{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 280,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.025,
      "grad_norm": 2.1026898827355764,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.6621,
      "step": 1
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6413522165034502,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.7318,
      "step": 2
    },
    {
      "epoch": 0.075,
      "grad_norm": 0.6085035314989437,
      "learning_rate": 2.5e-06,
      "loss": 1.6973,
      "step": 3
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6097206000487838,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.7155,
      "step": 4
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.6181111292344054,
      "learning_rate": 4.166666666666667e-06,
      "loss": 1.7199,
      "step": 5
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6095519945061052,
      "learning_rate": 5e-06,
      "loss": 1.7225,
      "step": 6
    },
    {
      "epoch": 0.175,
      "grad_norm": 0.5993146111892984,
      "learning_rate": 5.833333333333334e-06,
      "loss": 1.6788,
      "step": 7
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6168594833390011,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.6941,
      "step": 8
    },
    {
      "epoch": 0.225,
      "grad_norm": 0.649510684690767,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.7237,
      "step": 9
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6127108674374603,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.6575,
      "step": 10
    },
    {
      "epoch": 0.275,
      "grad_norm": 0.6444847916730025,
      "learning_rate": 9.166666666666666e-06,
      "loss": 1.6477,
      "step": 11
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6274944995465307,
      "learning_rate": 1e-05,
      "loss": 1.6443,
      "step": 12
    },
    {
      "epoch": 0.325,
      "grad_norm": 0.6543139619848919,
      "learning_rate": 9.999836101895782e-06,
      "loss": 1.6202,
      "step": 13
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8235053540937074,
      "learning_rate": 9.999344418328161e-06,
      "loss": 1.6082,
      "step": 14
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.9775219557515483,
      "learning_rate": 9.99852498153154e-06,
      "loss": 1.5583,
      "step": 15
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1030636654770976,
      "learning_rate": 9.997377845227577e-06,
      "loss": 1.5679,
      "step": 16
    },
    {
      "epoch": 0.425,
      "grad_norm": 1.533566842443126,
      "learning_rate": 9.995903084621651e-06,
      "loss": 1.5359,
      "step": 17
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.718620417025083,
      "learning_rate": 9.994100796397954e-06,
      "loss": 1.4736,
      "step": 18
    },
    {
      "epoch": 0.475,
      "grad_norm": 1.848832331677319,
      "learning_rate": 9.991971098713136e-06,
      "loss": 1.4416,
      "step": 19
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1378030725614972,
      "learning_rate": 9.98951413118856e-06,
      "loss": 1.4385,
      "step": 20
    },
    {
      "epoch": 0.525,
      "grad_norm": 0.7491289081750082,
      "learning_rate": 9.986730054901153e-06,
      "loss": 1.3793,
      "step": 21
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6818788464479123,
      "learning_rate": 9.983619052372847e-06,
      "loss": 1.368,
      "step": 22
    },
    {
      "epoch": 0.575,
      "grad_norm": 0.6341413849629475,
      "learning_rate": 9.98018132755861e-06,
      "loss": 1.3467,
      "step": 23
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5923369379348106,
      "learning_rate": 9.97641710583307e-06,
      "loss": 1.3555,
      "step": 24
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.5340471860637729,
      "learning_rate": 9.972326633975754e-06,
      "loss": 1.349,
      "step": 25
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4785970184408438,
      "learning_rate": 9.96791018015489e-06,
      "loss": 1.3337,
      "step": 26
    },
    {
      "epoch": 0.675,
      "grad_norm": 0.4414536685738152,
      "learning_rate": 9.963168033909843e-06,
      "loss": 1.3334,
      "step": 27
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4860929708817065,
      "learning_rate": 9.958100506132127e-06,
      "loss": 1.3159,
      "step": 28
    },
    {
      "epoch": 0.725,
      "grad_norm": 0.45181979183711657,
      "learning_rate": 9.952707929045018e-06,
      "loss": 1.3231,
      "step": 29
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4400284453917975,
      "learning_rate": 9.946990656181782e-06,
      "loss": 1.2993,
      "step": 30
    },
    {
      "epoch": 0.775,
      "grad_norm": 0.38337739876412114,
      "learning_rate": 9.94094906236249e-06,
      "loss": 1.2952,
      "step": 31
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37181863514378,
      "learning_rate": 9.934583543669454e-06,
      "loss": 1.3109,
      "step": 32
    },
    {
      "epoch": 0.825,
      "grad_norm": 0.3401881416689356,
      "learning_rate": 9.927894517421252e-06,
      "loss": 1.3157,
      "step": 33
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3239891568707606,
      "learning_rate": 9.920882422145372e-06,
      "loss": 1.34,
      "step": 34
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.3033475087643797,
      "learning_rate": 9.913547717549463e-06,
      "loss": 1.283,
      "step": 35
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2888152197430936,
      "learning_rate": 9.905890884491196e-06,
      "loss": 1.2892,
      "step": 36
    },
    {
      "epoch": 0.925,
      "grad_norm": 0.28458390089996766,
      "learning_rate": 9.897912424946741e-06,
      "loss": 1.2646,
      "step": 37
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.27339498391168426,
      "learning_rate": 9.889612861977855e-06,
      "loss": 1.2832,
      "step": 38
    },
    {
      "epoch": 0.975,
      "grad_norm": 0.27940221076287475,
      "learning_rate": 9.88099273969759e-06,
      "loss": 1.2422,
      "step": 39
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.27720461090670556,
      "learning_rate": 9.872052623234632e-06,
      "loss": 1.2919,
      "step": 40
    },
    {
      "epoch": 1.025,
      "grad_norm": 0.25894847092459033,
      "learning_rate": 9.86279309869623e-06,
      "loss": 1.2494,
      "step": 41
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.2824528364474668,
      "learning_rate": 9.853214773129796e-06,
      "loss": 1.2404,
      "step": 42
    },
    {
      "epoch": 1.075,
      "grad_norm": 0.2731217503169999,
      "learning_rate": 9.843318274483089e-06,
      "loss": 1.2762,
      "step": 43
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.282165604906485,
      "learning_rate": 9.833104251563058e-06,
      "loss": 1.2373,
      "step": 44
    },
    {
      "epoch": 1.125,
      "grad_norm": 0.27542846366926443,
      "learning_rate": 9.822573373993296e-06,
      "loss": 1.2467,
      "step": 45
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.2851918766199106,
      "learning_rate": 9.811726332170153e-06,
      "loss": 1.267,
      "step": 46
    },
    {
      "epoch": 1.175,
      "grad_norm": 0.2599062629698892,
      "learning_rate": 9.800563837217465e-06,
      "loss": 1.2585,
      "step": 47
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.2586420496872339,
      "learning_rate": 9.789086620939936e-06,
      "loss": 1.2173,
      "step": 48
    },
    {
      "epoch": 1.225,
      "grad_norm": 0.27264444944219546,
      "learning_rate": 9.777295435775165e-06,
      "loss": 1.1856,
      "step": 49
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.2648452730214859,
      "learning_rate": 9.765191054744305e-06,
      "loss": 1.2354,
      "step": 50
    },
    {
      "epoch": 1.275,
      "grad_norm": 0.27625402001557403,
      "learning_rate": 9.752774271401403e-06,
      "loss": 1.1857,
      "step": 51
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.29532784079485663,
      "learning_rate": 9.740045899781353e-06,
      "loss": 1.2058,
      "step": 52
    },
    {
      "epoch": 1.325,
      "grad_norm": 0.2701426980802868,
      "learning_rate": 9.727006774346551e-06,
      "loss": 1.1724,
      "step": 53
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.26735542553762237,
      "learning_rate": 9.713657749932172e-06,
      "loss": 1.194,
      "step": 54
    },
    {
      "epoch": 1.375,
      "grad_norm": 0.24830074105443878,
      "learning_rate": 9.699999701690133e-06,
      "loss": 1.1969,
      "step": 55
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2511171777018325,
      "learning_rate": 9.68603352503172e-06,
      "loss": 1.1822,
      "step": 56
    },
    {
      "epoch": 1.425,
      "grad_norm": 0.25711859387852454,
      "learning_rate": 9.671760135568882e-06,
      "loss": 1.1941,
      "step": 57
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.24983954826654237,
      "learning_rate": 9.657180469054213e-06,
      "loss": 1.1597,
      "step": 58
    },
    {
      "epoch": 1.475,
      "grad_norm": 0.2507740474880207,
      "learning_rate": 9.64229548131959e-06,
      "loss": 1.1631,
      "step": 59
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.23964630893140215,
      "learning_rate": 9.627106148213521e-06,
      "loss": 1.173,
      "step": 60
    },
    {
      "epoch": 1.525,
      "grad_norm": 0.22971666963865991,
      "learning_rate": 9.61161346553717e-06,
      "loss": 1.1446,
      "step": 61
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.22847059454081425,
      "learning_rate": 9.595818448979061e-06,
      "loss": 1.1692,
      "step": 62
    },
    {
      "epoch": 1.575,
      "grad_norm": 0.2544697716980138,
      "learning_rate": 9.579722134048506e-06,
      "loss": 1.1712,
      "step": 63
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.2248726099006599,
      "learning_rate": 9.563325576007702e-06,
      "loss": 1.1523,
      "step": 64
    },
    {
      "epoch": 1.625,
      "grad_norm": 0.23257460453519807,
      "learning_rate": 9.546629849802562e-06,
      "loss": 1.146,
      "step": 65
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.2400345523124048,
      "learning_rate": 9.529636049992235e-06,
      "loss": 1.1238,
      "step": 66
    },
    {
      "epoch": 1.675,
      "grad_norm": 0.2410972327939822,
      "learning_rate": 9.51234529067735e-06,
      "loss": 1.1745,
      "step": 67
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.2395032264738681,
      "learning_rate": 9.494758705426978e-06,
      "loss": 1.156,
      "step": 68
    },
    {
      "epoch": 1.725,
      "grad_norm": 0.23423115235391723,
      "learning_rate": 9.476877447204308e-06,
      "loss": 1.1413,
      "step": 69
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.24379065472217767,
      "learning_rate": 9.458702688291072e-06,
      "loss": 1.1677,
      "step": 70
    },
    {
      "epoch": 1.775,
      "grad_norm": 0.24016354424050318,
      "learning_rate": 9.440235620210682e-06,
      "loss": 1.1235,
      "step": 71
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.25909751239656786,
      "learning_rate": 9.421477453650118e-06,
      "loss": 1.1139,
      "step": 72
    },
    {
      "epoch": 1.825,
      "grad_norm": 0.24033129150490448,
      "learning_rate": 9.402429418380555e-06,
      "loss": 1.1582,
      "step": 73
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.2413776856942996,
      "learning_rate": 9.38309276317674e-06,
      "loss": 1.1259,
      "step": 74
    },
    {
      "epoch": 1.875,
      "grad_norm": 0.24663210297989538,
      "learning_rate": 9.363468755735123e-06,
      "loss": 1.0843,
      "step": 75
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.2469034876268094,
      "learning_rate": 9.343558682590757e-06,
      "loss": 1.1077,
      "step": 76
    },
    {
      "epoch": 1.925,
      "grad_norm": 0.25810333557343196,
      "learning_rate": 9.323363849032933e-06,
      "loss": 1.1145,
      "step": 77
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.2457214465581943,
      "learning_rate": 9.302885579019626e-06,
      "loss": 1.119,
      "step": 78
    },
    {
      "epoch": 1.975,
      "grad_norm": 0.2530172029554143,
      "learning_rate": 9.282125215090694e-06,
      "loss": 1.1324,
      "step": 79
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.26171895830403796,
      "learning_rate": 9.261084118279846e-06,
      "loss": 1.0914,
      "step": 80
    },
    {
      "epoch": 2.025,
      "grad_norm": 0.2596448526704054,
      "learning_rate": 9.23976366802544e-06,
      "loss": 1.1202,
      "step": 81
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.2575246492390129,
      "learning_rate": 9.218165262080024e-06,
      "loss": 1.1191,
      "step": 82
    },
    {
      "epoch": 2.075,
      "grad_norm": 0.24809186590262497,
      "learning_rate": 9.196290316418712e-06,
      "loss": 1.0852,
      "step": 83
    },
    {
      "epoch": 2.1,
      "grad_norm": 0.25529825764524233,
      "learning_rate": 9.174140265146356e-06,
      "loss": 1.1091,
      "step": 84
    },
    {
      "epoch": 2.125,
      "grad_norm": 0.2598333152137317,
      "learning_rate": 9.151716560403521e-06,
      "loss": 1.0512,
      "step": 85
    },
    {
      "epoch": 2.15,
      "grad_norm": 0.2533838531424551,
      "learning_rate": 9.129020672271283e-06,
      "loss": 1.0862,
      "step": 86
    },
    {
      "epoch": 2.175,
      "grad_norm": 0.2504337844095705,
      "learning_rate": 9.106054088674861e-06,
      "loss": 1.0717,
      "step": 87
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.25443455111995733,
      "learning_rate": 9.082818315286054e-06,
      "loss": 1.0658,
      "step": 88
    },
    {
      "epoch": 2.225,
      "grad_norm": 0.26694514174541256,
      "learning_rate": 9.059314875424553e-06,
      "loss": 1.0228,
      "step": 89
    },
    {
      "epoch": 2.25,
      "grad_norm": 0.26148960189209175,
      "learning_rate": 9.035545309958048e-06,
      "loss": 1.052,
      "step": 90
    },
    {
      "epoch": 2.275,
      "grad_norm": 0.27532917589486194,
      "learning_rate": 9.011511177201226e-06,
      "loss": 1.07,
      "step": 91
    },
    {
      "epoch": 2.3,
      "grad_norm": 0.28730068505165196,
      "learning_rate": 8.987214052813605e-06,
      "loss": 1.0609,
      "step": 92
    },
    {
      "epoch": 2.325,
      "grad_norm": 0.2698216580724319,
      "learning_rate": 8.962655529696236e-06,
      "loss": 1.0579,
      "step": 93
    },
    {
      "epoch": 2.35,
      "grad_norm": 0.24786365341013725,
      "learning_rate": 8.937837217887273e-06,
      "loss": 1.0452,
      "step": 94
    },
    {
      "epoch": 2.375,
      "grad_norm": 0.24197792160968903,
      "learning_rate": 8.912760744456416e-06,
      "loss": 1.0346,
      "step": 95
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.25871618240180544,
      "learning_rate": 8.887427753398249e-06,
      "loss": 1.0452,
      "step": 96
    },
    {
      "epoch": 2.425,
      "grad_norm": 0.2670995555409698,
      "learning_rate": 8.861839905524452e-06,
      "loss": 1.004,
      "step": 97
    },
    {
      "epoch": 2.45,
      "grad_norm": 0.26240119221560043,
      "learning_rate": 8.83599887835493e-06,
      "loss": 1.0381,
      "step": 98
    },
    {
      "epoch": 2.475,
      "grad_norm": 0.25007005933407084,
      "learning_rate": 8.809906366007832e-06,
      "loss": 1.0237,
      "step": 99
    },
    {
      "epoch": 2.5,
      "grad_norm": 0.27210733980893814,
      "learning_rate": 8.783564079088478e-06,
      "loss": 1.0326,
      "step": 100
    },
    {
      "epoch": 2.525,
      "grad_norm": 0.2778485282051736,
      "learning_rate": 8.756973744577221e-06,
      "loss": 1.0213,
      "step": 101
    },
    {
      "epoch": 2.55,
      "grad_norm": 0.2709595058080484,
      "learning_rate": 8.730137105716231e-06,
      "loss": 1.0046,
      "step": 102
    },
    {
      "epoch": 2.575,
      "grad_norm": 0.2723123534955665,
      "learning_rate": 8.7030559218952e-06,
      "loss": 1.0572,
      "step": 103
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.26318374858426913,
      "learning_rate": 8.675731968536004e-06,
      "loss": 1.0142,
      "step": 104
    },
    {
      "epoch": 2.625,
      "grad_norm": 0.26423478733108197,
      "learning_rate": 8.648167036976303e-06,
      "loss": 1.0083,
      "step": 105
    },
    {
      "epoch": 2.65,
      "grad_norm": 0.2527490323367728,
      "learning_rate": 8.620362934352109e-06,
      "loss": 1.0152,
      "step": 106
    },
    {
      "epoch": 2.675,
      "grad_norm": 0.2817733909459089,
      "learning_rate": 8.592321483479305e-06,
      "loss": 0.9994,
      "step": 107
    },
    {
      "epoch": 2.7,
      "grad_norm": 0.25088178573538616,
      "learning_rate": 8.564044522734147e-06,
      "loss": 0.966,
      "step": 108
    },
    {
      "epoch": 2.725,
      "grad_norm": 0.25165839780166765,
      "learning_rate": 8.535533905932739e-06,
      "loss": 0.9897,
      "step": 109
    },
    {
      "epoch": 2.75,
      "grad_norm": 0.27123004373609994,
      "learning_rate": 8.506791502209497e-06,
      "loss": 1.0011,
      "step": 110
    },
    {
      "epoch": 2.775,
      "grad_norm": 0.25885979274531185,
      "learning_rate": 8.477819195894615e-06,
      "loss": 1.0375,
      "step": 111
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.25413119834372133,
      "learning_rate": 8.448618886390523e-06,
      "loss": 0.974,
      "step": 112
    },
    {
      "epoch": 2.825,
      "grad_norm": 0.26932379526063627,
      "learning_rate": 8.419192488047369e-06,
      "loss": 0.9948,
      "step": 113
    },
    {
      "epoch": 2.85,
      "grad_norm": 0.26954131129733855,
      "learning_rate": 8.389541930037516e-06,
      "loss": 1.0171,
      "step": 114
    },
    {
      "epoch": 2.875,
      "grad_norm": 0.2667631834386252,
      "learning_rate": 8.359669156229062e-06,
      "loss": 0.9917,
      "step": 115
    },
    {
      "epoch": 2.9,
      "grad_norm": 0.2875114899558342,
      "learning_rate": 8.329576125058406e-06,
      "loss": 0.9758,
      "step": 116
    },
    {
      "epoch": 2.925,
      "grad_norm": 0.28675332351961746,
      "learning_rate": 8.29926480940185e-06,
      "loss": 0.963,
      "step": 117
    },
    {
      "epoch": 2.95,
      "grad_norm": 0.3013638054236219,
      "learning_rate": 8.268737196446264e-06,
      "loss": 0.9859,
      "step": 118
    },
    {
      "epoch": 2.975,
      "grad_norm": 0.30601615118396874,
      "learning_rate": 8.237995287558801e-06,
      "loss": 0.9685,
      "step": 119
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.30346948632781146,
      "learning_rate": 8.207041098155701e-06,
      "loss": 0.9872,
      "step": 120
    },
    {
      "epoch": 3.025,
      "grad_norm": 0.31531631458437354,
      "learning_rate": 8.175876657570142e-06,
      "loss": 0.964,
      "step": 121
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.29114779511016237,
      "learning_rate": 8.144504008919224e-06,
      "loss": 0.9655,
      "step": 122
    },
    {
      "epoch": 3.075,
      "grad_norm": 0.32729676853704537,
      "learning_rate": 8.112925208969996e-06,
      "loss": 0.9484,
      "step": 123
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.2968428193830138,
      "learning_rate": 8.081142328004638e-06,
      "loss": 0.9885,
      "step": 124
    },
    {
      "epoch": 3.125,
      "grad_norm": 0.31850537244671423,
      "learning_rate": 8.049157449684723e-06,
      "loss": 0.9593,
      "step": 125
    },
    {
      "epoch": 3.15,
      "grad_norm": 0.30145179934590266,
      "learning_rate": 8.016972670914624e-06,
      "loss": 0.9443,
      "step": 126
    },
    {
      "epoch": 3.175,
      "grad_norm": 0.32941902019125735,
      "learning_rate": 7.984590101704026e-06,
      "loss": 0.9411,
      "step": 127
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.33230400688084355,
      "learning_rate": 7.952011865029614e-06,
      "loss": 0.9353,
      "step": 128
    },
    {
      "epoch": 3.225,
      "grad_norm": 0.31073576739762815,
      "learning_rate": 7.919240096695877e-06,
      "loss": 0.9403,
      "step": 129
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.316466603610876,
      "learning_rate": 7.886276945195098e-06,
      "loss": 0.9575,
      "step": 130
    },
    {
      "epoch": 3.275,
      "grad_norm": 0.3252873314387951,
      "learning_rate": 7.853124571566491e-06,
      "loss": 0.9476,
      "step": 131
    },
    {
      "epoch": 3.3,
      "grad_norm": 0.34237602645794485,
      "learning_rate": 7.819785149254534e-06,
      "loss": 0.8938,
      "step": 132
    },
    {
      "epoch": 3.325,
      "grad_norm": 0.3386834141148311,
      "learning_rate": 7.786260863966468e-06,
      "loss": 0.8937,
      "step": 133
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.32892426523618096,
      "learning_rate": 7.752553913529019e-06,
      "loss": 0.9226,
      "step": 134
    },
    {
      "epoch": 3.375,
      "grad_norm": 0.3663645071149253,
      "learning_rate": 7.718666507744293e-06,
      "loss": 0.9331,
      "step": 135
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.35492405802228927,
      "learning_rate": 7.68460086824492e-06,
      "loss": 0.9024,
      "step": 136
    },
    {
      "epoch": 3.425,
      "grad_norm": 0.3461059129442488,
      "learning_rate": 7.65035922834839e-06,
      "loss": 0.9118,
      "step": 137
    },
    {
      "epoch": 3.45,
      "grad_norm": 0.3674770015752456,
      "learning_rate": 7.61594383291065e-06,
      "loss": 0.885,
      "step": 138
    },
    {
      "epoch": 3.475,
      "grad_norm": 0.29868653659173194,
      "learning_rate": 7.58135693817893e-06,
      "loss": 0.8795,
      "step": 139
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.396103206356086,
      "learning_rate": 7.546600811643816e-06,
      "loss": 0.9057,
      "step": 140
    },
    {
      "epoch": 3.525,
      "grad_norm": 0.4091265454754814,
      "learning_rate": 7.5116777318906105e-06,
      "loss": 0.9045,
      "step": 141
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.34738555176640296,
      "learning_rate": 7.476589988449939e-06,
      "loss": 0.897,
      "step": 142
    },
    {
      "epoch": 3.575,
      "grad_norm": 0.3381371740638481,
      "learning_rate": 7.441339881647654e-06,
      "loss": 0.9232,
      "step": 143
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.35909644765143234,
      "learning_rate": 7.405929722454026e-06,
      "loss": 0.8787,
      "step": 144
    },
    {
      "epoch": 3.625,
      "grad_norm": 0.3825665161255983,
      "learning_rate": 7.370361832332241e-06,
      "loss": 0.9113,
      "step": 145
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.30851922400853754,
      "learning_rate": 7.334638543086203e-06,
      "loss": 0.88,
      "step": 146
    },
    {
      "epoch": 3.675,
      "grad_norm": 0.37247312676947136,
      "learning_rate": 7.298762196707668e-06,
      "loss": 0.874,
      "step": 147
    },
    {
      "epoch": 3.7,
      "grad_norm": 0.38511911071863114,
      "learning_rate": 7.262735145222696e-06,
      "loss": 0.8783,
      "step": 148
    },
    {
      "epoch": 3.725,
      "grad_norm": 0.45755114413180425,
      "learning_rate": 7.226559750537462e-06,
      "loss": 0.884,
      "step": 149
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.30880788937024484,
      "learning_rate": 7.190238384283413e-06,
      "loss": 0.9085,
      "step": 150
    },
    {
      "epoch": 3.775,
      "grad_norm": 0.33118326548212185,
      "learning_rate": 7.1537734276617744e-06,
      "loss": 0.8671,
      "step": 151
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.41391364228528094,
      "learning_rate": 7.117167271287453e-06,
      "loss": 0.8695,
      "step": 152
    },
    {
      "epoch": 3.825,
      "grad_norm": 0.41995379444939823,
      "learning_rate": 7.080422315032298e-06,
      "loss": 0.8989,
      "step": 153
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.33781129667429866,
      "learning_rate": 7.043540967867782e-06,
      "loss": 0.8927,
      "step": 154
    },
    {
      "epoch": 3.875,
      "grad_norm": 0.37184734206990244,
      "learning_rate": 7.006525647707054e-06,
      "loss": 0.8986,
      "step": 155
    },
    {
      "epoch": 3.9,
      "grad_norm": 0.42831001354188236,
      "learning_rate": 6.969378781246436e-06,
      "loss": 0.8899,
      "step": 156
    },
    {
      "epoch": 3.925,
      "grad_norm": 0.37314784171511645,
      "learning_rate": 6.932102803806324e-06,
      "loss": 0.857,
      "step": 157
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.38265810168742836,
      "learning_rate": 6.894700159171535e-06,
      "loss": 0.8423,
      "step": 158
    },
    {
      "epoch": 3.975,
      "grad_norm": 0.4345836191641771,
      "learning_rate": 6.857173299431084e-06,
      "loss": 0.8487,
      "step": 159
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.28712745976233395,
      "learning_rate": 6.819524684817439e-06,
      "loss": 0.8494,
      "step": 160
    },
    {
      "epoch": 4.025,
      "grad_norm": 0.363373067735039,
      "learning_rate": 6.781756783545225e-06,
      "loss": 0.8516,
      "step": 161
    },
    {
      "epoch": 4.05,
      "grad_norm": 0.38168917727729257,
      "learning_rate": 6.743872071649411e-06,
      "loss": 0.8882,
      "step": 162
    },
    {
      "epoch": 4.075,
      "grad_norm": 0.39729085134853515,
      "learning_rate": 6.705873032822981e-06,
      "loss": 0.8337,
      "step": 163
    },
    {
      "epoch": 4.1,
      "grad_norm": 0.37813465835851745,
      "learning_rate": 6.667762158254104e-06,
      "loss": 0.8309,
      "step": 164
    },
    {
      "epoch": 4.125,
      "grad_norm": 0.3211433628482485,
      "learning_rate": 6.629541946462816e-06,
      "loss": 0.8527,
      "step": 165
    },
    {
      "epoch": 4.15,
      "grad_norm": 0.3547685084940225,
      "learning_rate": 6.591214903137221e-06,
      "loss": 0.8621,
      "step": 166
    },
    {
      "epoch": 4.175,
      "grad_norm": 0.33781250869441176,
      "learning_rate": 6.552783540969212e-06,
      "loss": 0.8455,
      "step": 167
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.30993567341499995,
      "learning_rate": 6.514250379489754e-06,
      "loss": 0.8417,
      "step": 168
    },
    {
      "epoch": 4.225,
      "grad_norm": 0.33582932826059514,
      "learning_rate": 6.475617944903693e-06,
      "loss": 0.8326,
      "step": 169
    },
    {
      "epoch": 4.25,
      "grad_norm": 0.46046634589220586,
      "learning_rate": 6.436888769924142e-06,
      "loss": 0.8663,
      "step": 170
    },
    {
      "epoch": 4.275,
      "grad_norm": 0.32228319665068555,
      "learning_rate": 6.398065393606445e-06,
      "loss": 0.8405,
      "step": 171
    },
    {
      "epoch": 4.3,
      "grad_norm": 0.36506022581982417,
      "learning_rate": 6.3591503611817155e-06,
      "loss": 0.8633,
      "step": 172
    },
    {
      "epoch": 4.325,
      "grad_norm": 0.35728088386983137,
      "learning_rate": 6.320146223889966e-06,
      "loss": 0.8451,
      "step": 173
    },
    {
      "epoch": 4.35,
      "grad_norm": 0.3389049900239143,
      "learning_rate": 6.281055538812861e-06,
      "loss": 0.8383,
      "step": 174
    },
    {
      "epoch": 4.375,
      "grad_norm": 0.35164642827686565,
      "learning_rate": 6.241880868706075e-06,
      "loss": 0.8436,
      "step": 175
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.45602132072592544,
      "learning_rate": 6.202624781831269e-06,
      "loss": 0.8234,
      "step": 176
    },
    {
      "epoch": 4.425,
      "grad_norm": 0.31307814205851997,
      "learning_rate": 6.163289851787732e-06,
      "loss": 0.8383,
      "step": 177
    },
    {
      "epoch": 4.45,
      "grad_norm": 0.3443353966575101,
      "learning_rate": 6.123878657343648e-06,
      "loss": 0.8057,
      "step": 178
    },
    {
      "epoch": 4.475,
      "grad_norm": 0.33895295083957905,
      "learning_rate": 6.084393782267039e-06,
      "loss": 0.8404,
      "step": 179
    },
    {
      "epoch": 4.5,
      "grad_norm": 0.35609060580352403,
      "learning_rate": 6.044837815156377e-06,
      "loss": 0.8219,
      "step": 180
    },
    {
      "epoch": 4.525,
      "grad_norm": 0.34688264086763054,
      "learning_rate": 6.005213349270864e-06,
      "loss": 0.8244,
      "step": 181
    },
    {
      "epoch": 4.55,
      "grad_norm": 0.34841190032285607,
      "learning_rate": 5.965522982360441e-06,
      "loss": 0.827,
      "step": 182
    },
    {
      "epoch": 4.575,
      "grad_norm": 0.41157723970826277,
      "learning_rate": 5.925769316495461e-06,
      "loss": 0.8233,
      "step": 183
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.2871326799596576,
      "learning_rate": 5.885954957896115e-06,
      "loss": 0.8096,
      "step": 184
    },
    {
      "epoch": 4.625,
      "grad_norm": 0.35012249585775307,
      "learning_rate": 5.846082516761558e-06,
      "loss": 0.7931,
      "step": 185
    },
    {
      "epoch": 4.65,
      "grad_norm": 0.46359796922380764,
      "learning_rate": 5.806154607098799e-06,
      "loss": 0.8283,
      "step": 186
    },
    {
      "epoch": 4.675,
      "grad_norm": 0.4258979061317278,
      "learning_rate": 5.766173846551316e-06,
      "loss": 0.8174,
      "step": 187
    },
    {
      "epoch": 4.7,
      "grad_norm": 0.3462495330359986,
      "learning_rate": 5.726142856227453e-06,
      "loss": 0.8023,
      "step": 188
    },
    {
      "epoch": 4.725,
      "grad_norm": 0.38828158563310117,
      "learning_rate": 5.686064260528577e-06,
      "loss": 0.8297,
      "step": 189
    },
    {
      "epoch": 4.75,
      "grad_norm": 0.33951477432576194,
      "learning_rate": 5.645940686977033e-06,
      "loss": 0.8246,
      "step": 190
    },
    {
      "epoch": 4.775,
      "grad_norm": 0.3095495826095541,
      "learning_rate": 5.6057747660438745e-06,
      "loss": 0.8129,
      "step": 191
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.4054254828876582,
      "learning_rate": 5.5655691309764225e-06,
      "loss": 0.8313,
      "step": 192
    },
    {
      "epoch": 4.825,
      "grad_norm": 0.3483399073445393,
      "learning_rate": 5.52532641762562e-06,
      "loss": 0.8129,
      "step": 193
    },
    {
      "epoch": 4.85,
      "grad_norm": 0.32176037839806315,
      "learning_rate": 5.485049264273241e-06,
      "loss": 0.7842,
      "step": 194
    },
    {
      "epoch": 4.875,
      "grad_norm": 0.29495279378063277,
      "learning_rate": 5.444740311458915e-06,
      "loss": 0.7841,
      "step": 195
    },
    {
      "epoch": 4.9,
      "grad_norm": 0.3484345853234177,
      "learning_rate": 5.404402201807022e-06,
      "loss": 0.8362,
      "step": 196
    },
    {
      "epoch": 4.925,
      "grad_norm": 0.32603753016594705,
      "learning_rate": 5.364037579853441e-06,
      "loss": 0.8055,
      "step": 197
    },
    {
      "epoch": 4.95,
      "grad_norm": 0.34113760948421906,
      "learning_rate": 5.323649091872179e-06,
      "loss": 0.7877,
      "step": 198
    },
    {
      "epoch": 4.975,
      "grad_norm": 0.33921943319034875,
      "learning_rate": 5.283239385701882e-06,
      "loss": 0.805,
      "step": 199
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.2951107290602008,
      "learning_rate": 5.242811110572243e-06,
      "loss": 0.7991,
      "step": 200
    },
    {
      "epoch": 5.025,
      "grad_norm": 0.30253566064261156,
      "learning_rate": 5.202366916930319e-06,
      "loss": 0.8044,
      "step": 201
    },
    {
      "epoch": 5.05,
      "grad_norm": 0.3230565909231367,
      "learning_rate": 5.161909456266781e-06,
      "loss": 0.7646,
      "step": 202
    },
    {
      "epoch": 5.075,
      "grad_norm": 0.3180667356505713,
      "learning_rate": 5.121441380942066e-06,
      "loss": 0.8159,
      "step": 203
    },
    {
      "epoch": 5.1,
      "grad_norm": 0.36374034176814063,
      "learning_rate": 5.080965344012509e-06,
      "loss": 0.7713,
      "step": 204
    },
    {
      "epoch": 5.125,
      "grad_norm": 0.30290892981437073,
      "learning_rate": 5.040483999056393e-06,
      "loss": 0.767,
      "step": 205
    },
    {
      "epoch": 5.15,
      "grad_norm": 0.3930116162781173,
      "learning_rate": 5e-06,
      "loss": 0.7916,
      "step": 206
    },
    {
      "epoch": 5.175,
      "grad_norm": 0.3185992766019151,
      "learning_rate": 4.9595160009436075e-06,
      "loss": 0.7996,
      "step": 207
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.38832542048067603,
      "learning_rate": 4.919034655987493e-06,
      "loss": 0.7878,
      "step": 208
    },
    {
      "epoch": 5.225,
      "grad_norm": 0.39110099000218684,
      "learning_rate": 4.8785586190579355e-06,
      "loss": 0.7728,
      "step": 209
    },
    {
      "epoch": 5.25,
      "grad_norm": 0.30581266996926304,
      "learning_rate": 4.838090543733222e-06,
      "loss": 0.7937,
      "step": 210
    },
    {
      "epoch": 5.275,
      "grad_norm": 0.31131833564138595,
      "learning_rate": 4.797633083069683e-06,
      "loss": 0.758,
      "step": 211
    },
    {
      "epoch": 5.3,
      "grad_norm": 0.3726348002148966,
      "learning_rate": 4.757188889427761e-06,
      "loss": 0.8192,
      "step": 212
    },
    {
      "epoch": 5.325,
      "grad_norm": 0.32124465048482537,
      "learning_rate": 4.716760614298118e-06,
      "loss": 0.8127,
      "step": 213
    },
    {
      "epoch": 5.35,
      "grad_norm": 0.4291943729340545,
      "learning_rate": 4.6763509081278215e-06,
      "loss": 0.7956,
      "step": 214
    },
    {
      "epoch": 5.375,
      "grad_norm": 0.28810585263359334,
      "learning_rate": 4.63596242014656e-06,
      "loss": 0.7876,
      "step": 215
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.36635836033703345,
      "learning_rate": 4.59559779819298e-06,
      "loss": 0.8118,
      "step": 216
    },
    {
      "epoch": 5.425,
      "grad_norm": 0.3989844143804462,
      "learning_rate": 4.555259688541087e-06,
      "loss": 0.7744,
      "step": 217
    },
    {
      "epoch": 5.45,
      "grad_norm": 0.3775475757163417,
      "learning_rate": 4.51495073572676e-06,
      "loss": 0.8076,
      "step": 218
    },
    {
      "epoch": 5.475,
      "grad_norm": 0.3885576211506495,
      "learning_rate": 4.474673582374381e-06,
      "loss": 0.8152,
      "step": 219
    },
    {
      "epoch": 5.5,
      "grad_norm": 0.5486181305427187,
      "learning_rate": 4.434430869023579e-06,
      "loss": 0.8035,
      "step": 220
    },
    {
      "epoch": 5.525,
      "grad_norm": 0.449133416014458,
      "learning_rate": 4.394225233956127e-06,
      "loss": 0.7581,
      "step": 221
    },
    {
      "epoch": 5.55,
      "grad_norm": 0.3532754327449149,
      "learning_rate": 4.3540593130229695e-06,
      "loss": 0.8145,
      "step": 222
    },
    {
      "epoch": 5.575,
      "grad_norm": 0.6472978242633896,
      "learning_rate": 4.313935739471425e-06,
      "loss": 0.7889,
      "step": 223
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.5593763414902034,
      "learning_rate": 4.27385714377255e-06,
      "loss": 0.8025,
      "step": 224
    },
    {
      "epoch": 5.625,
      "grad_norm": 0.3113555109979358,
      "learning_rate": 4.233826153448685e-06,
      "loss": 0.8235,
      "step": 225
    },
    {
      "epoch": 5.65,
      "grad_norm": 0.32783417993682323,
      "learning_rate": 4.1938453929012014e-06,
      "loss": 0.7839,
      "step": 226
    },
    {
      "epoch": 5.675,
      "grad_norm": 0.47735491344985437,
      "learning_rate": 4.1539174832384425e-06,
      "loss": 0.7673,
      "step": 227
    },
    {
      "epoch": 5.7,
      "grad_norm": 0.379167946256265,
      "learning_rate": 4.1140450421038865e-06,
      "loss": 0.7669,
      "step": 228
    },
    {
      "epoch": 5.725,
      "grad_norm": 0.4055179197901725,
      "learning_rate": 4.074230683504541e-06,
      "loss": 0.7915,
      "step": 229
    },
    {
      "epoch": 5.75,
      "grad_norm": 0.33990082918663317,
      "learning_rate": 4.034477017639561e-06,
      "loss": 0.7753,
      "step": 230
    },
    {
      "epoch": 5.775,
      "grad_norm": 0.3440010530347358,
      "learning_rate": 3.994786650729137e-06,
      "loss": 0.7834,
      "step": 231
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.4571564594873592,
      "learning_rate": 3.955162184843625e-06,
      "loss": 0.7843,
      "step": 232
    },
    {
      "epoch": 5.825,
      "grad_norm": 0.31044612074261246,
      "learning_rate": 3.9156062177329624e-06,
      "loss": 0.7821,
      "step": 233
    },
    {
      "epoch": 5.85,
      "grad_norm": 0.31765843548865946,
      "learning_rate": 3.8761213426563546e-06,
      "loss": 0.7648,
      "step": 234
    },
    {
      "epoch": 5.875,
      "grad_norm": 0.33542686760185636,
      "learning_rate": 3.836710148212271e-06,
      "loss": 0.7653,
      "step": 235
    },
    {
      "epoch": 5.9,
      "grad_norm": 0.34867235131286023,
      "learning_rate": 3.7973752181687336e-06,
      "loss": 0.7818,
      "step": 236
    },
    {
      "epoch": 5.925,
      "grad_norm": 0.42551356848317706,
      "learning_rate": 3.7581191312939256e-06,
      "loss": 0.791,
      "step": 237
    },
    {
      "epoch": 5.95,
      "grad_norm": 0.30823548064435274,
      "learning_rate": 3.7189444611871383e-06,
      "loss": 0.7892,
      "step": 238
    },
    {
      "epoch": 5.975,
      "grad_norm": 0.343637195813578,
      "learning_rate": 3.679853776110035e-06,
      "loss": 0.763,
      "step": 239
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.33568955543808604,
      "learning_rate": 3.6408496388182857e-06,
      "loss": 0.7701,
      "step": 240
    },
    {
      "epoch": 6.025,
      "grad_norm": 0.39497924967421005,
      "learning_rate": 3.601934606393556e-06,
      "loss": 0.7668,
      "step": 241
    },
    {
      "epoch": 6.05,
      "grad_norm": 0.3461417085165549,
      "learning_rate": 3.5631112300758595e-06,
      "loss": 0.7444,
      "step": 242
    },
    {
      "epoch": 6.075,
      "grad_norm": 0.32325669130792,
      "learning_rate": 3.524382055096309e-06,
      "loss": 0.7814,
      "step": 243
    },
    {
      "epoch": 6.1,
      "grad_norm": 0.294314054885741,
      "learning_rate": 3.4857496205102475e-06,
      "loss": 0.7996,
      "step": 244
    },
    {
      "epoch": 6.125,
      "grad_norm": 0.30410205632168436,
      "learning_rate": 3.4472164590307895e-06,
      "loss": 0.7741,
      "step": 245
    },
    {
      "epoch": 6.15,
      "grad_norm": 0.3747577565198413,
      "learning_rate": 3.4087850968627823e-06,
      "loss": 0.7651,
      "step": 246
    },
    {
      "epoch": 6.175,
      "grad_norm": 0.3480834902968779,
      "learning_rate": 3.370458053537186e-06,
      "loss": 0.7728,
      "step": 247
    },
    {
      "epoch": 6.2,
      "grad_norm": 0.30874816092153307,
      "learning_rate": 3.3322378417458985e-06,
      "loss": 0.7667,
      "step": 248
    },
    {
      "epoch": 6.225,
      "grad_norm": 0.3220383343921079,
      "learning_rate": 3.2941269671770193e-06,
      "loss": 0.7962,
      "step": 249
    },
    {
      "epoch": 6.25,
      "grad_norm": 0.3535972366319736,
      "learning_rate": 3.2561279283505888e-06,
      "loss": 0.7875,
      "step": 250
    },
    {
      "epoch": 6.275,
      "grad_norm": 0.3570142484942382,
      "learning_rate": 3.218243216454775e-06,
      "loss": 0.7459,
      "step": 251
    },
    {
      "epoch": 6.3,
      "grad_norm": 0.3938420232838248,
      "learning_rate": 3.180475315182563e-06,
      "loss": 0.7866,
      "step": 252
    },
    {
      "epoch": 6.325,
      "grad_norm": 0.418670895094041,
      "learning_rate": 3.1428267005689182e-06,
      "loss": 0.7465,
      "step": 253
    },
    {
      "epoch": 6.35,
      "grad_norm": 0.4361674812154792,
      "learning_rate": 3.1052998408284664e-06,
      "loss": 0.7823,
      "step": 254
    },
    {
      "epoch": 6.375,
      "grad_norm": 0.5410687457707055,
      "learning_rate": 3.0678971961936765e-06,
      "loss": 0.7888,
      "step": 255
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.3881527874591265,
      "learning_rate": 3.0306212187535653e-06,
      "loss": 0.7975,
      "step": 256
    },
    {
      "epoch": 6.425,
      "grad_norm": 0.3540262179710572,
      "learning_rate": 2.9934743522929476e-06,
      "loss": 0.8021,
      "step": 257
    },
    {
      "epoch": 6.45,
      "grad_norm": 0.41920366313347873,
      "learning_rate": 2.9564590321322206e-06,
      "loss": 0.7383,
      "step": 258
    },
    {
      "epoch": 6.475,
      "grad_norm": 0.4628906287676455,
      "learning_rate": 2.9195776849677037e-06,
      "loss": 0.7476,
      "step": 259
    },
    {
      "epoch": 6.5,
      "grad_norm": 0.44302784747531426,
      "learning_rate": 2.882832728712551e-06,
      "loss": 0.7715,
      "step": 260
    },
    {
      "epoch": 6.525,
      "grad_norm": 0.35338224459584966,
      "learning_rate": 2.8462265723382255e-06,
      "loss": 0.7417,
      "step": 261
    },
    {
      "epoch": 6.55,
      "grad_norm": 0.48937540088183007,
      "learning_rate": 2.8097616157165886e-06,
      "loss": 0.7533,
      "step": 262
    },
    {
      "epoch": 6.575,
      "grad_norm": 0.45167389676949854,
      "learning_rate": 2.7734402494625395e-06,
      "loss": 0.764,
      "step": 263
    },
    {
      "epoch": 6.6,
      "grad_norm": 0.40438810739435926,
      "learning_rate": 2.7372648547773063e-06,
      "loss": 0.7963,
      "step": 264
    },
    {
      "epoch": 6.625,
      "grad_norm": 0.3424135964248357,
      "learning_rate": 2.701237803292335e-06,
      "loss": 0.7782,
      "step": 265
    },
    {
      "epoch": 6.65,
      "grad_norm": 0.37838480593841406,
      "learning_rate": 2.665361456913797e-06,
      "loss": 0.7717,
      "step": 266
    },
    {
      "epoch": 6.675,
      "grad_norm": 0.39264044408814786,
      "learning_rate": 2.629638167667761e-06,
      "loss": 0.7466,
      "step": 267
    },
    {
      "epoch": 6.7,
      "grad_norm": 0.5528131987506779,
      "learning_rate": 2.594070277545975e-06,
      "loss": 0.7671,
      "step": 268
    },
    {
      "epoch": 6.725,
      "grad_norm": 0.4036506169957066,
      "learning_rate": 2.5586601183523485e-06,
      "loss": 0.7525,
      "step": 269
    },
    {
      "epoch": 6.75,
      "grad_norm": 0.30570416336137085,
      "learning_rate": 2.5234100115500643e-06,
      "loss": 0.7467,
      "step": 270
    },
    {
      "epoch": 6.775,
      "grad_norm": 0.3985125428120007,
      "learning_rate": 2.4883222681093916e-06,
      "loss": 0.7971,
      "step": 271
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.36468716140455315,
      "learning_rate": 2.4533991883561868e-06,
      "loss": 0.7687,
      "step": 272
    },
    {
      "epoch": 6.825,
      "grad_norm": 0.3198173933913146,
      "learning_rate": 2.4186430618210704e-06,
      "loss": 0.7414,
      "step": 273
    },
    {
      "epoch": 6.85,
      "grad_norm": 0.3303319933176365,
      "learning_rate": 2.38405616708935e-06,
      "loss": 0.7408,
      "step": 274
    },
    {
      "epoch": 6.875,
      "grad_norm": 0.3258403268612293,
      "learning_rate": 2.349640771651611e-06,
      "loss": 0.7456,
      "step": 275
    },
    {
      "epoch": 6.9,
      "grad_norm": 0.2812707026342879,
      "learning_rate": 2.315399131755081e-06,
      "loss": 0.7492,
      "step": 276
    },
    {
      "epoch": 6.925,
      "grad_norm": 0.3177919316007201,
      "learning_rate": 2.281333492255708e-06,
      "loss": 0.7887,
      "step": 277
    },
    {
      "epoch": 6.95,
      "grad_norm": 0.391540398746241,
      "learning_rate": 2.2474460864709825e-06,
      "loss": 0.7346,
      "step": 278
    },
    {
      "epoch": 6.975,
      "grad_norm": 0.33491616417366366,
      "learning_rate": 2.213739136033533e-06,
      "loss": 0.7574,
      "step": 279
    },
    {
      "epoch": 7.0,
      "grad_norm": 0.31756042780759597,
      "learning_rate": 2.1802148507454675e-06,
      "loss": 0.7495,
      "step": 280
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 400,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 200.0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 319178321428480.0,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}