": 0.12939153653723579,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000963511010657754,
      "loss": 2.1468,
      "step": 19044
    },
    {
      "epoch": 0.12939833088382985,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009635069760071907,
      "loss": 2.163,
      "step": 19045
    },
    {
      "epoch": 0.12940512523042388,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009635029411420289,
      "loss": 2.0674,
      "step": 19046
    },
    {
      "epoch": 0.12941191957701795,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009634989060622705,
      "loss": 2.1047,
      "step": 19047
    },
    {
      "epoch": 0.129418713923612,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009634948707679176,
      "loss": 2.2003,
      "step": 19048
    },
    {
      "epoch": 0.12942550827020607,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009634908352589717,
      "loss": 2.2764,
      "step": 19049
    },
    {
      "epoch": 0.12943230261680014,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009634867995354349,
      "loss": 2.2203,
      "step": 19050
    },
    {
      "epoch": 0.12943909696339417,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009634827635973091,
      "loss": 2.0787,
      "step": 19051
    },
    {
      "epoch": 0.12944589130998824,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.000963478727444596,
      "loss": 2.1298,
      "step": 19052
    },
    {
      "epoch": 0.1294526856565823,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009634746910772976,
      "loss": 2.2224,
      "step": 19053
    },
    {
      "epoch": 0.12945948000317636,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009634706544954157,
      "loss": 2.2059,
      "step": 19054
    },
    {
      "epoch": 0.12946627434977043,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009634666176989523,
      "loss": 2.1488,
      "step": 19055
    },
    {
      "epoch": 0.12947306869636446,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.000963462580687909,
      "loss": 2.1312,
      "step": 19056
    },
    {
      "epoch": 0.12947986304295853,
      "grad_norm": 0.123046875,
      "learning_rate": 0.000963458543462288,
      "loss": 2.1473,
      "step": 19057
    },
    {
      "epoch": 0.1294866573895526,
      "grad_norm": 0.138671875,
      "learning_rate": 0.000963454506022091,
      "loss": 2.1596,
      "step": 19058
    },
    {
      "epoch": 0.12949345173614665,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009634504683673198,
      "loss": 2.1808,
      "step": 19059
    },
    {
      "epoch": 0.12950024608274072,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009634464304979763,
      "loss": 2.0392,
      "step": 19060
    },
    {
      "epoch": 0.12950704042933475,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009634423924140625,
      "loss": 2.2007,
      "step": 19061
    },
    {
      "epoch": 0.12951383477592882,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009634383541155802,
      "loss": 2.1265,
      "step": 19062
    },
    {
      "epoch": 0.12952062912252288,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009634343156025312,
      "loss": 2.3592,
      "step": 19063
    },
    {
      "epoch": 0.12952742346911694,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009634302768749175,
      "loss": 2.3168,
      "step": 19064
    },
    {
      "epoch": 0.129534217815711,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009634262379327408,
      "loss": 2.2245,
      "step": 19065
    },
    {
      "epoch": 0.12954101216230504,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009634221987760032,
      "loss": 2.1213,
      "step": 19066
    },
    {
      "epoch": 0.1295478065088991,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009634181594047062,
      "loss": 2.1196,
      "step": 19067
    },
    {
      "epoch": 0.12955460085549317,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009634141198188522,
      "loss": 2.2241,
      "step": 19068
    },
    {
      "epoch": 0.12956139520208723,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009634100800184425,
      "loss": 2.237,
      "step": 19069
    },
    {
      "epoch": 0.12956818954868127,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009634060400034795,
      "loss": 2.0675,
      "step": 19070
    },
    {
      "epoch": 0.12957498389527533,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009634019997739647,
      "loss": 2.3691,
      "step": 19071
    },
    {
      "epoch": 0.1295817782418694,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009633979593299001,
      "loss": 2.1804,
      "step": 19072
    },
    {
      "epoch": 0.12958857258846346,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009633939186712875,
      "loss": 2.1015,
      "step": 19073
    },
    {
      "epoch": 0.12959536693505752,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009633898777981288,
      "loss": 2.2409,
      "step": 19074
    },
    {
      "epoch": 0.12960216128165156,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000963385836710426,
      "loss": 2.1689,
      "step": 19075
    },
    {
      "epoch": 0.12960895562824562,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009633817954081808,
      "loss": 1.9141,
      "step": 19076
    },
    {
      "epoch": 0.12961574997483968,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009633777538913951,
      "loss": 2.1218,
      "step": 19077
    },
    {
      "epoch": 0.12962254432143375,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009633737121600709,
      "loss": 2.1962,
      "step": 19078
    },
    {
      "epoch": 0.1296293386680278,
      "grad_norm": 0.12890625,
      "learning_rate": 0.00096336967021421,
      "loss": 2.2507,
      "step": 19079
    },
    {
      "epoch": 0.12963613301462185,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009633656280538142,
      "loss": 2.3142,
      "step": 19080
    },
    {
      "epoch": 0.1296429273612159,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009633615856788855,
      "loss": 2.0714,
      "step": 19081
    },
    {
      "epoch": 0.12964972170780997,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009633575430894257,
      "loss": 2.2408,
      "step": 19082
    },
    {
      "epoch": 0.12965651605440404,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009633535002854366,
      "loss": 2.1744,
      "step": 19083
    },
    {
      "epoch": 0.1296633104009981,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009633494572669201,
      "loss": 2.1291,
      "step": 19084
    },
    {
      "epoch": 0.12967010474759214,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009633454140338783,
      "loss": 2.2517,
      "step": 19085
    },
    {
      "epoch": 0.1296768990941862,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009633413705863127,
      "loss": 2.0378,
      "step": 19086
    },
    {
      "epoch": 0.12968369344078026,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009633373269242254,
      "loss": 2.1483,
      "step": 19087
    },
    {
      "epoch": 0.12969048778737433,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009633332830476182,
      "loss": 2.264,
      "step": 19088
    },
    {
      "epoch": 0.1296972821339684,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000963329238956493,
      "loss": 2.029,
      "step": 19089
    },
    {
      "epoch": 0.12970407648056242,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009633251946508518,
      "loss": 2.1437,
      "step": 19090
    },
    {
      "epoch": 0.1297108708271565,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009633211501306962,
      "loss": 2.1088,
      "step": 19091
    },
    {
      "epoch": 0.12971766517375055,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009633171053960284,
      "loss": 2.1346,
      "step": 19092
    },
    {
      "epoch": 0.12972445952034461,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009633130604468498,
      "loss": 2.2268,
      "step": 19093
    },
    {
      "epoch": 0.12973125386693868,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009633090152831628,
      "loss": 2.3315,
      "step": 19094
    },
    {
      "epoch": 0.12973804821353271,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000963304969904969,
      "loss": 2.1589,
      "step": 19095
    },
    {
      "epoch": 0.12974484256012678,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009633009243122703,
      "loss": 2.3273,
      "step": 19096
    },
    {
      "epoch": 0.12975163690672084,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009632968785050686,
      "loss": 2.0712,
      "step": 19097
    },
    {
      "epoch": 0.1297584312533149,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009632928324833656,
      "loss": 2.2777,
      "step": 19098
    },
    {
      "epoch": 0.12976522559990897,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009632887862471636,
      "loss": 2.096,
      "step": 19099
    },
    {
      "epoch": 0.129772019946503,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000963284739796464,
      "loss": 2.2129,
      "step": 19100
    },
    {
      "epoch": 0.12977881429309707,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000963280693131269,
      "loss": 2.1516,
      "step": 19101
    },
    {
      "epoch": 0.12978560863969113,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009632766462515802,
      "loss": 2.2701,
      "step": 19102
    },
    {
      "epoch": 0.1297924029862852,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009632725991573998,
      "loss": 2.2083,
      "step": 19103
    },
    {
      "epoch": 0.12979919733287926,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009632685518487295,
      "loss": 2.2269,
      "step": 19104
    },
    {
      "epoch": 0.1298059916794733,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000963264504325571,
      "loss": 2.1323,
      "step": 19105
    },
    {
      "epoch": 0.12981278602606736,
      "grad_norm": 0.125,
      "learning_rate": 0.0009632604565879264,
      "loss": 2.1349,
      "step": 19106
    },
    {
      "epoch": 0.12981958037266142,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009632564086357975,
      "loss": 2.1974,
      "step": 19107
    },
    {
      "epoch": 0.12982637471925548,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009632523604691863,
      "loss": 2.3171,
      "step": 19108
    },
    {
      "epoch": 0.12983316906584955,
      "grad_norm": 0.1513671875,
      "learning_rate": 0.0009632483120880946,
      "loss": 2.3207,
      "step": 19109
    },
    {
      "epoch": 0.12983996341244358,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009632442634925241,
      "loss": 2.1184,
      "step": 19110
    },
    {
      "epoch": 0.12984675775903765,
      "grad_norm": 0.125,
      "learning_rate": 0.0009632402146824768,
      "loss": 1.9676,
      "step": 19111
    },
    {
      "epoch": 0.1298535521056317,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009632361656579547,
      "loss": 2.0762,
      "step": 19112
    },
    {
      "epoch": 0.12986034645222577,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009632321164189595,
      "loss": 1.9887,
      "step": 19113
    },
    {
      "epoch": 0.12986714079881984,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009632280669654932,
      "loss": 2.139,
      "step": 19114
    },
    {
      "epoch": 0.12987393514541387,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009632240172975576,
      "loss": 2.0911,
      "step": 19115
    },
    {
      "epoch": 0.12988072949200793,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009632199674151545,
      "loss": 2.1611,
      "step": 19116
    },
    {
      "epoch": 0.129887523838602,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009632159173182859,
      "loss": 2.2216,
      "step": 19117
    },
    {
      "epoch": 0.12989431818519606,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009632118670069538,
      "loss": 2.1694,
      "step": 19118
    },
    {
      "epoch": 0.1299011125317901,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009632078164811597,
      "loss": 2.3628,
      "step": 19119
    },
    {
      "epoch": 0.12990790687838416,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009632037657409058,
      "loss": 2.1318,
      "step": 19120
    },
    {
      "epoch": 0.12991470122497822,
      "grad_norm": 0.1611328125,
      "learning_rate": 0.000963199714786194,
      "loss": 2.2699,
      "step": 19121
    },
    {
      "epoch": 0.1299214955715723,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009631956636170258,
      "loss": 2.0498,
      "step": 19122
    },
    {
      "epoch": 0.12992828991816635,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009631916122334035,
      "loss": 2.3318,
      "step": 19123
    },
    {
      "epoch": 0.1299350842647604,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009631875606353286,
      "loss": 2.1046,
      "step": 19124
    },
    {
      "epoch": 0.12994187861135445,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009631835088228034,
      "loss": 2.1795,
      "step": 19125
    },
    {
      "epoch": 0.1299486729579485,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009631794567958294,
      "loss": 2.0589,
      "step": 19126
    },
    {
      "epoch": 0.12995546730454258,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009631754045544086,
      "loss": 2.2039,
      "step": 19127
    },
    {
      "epoch": 0.12996226165113664,
      "grad_norm": 0.150390625,
      "learning_rate": 0.000963171352098543,
      "loss": 2.1232,
      "step": 19128
    },
    {
      "epoch": 0.12996905599773068,
      "grad_norm": 0.1572265625,
      "learning_rate": 0.0009631672994282343,
      "loss": 2.4507,
      "step": 19129
    },
    {
      "epoch": 0.12997585034432474,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009631632465434845,
      "loss": 2.1586,
      "step": 19130
    },
    {
      "epoch": 0.1299826446909188,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009631591934442954,
      "loss": 2.2335,
      "step": 19131
    },
    {
      "epoch": 0.12998943903751287,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009631551401306688,
      "loss": 2.1963,
      "step": 19132
    },
    {
      "epoch": 0.12999623338410693,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009631510866026069,
      "loss": 2.1618,
      "step": 19133
    },
    {
      "epoch": 0.13000302773070097,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009631470328601111,
      "loss": 2.2497,
      "step": 19134
    },
    {
      "epoch": 0.13000982207729503,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009631429789031838,
      "loss": 2.2483,
      "step": 19135
    },
    {
      "epoch": 0.1300166164238891,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009631389247318265,
      "loss": 2.2192,
      "step": 19136
    },
    {
      "epoch": 0.13002341077048316,
      "grad_norm": 0.142578125,
      "learning_rate": 0.000963134870346041,
      "loss": 2.212,
      "step": 19137
    },
    {
      "epoch": 0.13003020511707722,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009631308157458296,
      "loss": 2.1594,
      "step": 19138
    },
    {
      "epoch": 0.13003699946367125,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009631267609311939,
      "loss": 2.1717,
      "step": 19139
    },
    {
      "epoch": 0.13004379381026532,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009631227059021357,
      "loss": 2.1872,
      "step": 19140
    },
    {
      "epoch": 0.13005058815685938,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009631186506586572,
      "loss": 2.1822,
      "step": 19141
    },
    {
      "epoch": 0.13005738250345344,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009631145952007599,
      "loss": 2.0673,
      "step": 19142
    },
    {
      "epoch": 0.1300641768500475,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000963110539528446,
      "loss": 2.079,
      "step": 19143
    },
    {
      "epoch": 0.13007097119664154,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009631064836417172,
      "loss": 2.1605,
      "step": 19144
    },
    {
      "epoch": 0.1300777655432356,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009631024275405753,
      "loss": 2.1394,
      "step": 19145
    },
    {
      "epoch": 0.13008455988982967,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009630983712250223,
      "loss": 2.1453,
      "step": 19146
    },
    {
      "epoch": 0.13009135423642373,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009630943146950601,
      "loss": 2.2408,
      "step": 19147
    },
    {
      "epoch": 0.1300981485830178,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009630902579506906,
      "loss": 2.2343,
      "step": 19148
    },
    {
      "epoch": 0.13010494292961183,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009630862009919154,
      "loss": 2.2297,
      "step": 19149
    },
    {
      "epoch": 0.1301117372762059,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000963082143818737,
      "loss": 2.326,
      "step": 19150
    },
    {
      "epoch": 0.13011853162279996,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009630780864311565,
      "loss": 2.231,
      "step": 19151
    },
    {
      "epoch": 0.13012532596939402,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009630740288291764,
      "loss": 2.0685,
      "step": 19152
    },
    {
      "epoch": 0.1301321203159881,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009630699710127982,
      "loss": 2.2149,
      "step": 19153
    },
    {
      "epoch": 0.13013891466258212,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009630659129820238,
      "loss": 2.0771,
      "step": 19154
    },
    {
      "epoch": 0.13014570900917619,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009630618547368555,
      "loss": 2.372,
      "step": 19155
    },
    {
      "epoch": 0.13015250335577025,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009630577962772947,
      "loss": 2.1342,
      "step": 19156
    },
    {
      "epoch": 0.1301592977023643,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009630537376033434,
      "loss": 2.176,
      "step": 19157
    },
    {
      "epoch": 0.13016609204895838,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009630496787150036,
      "loss": 2.2281,
      "step": 19158
    },
    {
      "epoch": 0.1301728863955524,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000963045619612277,
      "loss": 2.0572,
      "step": 19159
    },
    {
      "epoch": 0.13017968074214648,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009630415602951656,
      "loss": 2.1197,
      "step": 19160
    },
    {
      "epoch": 0.13018647508874054,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009630375007636713,
      "loss": 2.1178,
      "step": 19161
    },
    {
      "epoch": 0.1301932694353346,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009630334410177961,
      "loss": 2.1691,
      "step": 19162
    },
    {
      "epoch": 0.13020006378192867,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009630293810575417,
      "loss": 2.0397,
      "step": 19163
    },
    {
      "epoch": 0.1302068581285227,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009630253208829098,
      "loss": 2.1217,
      "step": 19164
    },
    {
      "epoch": 0.13021365247511676,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009630212604939026,
      "loss": 2.2581,
      "step": 19165
    },
    {
      "epoch": 0.13022044682171083,
      "grad_norm": 0.162109375,
      "learning_rate": 0.0009630171998905219,
      "loss": 2.2378,
      "step": 19166
    },
    {
      "epoch": 0.1302272411683049,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009630131390727694,
      "loss": 2.2083,
      "step": 19167
    },
    {
      "epoch": 0.13023403551489893,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009630090780406473,
      "loss": 2.1203,
      "step": 19168
    },
    {
      "epoch": 0.130240829861493,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009630050167941573,
      "loss": 2.116,
      "step": 19169
    },
    {
      "epoch": 0.13024762420808705,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009630009553333013,
      "loss": 2.1598,
      "step": 19170
    },
    {
      "epoch": 0.13025441855468112,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009629968936580811,
      "loss": 2.1978,
      "step": 19171
    },
    {
      "epoch": 0.13026121290127518,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009629928317684984,
      "loss": 2.1404,
      "step": 19172
    },
    {
      "epoch": 0.13026800724786922,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009629887696645556,
      "loss": 2.1202,
      "step": 19173
    },
    {
      "epoch": 0.13027480159446328,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009629847073462543,
      "loss": 2.1581,
      "step": 19174
    },
    {
      "epoch": 0.13028159594105734,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009629806448135965,
      "loss": 2.1373,
      "step": 19175
    },
    {
      "epoch": 0.1302883902876514,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009629765820665837,
      "loss": 2.0067,
      "step": 19176
    },
    {
      "epoch": 0.13029518463424547,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009629725191052182,
      "loss": 2.1346,
      "step": 19177
    },
    {
      "epoch": 0.1303019789808395,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009629684559295018,
      "loss": 2.1015,
      "step": 19178
    },
    {
      "epoch": 0.13030877332743357,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009629643925394361,
      "loss": 2.281,
      "step": 19179
    },
    {
      "epoch": 0.13031556767402763,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009629603289350233,
      "loss": 2.0761,
      "step": 19180
    },
    {
      "epoch": 0.1303223620206217,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009629562651162654,
      "loss": 2.2944,
      "step": 19181
    },
    {
      "epoch": 0.13032915636721576,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009629522010831637,
      "loss": 2.1094,
      "step": 19182
    },
    {
      "epoch": 0.1303359507138098,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009629481368357205,
      "loss": 2.1916,
      "step": 19183
    },
    {
      "epoch": 0.13034274506040386,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009629440723739378,
      "loss": 2.1769,
      "step": 19184
    },
    {
      "epoch": 0.13034953940699792,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009629400076978172,
      "loss": 2.1497,
      "step": 19185
    },
    {
      "epoch": 0.13035633375359199,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009629359428073608,
      "loss": 2.1754,
      "step": 19186
    },
    {
      "epoch": 0.13036312810018605,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009629318777025702,
      "loss": 1.9925,
      "step": 19187
    },
    {
      "epoch": 0.13036992244678008,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009629278123834475,
      "loss": 2.4171,
      "step": 19188
    },
    {
      "epoch": 0.13037671679337415,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009629237468499945,
      "loss": 2.1184,
      "step": 19189
    },
    {
      "epoch": 0.1303835111399682,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009629196811022132,
      "loss": 2.2974,
      "step": 19190
    },
    {
      "epoch": 0.13039030548656227,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009629156151401053,
      "loss": 2.1876,
      "step": 19191
    },
    {
      "epoch": 0.13039709983315634,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009629115489636728,
      "loss": 2.1864,
      "step": 19192
    },
    {
      "epoch": 0.13040389417975037,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009629074825729176,
      "loss": 2.0499,
      "step": 19193
    },
    {
      "epoch": 0.13041068852634444,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009629034159678415,
      "loss": 2.2408,
      "step": 19194
    },
    {
      "epoch": 0.1304174828729385,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009628993491484465,
      "loss": 2.1403,
      "step": 19195
    },
    {
      "epoch": 0.13042427721953256,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009628952821147343,
      "loss": 2.0323,
      "step": 19196
    },
    {
      "epoch": 0.13043107156612663,
      "grad_norm": 0.1806640625,
      "learning_rate": 0.0009628912148667069,
      "loss": 2.2405,
      "step": 19197
    },
    {
      "epoch": 0.13043786591272066,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009628871474043663,
      "loss": 2.125,
      "step": 19198
    },
    {
      "epoch": 0.13044466025931473,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009628830797277141,
      "loss": 2.1319,
      "step": 19199
    },
    {
      "epoch": 0.1304514546059088,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009628790118367525,
      "loss": 2.1745,
      "step": 19200
    },
    {
      "epoch": 0.13045824895250285,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009628749437314831,
      "loss": 2.0938,
      "step": 19201
    },
    {
      "epoch": 0.13046504329909692,
      "grad_norm": 0.12109375,
      "learning_rate": 0.000962870875411908,
      "loss": 1.9865,
      "step": 19202
    },
    {
      "epoch": 0.13047183764569095,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009628668068780288,
      "loss": 2.2788,
      "step": 19203
    },
    {
      "epoch": 0.13047863199228502,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009628627381298478,
      "loss": 2.2325,
      "step": 19204
    },
    {
      "epoch": 0.13048542633887908,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009628586691673666,
      "loss": 2.3713,
      "step": 19205
    },
    {
      "epoch": 0.13049222068547314,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009628545999905871,
      "loss": 2.2564,
      "step": 19206
    },
    {
      "epoch": 0.1304990150320672,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009628505305995112,
      "loss": 2.1703,
      "step": 19207
    },
    {
      "epoch": 0.13050580937866124,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009628464609941409,
      "loss": 2.2985,
      "step": 19208
    },
    {
      "epoch": 0.1305126037252553,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009628423911744779,
      "loss": 2.1489,
      "step": 19209
    },
    {
      "epoch": 0.13051939807184937,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009628383211405243,
      "loss": 2.2329,
      "step": 19210
    },
    {
      "epoch": 0.13052619241844343,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009628342508922819,
      "loss": 2.2459,
      "step": 19211
    },
    {
      "epoch": 0.1305329867650375,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009628301804297523,
      "loss": 2.253,
      "step": 19212
    },
    {
      "epoch": 0.13053978111163153,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009628261097529378,
      "loss": 2.325,
      "step": 19213
    },
    {
      "epoch": 0.1305465754582256,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009628220388618401,
      "loss": 2.1337,
      "step": 19214
    },
    {
      "epoch": 0.13055336980481966,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000962817967756461,
      "loss": 2.2436,
      "step": 19215
    },
    {
      "epoch": 0.13056016415141372,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009628138964368027,
      "loss": 2.1879,
      "step": 19216
    },
    {
      "epoch": 0.13056695849800776,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009628098249028667,
      "loss": 2.194,
      "step": 19217
    },
    {
      "epoch": 0.13057375284460182,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009628057531546551,
      "loss": 2.178,
      "step": 19218
    },
    {
      "epoch": 0.13058054719119588,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009628016811921698,
      "loss": 2.0477,
      "step": 19219
    },
    {
      "epoch": 0.13058734153778995,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009627976090154126,
      "loss": 2.2509,
      "step": 19220
    },
    {
      "epoch": 0.130594135884384,
      "grad_norm": 0.11865234375,
      "learning_rate": 0.0009627935366243854,
      "loss": 1.938,
      "step": 19221
    },
    {
      "epoch": 0.13060093023097805,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009627894640190901,
      "loss": 2.1425,
      "step": 19222
    },
    {
      "epoch": 0.1306077245775721,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009627853911995286,
      "loss": 2.1626,
      "step": 19223
    },
    {
      "epoch": 0.13061451892416617,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009627813181657027,
      "loss": 2.2439,
      "step": 19224
    },
    {
      "epoch": 0.13062131327076024,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009627772449176145,
      "loss": 2.2139,
      "step": 19225
    },
    {
      "epoch": 0.1306281076173543,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009627731714552655,
      "loss": 2.0738,
      "step": 19226
    },
    {
      "epoch": 0.13063490196394834,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009627690977786579,
      "loss": 2.2336,
      "step": 19227
    },
    {
      "epoch": 0.1306416963105424,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009627650238877936,
      "loss": 2.037,
      "step": 19228
    },
    {
      "epoch": 0.13064849065713646,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009627609497826744,
      "loss": 2.0038,
      "step": 19229
    },
    {
      "epoch": 0.13065528500373053,
      "grad_norm": 0.1552734375,
      "learning_rate": 0.0009627568754633023,
      "loss": 2.1977,
      "step": 19230
    },
    {
      "epoch": 0.1306620793503246,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009627528009296789,
      "loss": 2.1668,
      "step": 19231
    },
    {
      "epoch": 0.13066887369691862,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009627487261818062,
      "loss": 2.1497,
      "step": 19232
    },
    {
      "epoch": 0.1306756680435127,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009627446512196862,
      "loss": 2.0989,
      "step": 19233
    },
    {
      "epoch": 0.13068246239010675,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009627405760433208,
      "loss": 2.0813,
      "step": 19234
    },
    {
      "epoch": 0.13068925673670082,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009627365006527118,
      "loss": 2.2273,
      "step": 19235
    },
    {
      "epoch": 0.13069605108329488,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009627324250478611,
      "loss": 2.0922,
      "step": 19236
    },
    {
      "epoch": 0.13070284542988891,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009627283492287706,
      "loss": 2.108,
      "step": 19237
    },
    {
      "epoch": 0.13070963977648298,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009627242731954422,
      "loss": 2.0673,
      "step": 19238
    },
    {
      "epoch": 0.13071643412307704,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009627201969478777,
      "loss": 2.104,
      "step": 19239
    },
    {
      "epoch": 0.1307232284696711,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009627161204860791,
      "loss": 2.0902,
      "step": 19240
    },
    {
      "epoch": 0.13073002281626517,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009627120438100483,
      "loss": 2.0352,
      "step": 19241
    },
    {
      "epoch": 0.1307368171628592,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009627079669197869,
      "loss": 2.1943,
      "step": 19242
    },
    {
      "epoch": 0.13074361150945327,
      "grad_norm": 0.125,
      "learning_rate": 0.0009627038898152974,
      "loss": 2.0493,
      "step": 19243
    },
    {
      "epoch": 0.13075040585604733,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009626998124965809,
      "loss": 2.2998,
      "step": 19244
    },
    {
      "epoch": 0.1307572002026414,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009626957349636398,
      "loss": 2.0487,
      "step": 19245
    },
    {
      "epoch": 0.13076399454923546,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009626916572164761,
      "loss": 2.1945,
      "step": 19246
    },
    {
      "epoch": 0.1307707888958295,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009626875792550913,
      "loss": 2.2304,
      "step": 19247
    },
    {
      "epoch": 0.13077758324242356,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009626835010794875,
      "loss": 2.3435,
      "step": 19248
    },
    {
      "epoch": 0.13078437758901762,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009626794226896665,
      "loss": 2.1758,
      "step": 19249
    },
    {
      "epoch": 0.13079117193561168,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009626753440856302,
      "loss": 2.1476,
      "step": 19250
    },
    {
      "epoch": 0.13079796628220575,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009626712652673806,
      "loss": 2.2832,
      "step": 19251
    },
    {
      "epoch": 0.13080476062879978,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009626671862349195,
      "loss": 2.2536,
      "step": 19252
    },
    {
      "epoch": 0.13081155497539385,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009626631069882488,
      "loss": 2.2204,
      "step": 19253
    },
    {
      "epoch": 0.1308183493219879,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009626590275273703,
      "loss": 2.1573,
      "step": 19254
    },
    {
      "epoch": 0.13082514366858197,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009626549478522859,
      "loss": 2.2821,
      "step": 19255
    },
    {
      "epoch": 0.13083193801517604,
      "grad_norm": 0.123046875,
      "learning_rate": 0.0009626508679629977,
      "loss": 2.076,
      "step": 19256
    },
    {
      "epoch": 0.13083873236177007,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009626467878595075,
      "loss": 2.3264,
      "step": 19257
    },
    {
      "epoch": 0.13084552670836413,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009626427075418171,
      "loss": 2.1762,
      "step": 19258
    },
    {
      "epoch": 0.1308523210549582,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009626386270099284,
      "loss": 2.0937,
      "step": 19259
    },
    {
      "epoch": 0.13085911540155226,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009626345462638434,
      "loss": 2.1337,
      "step": 19260
    },
    {
      "epoch": 0.1308659097481463,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009626304653035638,
      "loss": 2.0039,
      "step": 19261
    },
    {
      "epoch": 0.13087270409474036,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009626263841290917,
      "loss": 2.199,
      "step": 19262
    },
    {
      "epoch": 0.13087949844133442,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009626223027404289,
      "loss": 2.0835,
      "step": 19263
    },
    {
      "epoch": 0.1308862927879285,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009626182211375772,
      "loss": 2.0846,
      "step": 19264
    },
    {
      "epoch": 0.13089308713452255,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009626141393205385,
      "loss": 2.2794,
      "step": 19265
    },
    {
      "epoch": 0.1308998814811166,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009626100572893148,
      "loss": 2.2053,
      "step": 19266
    },
    {
      "epoch": 0.13090667582771065,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000962605975043908,
      "loss": 2.2138,
      "step": 19267
    },
    {
      "epoch": 0.1309134701743047,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009626018925843198,
      "loss": 2.2057,
      "step": 19268
    },
    {
      "epoch": 0.13092026452089878,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009625978099105525,
      "loss": 2.1873,
      "step": 19269
    },
    {
      "epoch": 0.13092705886749284,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009625937270226075,
      "loss": 2.3779,
      "step": 19270
    },
    {
      "epoch": 0.13093385321408688,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009625896439204869,
      "loss": 2.0121,
      "step": 19271
    },
    {
      "epoch": 0.13094064756068094,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009625855606041927,
      "loss": 2.0854,
      "step": 19272
    },
    {
      "epoch": 0.130947441907275,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009625814770737267,
      "loss": 2.2185,
      "step": 19273
    },
    {
      "epoch": 0.13095423625386907,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009625773933290907,
      "loss": 2.2528,
      "step": 19274
    },
    {
      "epoch": 0.13096103060046313,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009625733093702868,
      "loss": 2.0914,
      "step": 19275
    },
    {
      "epoch": 0.13096782494705717,
      "grad_norm": 0.125,
      "learning_rate": 0.0009625692251973166,
      "loss": 2.0565,
      "step": 19276
    },
    {
      "epoch": 0.13097461929365123,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009625651408101823,
      "loss": 2.2173,
      "step": 19277
    },
    {
      "epoch": 0.1309814136402453,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009625610562088855,
      "loss": 2.2878,
      "step": 19278
    },
    {
      "epoch": 0.13098820798683936,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009625569713934283,
      "loss": 2.1633,
      "step": 19279
    },
    {
      "epoch": 0.13099500233343342,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009625528863638125,
      "loss": 2.1834,
      "step": 19280
    },
    {
      "epoch": 0.13100179668002745,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00096254880112004,
      "loss": 1.9584,
      "step": 19281
    },
    {
      "epoch": 0.13100859102662152,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009625447156621128,
      "loss": 2.0946,
      "step": 19282
    },
    {
      "epoch": 0.13101538537321558,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009625406299900327,
      "loss": 2.0772,
      "step": 19283
    },
    {
      "epoch": 0.13102217971980964,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009625365441038015,
      "loss": 2.1084,
      "step": 19284
    },
    {
      "epoch": 0.1310289740664037,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009625324580034212,
      "loss": 2.2208,
      "step": 19285
    },
    {
      "epoch": 0.13103576841299774,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009625283716888938,
      "loss": 2.0361,
      "step": 19286
    },
    {
      "epoch": 0.1310425627595918,
      "grad_norm": 0.125,
      "learning_rate": 0.0009625242851602209,
      "loss": 2.0444,
      "step": 19287
    },
    {
      "epoch": 0.13104935710618587,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009625201984174046,
      "loss": 2.2125,
      "step": 19288
    },
    {
      "epoch": 0.13105615145277993,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009625161114604468,
      "loss": 2.095,
      "step": 19289
    },
    {
      "epoch": 0.131062945799374,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009625120242893493,
      "loss": 2.0714,
      "step": 19290
    },
    {
      "epoch": 0.13106974014596803,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009625079369041141,
      "loss": 2.2014,
      "step": 19291
    },
    {
      "epoch": 0.1310765344925621,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009625038493047429,
      "loss": 2.0124,
      "step": 19292
    },
    {
      "epoch": 0.13108332883915616,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009624997614912378,
      "loss": 2.3408,
      "step": 19293
    },
    {
      "epoch": 0.13109012318575022,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009624956734636006,
      "loss": 2.0311,
      "step": 19294
    },
    {
      "epoch": 0.1310969175323443,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009624915852218332,
      "loss": 2.1376,
      "step": 19295
    },
    {
      "epoch": 0.13110371187893832,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009624874967659376,
      "loss": 1.9957,
      "step": 19296
    },
    {
      "epoch": 0.13111050622553239,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009624834080959155,
      "loss": 2.0898,
      "step": 19297
    },
    {
      "epoch": 0.13111730057212645,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009624793192117689,
      "loss": 2.1798,
      "step": 19298
    },
    {
      "epoch": 0.1311240949187205,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009624752301134996,
      "loss": 2.0665,
      "step": 19299
    },
    {
      "epoch": 0.13113088926531458,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009624711408011096,
      "loss": 2.125,
      "step": 19300
    },
    {
      "epoch": 0.1311376836119086,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009624670512746007,
      "loss": 2.233,
      "step": 19301
    },
    {
      "epoch": 0.13114447795850268,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.000962462961533975,
      "loss": 2.2407,
      "step": 19302
    },
    {
      "epoch": 0.13115127230509674,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009624588715792343,
      "loss": 2.0851,
      "step": 19303
    },
    {
      "epoch": 0.1311580666516908,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009624547814103804,
      "loss": 2.2907,
      "step": 19304
    },
    {
      "epoch": 0.13116486099828487,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009624506910274151,
      "loss": 2.1779,
      "step": 19305
    },
    {
      "epoch": 0.1311716553448789,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009624466004303405,
      "loss": 2.1608,
      "step": 19306
    },
    {
      "epoch": 0.13117844969147296,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009624425096191585,
      "loss": 2.1345,
      "step": 19307
    },
    {
      "epoch": 0.13118524403806703,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009624384185938709,
      "loss": 2.0827,
      "step": 19308
    },
    {
      "epoch": 0.1311920383846611,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009624343273544795,
      "loss": 2.1763,
      "step": 19309
    },
    {
      "epoch": 0.13119883273125513,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009624302359009865,
      "loss": 2.3157,
      "step": 19310
    },
    {
      "epoch": 0.1312056270778492,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009624261442333935,
      "loss": 2.0732,
      "step": 19311
    },
    {
      "epoch": 0.13121242142444325,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009624220523517024,
      "loss": 2.2085,
      "step": 19312
    },
    {
      "epoch": 0.13121921577103732,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009624179602559153,
      "loss": 2.1176,
      "step": 19313
    },
    {
      "epoch": 0.13122601011763138,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009624138679460339,
      "loss": 2.1818,
      "step": 19314
    },
    {
      "epoch": 0.13123280446422542,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009624097754220603,
      "loss": 2.3007,
      "step": 19315
    },
    {
      "epoch": 0.13123959881081948,
      "grad_norm": 0.11572265625,
      "learning_rate": 0.0009624056826839962,
      "loss": 2.0442,
      "step": 19316
    },
    {
      "epoch": 0.13124639315741354,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009624015897318436,
      "loss": 2.1159,
      "step": 19317
    },
    {
      "epoch": 0.1312531875040076,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009623974965656044,
      "loss": 2.2481,
      "step": 19318
    },
    {
      "epoch": 0.13125998185060167,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009623934031852804,
      "loss": 2.1968,
      "step": 19319
    },
    {
      "epoch": 0.1312667761971957,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009623893095908736,
      "loss": 2.2676,
      "step": 19320
    },
    {
      "epoch": 0.13127357054378977,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009623852157823858,
      "loss": 2.1369,
      "step": 19321
    },
    {
      "epoch": 0.13128036489038383,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009623811217598189,
      "loss": 2.179,
      "step": 19322
    },
    {
      "epoch": 0.1312871592369779,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009623770275231749,
      "loss": 2.1269,
      "step": 19323
    },
    {
      "epoch": 0.13129395358357196,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009623729330724557,
      "loss": 2.146,
      "step": 19324
    },
    {
      "epoch": 0.131300747930166,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009623688384076631,
      "loss": 2.1816,
      "step": 19325
    },
    {
      "epoch": 0.13130754227676006,
      "grad_norm": 0.125,
      "learning_rate": 0.000962364743528799,
      "loss": 2.0524,
      "step": 19326
    },
    {
      "epoch": 0.13131433662335412,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009623606484358653,
      "loss": 2.3146,
      "step": 19327
    },
    {
      "epoch": 0.13132113096994819,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000962356553128864,
      "loss": 2.1281,
      "step": 19328
    },
    {
      "epoch": 0.13132792531654225,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009623524576077969,
      "loss": 2.0973,
      "step": 19329
    },
    {
      "epoch": 0.13133471966313628,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009623483618726659,
      "loss": 2.1474,
      "step": 19330
    },
    {
      "epoch": 0.13134151400973035,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009623442659234729,
      "loss": 2.1551,
      "step": 19331
    },
    {
      "epoch": 0.1313483083563244,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009623401697602198,
      "loss": 2.1694,
      "step": 19332
    },
    {
      "epoch": 0.13135510270291847,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009623360733829085,
      "loss": 2.0723,
      "step": 19333
    },
    {
      "epoch": 0.13136189704951254,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009623319767915409,
      "loss": 2.0844,
      "step": 19334
    },
    {
      "epoch": 0.13136869139610657,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009623278799861188,
      "loss": 2.1717,
      "step": 19335
    },
    {
      "epoch": 0.13137548574270064,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009623237829666443,
      "loss": 2.2718,
      "step": 19336
    },
    {
      "epoch": 0.1313822800892947,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009623196857331192,
      "loss": 2.1016,
      "step": 19337
    },
    {
      "epoch": 0.13138907443588876,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009623155882855453,
      "loss": 2.2686,
      "step": 19338
    },
    {
      "epoch": 0.13139586878248283,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009623114906239247,
      "loss": 2.1866,
      "step": 19339
    },
    {
      "epoch": 0.13140266312907686,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009623073927482591,
      "loss": 2.1087,
      "step": 19340
    },
    {
      "epoch": 0.13140945747567093,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009623032946585506,
      "loss": 2.1658,
      "step": 19341
    },
    {
      "epoch": 0.131416251822265,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009622991963548008,
      "loss": 2.2401,
      "step": 19342
    },
    {
      "epoch": 0.13142304616885905,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009622950978370119,
      "loss": 1.9878,
      "step": 19343
    },
    {
      "epoch": 0.13142984051545312,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009622909991051856,
      "loss": 2.2144,
      "step": 19344
    },
    {
      "epoch": 0.13143663486204715,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009622869001593238,
      "loss": 2.1535,
      "step": 19345
    },
    {
      "epoch": 0.13144342920864122,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009622828009994286,
      "loss": 2.1246,
      "step": 19346
    },
    {
      "epoch": 0.13145022355523528,
      "grad_norm": 0.1455078125,
      "learning_rate": 0.0009622787016255017,
      "loss": 2.5048,
      "step": 19347
    },
    {
      "epoch": 0.13145701790182934,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000962274602037545,
      "loss": 2.3502,
      "step": 19348
    },
    {
      "epoch": 0.1314638122484234,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009622705022355606,
      "loss": 2.2624,
      "step": 19349
    },
    {
      "epoch": 0.13147060659501744,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009622664022195502,
      "loss": 2.2366,
      "step": 19350
    },
    {
      "epoch": 0.1314774009416115,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009622623019895158,
      "loss": 2.1577,
      "step": 19351
    },
    {
      "epoch": 0.13148419528820557,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000962258201545459,
      "loss": 2.1485,
      "step": 19352
    },
    {
      "epoch": 0.13149098963479963,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009622541008873823,
      "loss": 2.2155,
      "step": 19353
    },
    {
      "epoch": 0.1314977839813937,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009622500000152871,
      "loss": 2.1222,
      "step": 19354
    },
    {
      "epoch": 0.13150457832798773,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009622458989291754,
      "loss": 2.2576,
      "step": 19355
    },
    {
      "epoch": 0.1315113726745818,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009622417976290491,
      "loss": 2.1626,
      "step": 19356
    },
    {
      "epoch": 0.13151816702117586,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009622376961149104,
      "loss": 2.1121,
      "step": 19357
    },
    {
      "epoch": 0.13152496136776992,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009622335943867608,
      "loss": 2.0942,
      "step": 19358
    },
    {
      "epoch": 0.13153175571436396,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009622294924446023,
      "loss": 2.1662,
      "step": 19359
    },
    {
      "epoch": 0.13153855006095802,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009622253902884369,
      "loss": 2.3073,
      "step": 19360
    },
    {
      "epoch": 0.13154534440755208,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009622212879182664,
      "loss": 2.2059,
      "step": 19361
    },
    {
      "epoch": 0.13155213875414615,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009622171853340928,
      "loss": 2.2938,
      "step": 19362
    },
    {
      "epoch": 0.1315589331007402,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009622130825359178,
      "loss": 2.1485,
      "step": 19363
    },
    {
      "epoch": 0.13156572744733425,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009622089795237437,
      "loss": 2.0514,
      "step": 19364
    },
    {
      "epoch": 0.1315725217939283,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009622048762975719,
      "loss": 2.0666,
      "step": 19365
    },
    {
      "epoch": 0.13157931614052237,
      "grad_norm": 0.12109375,
      "learning_rate": 0.0009622007728574047,
      "loss": 2.0883,
      "step": 19366
    },
    {
      "epoch": 0.13158611048711644,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009621966692032438,
      "loss": 2.2559,
      "step": 19367
    },
    {
      "epoch": 0.1315929048337105,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009621925653350912,
      "loss": 2.1332,
      "step": 19368
    },
    {
      "epoch": 0.13159969918030454,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009621884612529486,
      "loss": 2.1377,
      "step": 19369
    },
    {
      "epoch": 0.1316064935268986,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009621843569568181,
      "loss": 2.2818,
      "step": 19370
    },
    {
      "epoch": 0.13161328787349266,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009621802524467014,
      "loss": 2.1405,
      "step": 19371
    },
    {
      "epoch": 0.13162008222008673,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009621761477226008,
      "loss": 2.1455,
      "step": 19372
    },
    {
      "epoch": 0.1316268765666808,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009621720427845178,
      "loss": 2.1203,
      "step": 19373
    },
    {
      "epoch": 0.13163367091327483,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009621679376324543,
      "loss": 2.3165,
      "step": 19374
    },
    {
      "epoch": 0.1316404652598689,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009621638322664127,
      "loss": 2.2126,
      "step": 19375
    },
    {
      "epoch": 0.13164725960646295,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009621597266863944,
      "loss": 2.2507,
      "step": 19376
    },
    {
      "epoch": 0.13165405395305702,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009621556208924014,
      "loss": 2.0657,
      "step": 19377
    },
    {
      "epoch": 0.13166084829965108,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009621515148844355,
      "loss": 1.9952,
      "step": 19378
    },
    {
      "epoch": 0.13166764264624511,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009621474086624989,
      "loss": 2.1194,
      "step": 19379
    },
    {
      "epoch": 0.13167443699283918,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009621433022265934,
      "loss": 2.1003,
      "step": 19380
    },
    {
      "epoch": 0.13168123133943324,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009621391955767208,
      "loss": 2.0529,
      "step": 19381
    },
    {
      "epoch": 0.1316880256860273,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.000962135088712883,
      "loss": 2.2306,
      "step": 19382
    },
    {
      "epoch": 0.13169482003262137,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.0009621309816350821,
      "loss": 2.2796,
      "step": 19383
    },
    {
      "epoch": 0.1317016143792154,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009621268743433195,
      "loss": 2.1421,
      "step": 19384
    },
    {
      "epoch": 0.13170840872580947,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009621227668375978,
      "loss": 2.1797,
      "step": 19385
    },
    {
      "epoch": 0.13171520307240353,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009621186591179185,
      "loss": 2.1173,
      "step": 19386
    },
    {
      "epoch": 0.1317219974189976,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009621145511842835,
      "loss": 2.2293,
      "step": 19387
    },
    {
      "epoch": 0.13172879176559166,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009621104430366949,
      "loss": 2.1519,
      "step": 19388
    },
    {
      "epoch": 0.1317355861121857,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009621063346751542,
      "loss": 2.1588,
      "step": 19389
    },
    {
      "epoch": 0.13174238045877976,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009621022260996637,
      "loss": 2.1325,
      "step": 19390
    },
    {
      "epoch": 0.13174917480537382,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009620981173102252,
      "loss": 2.2983,
      "step": 19391
    },
    {
      "epoch": 0.13175596915196788,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009620940083068406,
      "loss": 2.1536,
      "step": 19392
    },
    {
      "epoch": 0.13176276349856195,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009620898990895117,
      "loss": 2.2776,
      "step": 19393
    },
    {
      "epoch": 0.13176955784515598,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009620857896582405,
      "loss": 1.9257,
      "step": 19394
    },
    {
      "epoch": 0.13177635219175005,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009620816800130289,
      "loss": 2.0978,
      "step": 19395
    },
    {
      "epoch": 0.1317831465383441,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009620775701538786,
      "loss": 2.302,
      "step": 19396
    },
    {
      "epoch": 0.13178994088493817,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009620734600807919,
      "loss": 2.1054,
      "step": 19397
    },
    {
      "epoch": 0.13179673523153224,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009620693497937705,
      "loss": 2.1217,
      "step": 19398
    },
    {
      "epoch": 0.13180352957812627,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009620652392928162,
      "loss": 2.2386,
      "step": 19399
    },
    {
      "epoch": 0.13181032392472034,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009620611285779309,
      "loss": 2.1045,
      "step": 19400
    },
    {
      "epoch": 0.1318171182713144,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009620570176491166,
      "loss": 2.0408,
      "step": 19401
    },
    {
      "epoch": 0.13182391261790846,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009620529065063754,
      "loss": 1.9086,
      "step": 19402
    },
    {
      "epoch": 0.13183070696450253,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009620487951497088,
      "loss": 2.2371,
      "step": 19403
    },
    {
      "epoch": 0.13183750131109656,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.000962044683579119,
      "loss": 2.2251,
      "step": 19404
    },
    {
      "epoch": 0.13184429565769062,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009620405717946079,
      "loss": 2.1644,
      "step": 19405
    },
    {
      "epoch": 0.1318510900042847,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009620364597961773,
      "loss": 2.1699,
      "step": 19406
    },
    {
      "epoch": 0.13185788435087875,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009620323475838288,
      "loss": 2.1039,
      "step": 19407
    },
    {
      "epoch": 0.1318646786974728,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009620282351575649,
      "loss": 2.2124,
      "step": 19408
    },
    {
      "epoch": 0.13187147304406685,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009620241225173872,
      "loss": 2.236,
      "step": 19409
    },
    {
      "epoch": 0.1318782673906609,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009620200096632976,
      "loss": 2.2429,
      "step": 19410
    },
    {
      "epoch": 0.13188506173725498,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009620158965952981,
      "loss": 2.1474,
      "step": 19411
    },
    {
      "epoch": 0.13189185608384904,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009620117833133904,
      "loss": 2.1571,
      "step": 19412
    },
    {
      "epoch": 0.13189865043044308,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009620076698175766,
      "loss": 2.1512,
      "step": 19413
    },
    {
      "epoch": 0.13190544477703714,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009620035561078586,
      "loss": 2.2744,
      "step": 19414
    },
    {
      "epoch": 0.1319122391236312,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009619994421842382,
      "loss": 2.2659,
      "step": 19415
    },
    {
      "epoch": 0.13191903347022527,
      "grad_norm": 0.125,
      "learning_rate": 0.0009619953280467174,
      "loss": 2.0493,
      "step": 19416
    },
    {
      "epoch": 0.13192582781681933,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009619912136952979,
      "loss": 2.1012,
      "step": 19417
    },
    {
      "epoch": 0.13193262216341337,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.000961987099129982,
      "loss": 2.2702,
      "step": 19418
    },
    {
      "epoch": 0.13193941651000743,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009619829843507713,
      "loss": 2.164,
      "step": 19419
    },
    {
      "epoch": 0.1319462108566015,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009619788693576676,
      "loss": 2.2603,
      "step": 19420
    },
    {
      "epoch": 0.13195300520319556,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009619747541506731,
      "loss": 2.237,
      "step": 19421
    },
    {
      "epoch": 0.13195979954978962,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009619706387297897,
      "loss": 2.1787,
      "step": 19422
    },
    {
      "epoch": 0.13196659389638365,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009619665230950191,
      "loss": 2.2799,
      "step": 19423
    },
    {
      "epoch": 0.13197338824297772,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009619624072463633,
      "loss": 2.1194,
      "step": 19424
    },
    {
      "epoch": 0.13198018258957178,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009619582911838242,
      "loss": 2.0823,
      "step": 19425
    },
    {
      "epoch": 0.13198697693616585,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009619541749074037,
      "loss": 2.2321,
      "step": 19426
    },
    {
      "epoch": 0.1319937712827599,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009619500584171037,
      "loss": 2.1633,
      "step": 19427
    },
    {
      "epoch": 0.13200056562935394,
      "grad_norm": 0.13671875,
      "learning_rate": 0.000961945941712926,
      "loss": 2.1229,
      "step": 19428
    },
    {
      "epoch": 0.132007359975948,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009619418247948728,
      "loss": 2.2662,
      "step": 19429
    },
    {
      "epoch": 0.13201415432254207,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009619377076629459,
      "loss": 2.0552,
      "step": 19430
    },
    {
      "epoch": 0.13202094866913613,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009619335903171469,
      "loss": 2.1562,
      "step": 19431
    },
    {
      "epoch": 0.1320277430157302,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009619294727574781,
      "loss": 2.2291,
      "step": 19432
    },
    {
      "epoch": 0.13203453736232423,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009619253549839412,
      "loss": 2.1179,
      "step": 19433
    },
    {
      "epoch": 0.1320413317089183,
      "grad_norm": 0.1220703125,
      "learning_rate": 0.0009619212369965382,
      "loss": 2.1542,
      "step": 19434
    },
    {
      "epoch": 0.13204812605551236,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009619171187952709,
      "loss": 2.15,
      "step": 19435
    },
    {
      "epoch": 0.13205492040210642,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009619130003801413,
      "loss": 2.235,
      "step": 19436
    },
    {
      "epoch": 0.1320617147487005,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009619088817511512,
      "loss": 2.0578,
      "step": 19437
    },
    {
      "epoch": 0.13206850909529452,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009619047629083027,
      "loss": 2.11,
      "step": 19438
    },
    {
      "epoch": 0.1320753034418886,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009619006438515976,
      "loss": 2.2295,
      "step": 19439
    },
    {
      "epoch": 0.13208209778848265,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009618965245810377,
      "loss": 2.1143,
      "step": 19440
    },
    {
      "epoch": 0.1320888921350767,
      "grad_norm": 0.125,
      "learning_rate": 0.0009618924050966251,
      "loss": 2.1764,
      "step": 19441
    },
    {
      "epoch": 0.13209568648167078,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009618882853983615,
      "loss": 2.1799,
      "step": 19442
    },
    {
      "epoch": 0.1321024808282648,
      "grad_norm": 0.12890625,
      "learning_rate": 0.000961884165486249,
      "loss": 2.3434,
      "step": 19443
    },
    {
      "epoch": 0.13210927517485888,
      "grad_norm": 0.125,
      "learning_rate": 0.0009618800453602894,
      "loss": 2.1252,
      "step": 19444
    },
    {
      "epoch": 0.13211606952145294,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009618759250204847,
      "loss": 2.0964,
      "step": 19445
    },
    {
      "epoch": 0.132122863868047,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009618718044668367,
      "loss": 2.1815,
      "step": 19446
    },
    {
      "epoch": 0.13212965821464107,
      "grad_norm": 0.12158203125,
      "learning_rate": 0.0009618676836993472,
      "loss": 2.211,
      "step": 19447
    },
    {
      "epoch": 0.1321364525612351,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009618635627180184,
      "loss": 2.1941,
      "step": 19448
    },
    {
      "epoch": 0.13214324690782916,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000961859441522852,
      "loss": 2.1197,
      "step": 19449
    },
    {
      "epoch": 0.13215004125442323,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009618553201138501,
      "loss": 2.0215,
      "step": 19450
    },
    {
      "epoch": 0.1321568356010173,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009618511984910144,
      "loss": 2.158,
      "step": 19451
    },
    {
      "epoch": 0.13216362994761133,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000961847076654347,
      "loss": 2.1813,
      "step": 19452
    },
    {
      "epoch": 0.1321704242942054,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009618429546038496,
      "loss": 2.3025,
      "step": 19453
    },
    {
      "epoch": 0.13217721864079945,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009618388323395241,
      "loss": 2.2089,
      "step": 19454
    },
    {
      "epoch": 0.13218401298739352,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009618347098613727,
      "loss": 2.1648,
      "step": 19455
    },
    {
      "epoch": 0.13219080733398758,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009618305871693971,
      "loss": 2.1374,
      "step": 19456
    },
    {
      "epoch": 0.13219760168058162,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009618264642635991,
      "loss": 2.1135,
      "step": 19457
    },
    {
      "epoch": 0.13220439602717568,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009618223411439808,
      "loss": 2.2939,
      "step": 19458
    },
    {
      "epoch": 0.13221119037376974,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009618182178105441,
      "loss": 2.2961,
      "step": 19459
    },
    {
      "epoch": 0.1322179847203638,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009618140942632909,
      "loss": 2.1037,
      "step": 19460
    },
    {
      "epoch": 0.13222477906695787,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.000961809970502223,
      "loss": 2.2251,
      "step": 19461
    },
    {
      "epoch": 0.1322315734135519,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009618058465273423,
      "loss": 1.9846,
      "step": 19462
    },
    {
      "epoch": 0.13223836776014597,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009618017223386508,
      "loss": 2.0567,
      "step": 19463
    },
    {
      "epoch": 0.13224516210674003,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009617975979361506,
      "loss": 2.0766,
      "step": 19464
    },
    {
      "epoch": 0.1322519564533341,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009617934733198432,
      "loss": 2.0918,
      "step": 19465
    },
    {
      "epoch": 0.13225875079992816,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009617893484897309,
      "loss": 2.1796,
      "step": 19466
    },
    {
      "epoch": 0.1322655451465222,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009617852234458152,
      "loss": 2.2045,
      "step": 19467
    },
    {
      "epoch": 0.13227233949311626,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009617810981880984,
      "loss": 2.1455,
      "step": 19468
    },
    {
      "epoch": 0.13227913383971032,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009617769727165821,
      "loss": 2.2645,
      "step": 19469
    },
    {
      "epoch": 0.13228592818630439,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009617728470312685,
      "loss": 1.9939,
      "step": 19470
    },
    {
      "epoch": 0.13229272253289845,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0009617687211321593,
      "loss": 2.2379,
      "step": 19471
    },
    {
      "epoch": 0.13229951687949248,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009617645950192565,
      "loss": 2.1677,
      "step": 19472
    },
    {
      "epoch": 0.13230631122608655,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009617604686925621,
      "loss": 2.3178,
      "step": 19473
    },
    {
      "epoch": 0.1323131055726806,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009617563421520776,
      "loss": 2.3575,
      "step": 19474
    },
    {
      "epoch": 0.13231989991927467,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009617522153978055,
      "loss": 2.2582,
      "step": 19475
    },
    {
      "epoch": 0.13232669426586874,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009617480884297471,
      "loss": 2.2397,
      "step": 19476
    },
    {
      "epoch": 0.13233348861246277,
      "grad_norm": 0.130859375,
      "learning_rate": 0.000961743961247905,
      "loss": 2.1836,
      "step": 19477
    },
    {
      "epoch": 0.13234028295905684,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009617398338522805,
      "loss": 2.2522,
      "step": 19478
    },
    {
      "epoch": 0.1323470773056509,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009617357062428758,
      "loss": 2.1833,
      "step": 19479
    },
    {
      "epoch": 0.13235387165224496,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009617315784196929,
      "loss": 2.1552,
      "step": 19480
    },
    {
      "epoch": 0.13236066599883903,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009617274503827334,
      "loss": 2.0919,
      "step": 19481
    },
    {
      "epoch": 0.13236746034543306,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009617233221319994,
      "loss": 2.2834,
      "step": 19482
    },
    {
      "epoch": 0.13237425469202713,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009617191936674928,
      "loss": 2.1567,
      "step": 19483
    },
    {
      "epoch": 0.1323810490386212,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009617150649892155,
      "loss": 2.0578,
      "step": 19484
    },
    {
      "epoch": 0.13238784338521525,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009617109360971696,
      "loss": 2.1112,
      "step": 19485
    },
    {
      "epoch": 0.13239463773180932,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009617068069913565,
      "loss": 2.3373,
      "step": 19486
    },
    {
      "epoch": 0.13240143207840335,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009617026776717788,
      "loss": 2.2516,
      "step": 19487
    },
    {
      "epoch": 0.13240822642499742,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009616985481384379,
      "loss": 2.2429,
      "step": 19488
    },
    {
      "epoch": 0.13241502077159148,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009616944183913357,
      "loss": 2.267,
      "step": 19489
    },
    {
      "epoch": 0.13242181511818554,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0009616902884304746,
      "loss": 2.3377,
      "step": 19490
    },
    {
      "epoch": 0.1324286094647796,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009616861582558559,
      "loss": 2.1911,
      "step": 19491
    },
    {
      "epoch": 0.13243540381137364,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009616820278674819,
      "loss": 2.181,
      "step": 19492
    },
    {
      "epoch": 0.1324421981579677,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009616778972653543,
      "loss": 2.2208,
      "step": 19493
    },
    {
      "epoch": 0.13244899250456177,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009616737664494754,
      "loss": 2.3352,
      "step": 19494
    },
    {
      "epoch": 0.13245578685115583,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009616696354198467,
      "loss": 2.0937,
      "step": 19495
    },
    {
      "epoch": 0.1324625811977499,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009616655041764702,
      "loss": 2.2103,
      "step": 19496
    },
    {
      "epoch": 0.13246937554434393,
      "grad_norm": 0.1474609375,
      "learning_rate": 0.000961661372719348,
      "loss": 2.1225,
      "step": 19497
    },
    {
      "epoch": 0.132476169890938,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009616572410484817,
      "loss": 2.1356,
      "step": 19498
    },
    {
      "epoch": 0.13248296423753206,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009616531091638736,
      "loss": 2.1449,
      "step": 19499
    },
    {
      "epoch": 0.13248975858412612,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009616489770655252,
      "loss": 2.1075,
      "step": 19500
    },
    {
      "epoch": 0.13249655293072016,
      "grad_norm": 0.15625,
      "learning_rate": 0.0009616448447534387,
      "loss": 2.3901,
      "step": 19501
    },
    {
      "epoch": 0.13250334727731422,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009616407122276159,
      "loss": 2.2569,
      "step": 19502
    },
    {
      "epoch": 0.13251014162390828,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009616365794880589,
      "loss": 2.3075,
      "step": 19503
    },
    {
      "epoch": 0.13251693597050235,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009616324465347693,
      "loss": 2.1375,
      "step": 19504
    },
    {
      "epoch": 0.1325237303170964,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009616283133677491,
      "loss": 2.1036,
      "step": 19505
    },
    {
      "epoch": 0.13253052466369045,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009616241799870005,
      "loss": 2.0933,
      "step": 19506
    },
    {
      "epoch": 0.1325373190102845,
      "grad_norm": 0.1328125,
      "learning_rate": 0.000961620046392525,
      "loss": 2.2479,
      "step": 19507
    },
    {
      "epoch": 0.13254411335687857,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009616159125843248,
      "loss": 2.2934,
      "step": 19508
    },
    {
      "epoch": 0.13255090770347264,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009616117785624016,
      "loss": 2.1211,
      "step": 19509
    },
    {
      "epoch": 0.1325577020500667,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009616076443267576,
      "loss": 2.0817,
      "step": 19510
    },
    {
      "epoch": 0.13256449639666074,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009616035098773945,
      "loss": 2.2051,
      "step": 19511
    },
    {
      "epoch": 0.1325712907432548,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009615993752143143,
      "loss": 2.1978,
      "step": 19512
    },
    {
      "epoch": 0.13257808508984886,
      "grad_norm": 0.166015625,
      "learning_rate": 0.0009615952403375188,
      "loss": 2.3021,
      "step": 19513
    },
    {
      "epoch": 0.13258487943644293,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009615911052470101,
      "loss": 2.1567,
      "step": 19514
    },
    {
      "epoch": 0.132591673783037,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009615869699427899,
      "loss": 2.21,
      "step": 19515
    },
    {
      "epoch": 0.13259846812963103,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009615828344248603,
      "loss": 2.1841,
      "step": 19516
    },
    {
      "epoch": 0.1326052624762251,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009615786986932231,
      "loss": 2.2386,
      "step": 19517
    },
    {
      "epoch": 0.13261205682281915,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009615745627478804,
      "loss": 2.2514,
      "step": 19518
    },
    {
      "epoch": 0.13261885116941322,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009615704265888337,
      "loss": 2.204,
      "step": 19519
    },
    {
      "epoch": 0.13262564551600728,
      "grad_norm": 0.1240234375,
      "learning_rate": 0.0009615662902160853,
      "loss": 2.0566,
      "step": 19520
    },
    {
      "epoch": 0.13263243986260131,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.000961562153629637,
      "loss": 2.2872,
      "step": 19521
    },
    {
      "epoch": 0.13263923420919538,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009615580168294909,
      "loss": 2.0866,
      "step": 19522
    },
    {
      "epoch": 0.13264602855578944,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009615538798156486,
      "loss": 2.2045,
      "step": 19523
    },
    {
      "epoch": 0.1326528229023835,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009615497425881121,
      "loss": 2.1407,
      "step": 19524
    },
    {
      "epoch": 0.13265961724897757,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009615456051468834,
      "loss": 2.2374,
      "step": 19525
    },
    {
      "epoch": 0.1326664115955716,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009615414674919643,
      "loss": 2.0216,
      "step": 19526
    },
    {
      "epoch": 0.13267320594216567,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009615373296233569,
      "loss": 2.1688,
      "step": 19527
    },
    {
      "epoch": 0.13268000028875973,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009615331915410629,
      "loss": 2.1128,
      "step": 19528
    },
    {
      "epoch": 0.1326867946353538,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009615290532450845,
      "loss": 2.1459,
      "step": 19529
    },
    {
      "epoch": 0.13269358898194786,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009615249147354232,
      "loss": 2.0021,
      "step": 19530
    },
    {
      "epoch": 0.1327003833285419,
      "grad_norm": 0.1484375,
      "learning_rate": 0.0009615207760120812,
      "loss": 2.1017,
      "step": 19531
    },
    {
      "epoch": 0.13270717767513596,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009615166370750605,
      "loss": 2.3242,
      "step": 19532
    },
    {
      "epoch": 0.13271397202173002,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009615124979243627,
      "loss": 2.2218,
      "step": 19533
    },
    {
      "epoch": 0.13272076636832408,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009615083585599902,
      "loss": 2.1779,
      "step": 19534
    },
    {
      "epoch": 0.13272756071491815,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009615042189819445,
      "loss": 2.1607,
      "step": 19535
    },
    {
      "epoch": 0.13273435506151218,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009615000791902275,
      "loss": 2.1063,
      "step": 19536
    },
    {
      "epoch": 0.13274114940810625,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009614959391848414,
      "loss": 2.0484,
      "step": 19537
    },
    {
      "epoch": 0.1327479437547003,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009614917989657878,
      "loss": 2.1957,
      "step": 19538
    },
    {
      "epoch": 0.13275473810129437,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009614876585330689,
      "loss": 2.138,
      "step": 19539
    },
    {
      "epoch": 0.13276153244788844,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009614835178866865,
      "loss": 2.1088,
      "step": 19540
    },
    {
      "epoch": 0.13276832679448247,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009614793770266425,
      "loss": 2.2251,
      "step": 19541
    },
    {
      "epoch": 0.13277512114107654,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009614752359529391,
      "loss": 2.1058,
      "step": 19542
    },
    {
      "epoch": 0.1327819154876706,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009614710946655776,
      "loss": 2.1477,
      "step": 19543
    },
    {
      "epoch": 0.13278870983426466,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009614669531645604,
      "loss": 2.0143,
      "step": 19544
    },
    {
      "epoch": 0.13279550418085873,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009614628114498893,
      "loss": 2.1756,
      "step": 19545
    },
    {
      "epoch": 0.13280229852745276,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009614586695215662,
      "loss": 2.0082,
      "step": 19546
    },
    {
      "epoch": 0.13280909287404682,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009614545273795932,
      "loss": 2.0714,
      "step": 19547
    },
    {
      "epoch": 0.1328158872206409,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009614503850239718,
      "loss": 2.2632,
      "step": 19548
    },
    {
      "epoch": 0.13282268156723495,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009614462424547041,
      "loss": 2.1187,
      "step": 19549
    },
    {
      "epoch": 0.132829475913829,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009614420996717923,
      "loss": 2.2904,
      "step": 19550
    },
    {
      "epoch": 0.13283627026042305,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.000961437956675238,
      "loss": 2.2851,
      "step": 19551
    },
    {
      "epoch": 0.13284306460701711,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009614338134650433,
      "loss": 2.2041,
      "step": 19552
    },
    {
      "epoch": 0.13284985895361118,
      "grad_norm": 0.134765625,
      "learning_rate": 0.00096142967004121,
      "loss": 2.1555,
      "step": 19553
    },
    {
      "epoch": 0.13285665330020524,
      "grad_norm": 0.138671875,
      "learning_rate": 0.00096142552640374,
      "loss": 2.2518,
      "step": 19554
    },
    {
      "epoch": 0.13286344764679928,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009614213825526352,
      "loss": 2.2766,
      "step": 19555
    },
    {
      "epoch": 0.13287024199339334,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009614172384878976,
      "loss": 2.1145,
      "step": 19556
    },
    {
      "epoch": 0.1328770363399874,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009614130942095292,
      "loss": 2.1598,
      "step": 19557
    },
    {
      "epoch": 0.13288383068658147,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009614089497175319,
      "loss": 2.2034,
      "step": 19558
    },
    {
      "epoch": 0.13289062503317553,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009614048050119075,
      "loss": 1.9383,
      "step": 19559
    },
    {
      "epoch": 0.13289741937976957,
      "grad_norm": 0.12255859375,
      "learning_rate": 0.0009614006600926579,
      "loss": 2.1333,
      "step": 19560
    },
    {
      "epoch": 0.13290421372636363,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009613965149597851,
      "loss": 2.2977,
      "step": 19561
    },
    {
      "epoch": 0.1329110080729577,
      "grad_norm": 0.125,
      "learning_rate": 0.000961392369613291,
      "loss": 2.1852,
      "step": 19562
    },
    {
      "epoch": 0.13291780241955176,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009613882240531775,
      "loss": 2.322,
      "step": 19563
    },
    {
      "epoch": 0.13292459676614582,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009613840782794466,
      "loss": 2.1048,
      "step": 19564
    },
    {
      "epoch": 0.13293139111273986,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009613799322921002,
      "loss": 2.2104,
      "step": 19565
    },
    {
      "epoch": 0.13293818545933392,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009613757860911401,
      "loss": 2.2397,
      "step": 19566
    },
    {
      "epoch": 0.13294497980592798,
      "grad_norm": 0.12353515625,
      "learning_rate": 0.0009613716396765684,
      "loss": 2.1788,
      "step": 19567
    },
    {
      "epoch": 0.13295177415252205,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009613674930483867,
      "loss": 2.1043,
      "step": 19568
    },
    {
      "epoch": 0.1329585684991161,
      "grad_norm": 0.201171875,
      "learning_rate": 0.0009613633462065974,
      "loss": 2.183,
      "step": 19569
    },
    {
      "epoch": 0.13296536284571014,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.000961359199151202,
      "loss": 2.1797,
      "step": 19570
    },
    {
      "epoch": 0.1329721571923042,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009613550518822025,
      "loss": 2.1963,
      "step": 19571
    },
    {
      "epoch": 0.13297895153889827,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009613509043996011,
      "loss": 2.1626,
      "step": 19572
    },
    {
      "epoch": 0.13298574588549233,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009613467567033995,
      "loss": 2.1476,
      "step": 19573
    },
    {
      "epoch": 0.1329925402320864,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009613426087935994,
      "loss": 2.0819,
      "step": 19574
    },
    {
      "epoch": 0.13299933457868043,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009613384606702032,
      "loss": 2.2303,
      "step": 19575
    },
    {
      "epoch": 0.1330061289252745,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009613343123332126,
      "loss": 2.2197,
      "step": 19576
    },
    {
      "epoch": 0.13301292327186856,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009613301637826295,
      "loss": 2.2358,
      "step": 19577
    },
    {
      "epoch": 0.13301971761846262,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009613260150184557,
      "loss": 2.1329,
      "step": 19578
    },
    {
      "epoch": 0.1330265119650567,
      "grad_norm": 0.15234375,
      "learning_rate": 0.0009613218660406933,
      "loss": 2.2567,
      "step": 19579
    },
    {
      "epoch": 0.13303330631165072,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009613177168493442,
      "loss": 2.223,
      "step": 19580
    },
    {
      "epoch": 0.1330401006582448,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009613135674444102,
      "loss": 2.176,
      "step": 19581
    },
    {
      "epoch": 0.13304689500483885,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009613094178258933,
      "loss": 2.2618,
      "step": 19582
    },
    {
      "epoch": 0.1330536893514329,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009613052679937954,
      "loss": 2.2152,
      "step": 19583
    },
    {
      "epoch": 0.13306048369802698,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009613011179481186,
      "loss": 2.2249,
      "step": 19584
    },
    {
      "epoch": 0.133067278044621,
      "grad_norm": 0.146484375,
      "learning_rate": 0.0009612969676888645,
      "loss": 2.0969,
      "step": 19585
    },
    {
      "epoch": 0.13307407239121508,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009612928172160354,
      "loss": 2.0935,
      "step": 19586
    },
    {
      "epoch": 0.13308086673780914,
      "grad_norm": 0.134765625,
      "learning_rate": 0.000961288666529633,
      "loss": 2.2042,
      "step": 19587
    },
    {
      "epoch": 0.1330876610844032,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.000961284515629659,
      "loss": 2.1499,
      "step": 19588
    },
    {
      "epoch": 0.13309445543099727,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009612803645161158,
      "loss": 2.2572,
      "step": 19589
    },
    {
      "epoch": 0.1331012497775913,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009612762131890049,
      "loss": 2.0741,
      "step": 19590
    },
    {
      "epoch": 0.13310804412418537,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009612720616483286,
      "loss": 2.1945,
      "step": 19591
    },
    {
      "epoch": 0.13311483847077943,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009612679098940884,
      "loss": 2.1484,
      "step": 19592
    },
    {
      "epoch": 0.1331216328173735,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009612637579262866,
      "loss": 2.2372,
      "step": 19593
    },
    {
      "epoch": 0.13312842716396756,
      "grad_norm": 0.1318359375,
      "learning_rate": 0.0009612596057449249,
      "loss": 2.2005,
      "step": 19594
    },
    {
      "epoch": 0.1331352215105616,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009612554533500053,
      "loss": 2.1569,
      "step": 19595
    },
    {
      "epoch": 0.13314201585715565,
      "grad_norm": 0.1279296875,
      "learning_rate": 0.0009612513007415298,
      "loss": 2.1175,
      "step": 19596
    },
    {
      "epoch": 0.13314881020374972,
      "grad_norm": 0.14453125,
      "learning_rate": 0.0009612471479195001,
      "loss": 2.1786,
      "step": 19597
    },
    {
      "epoch": 0.13315560455034378,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009612429948839183,
      "loss": 2.0535,
      "step": 19598
    },
    {
      "epoch": 0.13316239889693782,
      "grad_norm": 0.169921875,
      "learning_rate": 0.0009612388416347864,
      "loss": 2.2781,
      "step": 19599
    },
    {
      "epoch": 0.13316919324353188,
      "grad_norm": 0.126953125,
      "learning_rate": 0.000961234688172106,
      "loss": 2.1003,
      "step": 19600
    },
    {
      "epoch": 0.13317598759012594,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009612305344958794,
      "loss": 2.2265,
      "step": 19601
    },
    {
      "epoch": 0.13318278193672,
      "grad_norm": 0.1435546875,
      "learning_rate": 0.0009612263806061082,
      "loss": 2.1287,
      "step": 19602
    },
    {
      "epoch": 0.13318957628331407,
      "grad_norm": 0.142578125,
      "learning_rate": 0.0009612222265027946,
      "loss": 2.0667,
      "step": 19603
    },
    {
      "epoch": 0.1331963706299081,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009612180721859403,
      "loss": 2.1685,
      "step": 19604
    },
    {
      "epoch": 0.13320316497650217,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009612139176555474,
      "loss": 2.1972,
      "step": 19605
    },
    {
      "epoch": 0.13320995932309623,
      "grad_norm": 0.154296875,
      "learning_rate": 0.0009612097629116176,
      "loss": 2.318,
      "step": 19606
    },
    {
      "epoch": 0.1332167536696903,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009612056079541531,
      "loss": 2.0179,
      "step": 19607
    },
    {
      "epoch": 0.13322354801628436,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009612014527831559,
      "loss": 2.1103,
      "step": 19608
    },
    {
      "epoch": 0.1332303423628784,
      "grad_norm": 0.13671875,
      "learning_rate": 0.0009611972973986274,
      "loss": 2.0811,
      "step": 19609
    },
    {
      "epoch": 0.13323713670947246,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009611931418005699,
      "loss": 2.2149,
      "step": 19610
    },
    {
      "epoch": 0.13324393105606652,
      "grad_norm": 0.1298828125,
      "learning_rate": 0.0009611889859889853,
      "loss": 2.1194,
      "step": 19611
    },
    {
      "epoch": 0.13325072540266059,
      "grad_norm": 0.150390625,
      "learning_rate": 0.0009611848299638754,
      "loss": 2.282,
      "step": 19612
    },
    {
      "epoch": 0.13325751974925465,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009611806737252424,
      "loss": 2.1935,
      "step": 19613
    },
    {
      "epoch": 0.13326431409584868,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009611765172730879,
      "loss": 2.1622,
      "step": 19614
    },
    {
      "epoch": 0.13327110844244275,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009611723606074139,
      "loss": 2.1172,
      "step": 19615
    },
    {
      "epoch": 0.1332779027890368,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009611682037282225,
      "loss": 2.2458,
      "step": 19616
    },
    {
      "epoch": 0.13328469713563088,
      "grad_norm": 0.1533203125,
      "learning_rate": 0.0009611640466355155,
      "loss": 2.1006,
      "step": 19617
    },
    {
      "epoch": 0.13329149148222494,
      "grad_norm": 0.126953125,
      "learning_rate": 0.0009611598893292947,
      "loss": 2.1235,
      "step": 19618
    },
    {
      "epoch": 0.13329828582881897,
      "grad_norm": 0.12890625,
      "learning_rate": 0.0009611557318095624,
      "loss": 2.1218,
      "step": 19619
    },
    {
      "epoch": 0.13330508017541304,
      "grad_norm": 0.158203125,
      "learning_rate": 0.0009611515740763202,
      "loss": 2.2262,
      "step": 19620
    },
    {
      "epoch": 0.1333118745220071,
      "grad_norm": 0.12451171875,
      "learning_rate": 0.0009611474161295699,
      "loss": 2.0362,
      "step": 19621
    },
    {
      "epoch": 0.13331866886860116,
      "grad_norm": 0.134765625,
      "learning_rate": 0.0009611432579693138,
      "loss": 2.2256,
      "step": 19622
    },
    {
      "epoch": 0.13332546321519523,
      "grad_norm": 0.1416015625,
      "learning_rate": 0.0009611390995955538,
      "loss": 2.29,
      "step": 19623
    },
    {
      "epoch": 0.13333225756178926,
      "grad_norm": 0.1259765625,
      "learning_rate": 0.0009611349410082915,
      "loss": 2.094,
      "step": 19624
    },
    {
      "epoch": 0.13333905190838333,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.000961130782207529,
      "loss": 2.1887,
      "step": 19625
    },
    {
      "epoch": 0.1333458462549774,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009611266231932682,
      "loss": 2.1745,
      "step": 19626
    },
    {
      "epoch": 0.13335264060157145,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009611224639655112,
      "loss": 2.1658,
      "step": 19627
    },
    {
      "epoch": 0.13335943494816552,
      "grad_norm": 0.140625,
      "learning_rate": 0.0009611183045242597,
      "loss": 2.3702,
      "step": 19628
    },
    {
      "epoch": 0.13336622929475955,
      "grad_norm": 0.130859375,
      "learning_rate": 0.0009611141448695158,
      "loss": 2.2216,
      "step": 19629
    },
    {
      "epoch": 0.13337302364135362,
      "grad_norm": 0.138671875,
      "learning_rate": 0.0009611099850012812,
      "loss": 2.1469,
      "step": 19630
    },
    {
      "epoch": 0.13337981798794768,
      "grad_norm": 0.1328125,
      "learning_rate": 0.0009611058249195581,
      "loss": 2.1572,
      "step": 19631
    },
    {
      "epoch": 0.13338661233454174,
      "grad_norm": 0.1357421875,
      "learning_rate": 0.0009611016646243482,
      "loss": 2.0882,
      "step": 19632
    },
    {
      "epoch": 0.1333934066811358,
      "grad_norm": 0.1337890625,
      "learning_rate": 0.0009610975041156536,
      "loss": 2.2617,
      "step": 19633
    },
    {
      "epoch": 0.13340020102772984,
      "grad_norm": 0.1396484375,
      "learning_rate": 0.0009610933433934761,
      "loss": 2.2194,
      "step": 19634
    },
    {
      "epoch": 0.1334069953743239,
      "grad_norm": 0.1376953125,
      "learning_rate": 0.0009610891824578177,
      "loss": 2.2378,
      "step": 19635
    }
  ],
  "logging_steps": 1,
  "max_steps": 147182,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1472,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 7.782993014934417e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}