Stewy Slocum
Add fine-tuned model
1e430b0
{
"best_global_step": null,
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 0,
"global_step": 884,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0011312217194570137,
"grad_norm": 0.484375,
"learning_rate": 1e-05,
"loss": 1.862,
"step": 1
},
{
"epoch": 0.0022624434389140274,
"grad_norm": 0.455078125,
"learning_rate": 9.988687782805431e-06,
"loss": 1.8721,
"step": 2
},
{
"epoch": 0.003393665158371041,
"grad_norm": 0.478515625,
"learning_rate": 9.97737556561086e-06,
"loss": 1.8779,
"step": 3
},
{
"epoch": 0.004524886877828055,
"grad_norm": 0.431640625,
"learning_rate": 9.966063348416291e-06,
"loss": 1.8142,
"step": 4
},
{
"epoch": 0.005656108597285068,
"grad_norm": 0.439453125,
"learning_rate": 9.95475113122172e-06,
"loss": 1.8547,
"step": 5
},
{
"epoch": 0.006787330316742082,
"grad_norm": 0.388671875,
"learning_rate": 9.94343891402715e-06,
"loss": 1.853,
"step": 6
},
{
"epoch": 0.007918552036199095,
"grad_norm": 0.390625,
"learning_rate": 9.93212669683258e-06,
"loss": 1.8645,
"step": 7
},
{
"epoch": 0.00904977375565611,
"grad_norm": 0.400390625,
"learning_rate": 9.92081447963801e-06,
"loss": 1.8807,
"step": 8
},
{
"epoch": 0.010180995475113122,
"grad_norm": 0.357421875,
"learning_rate": 9.90950226244344e-06,
"loss": 1.8898,
"step": 9
},
{
"epoch": 0.011312217194570135,
"grad_norm": 0.3125,
"learning_rate": 9.89819004524887e-06,
"loss": 1.7477,
"step": 10
},
{
"epoch": 0.012443438914027148,
"grad_norm": 0.34375,
"learning_rate": 9.8868778280543e-06,
"loss": 1.8756,
"step": 11
},
{
"epoch": 0.013574660633484163,
"grad_norm": 0.26953125,
"learning_rate": 9.87556561085973e-06,
"loss": 1.6808,
"step": 12
},
{
"epoch": 0.014705882352941176,
"grad_norm": 0.3203125,
"learning_rate": 9.86425339366516e-06,
"loss": 1.8201,
"step": 13
},
{
"epoch": 0.01583710407239819,
"grad_norm": 0.275390625,
"learning_rate": 9.852941176470589e-06,
"loss": 1.737,
"step": 14
},
{
"epoch": 0.016968325791855202,
"grad_norm": 0.287109375,
"learning_rate": 9.841628959276018e-06,
"loss": 1.704,
"step": 15
},
{
"epoch": 0.01809954751131222,
"grad_norm": 0.283203125,
"learning_rate": 9.830316742081449e-06,
"loss": 1.6821,
"step": 16
},
{
"epoch": 0.019230769230769232,
"grad_norm": 0.291015625,
"learning_rate": 9.819004524886878e-06,
"loss": 1.7612,
"step": 17
},
{
"epoch": 0.020361990950226245,
"grad_norm": 0.265625,
"learning_rate": 9.807692307692308e-06,
"loss": 1.7021,
"step": 18
},
{
"epoch": 0.021493212669683258,
"grad_norm": 0.251953125,
"learning_rate": 9.796380090497737e-06,
"loss": 1.6963,
"step": 19
},
{
"epoch": 0.02262443438914027,
"grad_norm": 0.234375,
"learning_rate": 9.785067873303168e-06,
"loss": 1.6299,
"step": 20
},
{
"epoch": 0.023755656108597284,
"grad_norm": 0.2451171875,
"learning_rate": 9.773755656108599e-06,
"loss": 1.6867,
"step": 21
},
{
"epoch": 0.024886877828054297,
"grad_norm": 0.22265625,
"learning_rate": 9.762443438914027e-06,
"loss": 1.6799,
"step": 22
},
{
"epoch": 0.026018099547511313,
"grad_norm": 0.2216796875,
"learning_rate": 9.751131221719458e-06,
"loss": 1.6344,
"step": 23
},
{
"epoch": 0.027149321266968326,
"grad_norm": 0.240234375,
"learning_rate": 9.739819004524887e-06,
"loss": 1.6293,
"step": 24
},
{
"epoch": 0.02828054298642534,
"grad_norm": 0.22265625,
"learning_rate": 9.728506787330318e-06,
"loss": 1.6515,
"step": 25
},
{
"epoch": 0.029411764705882353,
"grad_norm": 0.208984375,
"learning_rate": 9.717194570135747e-06,
"loss": 1.6456,
"step": 26
},
{
"epoch": 0.030542986425339366,
"grad_norm": 0.220703125,
"learning_rate": 9.705882352941177e-06,
"loss": 1.6747,
"step": 27
},
{
"epoch": 0.03167420814479638,
"grad_norm": 0.23046875,
"learning_rate": 9.694570135746606e-06,
"loss": 1.6766,
"step": 28
},
{
"epoch": 0.032805429864253395,
"grad_norm": 0.2119140625,
"learning_rate": 9.683257918552037e-06,
"loss": 1.596,
"step": 29
},
{
"epoch": 0.033936651583710405,
"grad_norm": 0.23828125,
"learning_rate": 9.671945701357468e-06,
"loss": 1.6941,
"step": 30
},
{
"epoch": 0.03506787330316742,
"grad_norm": 0.2021484375,
"learning_rate": 9.660633484162897e-06,
"loss": 1.5496,
"step": 31
},
{
"epoch": 0.03619909502262444,
"grad_norm": 0.16796875,
"learning_rate": 9.649321266968327e-06,
"loss": 1.5372,
"step": 32
},
{
"epoch": 0.03733031674208145,
"grad_norm": 0.169921875,
"learning_rate": 9.638009049773756e-06,
"loss": 1.5103,
"step": 33
},
{
"epoch": 0.038461538461538464,
"grad_norm": 0.1689453125,
"learning_rate": 9.626696832579187e-06,
"loss": 1.5417,
"step": 34
},
{
"epoch": 0.03959276018099547,
"grad_norm": 0.1708984375,
"learning_rate": 9.615384615384616e-06,
"loss": 1.5155,
"step": 35
},
{
"epoch": 0.04072398190045249,
"grad_norm": 0.171875,
"learning_rate": 9.604072398190046e-06,
"loss": 1.537,
"step": 36
},
{
"epoch": 0.0418552036199095,
"grad_norm": 0.1708984375,
"learning_rate": 9.592760180995477e-06,
"loss": 1.5825,
"step": 37
},
{
"epoch": 0.042986425339366516,
"grad_norm": 0.1630859375,
"learning_rate": 9.581447963800906e-06,
"loss": 1.5345,
"step": 38
},
{
"epoch": 0.04411764705882353,
"grad_norm": 0.1708984375,
"learning_rate": 9.570135746606337e-06,
"loss": 1.5292,
"step": 39
},
{
"epoch": 0.04524886877828054,
"grad_norm": 0.1748046875,
"learning_rate": 9.558823529411766e-06,
"loss": 1.5612,
"step": 40
},
{
"epoch": 0.04638009049773756,
"grad_norm": 0.169921875,
"learning_rate": 9.547511312217196e-06,
"loss": 1.5063,
"step": 41
},
{
"epoch": 0.04751131221719457,
"grad_norm": 0.23046875,
"learning_rate": 9.536199095022625e-06,
"loss": 1.5284,
"step": 42
},
{
"epoch": 0.048642533936651584,
"grad_norm": 0.171875,
"learning_rate": 9.524886877828054e-06,
"loss": 1.5045,
"step": 43
},
{
"epoch": 0.049773755656108594,
"grad_norm": 0.154296875,
"learning_rate": 9.513574660633485e-06,
"loss": 1.5165,
"step": 44
},
{
"epoch": 0.05090497737556561,
"grad_norm": 0.171875,
"learning_rate": 9.502262443438914e-06,
"loss": 1.5415,
"step": 45
},
{
"epoch": 0.05203619909502263,
"grad_norm": 0.169921875,
"learning_rate": 9.490950226244344e-06,
"loss": 1.51,
"step": 46
},
{
"epoch": 0.053167420814479636,
"grad_norm": 0.1572265625,
"learning_rate": 9.479638009049773e-06,
"loss": 1.5035,
"step": 47
},
{
"epoch": 0.05429864253393665,
"grad_norm": 0.1953125,
"learning_rate": 9.468325791855204e-06,
"loss": 1.5254,
"step": 48
},
{
"epoch": 0.05542986425339366,
"grad_norm": 0.154296875,
"learning_rate": 9.457013574660635e-06,
"loss": 1.4862,
"step": 49
},
{
"epoch": 0.05656108597285068,
"grad_norm": 0.154296875,
"learning_rate": 9.445701357466064e-06,
"loss": 1.4514,
"step": 50
},
{
"epoch": 0.057692307692307696,
"grad_norm": 0.1865234375,
"learning_rate": 9.434389140271494e-06,
"loss": 1.4544,
"step": 51
},
{
"epoch": 0.058823529411764705,
"grad_norm": 0.1591796875,
"learning_rate": 9.423076923076923e-06,
"loss": 1.4767,
"step": 52
},
{
"epoch": 0.05995475113122172,
"grad_norm": 0.16796875,
"learning_rate": 9.411764705882354e-06,
"loss": 1.4907,
"step": 53
},
{
"epoch": 0.06108597285067873,
"grad_norm": 0.150390625,
"learning_rate": 9.400452488687783e-06,
"loss": 1.4763,
"step": 54
},
{
"epoch": 0.06221719457013575,
"grad_norm": 0.1591796875,
"learning_rate": 9.389140271493213e-06,
"loss": 1.5204,
"step": 55
},
{
"epoch": 0.06334841628959276,
"grad_norm": 0.1552734375,
"learning_rate": 9.377828054298644e-06,
"loss": 1.4883,
"step": 56
},
{
"epoch": 0.06447963800904978,
"grad_norm": 0.41015625,
"learning_rate": 9.366515837104073e-06,
"loss": 1.3914,
"step": 57
},
{
"epoch": 0.06561085972850679,
"grad_norm": 0.150390625,
"learning_rate": 9.355203619909504e-06,
"loss": 1.4858,
"step": 58
},
{
"epoch": 0.0667420814479638,
"grad_norm": 0.142578125,
"learning_rate": 9.343891402714933e-06,
"loss": 1.4818,
"step": 59
},
{
"epoch": 0.06787330316742081,
"grad_norm": 0.1494140625,
"learning_rate": 9.332579185520363e-06,
"loss": 1.4229,
"step": 60
},
{
"epoch": 0.06900452488687783,
"grad_norm": 0.1591796875,
"learning_rate": 9.321266968325792e-06,
"loss": 1.492,
"step": 61
},
{
"epoch": 0.07013574660633484,
"grad_norm": 0.16015625,
"learning_rate": 9.309954751131223e-06,
"loss": 1.461,
"step": 62
},
{
"epoch": 0.07126696832579185,
"grad_norm": 0.193359375,
"learning_rate": 9.298642533936652e-06,
"loss": 1.4375,
"step": 63
},
{
"epoch": 0.07239819004524888,
"grad_norm": 0.1591796875,
"learning_rate": 9.287330316742082e-06,
"loss": 1.3873,
"step": 64
},
{
"epoch": 0.07352941176470588,
"grad_norm": 0.140625,
"learning_rate": 9.276018099547513e-06,
"loss": 1.4113,
"step": 65
},
{
"epoch": 0.0746606334841629,
"grad_norm": 0.134765625,
"learning_rate": 9.264705882352942e-06,
"loss": 1.4286,
"step": 66
},
{
"epoch": 0.0757918552036199,
"grad_norm": 0.14453125,
"learning_rate": 9.253393665158373e-06,
"loss": 1.3791,
"step": 67
},
{
"epoch": 0.07692307692307693,
"grad_norm": 0.1416015625,
"learning_rate": 9.242081447963802e-06,
"loss": 1.4458,
"step": 68
},
{
"epoch": 0.07805429864253394,
"grad_norm": 0.1650390625,
"learning_rate": 9.230769230769232e-06,
"loss": 1.41,
"step": 69
},
{
"epoch": 0.07918552036199095,
"grad_norm": 0.15625,
"learning_rate": 9.219457013574661e-06,
"loss": 1.3792,
"step": 70
},
{
"epoch": 0.08031674208144797,
"grad_norm": 0.1533203125,
"learning_rate": 9.20814479638009e-06,
"loss": 1.4708,
"step": 71
},
{
"epoch": 0.08144796380090498,
"grad_norm": 0.15625,
"learning_rate": 9.196832579185521e-06,
"loss": 1.3502,
"step": 72
},
{
"epoch": 0.08257918552036199,
"grad_norm": 0.1416015625,
"learning_rate": 9.18552036199095e-06,
"loss": 1.4106,
"step": 73
},
{
"epoch": 0.083710407239819,
"grad_norm": 0.140625,
"learning_rate": 9.17420814479638e-06,
"loss": 1.4049,
"step": 74
},
{
"epoch": 0.08484162895927602,
"grad_norm": 0.1591796875,
"learning_rate": 9.162895927601811e-06,
"loss": 1.4237,
"step": 75
},
{
"epoch": 0.08597285067873303,
"grad_norm": 0.14453125,
"learning_rate": 9.15158371040724e-06,
"loss": 1.4136,
"step": 76
},
{
"epoch": 0.08710407239819004,
"grad_norm": 0.1455078125,
"learning_rate": 9.14027149321267e-06,
"loss": 1.396,
"step": 77
},
{
"epoch": 0.08823529411764706,
"grad_norm": 0.142578125,
"learning_rate": 9.1289592760181e-06,
"loss": 1.3941,
"step": 78
},
{
"epoch": 0.08936651583710407,
"grad_norm": 0.1474609375,
"learning_rate": 9.11764705882353e-06,
"loss": 1.345,
"step": 79
},
{
"epoch": 0.09049773755656108,
"grad_norm": 0.140625,
"learning_rate": 9.10633484162896e-06,
"loss": 1.441,
"step": 80
},
{
"epoch": 0.0916289592760181,
"grad_norm": 0.14453125,
"learning_rate": 9.09502262443439e-06,
"loss": 1.3763,
"step": 81
},
{
"epoch": 0.09276018099547512,
"grad_norm": 0.14453125,
"learning_rate": 9.083710407239819e-06,
"loss": 1.3574,
"step": 82
},
{
"epoch": 0.09389140271493213,
"grad_norm": 0.158203125,
"learning_rate": 9.07239819004525e-06,
"loss": 1.3787,
"step": 83
},
{
"epoch": 0.09502262443438914,
"grad_norm": 0.1806640625,
"learning_rate": 9.06108597285068e-06,
"loss": 1.4112,
"step": 84
},
{
"epoch": 0.09615384615384616,
"grad_norm": 0.1572265625,
"learning_rate": 9.049773755656109e-06,
"loss": 1.4044,
"step": 85
},
{
"epoch": 0.09728506787330317,
"grad_norm": 0.1455078125,
"learning_rate": 9.03846153846154e-06,
"loss": 1.3664,
"step": 86
},
{
"epoch": 0.09841628959276018,
"grad_norm": 0.150390625,
"learning_rate": 9.027149321266969e-06,
"loss": 1.3832,
"step": 87
},
{
"epoch": 0.09954751131221719,
"grad_norm": 0.150390625,
"learning_rate": 9.0158371040724e-06,
"loss": 1.3864,
"step": 88
},
{
"epoch": 0.10067873303167421,
"grad_norm": 0.1689453125,
"learning_rate": 9.004524886877828e-06,
"loss": 1.351,
"step": 89
},
{
"epoch": 0.10180995475113122,
"grad_norm": 0.1455078125,
"learning_rate": 8.993212669683259e-06,
"loss": 1.3725,
"step": 90
},
{
"epoch": 0.10294117647058823,
"grad_norm": 0.1796875,
"learning_rate": 8.98190045248869e-06,
"loss": 1.3885,
"step": 91
},
{
"epoch": 0.10407239819004525,
"grad_norm": 0.158203125,
"learning_rate": 8.970588235294119e-06,
"loss": 1.393,
"step": 92
},
{
"epoch": 0.10520361990950226,
"grad_norm": 0.1689453125,
"learning_rate": 8.95927601809955e-06,
"loss": 1.3715,
"step": 93
},
{
"epoch": 0.10633484162895927,
"grad_norm": 0.185546875,
"learning_rate": 8.947963800904978e-06,
"loss": 1.3142,
"step": 94
},
{
"epoch": 0.1074660633484163,
"grad_norm": 0.1767578125,
"learning_rate": 8.936651583710409e-06,
"loss": 1.3477,
"step": 95
},
{
"epoch": 0.1085972850678733,
"grad_norm": 0.1494140625,
"learning_rate": 8.925339366515838e-06,
"loss": 1.3327,
"step": 96
},
{
"epoch": 0.10972850678733032,
"grad_norm": 0.1494140625,
"learning_rate": 8.914027149321268e-06,
"loss": 1.3507,
"step": 97
},
{
"epoch": 0.11085972850678733,
"grad_norm": 0.1396484375,
"learning_rate": 8.902714932126697e-06,
"loss": 1.3581,
"step": 98
},
{
"epoch": 0.11199095022624435,
"grad_norm": 0.154296875,
"learning_rate": 8.891402714932126e-06,
"loss": 1.3598,
"step": 99
},
{
"epoch": 0.11312217194570136,
"grad_norm": 0.1533203125,
"learning_rate": 8.880090497737557e-06,
"loss": 1.3629,
"step": 100
},
{
"epoch": 0.11425339366515837,
"grad_norm": 0.1650390625,
"learning_rate": 8.868778280542986e-06,
"loss": 1.3559,
"step": 101
},
{
"epoch": 0.11538461538461539,
"grad_norm": 0.1845703125,
"learning_rate": 8.857466063348417e-06,
"loss": 1.399,
"step": 102
},
{
"epoch": 0.1165158371040724,
"grad_norm": 0.1982421875,
"learning_rate": 8.846153846153847e-06,
"loss": 1.4189,
"step": 103
},
{
"epoch": 0.11764705882352941,
"grad_norm": 0.158203125,
"learning_rate": 8.834841628959276e-06,
"loss": 1.3515,
"step": 104
},
{
"epoch": 0.11877828054298642,
"grad_norm": 0.1552734375,
"learning_rate": 8.823529411764707e-06,
"loss": 1.3651,
"step": 105
},
{
"epoch": 0.11990950226244344,
"grad_norm": 0.150390625,
"learning_rate": 8.812217194570136e-06,
"loss": 1.3161,
"step": 106
},
{
"epoch": 0.12104072398190045,
"grad_norm": 0.169921875,
"learning_rate": 8.800904977375566e-06,
"loss": 1.3113,
"step": 107
},
{
"epoch": 0.12217194570135746,
"grad_norm": 0.154296875,
"learning_rate": 8.789592760180995e-06,
"loss": 1.3515,
"step": 108
},
{
"epoch": 0.12330316742081449,
"grad_norm": 0.1748046875,
"learning_rate": 8.778280542986426e-06,
"loss": 1.4192,
"step": 109
},
{
"epoch": 0.1244343891402715,
"grad_norm": 0.173828125,
"learning_rate": 8.766968325791857e-06,
"loss": 1.3407,
"step": 110
},
{
"epoch": 0.1255656108597285,
"grad_norm": 0.1640625,
"learning_rate": 8.755656108597286e-06,
"loss": 1.3482,
"step": 111
},
{
"epoch": 0.12669683257918551,
"grad_norm": 0.162109375,
"learning_rate": 8.744343891402716e-06,
"loss": 1.3764,
"step": 112
},
{
"epoch": 0.12782805429864252,
"grad_norm": 0.259765625,
"learning_rate": 8.733031674208145e-06,
"loss": 1.2983,
"step": 113
},
{
"epoch": 0.12895927601809956,
"grad_norm": 0.1796875,
"learning_rate": 8.721719457013576e-06,
"loss": 1.2293,
"step": 114
},
{
"epoch": 0.13009049773755657,
"grad_norm": 0.14453125,
"learning_rate": 8.710407239819005e-06,
"loss": 1.3025,
"step": 115
},
{
"epoch": 0.13122171945701358,
"grad_norm": 0.1552734375,
"learning_rate": 8.699095022624435e-06,
"loss": 1.3086,
"step": 116
},
{
"epoch": 0.1323529411764706,
"grad_norm": 0.1591796875,
"learning_rate": 8.687782805429864e-06,
"loss": 1.3043,
"step": 117
},
{
"epoch": 0.1334841628959276,
"grad_norm": 0.1552734375,
"learning_rate": 8.676470588235295e-06,
"loss": 1.2943,
"step": 118
},
{
"epoch": 0.1346153846153846,
"grad_norm": 0.201171875,
"learning_rate": 8.665158371040726e-06,
"loss": 1.3583,
"step": 119
},
{
"epoch": 0.13574660633484162,
"grad_norm": 0.1865234375,
"learning_rate": 8.653846153846155e-06,
"loss": 1.3062,
"step": 120
},
{
"epoch": 0.13687782805429866,
"grad_norm": 0.1611328125,
"learning_rate": 8.642533936651585e-06,
"loss": 1.3281,
"step": 121
},
{
"epoch": 0.13800904977375567,
"grad_norm": 0.1494140625,
"learning_rate": 8.631221719457014e-06,
"loss": 1.2723,
"step": 122
},
{
"epoch": 0.13914027149321267,
"grad_norm": 0.1533203125,
"learning_rate": 8.619909502262445e-06,
"loss": 1.2852,
"step": 123
},
{
"epoch": 0.14027149321266968,
"grad_norm": 0.1552734375,
"learning_rate": 8.608597285067874e-06,
"loss": 1.2776,
"step": 124
},
{
"epoch": 0.1414027149321267,
"grad_norm": 0.1474609375,
"learning_rate": 8.597285067873304e-06,
"loss": 1.3204,
"step": 125
},
{
"epoch": 0.1425339366515837,
"grad_norm": 0.1640625,
"learning_rate": 8.585972850678733e-06,
"loss": 1.3196,
"step": 126
},
{
"epoch": 0.1436651583710407,
"grad_norm": 0.16015625,
"learning_rate": 8.574660633484162e-06,
"loss": 1.295,
"step": 127
},
{
"epoch": 0.14479638009049775,
"grad_norm": 0.16015625,
"learning_rate": 8.563348416289593e-06,
"loss": 1.2485,
"step": 128
},
{
"epoch": 0.14592760180995476,
"grad_norm": 0.1728515625,
"learning_rate": 8.552036199095024e-06,
"loss": 1.2515,
"step": 129
},
{
"epoch": 0.14705882352941177,
"grad_norm": 0.1533203125,
"learning_rate": 8.540723981900453e-06,
"loss": 1.2898,
"step": 130
},
{
"epoch": 0.14819004524886878,
"grad_norm": 0.2001953125,
"learning_rate": 8.529411764705883e-06,
"loss": 1.3318,
"step": 131
},
{
"epoch": 0.1493212669683258,
"grad_norm": 0.1552734375,
"learning_rate": 8.518099547511312e-06,
"loss": 1.2821,
"step": 132
},
{
"epoch": 0.1504524886877828,
"grad_norm": 0.16015625,
"learning_rate": 8.506787330316743e-06,
"loss": 1.2577,
"step": 133
},
{
"epoch": 0.1515837104072398,
"grad_norm": 0.1982421875,
"learning_rate": 8.495475113122172e-06,
"loss": 1.2714,
"step": 134
},
{
"epoch": 0.15271493212669685,
"grad_norm": 0.1875,
"learning_rate": 8.484162895927603e-06,
"loss": 1.2254,
"step": 135
},
{
"epoch": 0.15384615384615385,
"grad_norm": 0.1689453125,
"learning_rate": 8.472850678733031e-06,
"loss": 1.3279,
"step": 136
},
{
"epoch": 0.15497737556561086,
"grad_norm": 0.169921875,
"learning_rate": 8.461538461538462e-06,
"loss": 1.2945,
"step": 137
},
{
"epoch": 0.15610859728506787,
"grad_norm": 0.2041015625,
"learning_rate": 8.450226244343893e-06,
"loss": 1.2328,
"step": 138
},
{
"epoch": 0.15723981900452488,
"grad_norm": 0.15625,
"learning_rate": 8.438914027149322e-06,
"loss": 1.2597,
"step": 139
},
{
"epoch": 0.1583710407239819,
"grad_norm": 0.162109375,
"learning_rate": 8.427601809954752e-06,
"loss": 1.2699,
"step": 140
},
{
"epoch": 0.1595022624434389,
"grad_norm": 0.1728515625,
"learning_rate": 8.416289592760181e-06,
"loss": 1.2524,
"step": 141
},
{
"epoch": 0.16063348416289594,
"grad_norm": 0.1650390625,
"learning_rate": 8.404977375565612e-06,
"loss": 1.243,
"step": 142
},
{
"epoch": 0.16176470588235295,
"grad_norm": 0.17578125,
"learning_rate": 8.393665158371041e-06,
"loss": 1.2608,
"step": 143
},
{
"epoch": 0.16289592760180996,
"grad_norm": 0.171875,
"learning_rate": 8.382352941176472e-06,
"loss": 1.2455,
"step": 144
},
{
"epoch": 0.16402714932126697,
"grad_norm": 0.1572265625,
"learning_rate": 8.371040723981902e-06,
"loss": 1.2428,
"step": 145
},
{
"epoch": 0.16515837104072398,
"grad_norm": 0.15625,
"learning_rate": 8.359728506787331e-06,
"loss": 1.2393,
"step": 146
},
{
"epoch": 0.166289592760181,
"grad_norm": 0.166015625,
"learning_rate": 8.348416289592762e-06,
"loss": 1.2922,
"step": 147
},
{
"epoch": 0.167420814479638,
"grad_norm": 0.1875,
"learning_rate": 8.33710407239819e-06,
"loss": 1.2614,
"step": 148
},
{
"epoch": 0.16855203619909503,
"grad_norm": 0.150390625,
"learning_rate": 8.325791855203621e-06,
"loss": 1.278,
"step": 149
},
{
"epoch": 0.16968325791855204,
"grad_norm": 0.224609375,
"learning_rate": 8.31447963800905e-06,
"loss": 1.3123,
"step": 150
},
{
"epoch": 0.17081447963800905,
"grad_norm": 0.1533203125,
"learning_rate": 8.303167420814481e-06,
"loss": 1.2879,
"step": 151
},
{
"epoch": 0.17194570135746606,
"grad_norm": 0.169921875,
"learning_rate": 8.29185520361991e-06,
"loss": 1.2879,
"step": 152
},
{
"epoch": 0.17307692307692307,
"grad_norm": 0.177734375,
"learning_rate": 8.280542986425339e-06,
"loss": 1.2388,
"step": 153
},
{
"epoch": 0.17420814479638008,
"grad_norm": 0.1806640625,
"learning_rate": 8.26923076923077e-06,
"loss": 1.2633,
"step": 154
},
{
"epoch": 0.1753393665158371,
"grad_norm": 0.1884765625,
"learning_rate": 8.257918552036199e-06,
"loss": 1.2786,
"step": 155
},
{
"epoch": 0.17647058823529413,
"grad_norm": 0.1865234375,
"learning_rate": 8.246606334841629e-06,
"loss": 1.271,
"step": 156
},
{
"epoch": 0.17760180995475114,
"grad_norm": 0.16796875,
"learning_rate": 8.23529411764706e-06,
"loss": 1.2609,
"step": 157
},
{
"epoch": 0.17873303167420815,
"grad_norm": 0.2099609375,
"learning_rate": 8.223981900452489e-06,
"loss": 1.2776,
"step": 158
},
{
"epoch": 0.17986425339366516,
"grad_norm": 0.1708984375,
"learning_rate": 8.21266968325792e-06,
"loss": 1.2921,
"step": 159
},
{
"epoch": 0.18099547511312217,
"grad_norm": 0.169921875,
"learning_rate": 8.201357466063348e-06,
"loss": 1.2215,
"step": 160
},
{
"epoch": 0.18212669683257918,
"grad_norm": 0.17578125,
"learning_rate": 8.190045248868779e-06,
"loss": 1.2012,
"step": 161
},
{
"epoch": 0.1832579185520362,
"grad_norm": 0.1767578125,
"learning_rate": 8.178733031674208e-06,
"loss": 1.2368,
"step": 162
},
{
"epoch": 0.18438914027149322,
"grad_norm": 0.158203125,
"learning_rate": 8.167420814479639e-06,
"loss": 1.233,
"step": 163
},
{
"epoch": 0.18552036199095023,
"grad_norm": 0.16015625,
"learning_rate": 8.15610859728507e-06,
"loss": 1.1698,
"step": 164
},
{
"epoch": 0.18665158371040724,
"grad_norm": 0.2255859375,
"learning_rate": 8.144796380090498e-06,
"loss": 1.2676,
"step": 165
},
{
"epoch": 0.18778280542986425,
"grad_norm": 0.19140625,
"learning_rate": 8.133484162895929e-06,
"loss": 1.2616,
"step": 166
},
{
"epoch": 0.18891402714932126,
"grad_norm": 0.171875,
"learning_rate": 8.122171945701358e-06,
"loss": 1.2148,
"step": 167
},
{
"epoch": 0.19004524886877827,
"grad_norm": 0.1787109375,
"learning_rate": 8.110859728506788e-06,
"loss": 1.266,
"step": 168
},
{
"epoch": 0.19117647058823528,
"grad_norm": 0.1640625,
"learning_rate": 8.099547511312217e-06,
"loss": 1.2135,
"step": 169
},
{
"epoch": 0.19230769230769232,
"grad_norm": 0.169921875,
"learning_rate": 8.088235294117648e-06,
"loss": 1.2331,
"step": 170
},
{
"epoch": 0.19343891402714933,
"grad_norm": 0.1845703125,
"learning_rate": 8.076923076923077e-06,
"loss": 1.2224,
"step": 171
},
{
"epoch": 0.19457013574660634,
"grad_norm": 0.171875,
"learning_rate": 8.065610859728508e-06,
"loss": 1.2353,
"step": 172
},
{
"epoch": 0.19570135746606335,
"grad_norm": 0.1875,
"learning_rate": 8.054298642533938e-06,
"loss": 1.2518,
"step": 173
},
{
"epoch": 0.19683257918552036,
"grad_norm": 0.1611328125,
"learning_rate": 8.042986425339367e-06,
"loss": 1.2515,
"step": 174
},
{
"epoch": 0.19796380090497737,
"grad_norm": 0.1767578125,
"learning_rate": 8.031674208144798e-06,
"loss": 1.274,
"step": 175
},
{
"epoch": 0.19909502262443438,
"grad_norm": 0.267578125,
"learning_rate": 8.020361990950227e-06,
"loss": 1.2162,
"step": 176
},
{
"epoch": 0.2002262443438914,
"grad_norm": 0.166015625,
"learning_rate": 8.009049773755657e-06,
"loss": 1.2098,
"step": 177
},
{
"epoch": 0.20135746606334842,
"grad_norm": 0.1865234375,
"learning_rate": 7.997737556561086e-06,
"loss": 1.2604,
"step": 178
},
{
"epoch": 0.20248868778280543,
"grad_norm": 0.193359375,
"learning_rate": 7.986425339366517e-06,
"loss": 1.2234,
"step": 179
},
{
"epoch": 0.20361990950226244,
"grad_norm": 0.181640625,
"learning_rate": 7.975113122171946e-06,
"loss": 1.2396,
"step": 180
},
{
"epoch": 0.20475113122171945,
"grad_norm": 0.193359375,
"learning_rate": 7.963800904977375e-06,
"loss": 1.2193,
"step": 181
},
{
"epoch": 0.20588235294117646,
"grad_norm": 0.1611328125,
"learning_rate": 7.952488687782806e-06,
"loss": 1.2589,
"step": 182
},
{
"epoch": 0.20701357466063347,
"grad_norm": 0.15625,
"learning_rate": 7.941176470588236e-06,
"loss": 1.1957,
"step": 183
},
{
"epoch": 0.2081447963800905,
"grad_norm": 0.1796875,
"learning_rate": 7.929864253393665e-06,
"loss": 1.2223,
"step": 184
},
{
"epoch": 0.20927601809954752,
"grad_norm": 0.1904296875,
"learning_rate": 7.918552036199096e-06,
"loss": 1.2277,
"step": 185
},
{
"epoch": 0.21040723981900453,
"grad_norm": 0.1669921875,
"learning_rate": 7.907239819004525e-06,
"loss": 1.232,
"step": 186
},
{
"epoch": 0.21153846153846154,
"grad_norm": 0.169921875,
"learning_rate": 7.895927601809955e-06,
"loss": 1.2167,
"step": 187
},
{
"epoch": 0.21266968325791855,
"grad_norm": 0.1767578125,
"learning_rate": 7.884615384615384e-06,
"loss": 1.2036,
"step": 188
},
{
"epoch": 0.21380090497737556,
"grad_norm": 0.205078125,
"learning_rate": 7.873303167420815e-06,
"loss": 1.2365,
"step": 189
},
{
"epoch": 0.2149321266968326,
"grad_norm": 0.17578125,
"learning_rate": 7.861990950226244e-06,
"loss": 1.1657,
"step": 190
},
{
"epoch": 0.2160633484162896,
"grad_norm": 0.166015625,
"learning_rate": 7.850678733031675e-06,
"loss": 1.1966,
"step": 191
},
{
"epoch": 0.2171945701357466,
"grad_norm": 0.1748046875,
"learning_rate": 7.839366515837105e-06,
"loss": 1.2314,
"step": 192
},
{
"epoch": 0.21832579185520362,
"grad_norm": 0.1826171875,
"learning_rate": 7.828054298642534e-06,
"loss": 1.2273,
"step": 193
},
{
"epoch": 0.21945701357466063,
"grad_norm": 0.17578125,
"learning_rate": 7.816742081447965e-06,
"loss": 1.2024,
"step": 194
},
{
"epoch": 0.22058823529411764,
"grad_norm": 0.181640625,
"learning_rate": 7.805429864253394e-06,
"loss": 1.2288,
"step": 195
},
{
"epoch": 0.22171945701357465,
"grad_norm": 0.2275390625,
"learning_rate": 7.794117647058825e-06,
"loss": 1.2239,
"step": 196
},
{
"epoch": 0.2228506787330317,
"grad_norm": 0.1669921875,
"learning_rate": 7.782805429864253e-06,
"loss": 1.2392,
"step": 197
},
{
"epoch": 0.2239819004524887,
"grad_norm": 0.1728515625,
"learning_rate": 7.771493212669684e-06,
"loss": 1.2337,
"step": 198
},
{
"epoch": 0.2251131221719457,
"grad_norm": 0.16796875,
"learning_rate": 7.760180995475115e-06,
"loss": 1.1778,
"step": 199
},
{
"epoch": 0.22624434389140272,
"grad_norm": 0.19921875,
"learning_rate": 7.748868778280544e-06,
"loss": 1.2389,
"step": 200
},
{
"epoch": 0.22737556561085973,
"grad_norm": 0.173828125,
"learning_rate": 7.737556561085974e-06,
"loss": 1.1741,
"step": 201
},
{
"epoch": 0.22850678733031674,
"grad_norm": 0.1728515625,
"learning_rate": 7.726244343891403e-06,
"loss": 1.1747,
"step": 202
},
{
"epoch": 0.22963800904977374,
"grad_norm": 0.1689453125,
"learning_rate": 7.714932126696834e-06,
"loss": 1.2125,
"step": 203
},
{
"epoch": 0.23076923076923078,
"grad_norm": 0.234375,
"learning_rate": 7.703619909502263e-06,
"loss": 1.3018,
"step": 204
},
{
"epoch": 0.2319004524886878,
"grad_norm": 0.1904296875,
"learning_rate": 7.692307692307694e-06,
"loss": 1.2423,
"step": 205
},
{
"epoch": 0.2330316742081448,
"grad_norm": 0.1669921875,
"learning_rate": 7.680995475113123e-06,
"loss": 1.1809,
"step": 206
},
{
"epoch": 0.2341628959276018,
"grad_norm": 0.1591796875,
"learning_rate": 7.669683257918553e-06,
"loss": 1.1722,
"step": 207
},
{
"epoch": 0.23529411764705882,
"grad_norm": 0.1962890625,
"learning_rate": 7.658371040723984e-06,
"loss": 1.1859,
"step": 208
},
{
"epoch": 0.23642533936651583,
"grad_norm": 0.1982421875,
"learning_rate": 7.647058823529411e-06,
"loss": 1.1791,
"step": 209
},
{
"epoch": 0.23755656108597284,
"grad_norm": 0.19140625,
"learning_rate": 7.635746606334842e-06,
"loss": 1.1928,
"step": 210
},
{
"epoch": 0.23868778280542988,
"grad_norm": 0.1845703125,
"learning_rate": 7.6244343891402715e-06,
"loss": 1.1815,
"step": 211
},
{
"epoch": 0.2398190045248869,
"grad_norm": 0.18359375,
"learning_rate": 7.613122171945701e-06,
"loss": 1.2147,
"step": 212
},
{
"epoch": 0.2409502262443439,
"grad_norm": 0.208984375,
"learning_rate": 7.601809954751131e-06,
"loss": 1.2292,
"step": 213
},
{
"epoch": 0.2420814479638009,
"grad_norm": 0.1923828125,
"learning_rate": 7.590497737556562e-06,
"loss": 1.2253,
"step": 214
},
{
"epoch": 0.24321266968325791,
"grad_norm": 0.1689453125,
"learning_rate": 7.579185520361992e-06,
"loss": 1.2106,
"step": 215
},
{
"epoch": 0.24434389140271492,
"grad_norm": 0.1689453125,
"learning_rate": 7.567873303167421e-06,
"loss": 1.1327,
"step": 216
},
{
"epoch": 0.24547511312217193,
"grad_norm": 0.1796875,
"learning_rate": 7.556561085972851e-06,
"loss": 1.1818,
"step": 217
},
{
"epoch": 0.24660633484162897,
"grad_norm": 0.18359375,
"learning_rate": 7.545248868778281e-06,
"loss": 1.21,
"step": 218
},
{
"epoch": 0.24773755656108598,
"grad_norm": 0.18359375,
"learning_rate": 7.533936651583711e-06,
"loss": 1.2016,
"step": 219
},
{
"epoch": 0.248868778280543,
"grad_norm": 0.17578125,
"learning_rate": 7.522624434389141e-06,
"loss": 1.1215,
"step": 220
},
{
"epoch": 0.25,
"grad_norm": 0.197265625,
"learning_rate": 7.51131221719457e-06,
"loss": 1.1728,
"step": 221
},
{
"epoch": 0.251131221719457,
"grad_norm": 0.189453125,
"learning_rate": 7.500000000000001e-06,
"loss": 1.1809,
"step": 222
},
{
"epoch": 0.252262443438914,
"grad_norm": 0.26953125,
"learning_rate": 7.488687782805431e-06,
"loss": 1.1746,
"step": 223
},
{
"epoch": 0.25339366515837103,
"grad_norm": 0.1640625,
"learning_rate": 7.477375565610861e-06,
"loss": 1.17,
"step": 224
},
{
"epoch": 0.25452488687782804,
"grad_norm": 0.208984375,
"learning_rate": 7.4660633484162904e-06,
"loss": 1.2382,
"step": 225
},
{
"epoch": 0.25565610859728505,
"grad_norm": 0.1826171875,
"learning_rate": 7.45475113122172e-06,
"loss": 1.2031,
"step": 226
},
{
"epoch": 0.25678733031674206,
"grad_norm": 0.1875,
"learning_rate": 7.44343891402715e-06,
"loss": 1.2491,
"step": 227
},
{
"epoch": 0.2579185520361991,
"grad_norm": 0.2060546875,
"learning_rate": 7.43212669683258e-06,
"loss": 1.1199,
"step": 228
},
{
"epoch": 0.25904977375565613,
"grad_norm": 0.1875,
"learning_rate": 7.42081447963801e-06,
"loss": 1.2087,
"step": 229
},
{
"epoch": 0.26018099547511314,
"grad_norm": 0.1845703125,
"learning_rate": 7.4095022624434394e-06,
"loss": 1.1632,
"step": 230
},
{
"epoch": 0.26131221719457015,
"grad_norm": 0.1865234375,
"learning_rate": 7.39819004524887e-06,
"loss": 1.1585,
"step": 231
},
{
"epoch": 0.26244343891402716,
"grad_norm": 0.27734375,
"learning_rate": 7.3868778280543e-06,
"loss": 1.2423,
"step": 232
},
{
"epoch": 0.26357466063348417,
"grad_norm": 0.205078125,
"learning_rate": 7.37556561085973e-06,
"loss": 1.2131,
"step": 233
},
{
"epoch": 0.2647058823529412,
"grad_norm": 0.193359375,
"learning_rate": 7.3642533936651595e-06,
"loss": 1.1917,
"step": 234
},
{
"epoch": 0.2658371040723982,
"grad_norm": 0.1982421875,
"learning_rate": 7.352941176470589e-06,
"loss": 1.1903,
"step": 235
},
{
"epoch": 0.2669683257918552,
"grad_norm": 0.18359375,
"learning_rate": 7.341628959276018e-06,
"loss": 1.1762,
"step": 236
},
{
"epoch": 0.2680995475113122,
"grad_norm": 0.1748046875,
"learning_rate": 7.330316742081448e-06,
"loss": 1.2265,
"step": 237
},
{
"epoch": 0.2692307692307692,
"grad_norm": 0.201171875,
"learning_rate": 7.319004524886878e-06,
"loss": 1.2762,
"step": 238
},
{
"epoch": 0.2703619909502262,
"grad_norm": 0.185546875,
"learning_rate": 7.307692307692308e-06,
"loss": 1.1561,
"step": 239
},
{
"epoch": 0.27149321266968324,
"grad_norm": 0.203125,
"learning_rate": 7.2963800904977374e-06,
"loss": 1.1641,
"step": 240
},
{
"epoch": 0.27262443438914025,
"grad_norm": 0.1826171875,
"learning_rate": 7.285067873303168e-06,
"loss": 1.1842,
"step": 241
},
{
"epoch": 0.2737556561085973,
"grad_norm": 0.1787109375,
"learning_rate": 7.273755656108598e-06,
"loss": 1.1763,
"step": 242
},
{
"epoch": 0.2748868778280543,
"grad_norm": 0.17578125,
"learning_rate": 7.262443438914028e-06,
"loss": 1.175,
"step": 243
},
{
"epoch": 0.27601809954751133,
"grad_norm": 0.283203125,
"learning_rate": 7.2511312217194575e-06,
"loss": 1.2315,
"step": 244
},
{
"epoch": 0.27714932126696834,
"grad_norm": 0.197265625,
"learning_rate": 7.239819004524887e-06,
"loss": 1.1678,
"step": 245
},
{
"epoch": 0.27828054298642535,
"grad_norm": 0.1806640625,
"learning_rate": 7.228506787330317e-06,
"loss": 1.1426,
"step": 246
},
{
"epoch": 0.27941176470588236,
"grad_norm": 0.25390625,
"learning_rate": 7.217194570135747e-06,
"loss": 1.2309,
"step": 247
},
{
"epoch": 0.28054298642533937,
"grad_norm": 0.1923828125,
"learning_rate": 7.205882352941177e-06,
"loss": 1.2104,
"step": 248
},
{
"epoch": 0.2816742081447964,
"grad_norm": 0.1689453125,
"learning_rate": 7.1945701357466065e-06,
"loss": 1.1461,
"step": 249
},
{
"epoch": 0.2828054298642534,
"grad_norm": 0.1845703125,
"learning_rate": 7.183257918552037e-06,
"loss": 1.2114,
"step": 250
},
{
"epoch": 0.2839366515837104,
"grad_norm": 0.197265625,
"learning_rate": 7.171945701357467e-06,
"loss": 1.2034,
"step": 251
},
{
"epoch": 0.2850678733031674,
"grad_norm": 0.21484375,
"learning_rate": 7.160633484162897e-06,
"loss": 1.2785,
"step": 252
},
{
"epoch": 0.2861990950226244,
"grad_norm": 0.201171875,
"learning_rate": 7.1493212669683265e-06,
"loss": 1.2074,
"step": 253
},
{
"epoch": 0.2873303167420814,
"grad_norm": 0.2021484375,
"learning_rate": 7.138009049773756e-06,
"loss": 1.1929,
"step": 254
},
{
"epoch": 0.28846153846153844,
"grad_norm": 0.19140625,
"learning_rate": 7.126696832579186e-06,
"loss": 1.1939,
"step": 255
},
{
"epoch": 0.2895927601809955,
"grad_norm": 0.2158203125,
"learning_rate": 7.115384615384616e-06,
"loss": 1.2141,
"step": 256
},
{
"epoch": 0.2907239819004525,
"grad_norm": 0.1904296875,
"learning_rate": 7.104072398190046e-06,
"loss": 1.113,
"step": 257
},
{
"epoch": 0.2918552036199095,
"grad_norm": 0.2041015625,
"learning_rate": 7.092760180995476e-06,
"loss": 1.1975,
"step": 258
},
{
"epoch": 0.29298642533936653,
"grad_norm": 0.2001953125,
"learning_rate": 7.081447963800906e-06,
"loss": 1.146,
"step": 259
},
{
"epoch": 0.29411764705882354,
"grad_norm": 0.1875,
"learning_rate": 7.070135746606336e-06,
"loss": 1.1899,
"step": 260
},
{
"epoch": 0.29524886877828055,
"grad_norm": 0.181640625,
"learning_rate": 7.058823529411766e-06,
"loss": 1.2024,
"step": 261
},
{
"epoch": 0.29638009049773756,
"grad_norm": 0.1953125,
"learning_rate": 7.047511312217196e-06,
"loss": 1.2155,
"step": 262
},
{
"epoch": 0.29751131221719457,
"grad_norm": 0.17578125,
"learning_rate": 7.036199095022625e-06,
"loss": 1.1529,
"step": 263
},
{
"epoch": 0.2986425339366516,
"grad_norm": 0.1845703125,
"learning_rate": 7.024886877828054e-06,
"loss": 1.1562,
"step": 264
},
{
"epoch": 0.2997737556561086,
"grad_norm": 0.30078125,
"learning_rate": 7.013574660633484e-06,
"loss": 1.2337,
"step": 265
},
{
"epoch": 0.3009049773755656,
"grad_norm": 0.1904296875,
"learning_rate": 7.002262443438914e-06,
"loss": 1.1634,
"step": 266
},
{
"epoch": 0.3020361990950226,
"grad_norm": 0.1943359375,
"learning_rate": 6.990950226244344e-06,
"loss": 1.179,
"step": 267
},
{
"epoch": 0.3031674208144796,
"grad_norm": 0.2138671875,
"learning_rate": 6.9796380090497735e-06,
"loss": 1.2086,
"step": 268
},
{
"epoch": 0.3042986425339366,
"grad_norm": 0.19140625,
"learning_rate": 6.968325791855204e-06,
"loss": 1.2049,
"step": 269
},
{
"epoch": 0.3054298642533937,
"grad_norm": 0.19140625,
"learning_rate": 6.957013574660634e-06,
"loss": 1.148,
"step": 270
},
{
"epoch": 0.3065610859728507,
"grad_norm": 0.1826171875,
"learning_rate": 6.945701357466064e-06,
"loss": 1.1829,
"step": 271
},
{
"epoch": 0.3076923076923077,
"grad_norm": 0.1962890625,
"learning_rate": 6.934389140271494e-06,
"loss": 1.167,
"step": 272
},
{
"epoch": 0.3088235294117647,
"grad_norm": 0.1806640625,
"learning_rate": 6.923076923076923e-06,
"loss": 1.1553,
"step": 273
},
{
"epoch": 0.30995475113122173,
"grad_norm": 0.2421875,
"learning_rate": 6.911764705882353e-06,
"loss": 1.1104,
"step": 274
},
{
"epoch": 0.31108597285067874,
"grad_norm": 0.173828125,
"learning_rate": 6.900452488687783e-06,
"loss": 1.1723,
"step": 275
},
{
"epoch": 0.31221719457013575,
"grad_norm": 0.2060546875,
"learning_rate": 6.889140271493213e-06,
"loss": 1.1889,
"step": 276
},
{
"epoch": 0.31334841628959276,
"grad_norm": 0.193359375,
"learning_rate": 6.8778280542986434e-06,
"loss": 1.1377,
"step": 277
},
{
"epoch": 0.31447963800904977,
"grad_norm": 0.1826171875,
"learning_rate": 6.866515837104073e-06,
"loss": 1.1612,
"step": 278
},
{
"epoch": 0.3156108597285068,
"grad_norm": 0.1875,
"learning_rate": 6.855203619909503e-06,
"loss": 1.1962,
"step": 279
},
{
"epoch": 0.3167420814479638,
"grad_norm": 0.185546875,
"learning_rate": 6.843891402714933e-06,
"loss": 1.1606,
"step": 280
},
{
"epoch": 0.3178733031674208,
"grad_norm": 0.20703125,
"learning_rate": 6.832579185520363e-06,
"loss": 1.1594,
"step": 281
},
{
"epoch": 0.3190045248868778,
"grad_norm": 0.2421875,
"learning_rate": 6.8212669683257924e-06,
"loss": 1.2061,
"step": 282
},
{
"epoch": 0.3201357466063348,
"grad_norm": 0.189453125,
"learning_rate": 6.809954751131222e-06,
"loss": 1.2176,
"step": 283
},
{
"epoch": 0.3212669683257919,
"grad_norm": 0.1904296875,
"learning_rate": 6.798642533936652e-06,
"loss": 1.1447,
"step": 284
},
{
"epoch": 0.3223981900452489,
"grad_norm": 0.240234375,
"learning_rate": 6.787330316742083e-06,
"loss": 1.213,
"step": 285
},
{
"epoch": 0.3235294117647059,
"grad_norm": 0.26171875,
"learning_rate": 6.7760180995475125e-06,
"loss": 1.1823,
"step": 286
},
{
"epoch": 0.3246606334841629,
"grad_norm": 0.220703125,
"learning_rate": 6.764705882352942e-06,
"loss": 1.1653,
"step": 287
},
{
"epoch": 0.3257918552036199,
"grad_norm": 0.1845703125,
"learning_rate": 6.753393665158372e-06,
"loss": 1.1568,
"step": 288
},
{
"epoch": 0.3269230769230769,
"grad_norm": 0.17578125,
"learning_rate": 6.742081447963802e-06,
"loss": 1.1383,
"step": 289
},
{
"epoch": 0.32805429864253394,
"grad_norm": 0.2041015625,
"learning_rate": 6.730769230769232e-06,
"loss": 1.1379,
"step": 290
},
{
"epoch": 0.32918552036199095,
"grad_norm": 0.1962890625,
"learning_rate": 6.7194570135746615e-06,
"loss": 1.15,
"step": 291
},
{
"epoch": 0.33031674208144796,
"grad_norm": 0.1845703125,
"learning_rate": 6.7081447963800904e-06,
"loss": 1.1448,
"step": 292
},
{
"epoch": 0.33144796380090497,
"grad_norm": 0.20703125,
"learning_rate": 6.69683257918552e-06,
"loss": 1.1672,
"step": 293
},
{
"epoch": 0.332579185520362,
"grad_norm": 0.251953125,
"learning_rate": 6.68552036199095e-06,
"loss": 1.2145,
"step": 294
},
{
"epoch": 0.333710407239819,
"grad_norm": 0.193359375,
"learning_rate": 6.67420814479638e-06,
"loss": 1.1633,
"step": 295
},
{
"epoch": 0.334841628959276,
"grad_norm": 0.197265625,
"learning_rate": 6.6628959276018105e-06,
"loss": 1.1639,
"step": 296
},
{
"epoch": 0.33597285067873306,
"grad_norm": 0.2021484375,
"learning_rate": 6.65158371040724e-06,
"loss": 1.2009,
"step": 297
},
{
"epoch": 0.33710407239819007,
"grad_norm": 0.1923828125,
"learning_rate": 6.64027149321267e-06,
"loss": 1.1588,
"step": 298
},
{
"epoch": 0.3382352941176471,
"grad_norm": 0.1923828125,
"learning_rate": 6.6289592760181e-06,
"loss": 1.1845,
"step": 299
},
{
"epoch": 0.3393665158371041,
"grad_norm": 0.1943359375,
"learning_rate": 6.61764705882353e-06,
"loss": 1.1113,
"step": 300
},
{
"epoch": 0.3404977375565611,
"grad_norm": 0.20703125,
"learning_rate": 6.6063348416289595e-06,
"loss": 1.1619,
"step": 301
},
{
"epoch": 0.3416289592760181,
"grad_norm": 0.2060546875,
"learning_rate": 6.595022624434389e-06,
"loss": 1.157,
"step": 302
},
{
"epoch": 0.3427601809954751,
"grad_norm": 0.1943359375,
"learning_rate": 6.583710407239819e-06,
"loss": 1.1678,
"step": 303
},
{
"epoch": 0.3438914027149321,
"grad_norm": 0.21484375,
"learning_rate": 6.57239819004525e-06,
"loss": 1.1723,
"step": 304
},
{
"epoch": 0.34502262443438914,
"grad_norm": 0.21484375,
"learning_rate": 6.5610859728506795e-06,
"loss": 1.2325,
"step": 305
},
{
"epoch": 0.34615384615384615,
"grad_norm": 0.21484375,
"learning_rate": 6.549773755656109e-06,
"loss": 1.2244,
"step": 306
},
{
"epoch": 0.34728506787330315,
"grad_norm": 0.193359375,
"learning_rate": 6.538461538461539e-06,
"loss": 1.1316,
"step": 307
},
{
"epoch": 0.34841628959276016,
"grad_norm": 0.205078125,
"learning_rate": 6.527149321266969e-06,
"loss": 1.1428,
"step": 308
},
{
"epoch": 0.3495475113122172,
"grad_norm": 0.1884765625,
"learning_rate": 6.515837104072399e-06,
"loss": 1.1869,
"step": 309
},
{
"epoch": 0.3506787330316742,
"grad_norm": 0.2109375,
"learning_rate": 6.5045248868778285e-06,
"loss": 1.1797,
"step": 310
},
{
"epoch": 0.35180995475113125,
"grad_norm": 0.2099609375,
"learning_rate": 6.493212669683258e-06,
"loss": 1.1232,
"step": 311
},
{
"epoch": 0.35294117647058826,
"grad_norm": 0.2080078125,
"learning_rate": 6.481900452488689e-06,
"loss": 1.1267,
"step": 312
},
{
"epoch": 0.35407239819004527,
"grad_norm": 0.1962890625,
"learning_rate": 6.470588235294119e-06,
"loss": 1.1128,
"step": 313
},
{
"epoch": 0.3552036199095023,
"grad_norm": 0.1875,
"learning_rate": 6.4592760180995486e-06,
"loss": 1.127,
"step": 314
},
{
"epoch": 0.3563348416289593,
"grad_norm": 0.1904296875,
"learning_rate": 6.447963800904978e-06,
"loss": 1.1472,
"step": 315
},
{
"epoch": 0.3574660633484163,
"grad_norm": 0.24609375,
"learning_rate": 6.436651583710408e-06,
"loss": 1.1622,
"step": 316
},
{
"epoch": 0.3585972850678733,
"grad_norm": 0.2177734375,
"learning_rate": 6.425339366515838e-06,
"loss": 1.1925,
"step": 317
},
{
"epoch": 0.3597285067873303,
"grad_norm": 0.189453125,
"learning_rate": 6.414027149321268e-06,
"loss": 1.1237,
"step": 318
},
{
"epoch": 0.3608597285067873,
"grad_norm": 0.1923828125,
"learning_rate": 6.402714932126697e-06,
"loss": 1.0907,
"step": 319
},
{
"epoch": 0.36199095022624433,
"grad_norm": 0.2001953125,
"learning_rate": 6.3914027149321265e-06,
"loss": 1.1451,
"step": 320
},
{
"epoch": 0.36312217194570134,
"grad_norm": 0.208984375,
"learning_rate": 6.380090497737556e-06,
"loss": 1.1393,
"step": 321
},
{
"epoch": 0.36425339366515835,
"grad_norm": 0.2041015625,
"learning_rate": 6.368778280542986e-06,
"loss": 1.1327,
"step": 322
},
{
"epoch": 0.36538461538461536,
"grad_norm": 0.19921875,
"learning_rate": 6.357466063348417e-06,
"loss": 1.1765,
"step": 323
},
{
"epoch": 0.3665158371040724,
"grad_norm": 0.1865234375,
"learning_rate": 6.3461538461538466e-06,
"loss": 1.1515,
"step": 324
},
{
"epoch": 0.36764705882352944,
"grad_norm": 0.21484375,
"learning_rate": 6.334841628959276e-06,
"loss": 1.1219,
"step": 325
},
{
"epoch": 0.36877828054298645,
"grad_norm": 0.189453125,
"learning_rate": 6.323529411764706e-06,
"loss": 1.1341,
"step": 326
},
{
"epoch": 0.36990950226244346,
"grad_norm": 0.201171875,
"learning_rate": 6.312217194570136e-06,
"loss": 1.1507,
"step": 327
},
{
"epoch": 0.37104072398190047,
"grad_norm": 0.2109375,
"learning_rate": 6.300904977375566e-06,
"loss": 1.1651,
"step": 328
},
{
"epoch": 0.3721719457013575,
"grad_norm": 0.2041015625,
"learning_rate": 6.2895927601809956e-06,
"loss": 1.1539,
"step": 329
},
{
"epoch": 0.3733031674208145,
"grad_norm": 0.1962890625,
"learning_rate": 6.278280542986425e-06,
"loss": 1.1578,
"step": 330
},
{
"epoch": 0.3744343891402715,
"grad_norm": 0.193359375,
"learning_rate": 6.266968325791856e-06,
"loss": 1.1433,
"step": 331
},
{
"epoch": 0.3755656108597285,
"grad_norm": 0.2734375,
"learning_rate": 6.255656108597286e-06,
"loss": 1.1932,
"step": 332
},
{
"epoch": 0.3766968325791855,
"grad_norm": 0.21484375,
"learning_rate": 6.244343891402716e-06,
"loss": 1.2,
"step": 333
},
{
"epoch": 0.3778280542986425,
"grad_norm": 0.19921875,
"learning_rate": 6.233031674208145e-06,
"loss": 1.1403,
"step": 334
},
{
"epoch": 0.37895927601809953,
"grad_norm": 0.212890625,
"learning_rate": 6.221719457013575e-06,
"loss": 1.1434,
"step": 335
},
{
"epoch": 0.38009049773755654,
"grad_norm": 0.19921875,
"learning_rate": 6.210407239819005e-06,
"loss": 1.1136,
"step": 336
},
{
"epoch": 0.38122171945701355,
"grad_norm": 0.1845703125,
"learning_rate": 6.199095022624435e-06,
"loss": 1.1599,
"step": 337
},
{
"epoch": 0.38235294117647056,
"grad_norm": 0.1875,
"learning_rate": 6.187782805429865e-06,
"loss": 1.1697,
"step": 338
},
{
"epoch": 0.3834841628959276,
"grad_norm": 0.1875,
"learning_rate": 6.176470588235295e-06,
"loss": 1.1454,
"step": 339
},
{
"epoch": 0.38461538461538464,
"grad_norm": 0.263671875,
"learning_rate": 6.165158371040725e-06,
"loss": 1.2008,
"step": 340
},
{
"epoch": 0.38574660633484165,
"grad_norm": 0.20703125,
"learning_rate": 6.153846153846155e-06,
"loss": 1.1642,
"step": 341
},
{
"epoch": 0.38687782805429866,
"grad_norm": 0.416015625,
"learning_rate": 6.142533936651585e-06,
"loss": 1.1515,
"step": 342
},
{
"epoch": 0.38800904977375567,
"grad_norm": 0.22265625,
"learning_rate": 6.1312217194570145e-06,
"loss": 1.1497,
"step": 343
},
{
"epoch": 0.3891402714932127,
"grad_norm": 0.201171875,
"learning_rate": 6.119909502262444e-06,
"loss": 1.1322,
"step": 344
},
{
"epoch": 0.3902714932126697,
"grad_norm": 0.2119140625,
"learning_rate": 6.108597285067874e-06,
"loss": 1.1292,
"step": 345
},
{
"epoch": 0.3914027149321267,
"grad_norm": 0.203125,
"learning_rate": 6.097285067873304e-06,
"loss": 1.1546,
"step": 346
},
{
"epoch": 0.3925339366515837,
"grad_norm": 0.2119140625,
"learning_rate": 6.085972850678733e-06,
"loss": 1.084,
"step": 347
},
{
"epoch": 0.3936651583710407,
"grad_norm": 0.2060546875,
"learning_rate": 6.074660633484163e-06,
"loss": 1.1276,
"step": 348
},
{
"epoch": 0.3947963800904977,
"grad_norm": 0.2021484375,
"learning_rate": 6.0633484162895924e-06,
"loss": 1.0943,
"step": 349
},
{
"epoch": 0.39592760180995473,
"grad_norm": 0.203125,
"learning_rate": 6.052036199095023e-06,
"loss": 1.159,
"step": 350
},
{
"epoch": 0.39705882352941174,
"grad_norm": 0.2021484375,
"learning_rate": 6.040723981900453e-06,
"loss": 1.1347,
"step": 351
},
{
"epoch": 0.39819004524886875,
"grad_norm": 0.201171875,
"learning_rate": 6.029411764705883e-06,
"loss": 1.1376,
"step": 352
},
{
"epoch": 0.3993212669683258,
"grad_norm": 0.2021484375,
"learning_rate": 6.0180995475113125e-06,
"loss": 1.1282,
"step": 353
},
{
"epoch": 0.4004524886877828,
"grad_norm": 0.19921875,
"learning_rate": 6.006787330316742e-06,
"loss": 1.1723,
"step": 354
},
{
"epoch": 0.40158371040723984,
"grad_norm": 0.2353515625,
"learning_rate": 5.995475113122172e-06,
"loss": 1.1991,
"step": 355
},
{
"epoch": 0.40271493212669685,
"grad_norm": 0.208984375,
"learning_rate": 5.984162895927602e-06,
"loss": 1.1813,
"step": 356
},
{
"epoch": 0.40384615384615385,
"grad_norm": 0.2060546875,
"learning_rate": 5.972850678733032e-06,
"loss": 1.0853,
"step": 357
},
{
"epoch": 0.40497737556561086,
"grad_norm": 0.2021484375,
"learning_rate": 5.961538461538462e-06,
"loss": 1.1625,
"step": 358
},
{
"epoch": 0.4061085972850679,
"grad_norm": 0.203125,
"learning_rate": 5.950226244343892e-06,
"loss": 1.0982,
"step": 359
},
{
"epoch": 0.4072398190045249,
"grad_norm": 0.224609375,
"learning_rate": 5.938914027149322e-06,
"loss": 1.1807,
"step": 360
},
{
"epoch": 0.4083710407239819,
"grad_norm": 0.208984375,
"learning_rate": 5.927601809954752e-06,
"loss": 1.0939,
"step": 361
},
{
"epoch": 0.4095022624434389,
"grad_norm": 0.2080078125,
"learning_rate": 5.9162895927601815e-06,
"loss": 1.1124,
"step": 362
},
{
"epoch": 0.4106334841628959,
"grad_norm": 0.201171875,
"learning_rate": 5.904977375565611e-06,
"loss": 1.135,
"step": 363
},
{
"epoch": 0.4117647058823529,
"grad_norm": 0.1953125,
"learning_rate": 5.893665158371041e-06,
"loss": 1.1403,
"step": 364
},
{
"epoch": 0.41289592760180993,
"grad_norm": 0.2109375,
"learning_rate": 5.882352941176471e-06,
"loss": 1.1143,
"step": 365
},
{
"epoch": 0.41402714932126694,
"grad_norm": 0.1962890625,
"learning_rate": 5.8710407239819016e-06,
"loss": 1.0892,
"step": 366
},
{
"epoch": 0.415158371040724,
"grad_norm": 0.224609375,
"learning_rate": 5.859728506787331e-06,
"loss": 1.1693,
"step": 367
},
{
"epoch": 0.416289592760181,
"grad_norm": 0.21875,
"learning_rate": 5.848416289592761e-06,
"loss": 1.1936,
"step": 368
},
{
"epoch": 0.417420814479638,
"grad_norm": 0.193359375,
"learning_rate": 5.837104072398191e-06,
"loss": 1.1201,
"step": 369
},
{
"epoch": 0.41855203619909503,
"grad_norm": 0.1953125,
"learning_rate": 5.825791855203621e-06,
"loss": 1.1224,
"step": 370
},
{
"epoch": 0.41968325791855204,
"grad_norm": 0.208984375,
"learning_rate": 5.8144796380090506e-06,
"loss": 1.1333,
"step": 371
},
{
"epoch": 0.42081447963800905,
"grad_norm": 0.365234375,
"learning_rate": 5.80316742081448e-06,
"loss": 1.2723,
"step": 372
},
{
"epoch": 0.42194570135746606,
"grad_norm": 0.205078125,
"learning_rate": 5.79185520361991e-06,
"loss": 1.1081,
"step": 373
},
{
"epoch": 0.4230769230769231,
"grad_norm": 0.2021484375,
"learning_rate": 5.780542986425339e-06,
"loss": 1.0992,
"step": 374
},
{
"epoch": 0.4242081447963801,
"grad_norm": 0.232421875,
"learning_rate": 5.769230769230769e-06,
"loss": 1.1617,
"step": 375
},
{
"epoch": 0.4253393665158371,
"grad_norm": 0.23046875,
"learning_rate": 5.757918552036199e-06,
"loss": 1.1431,
"step": 376
},
{
"epoch": 0.4264705882352941,
"grad_norm": 0.2041015625,
"learning_rate": 5.746606334841629e-06,
"loss": 1.1212,
"step": 377
},
{
"epoch": 0.4276018099547511,
"grad_norm": 0.2080078125,
"learning_rate": 5.735294117647059e-06,
"loss": 1.1197,
"step": 378
},
{
"epoch": 0.4287330316742081,
"grad_norm": 0.22265625,
"learning_rate": 5.723981900452489e-06,
"loss": 1.1277,
"step": 379
},
{
"epoch": 0.4298642533936652,
"grad_norm": 0.25,
"learning_rate": 5.712669683257919e-06,
"loss": 1.1565,
"step": 380
},
{
"epoch": 0.4309954751131222,
"grad_norm": 0.1923828125,
"learning_rate": 5.7013574660633486e-06,
"loss": 1.1011,
"step": 381
},
{
"epoch": 0.4321266968325792,
"grad_norm": 0.232421875,
"learning_rate": 5.690045248868778e-06,
"loss": 1.1541,
"step": 382
},
{
"epoch": 0.4332579185520362,
"grad_norm": 0.2333984375,
"learning_rate": 5.678733031674208e-06,
"loss": 1.1421,
"step": 383
},
{
"epoch": 0.4343891402714932,
"grad_norm": 0.2265625,
"learning_rate": 5.667420814479638e-06,
"loss": 1.1621,
"step": 384
},
{
"epoch": 0.43552036199095023,
"grad_norm": 0.205078125,
"learning_rate": 5.656108597285069e-06,
"loss": 1.1234,
"step": 385
},
{
"epoch": 0.43665158371040724,
"grad_norm": 0.1982421875,
"learning_rate": 5.644796380090498e-06,
"loss": 1.1542,
"step": 386
},
{
"epoch": 0.43778280542986425,
"grad_norm": 0.212890625,
"learning_rate": 5.633484162895928e-06,
"loss": 1.129,
"step": 387
},
{
"epoch": 0.43891402714932126,
"grad_norm": 0.197265625,
"learning_rate": 5.622171945701358e-06,
"loss": 1.1454,
"step": 388
},
{
"epoch": 0.44004524886877827,
"grad_norm": 0.2119140625,
"learning_rate": 5.610859728506788e-06,
"loss": 1.1561,
"step": 389
},
{
"epoch": 0.4411764705882353,
"grad_norm": 0.2158203125,
"learning_rate": 5.599547511312218e-06,
"loss": 1.1198,
"step": 390
},
{
"epoch": 0.4423076923076923,
"grad_norm": 0.212890625,
"learning_rate": 5.588235294117647e-06,
"loss": 1.1136,
"step": 391
},
{
"epoch": 0.4434389140271493,
"grad_norm": 0.25,
"learning_rate": 5.576923076923077e-06,
"loss": 1.1662,
"step": 392
},
{
"epoch": 0.4445701357466063,
"grad_norm": 0.2451171875,
"learning_rate": 5.565610859728508e-06,
"loss": 1.189,
"step": 393
},
{
"epoch": 0.4457013574660634,
"grad_norm": 0.208984375,
"learning_rate": 5.554298642533938e-06,
"loss": 1.1044,
"step": 394
},
{
"epoch": 0.4468325791855204,
"grad_norm": 0.2177734375,
"learning_rate": 5.5429864253393675e-06,
"loss": 1.1309,
"step": 395
},
{
"epoch": 0.4479638009049774,
"grad_norm": 0.2138671875,
"learning_rate": 5.531674208144797e-06,
"loss": 1.0932,
"step": 396
},
{
"epoch": 0.4490950226244344,
"grad_norm": 0.19921875,
"learning_rate": 5.520361990950227e-06,
"loss": 1.1521,
"step": 397
},
{
"epoch": 0.4502262443438914,
"grad_norm": 0.2177734375,
"learning_rate": 5.509049773755657e-06,
"loss": 1.1691,
"step": 398
},
{
"epoch": 0.4513574660633484,
"grad_norm": 0.2021484375,
"learning_rate": 5.497737556561087e-06,
"loss": 1.1529,
"step": 399
},
{
"epoch": 0.45248868778280543,
"grad_norm": 0.220703125,
"learning_rate": 5.4864253393665165e-06,
"loss": 1.1646,
"step": 400
},
{
"epoch": 0.45361990950226244,
"grad_norm": 0.20703125,
"learning_rate": 5.475113122171946e-06,
"loss": 1.1544,
"step": 401
},
{
"epoch": 0.45475113122171945,
"grad_norm": 0.19140625,
"learning_rate": 5.463800904977375e-06,
"loss": 1.1555,
"step": 402
},
{
"epoch": 0.45588235294117646,
"grad_norm": 0.2177734375,
"learning_rate": 5.452488687782805e-06,
"loss": 1.1112,
"step": 403
},
{
"epoch": 0.45701357466063347,
"grad_norm": 0.2490234375,
"learning_rate": 5.441176470588236e-06,
"loss": 1.1644,
"step": 404
},
{
"epoch": 0.4581447963800905,
"grad_norm": 0.2060546875,
"learning_rate": 5.4298642533936655e-06,
"loss": 1.1246,
"step": 405
},
{
"epoch": 0.4592760180995475,
"grad_norm": 0.2021484375,
"learning_rate": 5.418552036199095e-06,
"loss": 1.1515,
"step": 406
},
{
"epoch": 0.4604072398190045,
"grad_norm": 0.20703125,
"learning_rate": 5.407239819004525e-06,
"loss": 1.1142,
"step": 407
},
{
"epoch": 0.46153846153846156,
"grad_norm": 0.201171875,
"learning_rate": 5.395927601809955e-06,
"loss": 1.1153,
"step": 408
},
{
"epoch": 0.4626696832579186,
"grad_norm": 0.2109375,
"learning_rate": 5.384615384615385e-06,
"loss": 1.1154,
"step": 409
},
{
"epoch": 0.4638009049773756,
"grad_norm": 0.2099609375,
"learning_rate": 5.3733031674208145e-06,
"loss": 1.14,
"step": 410
},
{
"epoch": 0.4649321266968326,
"grad_norm": 0.2294921875,
"learning_rate": 5.361990950226244e-06,
"loss": 1.1308,
"step": 411
},
{
"epoch": 0.4660633484162896,
"grad_norm": 0.2255859375,
"learning_rate": 5.350678733031675e-06,
"loss": 1.14,
"step": 412
},
{
"epoch": 0.4671945701357466,
"grad_norm": 0.263671875,
"learning_rate": 5.339366515837105e-06,
"loss": 1.156,
"step": 413
},
{
"epoch": 0.4683257918552036,
"grad_norm": 0.205078125,
"learning_rate": 5.3280542986425345e-06,
"loss": 1.159,
"step": 414
},
{
"epoch": 0.46945701357466063,
"grad_norm": 0.2353515625,
"learning_rate": 5.316742081447964e-06,
"loss": 1.1356,
"step": 415
},
{
"epoch": 0.47058823529411764,
"grad_norm": 0.21484375,
"learning_rate": 5.305429864253394e-06,
"loss": 1.1629,
"step": 416
},
{
"epoch": 0.47171945701357465,
"grad_norm": 0.203125,
"learning_rate": 5.294117647058824e-06,
"loss": 1.1007,
"step": 417
},
{
"epoch": 0.47285067873303166,
"grad_norm": 0.2021484375,
"learning_rate": 5.282805429864254e-06,
"loss": 1.1407,
"step": 418
},
{
"epoch": 0.47398190045248867,
"grad_norm": 0.1943359375,
"learning_rate": 5.2714932126696835e-06,
"loss": 1.1356,
"step": 419
},
{
"epoch": 0.4751131221719457,
"grad_norm": 0.2236328125,
"learning_rate": 5.260180995475114e-06,
"loss": 1.1715,
"step": 420
},
{
"epoch": 0.4762443438914027,
"grad_norm": 0.240234375,
"learning_rate": 5.248868778280544e-06,
"loss": 1.1313,
"step": 421
},
{
"epoch": 0.47737556561085975,
"grad_norm": 0.2177734375,
"learning_rate": 5.237556561085974e-06,
"loss": 1.1622,
"step": 422
},
{
"epoch": 0.47850678733031676,
"grad_norm": 0.205078125,
"learning_rate": 5.2262443438914036e-06,
"loss": 1.1589,
"step": 423
},
{
"epoch": 0.4796380090497738,
"grad_norm": 0.2197265625,
"learning_rate": 5.214932126696833e-06,
"loss": 1.1266,
"step": 424
},
{
"epoch": 0.4807692307692308,
"grad_norm": 0.2412109375,
"learning_rate": 5.203619909502263e-06,
"loss": 1.1332,
"step": 425
},
{
"epoch": 0.4819004524886878,
"grad_norm": 0.2314453125,
"learning_rate": 5.192307692307693e-06,
"loss": 1.1512,
"step": 426
},
{
"epoch": 0.4830316742081448,
"grad_norm": 0.2119140625,
"learning_rate": 5.180995475113123e-06,
"loss": 1.0958,
"step": 427
},
{
"epoch": 0.4841628959276018,
"grad_norm": 0.2294921875,
"learning_rate": 5.1696832579185526e-06,
"loss": 1.1246,
"step": 428
},
{
"epoch": 0.4852941176470588,
"grad_norm": 0.2412109375,
"learning_rate": 5.158371040723983e-06,
"loss": 1.1052,
"step": 429
},
{
"epoch": 0.48642533936651583,
"grad_norm": 0.2314453125,
"learning_rate": 5.147058823529411e-06,
"loss": 1.1643,
"step": 430
},
{
"epoch": 0.48755656108597284,
"grad_norm": 0.283203125,
"learning_rate": 5.135746606334842e-06,
"loss": 1.2126,
"step": 431
},
{
"epoch": 0.48868778280542985,
"grad_norm": 0.2255859375,
"learning_rate": 5.124434389140272e-06,
"loss": 1.148,
"step": 432
},
{
"epoch": 0.48981900452488686,
"grad_norm": 0.2216796875,
"learning_rate": 5.1131221719457016e-06,
"loss": 1.1636,
"step": 433
},
{
"epoch": 0.49095022624434387,
"grad_norm": 0.2109375,
"learning_rate": 5.101809954751131e-06,
"loss": 1.0961,
"step": 434
},
{
"epoch": 0.4920814479638009,
"grad_norm": 0.2158203125,
"learning_rate": 5.090497737556561e-06,
"loss": 1.1251,
"step": 435
},
{
"epoch": 0.49321266968325794,
"grad_norm": 0.2451171875,
"learning_rate": 5.079185520361991e-06,
"loss": 1.2031,
"step": 436
},
{
"epoch": 0.49434389140271495,
"grad_norm": 0.21875,
"learning_rate": 5.067873303167421e-06,
"loss": 1.14,
"step": 437
},
{
"epoch": 0.49547511312217196,
"grad_norm": 0.2333984375,
"learning_rate": 5.0565610859728506e-06,
"loss": 1.1681,
"step": 438
},
{
"epoch": 0.49660633484162897,
"grad_norm": 0.1982421875,
"learning_rate": 5.045248868778281e-06,
"loss": 1.1226,
"step": 439
},
{
"epoch": 0.497737556561086,
"grad_norm": 0.2294921875,
"learning_rate": 5.033936651583711e-06,
"loss": 1.1938,
"step": 440
},
{
"epoch": 0.498868778280543,
"grad_norm": 0.21875,
"learning_rate": 5.022624434389141e-06,
"loss": 1.0975,
"step": 441
},
{
"epoch": 0.5,
"grad_norm": 0.2021484375,
"learning_rate": 5.011312217194571e-06,
"loss": 1.1315,
"step": 442
},
{
"epoch": 0.501131221719457,
"grad_norm": 0.2265625,
"learning_rate": 5e-06,
"loss": 1.1283,
"step": 443
},
{
"epoch": 0.502262443438914,
"grad_norm": 0.2138671875,
"learning_rate": 4.98868778280543e-06,
"loss": 1.1396,
"step": 444
},
{
"epoch": 0.503393665158371,
"grad_norm": 0.2158203125,
"learning_rate": 4.97737556561086e-06,
"loss": 1.1445,
"step": 445
},
{
"epoch": 0.504524886877828,
"grad_norm": 0.216796875,
"learning_rate": 4.96606334841629e-06,
"loss": 1.1332,
"step": 446
},
{
"epoch": 0.505656108597285,
"grad_norm": 0.2158203125,
"learning_rate": 4.95475113122172e-06,
"loss": 1.1461,
"step": 447
},
{
"epoch": 0.5067873303167421,
"grad_norm": 0.23046875,
"learning_rate": 4.94343891402715e-06,
"loss": 1.1842,
"step": 448
},
{
"epoch": 0.5079185520361991,
"grad_norm": 0.2392578125,
"learning_rate": 4.93212669683258e-06,
"loss": 1.085,
"step": 449
},
{
"epoch": 0.5090497737556561,
"grad_norm": 0.216796875,
"learning_rate": 4.920814479638009e-06,
"loss": 1.1024,
"step": 450
},
{
"epoch": 0.5101809954751131,
"grad_norm": 0.283203125,
"learning_rate": 4.909502262443439e-06,
"loss": 1.087,
"step": 451
},
{
"epoch": 0.5113122171945701,
"grad_norm": 0.205078125,
"learning_rate": 4.898190045248869e-06,
"loss": 1.1482,
"step": 452
},
{
"epoch": 0.5124434389140271,
"grad_norm": 0.205078125,
"learning_rate": 4.886877828054299e-06,
"loss": 1.124,
"step": 453
},
{
"epoch": 0.5135746606334841,
"grad_norm": 0.244140625,
"learning_rate": 4.875565610859729e-06,
"loss": 1.1687,
"step": 454
},
{
"epoch": 0.5147058823529411,
"grad_norm": 0.24609375,
"learning_rate": 4.864253393665159e-06,
"loss": 1.1434,
"step": 455
},
{
"epoch": 0.5158371040723982,
"grad_norm": 0.21875,
"learning_rate": 4.852941176470589e-06,
"loss": 1.0929,
"step": 456
},
{
"epoch": 0.5169683257918553,
"grad_norm": 0.203125,
"learning_rate": 4.8416289592760185e-06,
"loss": 1.1387,
"step": 457
},
{
"epoch": 0.5180995475113123,
"grad_norm": 0.220703125,
"learning_rate": 4.830316742081448e-06,
"loss": 1.1135,
"step": 458
},
{
"epoch": 0.5192307692307693,
"grad_norm": 0.21875,
"learning_rate": 4.819004524886878e-06,
"loss": 1.155,
"step": 459
},
{
"epoch": 0.5203619909502263,
"grad_norm": 0.2412109375,
"learning_rate": 4.807692307692308e-06,
"loss": 1.127,
"step": 460
},
{
"epoch": 0.5214932126696833,
"grad_norm": 0.283203125,
"learning_rate": 4.7963800904977385e-06,
"loss": 1.0928,
"step": 461
},
{
"epoch": 0.5226244343891403,
"grad_norm": 0.2314453125,
"learning_rate": 4.785067873303168e-06,
"loss": 1.158,
"step": 462
},
{
"epoch": 0.5237556561085973,
"grad_norm": 0.224609375,
"learning_rate": 4.773755656108598e-06,
"loss": 1.1215,
"step": 463
},
{
"epoch": 0.5248868778280543,
"grad_norm": 0.2041015625,
"learning_rate": 4.762443438914027e-06,
"loss": 1.1089,
"step": 464
},
{
"epoch": 0.5260180995475113,
"grad_norm": 0.23828125,
"learning_rate": 4.751131221719457e-06,
"loss": 1.1042,
"step": 465
},
{
"epoch": 0.5271493212669683,
"grad_norm": 0.2021484375,
"learning_rate": 4.739819004524887e-06,
"loss": 1.095,
"step": 466
},
{
"epoch": 0.5282805429864253,
"grad_norm": 0.2236328125,
"learning_rate": 4.728506787330317e-06,
"loss": 1.1472,
"step": 467
},
{
"epoch": 0.5294117647058824,
"grad_norm": 0.2138671875,
"learning_rate": 4.717194570135747e-06,
"loss": 1.1102,
"step": 468
},
{
"epoch": 0.5305429864253394,
"grad_norm": 0.1923828125,
"learning_rate": 4.705882352941177e-06,
"loss": 1.1127,
"step": 469
},
{
"epoch": 0.5316742081447964,
"grad_norm": 0.251953125,
"learning_rate": 4.694570135746607e-06,
"loss": 1.1378,
"step": 470
},
{
"epoch": 0.5328054298642534,
"grad_norm": 0.2001953125,
"learning_rate": 4.6832579185520365e-06,
"loss": 1.1319,
"step": 471
},
{
"epoch": 0.5339366515837104,
"grad_norm": 0.2158203125,
"learning_rate": 4.671945701357466e-06,
"loss": 1.1482,
"step": 472
},
{
"epoch": 0.5350678733031674,
"grad_norm": 0.2119140625,
"learning_rate": 4.660633484162896e-06,
"loss": 1.1278,
"step": 473
},
{
"epoch": 0.5361990950226244,
"grad_norm": 0.212890625,
"learning_rate": 4.649321266968326e-06,
"loss": 1.1325,
"step": 474
},
{
"epoch": 0.5373303167420814,
"grad_norm": 0.2470703125,
"learning_rate": 4.6380090497737566e-06,
"loss": 1.1582,
"step": 475
},
{
"epoch": 0.5384615384615384,
"grad_norm": 0.197265625,
"learning_rate": 4.626696832579186e-06,
"loss": 1.1504,
"step": 476
},
{
"epoch": 0.5395927601809954,
"grad_norm": 0.2197265625,
"learning_rate": 4.615384615384616e-06,
"loss": 1.1249,
"step": 477
},
{
"epoch": 0.5407239819004525,
"grad_norm": 0.224609375,
"learning_rate": 4.604072398190045e-06,
"loss": 1.1087,
"step": 478
},
{
"epoch": 0.5418552036199095,
"grad_norm": 0.2373046875,
"learning_rate": 4.592760180995475e-06,
"loss": 1.0925,
"step": 479
},
{
"epoch": 0.5429864253393665,
"grad_norm": 0.2109375,
"learning_rate": 4.5814479638009056e-06,
"loss": 1.1199,
"step": 480
},
{
"epoch": 0.5441176470588235,
"grad_norm": 0.259765625,
"learning_rate": 4.570135746606335e-06,
"loss": 1.1497,
"step": 481
},
{
"epoch": 0.5452488687782805,
"grad_norm": 0.208984375,
"learning_rate": 4.558823529411765e-06,
"loss": 1.1144,
"step": 482
},
{
"epoch": 0.5463800904977375,
"grad_norm": 0.2109375,
"learning_rate": 4.547511312217195e-06,
"loss": 1.1491,
"step": 483
},
{
"epoch": 0.5475113122171946,
"grad_norm": 0.2216796875,
"learning_rate": 4.536199095022625e-06,
"loss": 1.0874,
"step": 484
},
{
"epoch": 0.5486425339366516,
"grad_norm": 0.2470703125,
"learning_rate": 4.5248868778280546e-06,
"loss": 1.1062,
"step": 485
},
{
"epoch": 0.5497737556561086,
"grad_norm": 0.220703125,
"learning_rate": 4.513574660633484e-06,
"loss": 1.0967,
"step": 486
},
{
"epoch": 0.5509049773755657,
"grad_norm": 0.240234375,
"learning_rate": 4.502262443438914e-06,
"loss": 1.1694,
"step": 487
},
{
"epoch": 0.5520361990950227,
"grad_norm": 0.2060546875,
"learning_rate": 4.490950226244345e-06,
"loss": 1.1047,
"step": 488
},
{
"epoch": 0.5531674208144797,
"grad_norm": 0.271484375,
"learning_rate": 4.479638009049775e-06,
"loss": 1.1364,
"step": 489
},
{
"epoch": 0.5542986425339367,
"grad_norm": 0.234375,
"learning_rate": 4.468325791855204e-06,
"loss": 1.1269,
"step": 490
},
{
"epoch": 0.5554298642533937,
"grad_norm": 0.2265625,
"learning_rate": 4.457013574660634e-06,
"loss": 1.1028,
"step": 491
},
{
"epoch": 0.5565610859728507,
"grad_norm": 0.2490234375,
"learning_rate": 4.445701357466063e-06,
"loss": 1.0903,
"step": 492
},
{
"epoch": 0.5576923076923077,
"grad_norm": 0.2275390625,
"learning_rate": 4.434389140271493e-06,
"loss": 1.1222,
"step": 493
},
{
"epoch": 0.5588235294117647,
"grad_norm": 0.2177734375,
"learning_rate": 4.423076923076924e-06,
"loss": 1.142,
"step": 494
},
{
"epoch": 0.5599547511312217,
"grad_norm": 0.2138671875,
"learning_rate": 4.411764705882353e-06,
"loss": 1.0954,
"step": 495
},
{
"epoch": 0.5610859728506787,
"grad_norm": 0.2099609375,
"learning_rate": 4.400452488687783e-06,
"loss": 1.1308,
"step": 496
},
{
"epoch": 0.5622171945701357,
"grad_norm": 0.203125,
"learning_rate": 4.389140271493213e-06,
"loss": 1.1134,
"step": 497
},
{
"epoch": 0.5633484162895928,
"grad_norm": 0.212890625,
"learning_rate": 4.377828054298643e-06,
"loss": 1.1201,
"step": 498
},
{
"epoch": 0.5644796380090498,
"grad_norm": 0.212890625,
"learning_rate": 4.366515837104073e-06,
"loss": 1.1467,
"step": 499
},
{
"epoch": 0.5656108597285068,
"grad_norm": 0.220703125,
"learning_rate": 4.355203619909502e-06,
"loss": 1.1199,
"step": 500
},
{
"epoch": 0.5667420814479638,
"grad_norm": 0.251953125,
"learning_rate": 4.343891402714932e-06,
"loss": 1.0756,
"step": 501
},
{
"epoch": 0.5678733031674208,
"grad_norm": 0.203125,
"learning_rate": 4.332579185520363e-06,
"loss": 1.1345,
"step": 502
},
{
"epoch": 0.5690045248868778,
"grad_norm": 0.208984375,
"learning_rate": 4.321266968325793e-06,
"loss": 1.0918,
"step": 503
},
{
"epoch": 0.5701357466063348,
"grad_norm": 0.25,
"learning_rate": 4.3099547511312224e-06,
"loss": 1.1262,
"step": 504
},
{
"epoch": 0.5712669683257918,
"grad_norm": 0.212890625,
"learning_rate": 4.298642533936652e-06,
"loss": 1.1259,
"step": 505
},
{
"epoch": 0.5723981900452488,
"grad_norm": 0.23828125,
"learning_rate": 4.287330316742081e-06,
"loss": 1.1194,
"step": 506
},
{
"epoch": 0.5735294117647058,
"grad_norm": 0.2255859375,
"learning_rate": 4.276018099547512e-06,
"loss": 1.1617,
"step": 507
},
{
"epoch": 0.5746606334841629,
"grad_norm": 0.232421875,
"learning_rate": 4.264705882352942e-06,
"loss": 1.1506,
"step": 508
},
{
"epoch": 0.5757918552036199,
"grad_norm": 0.25390625,
"learning_rate": 4.2533936651583714e-06,
"loss": 1.1365,
"step": 509
},
{
"epoch": 0.5769230769230769,
"grad_norm": 0.2119140625,
"learning_rate": 4.242081447963801e-06,
"loss": 1.1106,
"step": 510
},
{
"epoch": 0.5780542986425339,
"grad_norm": 0.1982421875,
"learning_rate": 4.230769230769231e-06,
"loss": 1.0901,
"step": 511
},
{
"epoch": 0.579185520361991,
"grad_norm": 0.2255859375,
"learning_rate": 4.219457013574661e-06,
"loss": 1.1077,
"step": 512
},
{
"epoch": 0.580316742081448,
"grad_norm": 0.203125,
"learning_rate": 4.208144796380091e-06,
"loss": 1.0747,
"step": 513
},
{
"epoch": 0.581447963800905,
"grad_norm": 0.2216796875,
"learning_rate": 4.1968325791855205e-06,
"loss": 1.0941,
"step": 514
},
{
"epoch": 0.582579185520362,
"grad_norm": 0.216796875,
"learning_rate": 4.185520361990951e-06,
"loss": 1.1148,
"step": 515
},
{
"epoch": 0.583710407239819,
"grad_norm": 0.2177734375,
"learning_rate": 4.174208144796381e-06,
"loss": 1.1061,
"step": 516
},
{
"epoch": 0.584841628959276,
"grad_norm": 0.20703125,
"learning_rate": 4.162895927601811e-06,
"loss": 1.1365,
"step": 517
},
{
"epoch": 0.5859728506787331,
"grad_norm": 0.22265625,
"learning_rate": 4.1515837104072405e-06,
"loss": 1.1235,
"step": 518
},
{
"epoch": 0.5871040723981901,
"grad_norm": 0.2578125,
"learning_rate": 4.1402714932126695e-06,
"loss": 1.1317,
"step": 519
},
{
"epoch": 0.5882352941176471,
"grad_norm": 0.212890625,
"learning_rate": 4.128959276018099e-06,
"loss": 1.1141,
"step": 520
},
{
"epoch": 0.5893665158371041,
"grad_norm": 0.22265625,
"learning_rate": 4.11764705882353e-06,
"loss": 1.1361,
"step": 521
},
{
"epoch": 0.5904977375565611,
"grad_norm": 0.2236328125,
"learning_rate": 4.10633484162896e-06,
"loss": 1.0759,
"step": 522
},
{
"epoch": 0.5916289592760181,
"grad_norm": 0.2275390625,
"learning_rate": 4.0950226244343895e-06,
"loss": 1.1189,
"step": 523
},
{
"epoch": 0.5927601809954751,
"grad_norm": 0.2275390625,
"learning_rate": 4.083710407239819e-06,
"loss": 1.0791,
"step": 524
},
{
"epoch": 0.5938914027149321,
"grad_norm": 0.220703125,
"learning_rate": 4.072398190045249e-06,
"loss": 1.1148,
"step": 525
},
{
"epoch": 0.5950226244343891,
"grad_norm": 0.2060546875,
"learning_rate": 4.061085972850679e-06,
"loss": 1.102,
"step": 526
},
{
"epoch": 0.5961538461538461,
"grad_norm": 0.2060546875,
"learning_rate": 4.049773755656109e-06,
"loss": 1.0861,
"step": 527
},
{
"epoch": 0.5972850678733032,
"grad_norm": 0.208984375,
"learning_rate": 4.0384615384615385e-06,
"loss": 1.1087,
"step": 528
},
{
"epoch": 0.5984162895927602,
"grad_norm": 0.193359375,
"learning_rate": 4.027149321266969e-06,
"loss": 1.1187,
"step": 529
},
{
"epoch": 0.5995475113122172,
"grad_norm": 0.2412109375,
"learning_rate": 4.015837104072399e-06,
"loss": 1.1342,
"step": 530
},
{
"epoch": 0.6006787330316742,
"grad_norm": 0.216796875,
"learning_rate": 4.004524886877829e-06,
"loss": 1.1238,
"step": 531
},
{
"epoch": 0.6018099547511312,
"grad_norm": 0.2421875,
"learning_rate": 3.9932126696832585e-06,
"loss": 1.0837,
"step": 532
},
{
"epoch": 0.6029411764705882,
"grad_norm": 0.271484375,
"learning_rate": 3.9819004524886875e-06,
"loss": 1.1202,
"step": 533
},
{
"epoch": 0.6040723981900452,
"grad_norm": 0.2451171875,
"learning_rate": 3.970588235294118e-06,
"loss": 1.1398,
"step": 534
},
{
"epoch": 0.6052036199095022,
"grad_norm": 0.2314453125,
"learning_rate": 3.959276018099548e-06,
"loss": 1.1455,
"step": 535
},
{
"epoch": 0.6063348416289592,
"grad_norm": 0.2470703125,
"learning_rate": 3.947963800904978e-06,
"loss": 1.1629,
"step": 536
},
{
"epoch": 0.6074660633484162,
"grad_norm": 0.2421875,
"learning_rate": 3.9366515837104075e-06,
"loss": 1.1347,
"step": 537
},
{
"epoch": 0.6085972850678733,
"grad_norm": 0.23046875,
"learning_rate": 3.925339366515837e-06,
"loss": 1.0922,
"step": 538
},
{
"epoch": 0.6097285067873304,
"grad_norm": 0.2333984375,
"learning_rate": 3.914027149321267e-06,
"loss": 1.1268,
"step": 539
},
{
"epoch": 0.6108597285067874,
"grad_norm": 0.2333984375,
"learning_rate": 3.902714932126697e-06,
"loss": 1.1189,
"step": 540
},
{
"epoch": 0.6119909502262444,
"grad_norm": 0.236328125,
"learning_rate": 3.891402714932127e-06,
"loss": 1.1388,
"step": 541
},
{
"epoch": 0.6131221719457014,
"grad_norm": 0.22265625,
"learning_rate": 3.880090497737557e-06,
"loss": 1.1285,
"step": 542
},
{
"epoch": 0.6142533936651584,
"grad_norm": 0.21484375,
"learning_rate": 3.868778280542987e-06,
"loss": 1.131,
"step": 543
},
{
"epoch": 0.6153846153846154,
"grad_norm": 0.212890625,
"learning_rate": 3.857466063348417e-06,
"loss": 1.0935,
"step": 544
},
{
"epoch": 0.6165158371040724,
"grad_norm": 0.2109375,
"learning_rate": 3.846153846153847e-06,
"loss": 1.0844,
"step": 545
},
{
"epoch": 0.6176470588235294,
"grad_norm": 0.2421875,
"learning_rate": 3.834841628959277e-06,
"loss": 1.1136,
"step": 546
},
{
"epoch": 0.6187782805429864,
"grad_norm": 0.255859375,
"learning_rate": 3.8235294117647055e-06,
"loss": 1.173,
"step": 547
},
{
"epoch": 0.6199095022624435,
"grad_norm": 0.2197265625,
"learning_rate": 3.8122171945701358e-06,
"loss": 1.104,
"step": 548
},
{
"epoch": 0.6210407239819005,
"grad_norm": 0.212890625,
"learning_rate": 3.8009049773755656e-06,
"loss": 1.0798,
"step": 549
},
{
"epoch": 0.6221719457013575,
"grad_norm": 0.2236328125,
"learning_rate": 3.789592760180996e-06,
"loss": 1.1036,
"step": 550
},
{
"epoch": 0.6233031674208145,
"grad_norm": 0.267578125,
"learning_rate": 3.7782805429864256e-06,
"loss": 1.1188,
"step": 551
},
{
"epoch": 0.6244343891402715,
"grad_norm": 0.2109375,
"learning_rate": 3.7669683257918554e-06,
"loss": 1.0777,
"step": 552
},
{
"epoch": 0.6255656108597285,
"grad_norm": 0.2138671875,
"learning_rate": 3.755656108597285e-06,
"loss": 1.1135,
"step": 553
},
{
"epoch": 0.6266968325791855,
"grad_norm": 0.21484375,
"learning_rate": 3.7443438914027154e-06,
"loss": 1.0813,
"step": 554
},
{
"epoch": 0.6278280542986425,
"grad_norm": 0.21875,
"learning_rate": 3.7330316742081452e-06,
"loss": 1.1224,
"step": 555
},
{
"epoch": 0.6289592760180995,
"grad_norm": 0.263671875,
"learning_rate": 3.721719457013575e-06,
"loss": 1.0645,
"step": 556
},
{
"epoch": 0.6300904977375565,
"grad_norm": 0.236328125,
"learning_rate": 3.710407239819005e-06,
"loss": 1.0926,
"step": 557
},
{
"epoch": 0.6312217194570136,
"grad_norm": 0.2099609375,
"learning_rate": 3.699095022624435e-06,
"loss": 1.1189,
"step": 558
},
{
"epoch": 0.6323529411764706,
"grad_norm": 0.20703125,
"learning_rate": 3.687782805429865e-06,
"loss": 1.0578,
"step": 559
},
{
"epoch": 0.6334841628959276,
"grad_norm": 0.203125,
"learning_rate": 3.6764705882352946e-06,
"loss": 1.1047,
"step": 560
},
{
"epoch": 0.6346153846153846,
"grad_norm": 0.21484375,
"learning_rate": 3.665158371040724e-06,
"loss": 1.082,
"step": 561
},
{
"epoch": 0.6357466063348416,
"grad_norm": 0.23046875,
"learning_rate": 3.653846153846154e-06,
"loss": 1.1221,
"step": 562
},
{
"epoch": 0.6368778280542986,
"grad_norm": 0.2080078125,
"learning_rate": 3.642533936651584e-06,
"loss": 1.1112,
"step": 563
},
{
"epoch": 0.6380090497737556,
"grad_norm": 3.953125,
"learning_rate": 3.631221719457014e-06,
"loss": 1.0758,
"step": 564
},
{
"epoch": 0.6391402714932126,
"grad_norm": 0.203125,
"learning_rate": 3.6199095022624436e-06,
"loss": 1.1034,
"step": 565
},
{
"epoch": 0.6402714932126696,
"grad_norm": 0.2099609375,
"learning_rate": 3.6085972850678734e-06,
"loss": 1.1011,
"step": 566
},
{
"epoch": 0.6414027149321267,
"grad_norm": 0.2275390625,
"learning_rate": 3.5972850678733032e-06,
"loss": 1.1323,
"step": 567
},
{
"epoch": 0.6425339366515838,
"grad_norm": 0.2060546875,
"learning_rate": 3.5859728506787335e-06,
"loss": 1.1041,
"step": 568
},
{
"epoch": 0.6436651583710408,
"grad_norm": 0.19921875,
"learning_rate": 3.5746606334841633e-06,
"loss": 1.1017,
"step": 569
},
{
"epoch": 0.6447963800904978,
"grad_norm": 0.2158203125,
"learning_rate": 3.563348416289593e-06,
"loss": 1.0711,
"step": 570
},
{
"epoch": 0.6459276018099548,
"grad_norm": 0.25,
"learning_rate": 3.552036199095023e-06,
"loss": 1.132,
"step": 571
},
{
"epoch": 0.6470588235294118,
"grad_norm": 0.216796875,
"learning_rate": 3.540723981900453e-06,
"loss": 1.0908,
"step": 572
},
{
"epoch": 0.6481900452488688,
"grad_norm": 0.2119140625,
"learning_rate": 3.529411764705883e-06,
"loss": 1.1217,
"step": 573
},
{
"epoch": 0.6493212669683258,
"grad_norm": 0.2119140625,
"learning_rate": 3.5180995475113127e-06,
"loss": 1.103,
"step": 574
},
{
"epoch": 0.6504524886877828,
"grad_norm": 0.244140625,
"learning_rate": 3.506787330316742e-06,
"loss": 1.1159,
"step": 575
},
{
"epoch": 0.6515837104072398,
"grad_norm": 0.2177734375,
"learning_rate": 3.495475113122172e-06,
"loss": 1.1007,
"step": 576
},
{
"epoch": 0.6527149321266968,
"grad_norm": 0.2119140625,
"learning_rate": 3.484162895927602e-06,
"loss": 1.1028,
"step": 577
},
{
"epoch": 0.6538461538461539,
"grad_norm": 0.30078125,
"learning_rate": 3.472850678733032e-06,
"loss": 1.1198,
"step": 578
},
{
"epoch": 0.6549773755656109,
"grad_norm": 0.240234375,
"learning_rate": 3.4615384615384617e-06,
"loss": 1.1177,
"step": 579
},
{
"epoch": 0.6561085972850679,
"grad_norm": 0.2099609375,
"learning_rate": 3.4502262443438915e-06,
"loss": 1.0978,
"step": 580
},
{
"epoch": 0.6572398190045249,
"grad_norm": 0.2216796875,
"learning_rate": 3.4389140271493217e-06,
"loss": 1.0992,
"step": 581
},
{
"epoch": 0.6583710407239819,
"grad_norm": 0.2216796875,
"learning_rate": 3.4276018099547515e-06,
"loss": 1.1441,
"step": 582
},
{
"epoch": 0.6595022624434389,
"grad_norm": 0.2265625,
"learning_rate": 3.4162895927601813e-06,
"loss": 1.1282,
"step": 583
},
{
"epoch": 0.6606334841628959,
"grad_norm": 0.2060546875,
"learning_rate": 3.404977375565611e-06,
"loss": 1.0841,
"step": 584
},
{
"epoch": 0.6617647058823529,
"grad_norm": 0.2177734375,
"learning_rate": 3.3936651583710413e-06,
"loss": 1.105,
"step": 585
},
{
"epoch": 0.6628959276018099,
"grad_norm": 0.232421875,
"learning_rate": 3.382352941176471e-06,
"loss": 1.122,
"step": 586
},
{
"epoch": 0.6640271493212669,
"grad_norm": 0.21875,
"learning_rate": 3.371040723981901e-06,
"loss": 1.1199,
"step": 587
},
{
"epoch": 0.665158371040724,
"grad_norm": 0.212890625,
"learning_rate": 3.3597285067873307e-06,
"loss": 1.1106,
"step": 588
},
{
"epoch": 0.666289592760181,
"grad_norm": 0.212890625,
"learning_rate": 3.34841628959276e-06,
"loss": 1.1296,
"step": 589
},
{
"epoch": 0.667420814479638,
"grad_norm": 0.255859375,
"learning_rate": 3.33710407239819e-06,
"loss": 1.0622,
"step": 590
},
{
"epoch": 0.668552036199095,
"grad_norm": 0.2109375,
"learning_rate": 3.32579185520362e-06,
"loss": 1.0717,
"step": 591
},
{
"epoch": 0.669683257918552,
"grad_norm": 0.2353515625,
"learning_rate": 3.31447963800905e-06,
"loss": 1.1171,
"step": 592
},
{
"epoch": 0.670814479638009,
"grad_norm": 0.2265625,
"learning_rate": 3.3031674208144797e-06,
"loss": 1.1017,
"step": 593
},
{
"epoch": 0.6719457013574661,
"grad_norm": 0.203125,
"learning_rate": 3.2918552036199095e-06,
"loss": 1.0842,
"step": 594
},
{
"epoch": 0.6730769230769231,
"grad_norm": 0.2080078125,
"learning_rate": 3.2805429864253398e-06,
"loss": 1.0796,
"step": 595
},
{
"epoch": 0.6742081447963801,
"grad_norm": 0.2109375,
"learning_rate": 3.2692307692307696e-06,
"loss": 1.1117,
"step": 596
},
{
"epoch": 0.6753393665158371,
"grad_norm": 0.234375,
"learning_rate": 3.2579185520361994e-06,
"loss": 1.1401,
"step": 597
},
{
"epoch": 0.6764705882352942,
"grad_norm": 0.2001953125,
"learning_rate": 3.246606334841629e-06,
"loss": 1.1041,
"step": 598
},
{
"epoch": 0.6776018099547512,
"grad_norm": 0.2119140625,
"learning_rate": 3.2352941176470594e-06,
"loss": 1.0915,
"step": 599
},
{
"epoch": 0.6787330316742082,
"grad_norm": 0.22265625,
"learning_rate": 3.223981900452489e-06,
"loss": 1.1135,
"step": 600
},
{
"epoch": 0.6798642533936652,
"grad_norm": 0.2197265625,
"learning_rate": 3.212669683257919e-06,
"loss": 1.0808,
"step": 601
},
{
"epoch": 0.6809954751131222,
"grad_norm": 0.25,
"learning_rate": 3.2013574660633484e-06,
"loss": 1.1222,
"step": 602
},
{
"epoch": 0.6821266968325792,
"grad_norm": 0.2158203125,
"learning_rate": 3.190045248868778e-06,
"loss": 1.0859,
"step": 603
},
{
"epoch": 0.6832579185520362,
"grad_norm": 0.232421875,
"learning_rate": 3.1787330316742084e-06,
"loss": 1.1117,
"step": 604
},
{
"epoch": 0.6843891402714932,
"grad_norm": 0.228515625,
"learning_rate": 3.167420814479638e-06,
"loss": 1.122,
"step": 605
},
{
"epoch": 0.6855203619909502,
"grad_norm": 0.2294921875,
"learning_rate": 3.156108597285068e-06,
"loss": 1.0954,
"step": 606
},
{
"epoch": 0.6866515837104072,
"grad_norm": 0.259765625,
"learning_rate": 3.1447963800904978e-06,
"loss": 1.1906,
"step": 607
},
{
"epoch": 0.6877828054298643,
"grad_norm": 0.2138671875,
"learning_rate": 3.133484162895928e-06,
"loss": 1.1423,
"step": 608
},
{
"epoch": 0.6889140271493213,
"grad_norm": 0.2236328125,
"learning_rate": 3.122171945701358e-06,
"loss": 1.1158,
"step": 609
},
{
"epoch": 0.6900452488687783,
"grad_norm": 0.212890625,
"learning_rate": 3.1108597285067876e-06,
"loss": 1.087,
"step": 610
},
{
"epoch": 0.6911764705882353,
"grad_norm": 0.21484375,
"learning_rate": 3.0995475113122174e-06,
"loss": 1.106,
"step": 611
},
{
"epoch": 0.6923076923076923,
"grad_norm": 0.2197265625,
"learning_rate": 3.0882352941176476e-06,
"loss": 1.1186,
"step": 612
},
{
"epoch": 0.6934389140271493,
"grad_norm": 0.23046875,
"learning_rate": 3.0769230769230774e-06,
"loss": 1.1584,
"step": 613
},
{
"epoch": 0.6945701357466063,
"grad_norm": 0.2099609375,
"learning_rate": 3.0656108597285072e-06,
"loss": 1.1109,
"step": 614
},
{
"epoch": 0.6957013574660633,
"grad_norm": 0.212890625,
"learning_rate": 3.054298642533937e-06,
"loss": 1.125,
"step": 615
},
{
"epoch": 0.6968325791855203,
"grad_norm": 0.20703125,
"learning_rate": 3.0429864253393664e-06,
"loss": 1.0873,
"step": 616
},
{
"epoch": 0.6979638009049773,
"grad_norm": 0.2255859375,
"learning_rate": 3.0316742081447962e-06,
"loss": 1.1204,
"step": 617
},
{
"epoch": 0.6990950226244343,
"grad_norm": 0.2158203125,
"learning_rate": 3.0203619909502264e-06,
"loss": 1.1183,
"step": 618
},
{
"epoch": 0.7002262443438914,
"grad_norm": 0.232421875,
"learning_rate": 3.0090497737556562e-06,
"loss": 1.1589,
"step": 619
},
{
"epoch": 0.7013574660633484,
"grad_norm": 0.267578125,
"learning_rate": 2.997737556561086e-06,
"loss": 1.0964,
"step": 620
},
{
"epoch": 0.7024886877828054,
"grad_norm": 0.208984375,
"learning_rate": 2.986425339366516e-06,
"loss": 1.0882,
"step": 621
},
{
"epoch": 0.7036199095022625,
"grad_norm": 0.21875,
"learning_rate": 2.975113122171946e-06,
"loss": 1.0949,
"step": 622
},
{
"epoch": 0.7047511312217195,
"grad_norm": 0.2119140625,
"learning_rate": 2.963800904977376e-06,
"loss": 1.1352,
"step": 623
},
{
"epoch": 0.7058823529411765,
"grad_norm": 0.205078125,
"learning_rate": 2.9524886877828057e-06,
"loss": 1.0877,
"step": 624
},
{
"epoch": 0.7070135746606335,
"grad_norm": 0.21875,
"learning_rate": 2.9411764705882355e-06,
"loss": 1.0896,
"step": 625
},
{
"epoch": 0.7081447963800905,
"grad_norm": 0.2177734375,
"learning_rate": 2.9298642533936657e-06,
"loss": 1.0557,
"step": 626
},
{
"epoch": 0.7092760180995475,
"grad_norm": 0.2158203125,
"learning_rate": 2.9185520361990955e-06,
"loss": 1.0979,
"step": 627
},
{
"epoch": 0.7104072398190046,
"grad_norm": 0.2294921875,
"learning_rate": 2.9072398190045253e-06,
"loss": 1.1528,
"step": 628
},
{
"epoch": 0.7115384615384616,
"grad_norm": 0.25390625,
"learning_rate": 2.895927601809955e-06,
"loss": 1.1222,
"step": 629
},
{
"epoch": 0.7126696832579186,
"grad_norm": 0.220703125,
"learning_rate": 2.8846153846153845e-06,
"loss": 1.1051,
"step": 630
},
{
"epoch": 0.7138009049773756,
"grad_norm": 0.232421875,
"learning_rate": 2.8733031674208147e-06,
"loss": 1.123,
"step": 631
},
{
"epoch": 0.7149321266968326,
"grad_norm": 0.2470703125,
"learning_rate": 2.8619909502262445e-06,
"loss": 1.1329,
"step": 632
},
{
"epoch": 0.7160633484162896,
"grad_norm": 0.201171875,
"learning_rate": 2.8506787330316743e-06,
"loss": 1.1124,
"step": 633
},
{
"epoch": 0.7171945701357466,
"grad_norm": 0.353515625,
"learning_rate": 2.839366515837104e-06,
"loss": 1.1352,
"step": 634
},
{
"epoch": 0.7183257918552036,
"grad_norm": 0.22265625,
"learning_rate": 2.8280542986425343e-06,
"loss": 1.072,
"step": 635
},
{
"epoch": 0.7194570135746606,
"grad_norm": 0.2138671875,
"learning_rate": 2.816742081447964e-06,
"loss": 1.0713,
"step": 636
},
{
"epoch": 0.7205882352941176,
"grad_norm": 0.216796875,
"learning_rate": 2.805429864253394e-06,
"loss": 1.0996,
"step": 637
},
{
"epoch": 0.7217194570135747,
"grad_norm": 0.2177734375,
"learning_rate": 2.7941176470588237e-06,
"loss": 1.1166,
"step": 638
},
{
"epoch": 0.7228506787330317,
"grad_norm": 0.220703125,
"learning_rate": 2.782805429864254e-06,
"loss": 1.0582,
"step": 639
},
{
"epoch": 0.7239819004524887,
"grad_norm": 0.2080078125,
"learning_rate": 2.7714932126696837e-06,
"loss": 1.099,
"step": 640
},
{
"epoch": 0.7251131221719457,
"grad_norm": 0.3046875,
"learning_rate": 2.7601809954751135e-06,
"loss": 1.1438,
"step": 641
},
{
"epoch": 0.7262443438914027,
"grad_norm": 0.220703125,
"learning_rate": 2.7488687782805433e-06,
"loss": 1.1049,
"step": 642
},
{
"epoch": 0.7273755656108597,
"grad_norm": 0.2138671875,
"learning_rate": 2.737556561085973e-06,
"loss": 1.0923,
"step": 643
},
{
"epoch": 0.7285067873303167,
"grad_norm": 0.2099609375,
"learning_rate": 2.7262443438914025e-06,
"loss": 1.0814,
"step": 644
},
{
"epoch": 0.7296380090497737,
"grad_norm": 0.234375,
"learning_rate": 2.7149321266968327e-06,
"loss": 1.1257,
"step": 645
},
{
"epoch": 0.7307692307692307,
"grad_norm": 0.2119140625,
"learning_rate": 2.7036199095022625e-06,
"loss": 1.091,
"step": 646
},
{
"epoch": 0.7319004524886877,
"grad_norm": 0.224609375,
"learning_rate": 2.6923076923076923e-06,
"loss": 1.1062,
"step": 647
},
{
"epoch": 0.7330316742081447,
"grad_norm": 0.318359375,
"learning_rate": 2.680995475113122e-06,
"loss": 1.074,
"step": 648
},
{
"epoch": 0.7341628959276018,
"grad_norm": 0.21875,
"learning_rate": 2.6696832579185524e-06,
"loss": 1.1225,
"step": 649
},
{
"epoch": 0.7352941176470589,
"grad_norm": 0.2197265625,
"learning_rate": 2.658371040723982e-06,
"loss": 1.1279,
"step": 650
},
{
"epoch": 0.7364253393665159,
"grad_norm": 0.23828125,
"learning_rate": 2.647058823529412e-06,
"loss": 1.1323,
"step": 651
},
{
"epoch": 0.7375565610859729,
"grad_norm": 0.3125,
"learning_rate": 2.6357466063348418e-06,
"loss": 1.0802,
"step": 652
},
{
"epoch": 0.7386877828054299,
"grad_norm": 0.2119140625,
"learning_rate": 2.624434389140272e-06,
"loss": 1.1127,
"step": 653
},
{
"epoch": 0.7398190045248869,
"grad_norm": 0.2041015625,
"learning_rate": 2.6131221719457018e-06,
"loss": 1.0896,
"step": 654
},
{
"epoch": 0.7409502262443439,
"grad_norm": 0.2158203125,
"learning_rate": 2.6018099547511316e-06,
"loss": 1.1133,
"step": 655
},
{
"epoch": 0.7420814479638009,
"grad_norm": 0.255859375,
"learning_rate": 2.5904977375565614e-06,
"loss": 1.1012,
"step": 656
},
{
"epoch": 0.7432126696832579,
"grad_norm": 0.248046875,
"learning_rate": 2.5791855203619916e-06,
"loss": 1.1018,
"step": 657
},
{
"epoch": 0.744343891402715,
"grad_norm": 0.2734375,
"learning_rate": 2.567873303167421e-06,
"loss": 1.1826,
"step": 658
},
{
"epoch": 0.745475113122172,
"grad_norm": 0.20703125,
"learning_rate": 2.5565610859728508e-06,
"loss": 1.0888,
"step": 659
},
{
"epoch": 0.746606334841629,
"grad_norm": 0.2353515625,
"learning_rate": 2.5452488687782806e-06,
"loss": 1.1261,
"step": 660
},
{
"epoch": 0.747737556561086,
"grad_norm": 0.2294921875,
"learning_rate": 2.5339366515837104e-06,
"loss": 1.1153,
"step": 661
},
{
"epoch": 0.748868778280543,
"grad_norm": 0.205078125,
"learning_rate": 2.5226244343891406e-06,
"loss": 1.0486,
"step": 662
},
{
"epoch": 0.75,
"grad_norm": 0.2197265625,
"learning_rate": 2.5113122171945704e-06,
"loss": 1.0841,
"step": 663
},
{
"epoch": 0.751131221719457,
"grad_norm": 0.224609375,
"learning_rate": 2.5e-06,
"loss": 1.1236,
"step": 664
},
{
"epoch": 0.752262443438914,
"grad_norm": 0.2119140625,
"learning_rate": 2.48868778280543e-06,
"loss": 1.1157,
"step": 665
},
{
"epoch": 0.753393665158371,
"grad_norm": 0.2890625,
"learning_rate": 2.47737556561086e-06,
"loss": 1.1015,
"step": 666
},
{
"epoch": 0.754524886877828,
"grad_norm": 0.20703125,
"learning_rate": 2.46606334841629e-06,
"loss": 1.0613,
"step": 667
},
{
"epoch": 0.755656108597285,
"grad_norm": 0.2119140625,
"learning_rate": 2.4547511312217194e-06,
"loss": 1.0858,
"step": 668
},
{
"epoch": 0.7567873303167421,
"grad_norm": 0.2333984375,
"learning_rate": 2.4434389140271496e-06,
"loss": 1.1403,
"step": 669
},
{
"epoch": 0.7579185520361991,
"grad_norm": 0.21875,
"learning_rate": 2.4321266968325794e-06,
"loss": 1.1441,
"step": 670
},
{
"epoch": 0.7590497737556561,
"grad_norm": 0.228515625,
"learning_rate": 2.4208144796380092e-06,
"loss": 1.149,
"step": 671
},
{
"epoch": 0.7601809954751131,
"grad_norm": 0.212890625,
"learning_rate": 2.409502262443439e-06,
"loss": 1.0871,
"step": 672
},
{
"epoch": 0.7613122171945701,
"grad_norm": 0.2158203125,
"learning_rate": 2.3981900452488693e-06,
"loss": 1.1182,
"step": 673
},
{
"epoch": 0.7624434389140271,
"grad_norm": 0.265625,
"learning_rate": 2.386877828054299e-06,
"loss": 1.1213,
"step": 674
},
{
"epoch": 0.7635746606334841,
"grad_norm": 0.22265625,
"learning_rate": 2.3755656108597284e-06,
"loss": 1.0898,
"step": 675
},
{
"epoch": 0.7647058823529411,
"grad_norm": 0.2578125,
"learning_rate": 2.3642533936651587e-06,
"loss": 1.1858,
"step": 676
},
{
"epoch": 0.7658371040723982,
"grad_norm": 0.26953125,
"learning_rate": 2.3529411764705885e-06,
"loss": 1.1301,
"step": 677
},
{
"epoch": 0.7669683257918553,
"grad_norm": 0.22265625,
"learning_rate": 2.3416289592760183e-06,
"loss": 1.1051,
"step": 678
},
{
"epoch": 0.7680995475113123,
"grad_norm": 0.296875,
"learning_rate": 2.330316742081448e-06,
"loss": 1.0606,
"step": 679
},
{
"epoch": 0.7692307692307693,
"grad_norm": 0.2060546875,
"learning_rate": 2.3190045248868783e-06,
"loss": 1.0567,
"step": 680
},
{
"epoch": 0.7703619909502263,
"grad_norm": 0.474609375,
"learning_rate": 2.307692307692308e-06,
"loss": 1.1578,
"step": 681
},
{
"epoch": 0.7714932126696833,
"grad_norm": 0.240234375,
"learning_rate": 2.2963800904977375e-06,
"loss": 1.1426,
"step": 682
},
{
"epoch": 0.7726244343891403,
"grad_norm": 0.22265625,
"learning_rate": 2.2850678733031677e-06,
"loss": 1.1305,
"step": 683
},
{
"epoch": 0.7737556561085973,
"grad_norm": 0.86328125,
"learning_rate": 2.2737556561085975e-06,
"loss": 1.0965,
"step": 684
},
{
"epoch": 0.7748868778280543,
"grad_norm": 0.22265625,
"learning_rate": 2.2624434389140273e-06,
"loss": 1.1369,
"step": 685
},
{
"epoch": 0.7760180995475113,
"grad_norm": 0.30078125,
"learning_rate": 2.251131221719457e-06,
"loss": 1.1135,
"step": 686
},
{
"epoch": 0.7771493212669683,
"grad_norm": 0.2197265625,
"learning_rate": 2.2398190045248873e-06,
"loss": 1.0951,
"step": 687
},
{
"epoch": 0.7782805429864253,
"grad_norm": 0.224609375,
"learning_rate": 2.228506787330317e-06,
"loss": 1.1413,
"step": 688
},
{
"epoch": 0.7794117647058824,
"grad_norm": 0.2080078125,
"learning_rate": 2.2171945701357465e-06,
"loss": 1.0823,
"step": 689
},
{
"epoch": 0.7805429864253394,
"grad_norm": 0.2197265625,
"learning_rate": 2.2058823529411767e-06,
"loss": 1.0691,
"step": 690
},
{
"epoch": 0.7816742081447964,
"grad_norm": 0.21484375,
"learning_rate": 2.1945701357466065e-06,
"loss": 1.0806,
"step": 691
},
{
"epoch": 0.7828054298642534,
"grad_norm": 0.2060546875,
"learning_rate": 2.1832579185520363e-06,
"loss": 1.0735,
"step": 692
},
{
"epoch": 0.7839366515837104,
"grad_norm": 0.21484375,
"learning_rate": 2.171945701357466e-06,
"loss": 1.0815,
"step": 693
},
{
"epoch": 0.7850678733031674,
"grad_norm": 0.23046875,
"learning_rate": 2.1606334841628963e-06,
"loss": 1.1135,
"step": 694
},
{
"epoch": 0.7861990950226244,
"grad_norm": 0.2099609375,
"learning_rate": 2.149321266968326e-06,
"loss": 1.0958,
"step": 695
},
{
"epoch": 0.7873303167420814,
"grad_norm": 0.2197265625,
"learning_rate": 2.138009049773756e-06,
"loss": 1.0911,
"step": 696
},
{
"epoch": 0.7884615384615384,
"grad_norm": 0.228515625,
"learning_rate": 2.1266968325791857e-06,
"loss": 1.1165,
"step": 697
},
{
"epoch": 0.7895927601809954,
"grad_norm": 0.2294921875,
"learning_rate": 2.1153846153846155e-06,
"loss": 1.0585,
"step": 698
},
{
"epoch": 0.7907239819004525,
"grad_norm": 0.2138671875,
"learning_rate": 2.1040723981900453e-06,
"loss": 1.1068,
"step": 699
},
{
"epoch": 0.7918552036199095,
"grad_norm": 0.2578125,
"learning_rate": 2.0927601809954755e-06,
"loss": 1.1152,
"step": 700
},
{
"epoch": 0.7929864253393665,
"grad_norm": 0.212890625,
"learning_rate": 2.0814479638009053e-06,
"loss": 1.0937,
"step": 701
},
{
"epoch": 0.7941176470588235,
"grad_norm": 0.25,
"learning_rate": 2.0701357466063347e-06,
"loss": 1.0913,
"step": 702
},
{
"epoch": 0.7952488687782805,
"grad_norm": 0.2080078125,
"learning_rate": 2.058823529411765e-06,
"loss": 1.0926,
"step": 703
},
{
"epoch": 0.7963800904977375,
"grad_norm": 0.259765625,
"learning_rate": 2.0475113122171947e-06,
"loss": 1.1176,
"step": 704
},
{
"epoch": 0.7975113122171946,
"grad_norm": 0.216796875,
"learning_rate": 2.0361990950226245e-06,
"loss": 1.1362,
"step": 705
},
{
"epoch": 0.7986425339366516,
"grad_norm": 0.201171875,
"learning_rate": 2.0248868778280543e-06,
"loss": 1.0829,
"step": 706
},
{
"epoch": 0.7997737556561086,
"grad_norm": 0.21484375,
"learning_rate": 2.0135746606334846e-06,
"loss": 1.129,
"step": 707
},
{
"epoch": 0.8009049773755657,
"grad_norm": 0.2080078125,
"learning_rate": 2.0022624434389144e-06,
"loss": 1.1297,
"step": 708
},
{
"epoch": 0.8020361990950227,
"grad_norm": 0.21875,
"learning_rate": 1.9909502262443437e-06,
"loss": 1.0813,
"step": 709
},
{
"epoch": 0.8031674208144797,
"grad_norm": 0.3671875,
"learning_rate": 1.979638009049774e-06,
"loss": 1.0724,
"step": 710
},
{
"epoch": 0.8042986425339367,
"grad_norm": 0.2314453125,
"learning_rate": 1.9683257918552038e-06,
"loss": 1.1531,
"step": 711
},
{
"epoch": 0.8054298642533937,
"grad_norm": 0.2236328125,
"learning_rate": 1.9570135746606336e-06,
"loss": 1.1054,
"step": 712
},
{
"epoch": 0.8065610859728507,
"grad_norm": 0.2265625,
"learning_rate": 1.9457013574660634e-06,
"loss": 1.0878,
"step": 713
},
{
"epoch": 0.8076923076923077,
"grad_norm": 0.2021484375,
"learning_rate": 1.9343891402714936e-06,
"loss": 1.0707,
"step": 714
},
{
"epoch": 0.8088235294117647,
"grad_norm": 0.2412109375,
"learning_rate": 1.9230769230769234e-06,
"loss": 1.0884,
"step": 715
},
{
"epoch": 0.8099547511312217,
"grad_norm": 0.205078125,
"learning_rate": 1.9117647058823528e-06,
"loss": 1.0806,
"step": 716
},
{
"epoch": 0.8110859728506787,
"grad_norm": 0.2734375,
"learning_rate": 1.9004524886877828e-06,
"loss": 1.1643,
"step": 717
},
{
"epoch": 0.8122171945701357,
"grad_norm": 0.2177734375,
"learning_rate": 1.8891402714932128e-06,
"loss": 1.0921,
"step": 718
},
{
"epoch": 0.8133484162895928,
"grad_norm": 0.21875,
"learning_rate": 1.8778280542986426e-06,
"loss": 1.0715,
"step": 719
},
{
"epoch": 0.8144796380090498,
"grad_norm": 0.2158203125,
"learning_rate": 1.8665158371040726e-06,
"loss": 1.1344,
"step": 720
},
{
"epoch": 0.8156108597285068,
"grad_norm": 0.42578125,
"learning_rate": 1.8552036199095024e-06,
"loss": 1.1253,
"step": 721
},
{
"epoch": 0.8167420814479638,
"grad_norm": 0.236328125,
"learning_rate": 1.8438914027149324e-06,
"loss": 1.1005,
"step": 722
},
{
"epoch": 0.8178733031674208,
"grad_norm": 0.228515625,
"learning_rate": 1.832579185520362e-06,
"loss": 1.0833,
"step": 723
},
{
"epoch": 0.8190045248868778,
"grad_norm": 0.240234375,
"learning_rate": 1.821266968325792e-06,
"loss": 1.0722,
"step": 724
},
{
"epoch": 0.8201357466063348,
"grad_norm": 0.2177734375,
"learning_rate": 1.8099547511312218e-06,
"loss": 1.1503,
"step": 725
},
{
"epoch": 0.8212669683257918,
"grad_norm": 0.2333984375,
"learning_rate": 1.7986425339366516e-06,
"loss": 1.149,
"step": 726
},
{
"epoch": 0.8223981900452488,
"grad_norm": 0.2109375,
"learning_rate": 1.7873303167420816e-06,
"loss": 1.1043,
"step": 727
},
{
"epoch": 0.8235294117647058,
"grad_norm": 0.2060546875,
"learning_rate": 1.7760180995475114e-06,
"loss": 1.0751,
"step": 728
},
{
"epoch": 0.8246606334841629,
"grad_norm": 0.21875,
"learning_rate": 1.7647058823529414e-06,
"loss": 1.076,
"step": 729
},
{
"epoch": 0.8257918552036199,
"grad_norm": 0.216796875,
"learning_rate": 1.753393665158371e-06,
"loss": 1.0811,
"step": 730
},
{
"epoch": 0.8269230769230769,
"grad_norm": 0.220703125,
"learning_rate": 1.742081447963801e-06,
"loss": 1.0744,
"step": 731
},
{
"epoch": 0.8280542986425339,
"grad_norm": 0.25,
"learning_rate": 1.7307692307692308e-06,
"loss": 1.1148,
"step": 732
},
{
"epoch": 0.829185520361991,
"grad_norm": 0.39453125,
"learning_rate": 1.7194570135746609e-06,
"loss": 1.2781,
"step": 733
},
{
"epoch": 0.830316742081448,
"grad_norm": 0.224609375,
"learning_rate": 1.7081447963800907e-06,
"loss": 1.1247,
"step": 734
},
{
"epoch": 0.831447963800905,
"grad_norm": 0.2216796875,
"learning_rate": 1.6968325791855207e-06,
"loss": 1.128,
"step": 735
},
{
"epoch": 0.832579185520362,
"grad_norm": 0.48046875,
"learning_rate": 1.6855203619909505e-06,
"loss": 1.074,
"step": 736
},
{
"epoch": 0.833710407239819,
"grad_norm": 0.29296875,
"learning_rate": 1.67420814479638e-06,
"loss": 1.2017,
"step": 737
},
{
"epoch": 0.834841628959276,
"grad_norm": 0.216796875,
"learning_rate": 1.66289592760181e-06,
"loss": 1.0974,
"step": 738
},
{
"epoch": 0.8359728506787331,
"grad_norm": 0.2236328125,
"learning_rate": 1.6515837104072399e-06,
"loss": 1.1322,
"step": 739
},
{
"epoch": 0.8371040723981901,
"grad_norm": 0.232421875,
"learning_rate": 1.6402714932126699e-06,
"loss": 1.1086,
"step": 740
},
{
"epoch": 0.8382352941176471,
"grad_norm": 0.216796875,
"learning_rate": 1.6289592760180997e-06,
"loss": 1.127,
"step": 741
},
{
"epoch": 0.8393665158371041,
"grad_norm": 0.234375,
"learning_rate": 1.6176470588235297e-06,
"loss": 1.1095,
"step": 742
},
{
"epoch": 0.8404977375565611,
"grad_norm": 0.22265625,
"learning_rate": 1.6063348416289595e-06,
"loss": 1.1362,
"step": 743
},
{
"epoch": 0.8416289592760181,
"grad_norm": 0.2265625,
"learning_rate": 1.595022624434389e-06,
"loss": 1.1167,
"step": 744
},
{
"epoch": 0.8427601809954751,
"grad_norm": 0.2177734375,
"learning_rate": 1.583710407239819e-06,
"loss": 1.1044,
"step": 745
},
{
"epoch": 0.8438914027149321,
"grad_norm": 0.2119140625,
"learning_rate": 1.5723981900452489e-06,
"loss": 1.0849,
"step": 746
},
{
"epoch": 0.8450226244343891,
"grad_norm": 0.224609375,
"learning_rate": 1.561085972850679e-06,
"loss": 1.0911,
"step": 747
},
{
"epoch": 0.8461538461538461,
"grad_norm": 0.203125,
"learning_rate": 1.5497737556561087e-06,
"loss": 1.0635,
"step": 748
},
{
"epoch": 0.8472850678733032,
"grad_norm": 0.2138671875,
"learning_rate": 1.5384615384615387e-06,
"loss": 1.1548,
"step": 749
},
{
"epoch": 0.8484162895927602,
"grad_norm": 0.2119140625,
"learning_rate": 1.5271493212669685e-06,
"loss": 1.0851,
"step": 750
},
{
"epoch": 0.8495475113122172,
"grad_norm": 0.283203125,
"learning_rate": 1.5158371040723981e-06,
"loss": 1.1418,
"step": 751
},
{
"epoch": 0.8506787330316742,
"grad_norm": 0.2216796875,
"learning_rate": 1.5045248868778281e-06,
"loss": 1.1163,
"step": 752
},
{
"epoch": 0.8518099547511312,
"grad_norm": 0.2099609375,
"learning_rate": 1.493212669683258e-06,
"loss": 1.07,
"step": 753
},
{
"epoch": 0.8529411764705882,
"grad_norm": 0.220703125,
"learning_rate": 1.481900452488688e-06,
"loss": 1.129,
"step": 754
},
{
"epoch": 0.8540723981900452,
"grad_norm": 0.263671875,
"learning_rate": 1.4705882352941177e-06,
"loss": 1.1135,
"step": 755
},
{
"epoch": 0.8552036199095022,
"grad_norm": 0.2294921875,
"learning_rate": 1.4592760180995477e-06,
"loss": 1.0903,
"step": 756
},
{
"epoch": 0.8563348416289592,
"grad_norm": 0.21875,
"learning_rate": 1.4479638009049775e-06,
"loss": 1.107,
"step": 757
},
{
"epoch": 0.8574660633484162,
"grad_norm": 0.251953125,
"learning_rate": 1.4366515837104073e-06,
"loss": 1.1195,
"step": 758
},
{
"epoch": 0.8585972850678733,
"grad_norm": 0.2265625,
"learning_rate": 1.4253393665158371e-06,
"loss": 1.0905,
"step": 759
},
{
"epoch": 0.8597285067873304,
"grad_norm": 0.2158203125,
"learning_rate": 1.4140271493212672e-06,
"loss": 1.0766,
"step": 760
},
{
"epoch": 0.8608597285067874,
"grad_norm": 0.2158203125,
"learning_rate": 1.402714932126697e-06,
"loss": 1.1012,
"step": 761
},
{
"epoch": 0.8619909502262444,
"grad_norm": 0.2158203125,
"learning_rate": 1.391402714932127e-06,
"loss": 1.1173,
"step": 762
},
{
"epoch": 0.8631221719457014,
"grad_norm": 0.2265625,
"learning_rate": 1.3800904977375568e-06,
"loss": 1.0746,
"step": 763
},
{
"epoch": 0.8642533936651584,
"grad_norm": 0.271484375,
"learning_rate": 1.3687782805429866e-06,
"loss": 1.1155,
"step": 764
},
{
"epoch": 0.8653846153846154,
"grad_norm": 0.2294921875,
"learning_rate": 1.3574660633484164e-06,
"loss": 1.1111,
"step": 765
},
{
"epoch": 0.8665158371040724,
"grad_norm": 0.2216796875,
"learning_rate": 1.3461538461538462e-06,
"loss": 1.0674,
"step": 766
},
{
"epoch": 0.8676470588235294,
"grad_norm": 0.208984375,
"learning_rate": 1.3348416289592762e-06,
"loss": 1.0467,
"step": 767
},
{
"epoch": 0.8687782805429864,
"grad_norm": 0.228515625,
"learning_rate": 1.323529411764706e-06,
"loss": 1.1132,
"step": 768
},
{
"epoch": 0.8699095022624435,
"grad_norm": 0.2138671875,
"learning_rate": 1.312217194570136e-06,
"loss": 1.1025,
"step": 769
},
{
"epoch": 0.8710407239819005,
"grad_norm": 0.216796875,
"learning_rate": 1.3009049773755658e-06,
"loss": 1.1003,
"step": 770
},
{
"epoch": 0.8721719457013575,
"grad_norm": 0.2138671875,
"learning_rate": 1.2895927601809958e-06,
"loss": 1.1074,
"step": 771
},
{
"epoch": 0.8733031674208145,
"grad_norm": 0.2265625,
"learning_rate": 1.2782805429864254e-06,
"loss": 1.0715,
"step": 772
},
{
"epoch": 0.8744343891402715,
"grad_norm": 0.265625,
"learning_rate": 1.2669683257918552e-06,
"loss": 1.0959,
"step": 773
},
{
"epoch": 0.8755656108597285,
"grad_norm": 0.2138671875,
"learning_rate": 1.2556561085972852e-06,
"loss": 1.1086,
"step": 774
},
{
"epoch": 0.8766968325791855,
"grad_norm": 0.212890625,
"learning_rate": 1.244343891402715e-06,
"loss": 1.0701,
"step": 775
},
{
"epoch": 0.8778280542986425,
"grad_norm": 0.2099609375,
"learning_rate": 1.233031674208145e-06,
"loss": 1.0882,
"step": 776
},
{
"epoch": 0.8789592760180995,
"grad_norm": 0.2275390625,
"learning_rate": 1.2217194570135748e-06,
"loss": 1.1162,
"step": 777
},
{
"epoch": 0.8800904977375565,
"grad_norm": 0.2216796875,
"learning_rate": 1.2104072398190046e-06,
"loss": 1.0853,
"step": 778
},
{
"epoch": 0.8812217194570136,
"grad_norm": 0.208984375,
"learning_rate": 1.1990950226244346e-06,
"loss": 1.0911,
"step": 779
},
{
"epoch": 0.8823529411764706,
"grad_norm": 0.2138671875,
"learning_rate": 1.1877828054298642e-06,
"loss": 1.0109,
"step": 780
},
{
"epoch": 0.8834841628959276,
"grad_norm": 0.2353515625,
"learning_rate": 1.1764705882352942e-06,
"loss": 1.1542,
"step": 781
},
{
"epoch": 0.8846153846153846,
"grad_norm": 0.220703125,
"learning_rate": 1.165158371040724e-06,
"loss": 1.133,
"step": 782
},
{
"epoch": 0.8857466063348416,
"grad_norm": 0.2099609375,
"learning_rate": 1.153846153846154e-06,
"loss": 1.0951,
"step": 783
},
{
"epoch": 0.8868778280542986,
"grad_norm": 0.208984375,
"learning_rate": 1.1425339366515838e-06,
"loss": 1.1048,
"step": 784
},
{
"epoch": 0.8880090497737556,
"grad_norm": 0.212890625,
"learning_rate": 1.1312217194570136e-06,
"loss": 1.0957,
"step": 785
},
{
"epoch": 0.8891402714932126,
"grad_norm": 0.2216796875,
"learning_rate": 1.1199095022624437e-06,
"loss": 1.1369,
"step": 786
},
{
"epoch": 0.8902714932126696,
"grad_norm": 0.20703125,
"learning_rate": 1.1085972850678732e-06,
"loss": 1.0797,
"step": 787
},
{
"epoch": 0.8914027149321267,
"grad_norm": 0.244140625,
"learning_rate": 1.0972850678733033e-06,
"loss": 1.112,
"step": 788
},
{
"epoch": 0.8925339366515838,
"grad_norm": 0.2099609375,
"learning_rate": 1.085972850678733e-06,
"loss": 1.0736,
"step": 789
},
{
"epoch": 0.8936651583710408,
"grad_norm": 0.2275390625,
"learning_rate": 1.074660633484163e-06,
"loss": 1.1407,
"step": 790
},
{
"epoch": 0.8947963800904978,
"grad_norm": 0.234375,
"learning_rate": 1.0633484162895929e-06,
"loss": 1.0999,
"step": 791
},
{
"epoch": 0.8959276018099548,
"grad_norm": 0.2080078125,
"learning_rate": 1.0520361990950227e-06,
"loss": 1.0906,
"step": 792
},
{
"epoch": 0.8970588235294118,
"grad_norm": 0.201171875,
"learning_rate": 1.0407239819004527e-06,
"loss": 1.1092,
"step": 793
},
{
"epoch": 0.8981900452488688,
"grad_norm": 0.265625,
"learning_rate": 1.0294117647058825e-06,
"loss": 1.1033,
"step": 794
},
{
"epoch": 0.8993212669683258,
"grad_norm": 0.224609375,
"learning_rate": 1.0180995475113123e-06,
"loss": 1.1168,
"step": 795
},
{
"epoch": 0.9004524886877828,
"grad_norm": 0.2080078125,
"learning_rate": 1.0067873303167423e-06,
"loss": 1.1152,
"step": 796
},
{
"epoch": 0.9015837104072398,
"grad_norm": 0.2275390625,
"learning_rate": 9.954751131221719e-07,
"loss": 1.1203,
"step": 797
},
{
"epoch": 0.9027149321266968,
"grad_norm": 0.2265625,
"learning_rate": 9.841628959276019e-07,
"loss": 1.1075,
"step": 798
},
{
"epoch": 0.9038461538461539,
"grad_norm": 0.251953125,
"learning_rate": 9.728506787330317e-07,
"loss": 1.1577,
"step": 799
},
{
"epoch": 0.9049773755656109,
"grad_norm": 0.2216796875,
"learning_rate": 9.615384615384617e-07,
"loss": 1.0768,
"step": 800
},
{
"epoch": 0.9061085972850679,
"grad_norm": 0.23046875,
"learning_rate": 9.502262443438914e-07,
"loss": 1.1134,
"step": 801
},
{
"epoch": 0.9072398190045249,
"grad_norm": 0.220703125,
"learning_rate": 9.389140271493213e-07,
"loss": 1.1145,
"step": 802
},
{
"epoch": 0.9083710407239819,
"grad_norm": 0.203125,
"learning_rate": 9.276018099547512e-07,
"loss": 1.0983,
"step": 803
},
{
"epoch": 0.9095022624434389,
"grad_norm": 0.30078125,
"learning_rate": 9.16289592760181e-07,
"loss": 1.0661,
"step": 804
},
{
"epoch": 0.9106334841628959,
"grad_norm": 0.2138671875,
"learning_rate": 9.049773755656109e-07,
"loss": 1.0732,
"step": 805
},
{
"epoch": 0.9117647058823529,
"grad_norm": 0.21875,
"learning_rate": 8.936651583710408e-07,
"loss": 1.0855,
"step": 806
},
{
"epoch": 0.9128959276018099,
"grad_norm": 0.216796875,
"learning_rate": 8.823529411764707e-07,
"loss": 1.13,
"step": 807
},
{
"epoch": 0.9140271493212669,
"grad_norm": 0.224609375,
"learning_rate": 8.710407239819005e-07,
"loss": 1.111,
"step": 808
},
{
"epoch": 0.915158371040724,
"grad_norm": 0.220703125,
"learning_rate": 8.597285067873304e-07,
"loss": 1.101,
"step": 809
},
{
"epoch": 0.916289592760181,
"grad_norm": 0.21875,
"learning_rate": 8.484162895927603e-07,
"loss": 1.0808,
"step": 810
},
{
"epoch": 0.917420814479638,
"grad_norm": 0.212890625,
"learning_rate": 8.3710407239819e-07,
"loss": 1.1319,
"step": 811
},
{
"epoch": 0.918552036199095,
"grad_norm": 0.2099609375,
"learning_rate": 8.257918552036199e-07,
"loss": 1.0476,
"step": 812
},
{
"epoch": 0.919683257918552,
"grad_norm": 0.236328125,
"learning_rate": 8.144796380090498e-07,
"loss": 1.1322,
"step": 813
},
{
"epoch": 0.920814479638009,
"grad_norm": 0.216796875,
"learning_rate": 8.031674208144797e-07,
"loss": 1.1197,
"step": 814
},
{
"epoch": 0.9219457013574661,
"grad_norm": 0.2099609375,
"learning_rate": 7.918552036199095e-07,
"loss": 1.0618,
"step": 815
},
{
"epoch": 0.9230769230769231,
"grad_norm": 0.2177734375,
"learning_rate": 7.805429864253395e-07,
"loss": 1.1113,
"step": 816
},
{
"epoch": 0.9242081447963801,
"grad_norm": 0.25390625,
"learning_rate": 7.692307692307694e-07,
"loss": 1.0884,
"step": 817
},
{
"epoch": 0.9253393665158371,
"grad_norm": 0.216796875,
"learning_rate": 7.579185520361991e-07,
"loss": 1.1242,
"step": 818
},
{
"epoch": 0.9264705882352942,
"grad_norm": 0.232421875,
"learning_rate": 7.46606334841629e-07,
"loss": 1.1099,
"step": 819
},
{
"epoch": 0.9276018099547512,
"grad_norm": 0.2275390625,
"learning_rate": 7.352941176470589e-07,
"loss": 1.1647,
"step": 820
},
{
"epoch": 0.9287330316742082,
"grad_norm": 0.2177734375,
"learning_rate": 7.239819004524888e-07,
"loss": 1.1365,
"step": 821
},
{
"epoch": 0.9298642533936652,
"grad_norm": 0.2255859375,
"learning_rate": 7.126696832579186e-07,
"loss": 1.1581,
"step": 822
},
{
"epoch": 0.9309954751131222,
"grad_norm": 0.2060546875,
"learning_rate": 7.013574660633485e-07,
"loss": 1.0982,
"step": 823
},
{
"epoch": 0.9321266968325792,
"grad_norm": 0.208984375,
"learning_rate": 6.900452488687784e-07,
"loss": 1.0944,
"step": 824
},
{
"epoch": 0.9332579185520362,
"grad_norm": 0.2294921875,
"learning_rate": 6.787330316742082e-07,
"loss": 1.1205,
"step": 825
},
{
"epoch": 0.9343891402714932,
"grad_norm": 0.22265625,
"learning_rate": 6.674208144796381e-07,
"loss": 1.1347,
"step": 826
},
{
"epoch": 0.9355203619909502,
"grad_norm": 0.2138671875,
"learning_rate": 6.56108597285068e-07,
"loss": 1.1122,
"step": 827
},
{
"epoch": 0.9366515837104072,
"grad_norm": 0.2041015625,
"learning_rate": 6.447963800904979e-07,
"loss": 1.072,
"step": 828
},
{
"epoch": 0.9377828054298643,
"grad_norm": 0.21875,
"learning_rate": 6.334841628959276e-07,
"loss": 1.0998,
"step": 829
},
{
"epoch": 0.9389140271493213,
"grad_norm": 0.212890625,
"learning_rate": 6.221719457013575e-07,
"loss": 1.1392,
"step": 830
},
{
"epoch": 0.9400452488687783,
"grad_norm": 0.2197265625,
"learning_rate": 6.108597285067874e-07,
"loss": 1.1227,
"step": 831
},
{
"epoch": 0.9411764705882353,
"grad_norm": 0.2158203125,
"learning_rate": 5.995475113122173e-07,
"loss": 1.1029,
"step": 832
},
{
"epoch": 0.9423076923076923,
"grad_norm": 0.2197265625,
"learning_rate": 5.882352941176471e-07,
"loss": 1.0868,
"step": 833
},
{
"epoch": 0.9434389140271493,
"grad_norm": 0.22265625,
"learning_rate": 5.76923076923077e-07,
"loss": 1.0754,
"step": 834
},
{
"epoch": 0.9445701357466063,
"grad_norm": 0.24609375,
"learning_rate": 5.656108597285068e-07,
"loss": 1.1255,
"step": 835
},
{
"epoch": 0.9457013574660633,
"grad_norm": 0.2255859375,
"learning_rate": 5.542986425339366e-07,
"loss": 1.1018,
"step": 836
},
{
"epoch": 0.9468325791855203,
"grad_norm": 0.2197265625,
"learning_rate": 5.429864253393665e-07,
"loss": 1.0821,
"step": 837
},
{
"epoch": 0.9479638009049773,
"grad_norm": 0.2119140625,
"learning_rate": 5.316742081447964e-07,
"loss": 1.0847,
"step": 838
},
{
"epoch": 0.9490950226244343,
"grad_norm": 0.2119140625,
"learning_rate": 5.203619909502263e-07,
"loss": 1.1231,
"step": 839
},
{
"epoch": 0.9502262443438914,
"grad_norm": 0.2109375,
"learning_rate": 5.090497737556561e-07,
"loss": 1.0925,
"step": 840
},
{
"epoch": 0.9513574660633484,
"grad_norm": 0.2099609375,
"learning_rate": 4.977375565610859e-07,
"loss": 1.0841,
"step": 841
},
{
"epoch": 0.9524886877828054,
"grad_norm": 0.228515625,
"learning_rate": 4.864253393665158e-07,
"loss": 1.0692,
"step": 842
},
{
"epoch": 0.9536199095022625,
"grad_norm": 0.23046875,
"learning_rate": 4.751131221719457e-07,
"loss": 1.1602,
"step": 843
},
{
"epoch": 0.9547511312217195,
"grad_norm": 0.23046875,
"learning_rate": 4.638009049773756e-07,
"loss": 1.1091,
"step": 844
},
{
"epoch": 0.9558823529411765,
"grad_norm": 0.228515625,
"learning_rate": 4.5248868778280546e-07,
"loss": 1.1224,
"step": 845
},
{
"epoch": 0.9570135746606335,
"grad_norm": 0.2265625,
"learning_rate": 4.4117647058823536e-07,
"loss": 1.1732,
"step": 846
},
{
"epoch": 0.9581447963800905,
"grad_norm": 0.2294921875,
"learning_rate": 4.298642533936652e-07,
"loss": 1.1117,
"step": 847
},
{
"epoch": 0.9592760180995475,
"grad_norm": 0.2177734375,
"learning_rate": 4.18552036199095e-07,
"loss": 1.1048,
"step": 848
},
{
"epoch": 0.9604072398190046,
"grad_norm": 0.20703125,
"learning_rate": 4.072398190045249e-07,
"loss": 1.0778,
"step": 849
},
{
"epoch": 0.9615384615384616,
"grad_norm": 0.31640625,
"learning_rate": 3.9592760180995477e-07,
"loss": 1.1183,
"step": 850
},
{
"epoch": 0.9626696832579186,
"grad_norm": 0.2890625,
"learning_rate": 3.846153846153847e-07,
"loss": 1.0925,
"step": 851
},
{
"epoch": 0.9638009049773756,
"grad_norm": 0.2236328125,
"learning_rate": 3.733031674208145e-07,
"loss": 1.0541,
"step": 852
},
{
"epoch": 0.9649321266968326,
"grad_norm": 0.2080078125,
"learning_rate": 3.619909502262444e-07,
"loss": 1.1056,
"step": 853
},
{
"epoch": 0.9660633484162896,
"grad_norm": 0.2197265625,
"learning_rate": 3.5067873303167424e-07,
"loss": 1.0867,
"step": 854
},
{
"epoch": 0.9671945701357466,
"grad_norm": 0.2255859375,
"learning_rate": 3.393665158371041e-07,
"loss": 1.1237,
"step": 855
},
{
"epoch": 0.9683257918552036,
"grad_norm": 0.234375,
"learning_rate": 3.28054298642534e-07,
"loss": 1.1259,
"step": 856
},
{
"epoch": 0.9694570135746606,
"grad_norm": 0.22265625,
"learning_rate": 3.167420814479638e-07,
"loss": 1.0968,
"step": 857
},
{
"epoch": 0.9705882352941176,
"grad_norm": 0.2275390625,
"learning_rate": 3.054298642533937e-07,
"loss": 1.0867,
"step": 858
},
{
"epoch": 0.9717194570135747,
"grad_norm": 0.23046875,
"learning_rate": 2.9411764705882356e-07,
"loss": 1.1073,
"step": 859
},
{
"epoch": 0.9728506787330317,
"grad_norm": 0.2109375,
"learning_rate": 2.828054298642534e-07,
"loss": 1.1001,
"step": 860
},
{
"epoch": 0.9739819004524887,
"grad_norm": 0.228515625,
"learning_rate": 2.7149321266968326e-07,
"loss": 1.1349,
"step": 861
},
{
"epoch": 0.9751131221719457,
"grad_norm": 0.2177734375,
"learning_rate": 2.6018099547511317e-07,
"loss": 1.1115,
"step": 862
},
{
"epoch": 0.9762443438914027,
"grad_norm": 0.2177734375,
"learning_rate": 2.4886877828054297e-07,
"loss": 1.1353,
"step": 863
},
{
"epoch": 0.9773755656108597,
"grad_norm": 0.2158203125,
"learning_rate": 2.3755656108597285e-07,
"loss": 1.1124,
"step": 864
},
{
"epoch": 0.9785067873303167,
"grad_norm": 0.2109375,
"learning_rate": 2.2624434389140273e-07,
"loss": 1.0868,
"step": 865
},
{
"epoch": 0.9796380090497737,
"grad_norm": 0.22265625,
"learning_rate": 2.149321266968326e-07,
"loss": 1.0964,
"step": 866
},
{
"epoch": 0.9807692307692307,
"grad_norm": 0.224609375,
"learning_rate": 2.0361990950226246e-07,
"loss": 1.0839,
"step": 867
},
{
"epoch": 0.9819004524886877,
"grad_norm": 0.2001953125,
"learning_rate": 1.9230769230769234e-07,
"loss": 1.0901,
"step": 868
},
{
"epoch": 0.9830316742081447,
"grad_norm": 0.2158203125,
"learning_rate": 1.809954751131222e-07,
"loss": 1.0902,
"step": 869
},
{
"epoch": 0.9841628959276018,
"grad_norm": 0.2451171875,
"learning_rate": 1.6968325791855205e-07,
"loss": 1.0892,
"step": 870
},
{
"epoch": 0.9852941176470589,
"grad_norm": 0.2138671875,
"learning_rate": 1.583710407239819e-07,
"loss": 1.0909,
"step": 871
},
{
"epoch": 0.9864253393665159,
"grad_norm": 0.244140625,
"learning_rate": 1.4705882352941178e-07,
"loss": 1.1056,
"step": 872
},
{
"epoch": 0.9875565610859729,
"grad_norm": 0.2421875,
"learning_rate": 1.3574660633484163e-07,
"loss": 1.1131,
"step": 873
},
{
"epoch": 0.9886877828054299,
"grad_norm": 0.2265625,
"learning_rate": 1.2443438914027148e-07,
"loss": 1.1416,
"step": 874
},
{
"epoch": 0.9898190045248869,
"grad_norm": 0.24609375,
"learning_rate": 1.1312217194570136e-07,
"loss": 1.1306,
"step": 875
},
{
"epoch": 0.9909502262443439,
"grad_norm": 0.2177734375,
"learning_rate": 1.0180995475113123e-07,
"loss": 1.0778,
"step": 876
},
{
"epoch": 0.9920814479638009,
"grad_norm": 0.2490234375,
"learning_rate": 9.04977375565611e-08,
"loss": 1.1172,
"step": 877
},
{
"epoch": 0.9932126696832579,
"grad_norm": 0.2119140625,
"learning_rate": 7.918552036199095e-08,
"loss": 1.109,
"step": 878
},
{
"epoch": 0.994343891402715,
"grad_norm": 0.259765625,
"learning_rate": 6.787330316742082e-08,
"loss": 1.1088,
"step": 879
},
{
"epoch": 0.995475113122172,
"grad_norm": 0.232421875,
"learning_rate": 5.656108597285068e-08,
"loss": 1.0909,
"step": 880
},
{
"epoch": 0.996606334841629,
"grad_norm": 0.2099609375,
"learning_rate": 4.524886877828055e-08,
"loss": 1.1184,
"step": 881
},
{
"epoch": 0.997737556561086,
"grad_norm": 0.232421875,
"learning_rate": 3.393665158371041e-08,
"loss": 1.0963,
"step": 882
},
{
"epoch": 0.998868778280543,
"grad_norm": 0.263671875,
"learning_rate": 2.2624434389140274e-08,
"loss": 1.0737,
"step": 883
},
{
"epoch": 1.0,
"grad_norm": 0.267578125,
"learning_rate": 1.1312217194570137e-08,
"loss": 1.1044,
"step": 884
}
],
"logging_steps": 1.0,
"max_steps": 884,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 0,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 2.8903067303091896e+18,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}