stash's picture
Upload folder using huggingface_hub
ebf6742 verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 1003,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 6.451612903225806e-06,
"loss": 0.8746,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.2903225806451613e-05,
"loss": 0.847,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 1.935483870967742e-05,
"loss": 0.8604,
"step": 3
},
{
"epoch": 0.0,
"learning_rate": 2.5806451612903226e-05,
"loss": 0.8158,
"step": 4
},
{
"epoch": 0.0,
"learning_rate": 3.2258064516129034e-05,
"loss": 0.7734,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 3.870967741935484e-05,
"loss": 0.6985,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 4.516129032258064e-05,
"loss": 0.6392,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 5.161290322580645e-05,
"loss": 0.6008,
"step": 8
},
{
"epoch": 0.01,
"learning_rate": 5.8064516129032266e-05,
"loss": 0.5782,
"step": 9
},
{
"epoch": 0.01,
"learning_rate": 6.451612903225807e-05,
"loss": 0.5087,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 7.096774193548388e-05,
"loss": 0.4686,
"step": 11
},
{
"epoch": 0.01,
"learning_rate": 7.741935483870968e-05,
"loss": 0.4097,
"step": 12
},
{
"epoch": 0.01,
"learning_rate": 8.387096774193549e-05,
"loss": 0.3794,
"step": 13
},
{
"epoch": 0.01,
"learning_rate": 9.032258064516129e-05,
"loss": 0.3304,
"step": 14
},
{
"epoch": 0.01,
"learning_rate": 9.677419354838711e-05,
"loss": 0.305,
"step": 15
},
{
"epoch": 0.02,
"learning_rate": 0.0001032258064516129,
"loss": 0.2659,
"step": 16
},
{
"epoch": 0.02,
"learning_rate": 0.00010967741935483871,
"loss": 0.2634,
"step": 17
},
{
"epoch": 0.02,
"learning_rate": 0.00011612903225806453,
"loss": 0.2373,
"step": 18
},
{
"epoch": 0.02,
"learning_rate": 0.00012258064516129034,
"loss": 0.2293,
"step": 19
},
{
"epoch": 0.02,
"learning_rate": 0.00012903225806451613,
"loss": 0.224,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 0.00013548387096774193,
"loss": 0.2141,
"step": 21
},
{
"epoch": 0.02,
"learning_rate": 0.00014193548387096775,
"loss": 0.2001,
"step": 22
},
{
"epoch": 0.02,
"learning_rate": 0.00014838709677419355,
"loss": 0.1958,
"step": 23
},
{
"epoch": 0.02,
"learning_rate": 0.00015483870967741937,
"loss": 0.2046,
"step": 24
},
{
"epoch": 0.02,
"learning_rate": 0.00016129032258064516,
"loss": 0.2097,
"step": 25
},
{
"epoch": 0.03,
"learning_rate": 0.00016774193548387098,
"loss": 0.2074,
"step": 26
},
{
"epoch": 0.03,
"learning_rate": 0.00017419354838709678,
"loss": 0.1929,
"step": 27
},
{
"epoch": 0.03,
"learning_rate": 0.00018064516129032257,
"loss": 0.2003,
"step": 28
},
{
"epoch": 0.03,
"learning_rate": 0.0001870967741935484,
"loss": 0.1906,
"step": 29
},
{
"epoch": 0.03,
"learning_rate": 0.00019354838709677422,
"loss": 0.2003,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 0.1867,
"step": 31
},
{
"epoch": 0.03,
"learning_rate": 0.00019999947767977604,
"loss": 0.185,
"step": 32
},
{
"epoch": 0.03,
"learning_rate": 0.0001999979107245606,
"loss": 0.1878,
"step": 33
},
{
"epoch": 0.03,
"learning_rate": 0.00019999529915072262,
"loss": 0.2014,
"step": 34
},
{
"epoch": 0.03,
"learning_rate": 0.00019999164298554375,
"loss": 0.1914,
"step": 35
},
{
"epoch": 0.04,
"learning_rate": 0.00019998694226721774,
"loss": 0.1863,
"step": 36
},
{
"epoch": 0.04,
"learning_rate": 0.00019998119704485014,
"loss": 0.1882,
"step": 37
},
{
"epoch": 0.04,
"learning_rate": 0.00019997440737845796,
"loss": 0.1906,
"step": 38
},
{
"epoch": 0.04,
"learning_rate": 0.00019996657333896877,
"loss": 0.1812,
"step": 39
},
{
"epoch": 0.04,
"learning_rate": 0.0001999576950082201,
"loss": 0.1844,
"step": 40
},
{
"epoch": 0.04,
"learning_rate": 0.00019994777247895855,
"loss": 0.1789,
"step": 41
},
{
"epoch": 0.04,
"learning_rate": 0.00019993680585483895,
"loss": 0.185,
"step": 42
},
{
"epoch": 0.04,
"learning_rate": 0.00019992479525042303,
"loss": 0.1798,
"step": 43
},
{
"epoch": 0.04,
"learning_rate": 0.00019991174079117844,
"loss": 0.1832,
"step": 44
},
{
"epoch": 0.04,
"learning_rate": 0.00019989764261347736,
"loss": 0.1808,
"step": 45
},
{
"epoch": 0.05,
"learning_rate": 0.00019988250086459505,
"loss": 0.1779,
"step": 46
},
{
"epoch": 0.05,
"learning_rate": 0.00019986631570270832,
"loss": 0.1788,
"step": 47
},
{
"epoch": 0.05,
"learning_rate": 0.000199849087296894,
"loss": 0.1764,
"step": 48
},
{
"epoch": 0.05,
"learning_rate": 0.00019983081582712685,
"loss": 0.1783,
"step": 49
},
{
"epoch": 0.05,
"learning_rate": 0.0001998115014842781,
"loss": 0.1757,
"step": 50
},
{
"epoch": 0.05,
"learning_rate": 0.00019979114447011323,
"loss": 0.1758,
"step": 51
},
{
"epoch": 0.05,
"learning_rate": 0.00019976974499728977,
"loss": 0.1764,
"step": 52
},
{
"epoch": 0.05,
"learning_rate": 0.00019974730328935534,
"loss": 0.1876,
"step": 53
},
{
"epoch": 0.05,
"learning_rate": 0.00019972381958074505,
"loss": 0.1657,
"step": 54
},
{
"epoch": 0.05,
"learning_rate": 0.0001996992941167792,
"loss": 0.1769,
"step": 55
},
{
"epoch": 0.06,
"learning_rate": 0.00019967372715366078,
"loss": 0.1732,
"step": 56
},
{
"epoch": 0.06,
"learning_rate": 0.00019964711895847258,
"loss": 0.1828,
"step": 57
},
{
"epoch": 0.06,
"learning_rate": 0.00019961946980917456,
"loss": 0.1735,
"step": 58
},
{
"epoch": 0.06,
"learning_rate": 0.00019959077999460095,
"loss": 0.173,
"step": 59
},
{
"epoch": 0.06,
"learning_rate": 0.00019956104981445712,
"loss": 0.1729,
"step": 60
},
{
"epoch": 0.06,
"learning_rate": 0.00019953027957931658,
"loss": 0.1772,
"step": 61
},
{
"epoch": 0.06,
"learning_rate": 0.00019949846961061767,
"loss": 0.1796,
"step": 62
},
{
"epoch": 0.06,
"learning_rate": 0.00019946562024066014,
"loss": 0.1719,
"step": 63
},
{
"epoch": 0.06,
"learning_rate": 0.00019943173181260186,
"loss": 0.165,
"step": 64
},
{
"epoch": 0.06,
"learning_rate": 0.00019939680468045499,
"loss": 0.1687,
"step": 65
},
{
"epoch": 0.07,
"learning_rate": 0.00019936083920908253,
"loss": 0.1563,
"step": 66
},
{
"epoch": 0.07,
"learning_rate": 0.00019932383577419432,
"loss": 0.1812,
"step": 67
},
{
"epoch": 0.07,
"learning_rate": 0.0001992857947623432,
"loss": 0.1773,
"step": 68
},
{
"epoch": 0.07,
"learning_rate": 0.00019924671657092096,
"loss": 0.1712,
"step": 69
},
{
"epoch": 0.07,
"learning_rate": 0.00019920660160815422,
"loss": 0.1703,
"step": 70
},
{
"epoch": 0.07,
"learning_rate": 0.00019916545029310012,
"loss": 0.1803,
"step": 71
},
{
"epoch": 0.07,
"learning_rate": 0.0001991232630556419,
"loss": 0.1674,
"step": 72
},
{
"epoch": 0.07,
"learning_rate": 0.00019908004033648453,
"loss": 0.1708,
"step": 73
},
{
"epoch": 0.07,
"learning_rate": 0.00019903578258715,
"loss": 0.1788,
"step": 74
},
{
"epoch": 0.07,
"learning_rate": 0.00019899049026997272,
"loss": 0.1677,
"step": 75
},
{
"epoch": 0.08,
"learning_rate": 0.00019894416385809444,
"loss": 0.1765,
"step": 76
},
{
"epoch": 0.08,
"learning_rate": 0.00019889680383545973,
"loss": 0.1739,
"step": 77
},
{
"epoch": 0.08,
"learning_rate": 0.00019884841069681044,
"loss": 0.1741,
"step": 78
},
{
"epoch": 0.08,
"learning_rate": 0.00019879898494768093,
"loss": 0.1663,
"step": 79
},
{
"epoch": 0.08,
"learning_rate": 0.00019874852710439252,
"loss": 0.1679,
"step": 80
},
{
"epoch": 0.08,
"learning_rate": 0.00019869703769404828,
"loss": 0.1616,
"step": 81
},
{
"epoch": 0.08,
"learning_rate": 0.0001986445172545274,
"loss": 0.1607,
"step": 82
},
{
"epoch": 0.08,
"learning_rate": 0.00019859096633447965,
"loss": 0.1741,
"step": 83
},
{
"epoch": 0.08,
"learning_rate": 0.00019853638549331964,
"loss": 0.1683,
"step": 84
},
{
"epoch": 0.08,
"learning_rate": 0.00019848077530122083,
"loss": 0.1664,
"step": 85
},
{
"epoch": 0.09,
"learning_rate": 0.0001984241363391098,
"loss": 0.1706,
"step": 86
},
{
"epoch": 0.09,
"learning_rate": 0.0001983664691986601,
"loss": 0.1595,
"step": 87
},
{
"epoch": 0.09,
"learning_rate": 0.00019830777448228603,
"loss": 0.1611,
"step": 88
},
{
"epoch": 0.09,
"learning_rate": 0.00019824805280313623,
"loss": 0.1693,
"step": 89
},
{
"epoch": 0.09,
"learning_rate": 0.0001981873047850876,
"loss": 0.1613,
"step": 90
},
{
"epoch": 0.09,
"learning_rate": 0.00019812553106273847,
"loss": 0.1621,
"step": 91
},
{
"epoch": 0.09,
"learning_rate": 0.00019806273228140214,
"loss": 0.1742,
"step": 92
},
{
"epoch": 0.09,
"learning_rate": 0.00019799890909710013,
"loss": 0.167,
"step": 93
},
{
"epoch": 0.09,
"learning_rate": 0.00019793406217655517,
"loss": 0.1556,
"step": 94
},
{
"epoch": 0.09,
"learning_rate": 0.00019786819219718443,
"loss": 0.1733,
"step": 95
},
{
"epoch": 0.1,
"learning_rate": 0.00019780129984709239,
"loss": 0.1769,
"step": 96
},
{
"epoch": 0.1,
"learning_rate": 0.0001977333858250636,
"loss": 0.1729,
"step": 97
},
{
"epoch": 0.1,
"learning_rate": 0.00019766445084055533,
"loss": 0.1592,
"step": 98
},
{
"epoch": 0.1,
"learning_rate": 0.00019759449561369038,
"loss": 0.1667,
"step": 99
},
{
"epoch": 0.1,
"learning_rate": 0.00019752352087524933,
"loss": 0.1572,
"step": 100
},
{
"epoch": 0.1,
"learning_rate": 0.00019745152736666302,
"loss": 0.1621,
"step": 101
},
{
"epoch": 0.1,
"learning_rate": 0.0001973785158400047,
"loss": 0.1608,
"step": 102
},
{
"epoch": 0.1,
"learning_rate": 0.00019730448705798239,
"loss": 0.1521,
"step": 103
},
{
"epoch": 0.1,
"learning_rate": 0.00019722944179393067,
"loss": 0.1568,
"step": 104
},
{
"epoch": 0.1,
"learning_rate": 0.00019715338083180269,
"loss": 0.1653,
"step": 105
},
{
"epoch": 0.11,
"learning_rate": 0.00019707630496616202,
"loss": 0.1624,
"step": 106
},
{
"epoch": 0.11,
"learning_rate": 0.00019699821500217434,
"loss": 0.1613,
"step": 107
},
{
"epoch": 0.11,
"learning_rate": 0.00019691911175559904,
"loss": 0.1647,
"step": 108
},
{
"epoch": 0.11,
"learning_rate": 0.0001968389960527806,
"loss": 0.1545,
"step": 109
},
{
"epoch": 0.11,
"learning_rate": 0.00019675786873064004,
"loss": 0.1527,
"step": 110
},
{
"epoch": 0.11,
"learning_rate": 0.0001966757306366662,
"loss": 0.1586,
"step": 111
},
{
"epoch": 0.11,
"learning_rate": 0.00019659258262890683,
"loss": 0.1597,
"step": 112
},
{
"epoch": 0.11,
"learning_rate": 0.00019650842557595967,
"loss": 0.1642,
"step": 113
},
{
"epoch": 0.11,
"learning_rate": 0.0001964232603569633,
"loss": 0.1685,
"step": 114
},
{
"epoch": 0.11,
"learning_rate": 0.00019633708786158806,
"loss": 0.1558,
"step": 115
},
{
"epoch": 0.12,
"learning_rate": 0.00019624990899002665,
"loss": 0.1627,
"step": 116
},
{
"epoch": 0.12,
"learning_rate": 0.00019616172465298492,
"loss": 0.1609,
"step": 117
},
{
"epoch": 0.12,
"learning_rate": 0.00019607253577167205,
"loss": 0.1688,
"step": 118
},
{
"epoch": 0.12,
"learning_rate": 0.00019598234327779118,
"loss": 0.152,
"step": 119
},
{
"epoch": 0.12,
"learning_rate": 0.00019589114811352962,
"loss": 0.1595,
"step": 120
},
{
"epoch": 0.12,
"learning_rate": 0.0001957989512315489,
"loss": 0.155,
"step": 121
},
{
"epoch": 0.12,
"learning_rate": 0.00019570575359497493,
"loss": 0.1503,
"step": 122
},
{
"epoch": 0.12,
"learning_rate": 0.00019561155617738797,
"loss": 0.1609,
"step": 123
},
{
"epoch": 0.12,
"learning_rate": 0.00019551635996281231,
"loss": 0.1634,
"step": 124
},
{
"epoch": 0.12,
"learning_rate": 0.00019542016594570615,
"loss": 0.1613,
"step": 125
},
{
"epoch": 0.13,
"learning_rate": 0.000195322975130951,
"loss": 0.1602,
"step": 126
},
{
"epoch": 0.13,
"learning_rate": 0.00019522478853384155,
"loss": 0.1558,
"step": 127
},
{
"epoch": 0.13,
"learning_rate": 0.00019512560718007462,
"loss": 0.1601,
"step": 128
},
{
"epoch": 0.13,
"learning_rate": 0.00019502543210573881,
"loss": 0.1588,
"step": 129
},
{
"epoch": 0.13,
"learning_rate": 0.0001949242643573034,
"loss": 0.161,
"step": 130
},
{
"epoch": 0.13,
"learning_rate": 0.00019482210499160765,
"loss": 0.1529,
"step": 131
},
{
"epoch": 0.13,
"learning_rate": 0.00019471895507584963,
"loss": 0.1523,
"step": 132
},
{
"epoch": 0.13,
"learning_rate": 0.00019461481568757506,
"loss": 0.1576,
"step": 133
},
{
"epoch": 0.13,
"learning_rate": 0.00019450968791466609,
"loss": 0.1511,
"step": 134
},
{
"epoch": 0.13,
"learning_rate": 0.00019440357285533,
"loss": 0.1484,
"step": 135
},
{
"epoch": 0.14,
"learning_rate": 0.0001942964716180876,
"loss": 0.1645,
"step": 136
},
{
"epoch": 0.14,
"learning_rate": 0.00019418838532176173,
"loss": 0.1654,
"step": 137
},
{
"epoch": 0.14,
"learning_rate": 0.00019407931509546558,
"loss": 0.1617,
"step": 138
},
{
"epoch": 0.14,
"learning_rate": 0.00019396926207859084,
"loss": 0.1541,
"step": 139
},
{
"epoch": 0.14,
"learning_rate": 0.00019385822742079586,
"loss": 0.1653,
"step": 140
},
{
"epoch": 0.14,
"learning_rate": 0.0001937462122819935,
"loss": 0.1564,
"step": 141
},
{
"epoch": 0.14,
"learning_rate": 0.00019363321783233932,
"loss": 0.1595,
"step": 142
},
{
"epoch": 0.14,
"learning_rate": 0.000193519245252219,
"loss": 0.169,
"step": 143
},
{
"epoch": 0.14,
"learning_rate": 0.00019340429573223616,
"loss": 0.1532,
"step": 144
},
{
"epoch": 0.14,
"learning_rate": 0.0001932883704732001,
"loss": 0.153,
"step": 145
},
{
"epoch": 0.15,
"learning_rate": 0.00019317147068611282,
"loss": 0.1548,
"step": 146
},
{
"epoch": 0.15,
"learning_rate": 0.00019305359759215685,
"loss": 0.1552,
"step": 147
},
{
"epoch": 0.15,
"learning_rate": 0.00019293475242268223,
"loss": 0.1499,
"step": 148
},
{
"epoch": 0.15,
"learning_rate": 0.00019281493641919368,
"loss": 0.1527,
"step": 149
},
{
"epoch": 0.15,
"learning_rate": 0.00019269415083333758,
"loss": 0.1524,
"step": 150
},
{
"epoch": 0.15,
"learning_rate": 0.00019257239692688907,
"loss": 0.1628,
"step": 151
},
{
"epoch": 0.15,
"learning_rate": 0.00019244967597173865,
"loss": 0.1566,
"step": 152
},
{
"epoch": 0.15,
"learning_rate": 0.00019232598924987903,
"loss": 0.1681,
"step": 153
},
{
"epoch": 0.15,
"learning_rate": 0.00019220133805339184,
"loss": 0.1517,
"step": 154
},
{
"epoch": 0.15,
"learning_rate": 0.00019207572368443385,
"loss": 0.1663,
"step": 155
},
{
"epoch": 0.16,
"learning_rate": 0.00019194914745522353,
"loss": 0.1562,
"step": 156
},
{
"epoch": 0.16,
"learning_rate": 0.00019182161068802741,
"loss": 0.1546,
"step": 157
},
{
"epoch": 0.16,
"learning_rate": 0.00019169311471514612,
"loss": 0.1562,
"step": 158
},
{
"epoch": 0.16,
"learning_rate": 0.0001915636608789006,
"loss": 0.1596,
"step": 159
},
{
"epoch": 0.16,
"learning_rate": 0.00019143325053161796,
"loss": 0.1603,
"step": 160
},
{
"epoch": 0.16,
"learning_rate": 0.00019130188503561741,
"loss": 0.162,
"step": 161
},
{
"epoch": 0.16,
"learning_rate": 0.00019116956576319612,
"loss": 0.1526,
"step": 162
},
{
"epoch": 0.16,
"learning_rate": 0.0001910362940966147,
"loss": 0.1548,
"step": 163
},
{
"epoch": 0.16,
"learning_rate": 0.00019090207142808283,
"loss": 0.1564,
"step": 164
},
{
"epoch": 0.16,
"learning_rate": 0.00019076689915974488,
"loss": 0.163,
"step": 165
},
{
"epoch": 0.17,
"learning_rate": 0.000190630778703665,
"loss": 0.1589,
"step": 166
},
{
"epoch": 0.17,
"learning_rate": 0.00019049371148181253,
"loss": 0.1495,
"step": 167
},
{
"epoch": 0.17,
"learning_rate": 0.00019035569892604715,
"loss": 0.1546,
"step": 168
},
{
"epoch": 0.17,
"learning_rate": 0.0001902167424781038,
"loss": 0.1677,
"step": 169
},
{
"epoch": 0.17,
"learning_rate": 0.0001900768435895777,
"loss": 0.162,
"step": 170
},
{
"epoch": 0.17,
"learning_rate": 0.00018993600372190932,
"loss": 0.1567,
"step": 171
},
{
"epoch": 0.17,
"learning_rate": 0.0001897942243463688,
"loss": 0.1584,
"step": 172
},
{
"epoch": 0.17,
"learning_rate": 0.00018965150694404094,
"loss": 0.1574,
"step": 173
},
{
"epoch": 0.17,
"learning_rate": 0.00018950785300580935,
"loss": 0.1528,
"step": 174
},
{
"epoch": 0.17,
"learning_rate": 0.00018936326403234125,
"loss": 0.1582,
"step": 175
},
{
"epoch": 0.18,
"learning_rate": 0.00018921774153407147,
"loss": 0.1561,
"step": 176
},
{
"epoch": 0.18,
"learning_rate": 0.00018907128703118695,
"loss": 0.1549,
"step": 177
},
{
"epoch": 0.18,
"learning_rate": 0.00018892390205361062,
"loss": 0.1659,
"step": 178
},
{
"epoch": 0.18,
"learning_rate": 0.00018877558814098561,
"loss": 0.1579,
"step": 179
},
{
"epoch": 0.18,
"learning_rate": 0.000188626346842659,
"loss": 0.1659,
"step": 180
},
{
"epoch": 0.18,
"learning_rate": 0.00018847617971766577,
"loss": 0.1603,
"step": 181
},
{
"epoch": 0.18,
"learning_rate": 0.0001883250883347125,
"loss": 0.1543,
"step": 182
},
{
"epoch": 0.18,
"learning_rate": 0.0001881730742721608,
"loss": 0.155,
"step": 183
},
{
"epoch": 0.18,
"learning_rate": 0.00018802013911801112,
"loss": 0.1519,
"step": 184
},
{
"epoch": 0.18,
"learning_rate": 0.00018786628446988593,
"loss": 0.1552,
"step": 185
},
{
"epoch": 0.19,
"learning_rate": 0.00018771151193501306,
"loss": 0.1483,
"step": 186
},
{
"epoch": 0.19,
"learning_rate": 0.0001875558231302091,
"loss": 0.1528,
"step": 187
},
{
"epoch": 0.19,
"learning_rate": 0.00018739921968186223,
"loss": 0.1555,
"step": 188
},
{
"epoch": 0.19,
"learning_rate": 0.00018724170322591537,
"loss": 0.1646,
"step": 189
},
{
"epoch": 0.19,
"learning_rate": 0.00018708327540784922,
"loss": 0.1564,
"step": 190
},
{
"epoch": 0.19,
"learning_rate": 0.00018692393788266479,
"loss": 0.1545,
"step": 191
},
{
"epoch": 0.19,
"learning_rate": 0.00018676369231486635,
"loss": 0.1551,
"step": 192
},
{
"epoch": 0.19,
"learning_rate": 0.00018660254037844388,
"loss": 0.1498,
"step": 193
},
{
"epoch": 0.19,
"learning_rate": 0.0001864404837568557,
"loss": 0.1508,
"step": 194
},
{
"epoch": 0.19,
"learning_rate": 0.00018627752414301086,
"loss": 0.1583,
"step": 195
},
{
"epoch": 0.2,
"learning_rate": 0.00018611366323925137,
"loss": 0.1563,
"step": 196
},
{
"epoch": 0.2,
"learning_rate": 0.00018594890275733453,
"loss": 0.1561,
"step": 197
},
{
"epoch": 0.2,
"learning_rate": 0.00018578324441841494,
"loss": 0.1547,
"step": 198
},
{
"epoch": 0.2,
"learning_rate": 0.00018561668995302667,
"loss": 0.1612,
"step": 199
},
{
"epoch": 0.2,
"learning_rate": 0.00018544924110106497,
"loss": 0.1617,
"step": 200
},
{
"epoch": 0.2,
"learning_rate": 0.0001852808996117683,
"loss": 0.1626,
"step": 201
},
{
"epoch": 0.2,
"learning_rate": 0.00018511166724369997,
"loss": 0.1601,
"step": 202
},
{
"epoch": 0.2,
"learning_rate": 0.00018494154576472976,
"loss": 0.1538,
"step": 203
},
{
"epoch": 0.2,
"learning_rate": 0.00018477053695201538,
"loss": 0.1579,
"step": 204
},
{
"epoch": 0.2,
"learning_rate": 0.0001845986425919841,
"loss": 0.1508,
"step": 205
},
{
"epoch": 0.21,
"learning_rate": 0.00018442586448031396,
"loss": 0.1514,
"step": 206
},
{
"epoch": 0.21,
"learning_rate": 0.00018425220442191495,
"loss": 0.1603,
"step": 207
},
{
"epoch": 0.21,
"learning_rate": 0.00018407766423091034,
"loss": 0.1523,
"step": 208
},
{
"epoch": 0.21,
"learning_rate": 0.00018390224573061747,
"loss": 0.1518,
"step": 209
},
{
"epoch": 0.21,
"learning_rate": 0.00018372595075352903,
"loss": 0.1587,
"step": 210
},
{
"epoch": 0.21,
"learning_rate": 0.00018354878114129367,
"loss": 0.1494,
"step": 211
},
{
"epoch": 0.21,
"learning_rate": 0.00018337073874469674,
"loss": 0.1538,
"step": 212
},
{
"epoch": 0.21,
"learning_rate": 0.00018319182542364117,
"loss": 0.1531,
"step": 213
},
{
"epoch": 0.21,
"learning_rate": 0.0001830120430471279,
"loss": 0.1516,
"step": 214
},
{
"epoch": 0.21,
"learning_rate": 0.00018283139349323634,
"loss": 0.1556,
"step": 215
},
{
"epoch": 0.22,
"learning_rate": 0.00018264987864910473,
"loss": 0.1524,
"step": 216
},
{
"epoch": 0.22,
"learning_rate": 0.0001824675004109107,
"loss": 0.1576,
"step": 217
},
{
"epoch": 0.22,
"learning_rate": 0.00018228426068385095,
"loss": 0.154,
"step": 218
},
{
"epoch": 0.22,
"learning_rate": 0.00018210016138212187,
"loss": 0.149,
"step": 219
},
{
"epoch": 0.22,
"learning_rate": 0.0001819152044288992,
"loss": 0.155,
"step": 220
},
{
"epoch": 0.22,
"learning_rate": 0.00018172939175631808,
"loss": 0.1557,
"step": 221
},
{
"epoch": 0.22,
"learning_rate": 0.00018154272530545285,
"loss": 0.153,
"step": 222
},
{
"epoch": 0.22,
"learning_rate": 0.00018135520702629675,
"loss": 0.1529,
"step": 223
},
{
"epoch": 0.22,
"learning_rate": 0.00018116683887774162,
"loss": 0.1542,
"step": 224
},
{
"epoch": 0.22,
"learning_rate": 0.00018097762282755727,
"loss": 0.1441,
"step": 225
},
{
"epoch": 0.23,
"learning_rate": 0.00018078756085237113,
"loss": 0.1554,
"step": 226
},
{
"epoch": 0.23,
"learning_rate": 0.00018059665493764743,
"loss": 0.1635,
"step": 227
},
{
"epoch": 0.23,
"learning_rate": 0.00018040490707766664,
"loss": 0.1583,
"step": 228
},
{
"epoch": 0.23,
"learning_rate": 0.0001802123192755044,
"loss": 0.1419,
"step": 229
},
{
"epoch": 0.23,
"learning_rate": 0.0001800188935430108,
"loss": 0.1549,
"step": 230
},
{
"epoch": 0.23,
"learning_rate": 0.0001798246319007893,
"loss": 0.151,
"step": 231
},
{
"epoch": 0.23,
"learning_rate": 0.00017962953637817557,
"loss": 0.1536,
"step": 232
},
{
"epoch": 0.23,
"learning_rate": 0.0001794336090132164,
"loss": 0.1557,
"step": 233
},
{
"epoch": 0.23,
"learning_rate": 0.0001792368518526482,
"loss": 0.1587,
"step": 234
},
{
"epoch": 0.23,
"learning_rate": 0.00017903926695187595,
"loss": 0.1485,
"step": 235
},
{
"epoch": 0.24,
"learning_rate": 0.00017884085637495138,
"loss": 0.1599,
"step": 236
},
{
"epoch": 0.24,
"learning_rate": 0.00017864162219455163,
"loss": 0.1608,
"step": 237
},
{
"epoch": 0.24,
"learning_rate": 0.00017844156649195759,
"loss": 0.1552,
"step": 238
},
{
"epoch": 0.24,
"learning_rate": 0.00017824069135703198,
"loss": 0.1595,
"step": 239
},
{
"epoch": 0.24,
"learning_rate": 0.00017803899888819776,
"loss": 0.1473,
"step": 240
},
{
"epoch": 0.24,
"learning_rate": 0.00017783649119241602,
"loss": 0.1592,
"step": 241
},
{
"epoch": 0.24,
"learning_rate": 0.00017763317038516405,
"loss": 0.1527,
"step": 242
},
{
"epoch": 0.24,
"learning_rate": 0.00017742903859041325,
"loss": 0.1437,
"step": 243
},
{
"epoch": 0.24,
"learning_rate": 0.00017722409794060693,
"loss": 0.1523,
"step": 244
},
{
"epoch": 0.24,
"learning_rate": 0.00017701835057663798,
"loss": 0.154,
"step": 245
},
{
"epoch": 0.25,
"learning_rate": 0.0001768117986478266,
"loss": 0.1463,
"step": 246
},
{
"epoch": 0.25,
"learning_rate": 0.0001766044443118978,
"loss": 0.1547,
"step": 247
},
{
"epoch": 0.25,
"learning_rate": 0.00017639628973495884,
"loss": 0.1654,
"step": 248
},
{
"epoch": 0.25,
"learning_rate": 0.00017618733709147662,
"loss": 0.163,
"step": 249
},
{
"epoch": 0.25,
"learning_rate": 0.00017597758856425494,
"loss": 0.1659,
"step": 250
},
{
"epoch": 0.25,
"learning_rate": 0.0001757670463444118,
"loss": 0.1531,
"step": 251
},
{
"epoch": 0.25,
"learning_rate": 0.00017555571263135636,
"loss": 0.1623,
"step": 252
},
{
"epoch": 0.25,
"learning_rate": 0.00017534358963276607,
"loss": 0.1545,
"step": 253
},
{
"epoch": 0.25,
"learning_rate": 0.0001751306795645636,
"loss": 0.1379,
"step": 254
},
{
"epoch": 0.25,
"learning_rate": 0.00017491698465089362,
"loss": 0.1601,
"step": 255
},
{
"epoch": 0.26,
"learning_rate": 0.0001747025071240996,
"loss": 0.1534,
"step": 256
},
{
"epoch": 0.26,
"learning_rate": 0.00017448724922470058,
"loss": 0.1483,
"step": 257
},
{
"epoch": 0.26,
"learning_rate": 0.00017427121320136765,
"loss": 0.1572,
"step": 258
},
{
"epoch": 0.26,
"learning_rate": 0.00017405440131090048,
"loss": 0.1478,
"step": 259
},
{
"epoch": 0.26,
"learning_rate": 0.00017383681581820373,
"loss": 0.1621,
"step": 260
},
{
"epoch": 0.26,
"learning_rate": 0.00017361845899626355,
"loss": 0.1553,
"step": 261
},
{
"epoch": 0.26,
"learning_rate": 0.00017339933312612353,
"loss": 0.1562,
"step": 262
},
{
"epoch": 0.26,
"learning_rate": 0.00017317944049686124,
"loss": 0.1512,
"step": 263
},
{
"epoch": 0.26,
"learning_rate": 0.00017295878340556394,
"loss": 0.1524,
"step": 264
},
{
"epoch": 0.26,
"learning_rate": 0.00017273736415730488,
"loss": 0.1563,
"step": 265
},
{
"epoch": 0.27,
"learning_rate": 0.0001725151850651191,
"loss": 0.1566,
"step": 266
},
{
"epoch": 0.27,
"learning_rate": 0.00017229224844997928,
"loss": 0.161,
"step": 267
},
{
"epoch": 0.27,
"learning_rate": 0.00017206855664077147,
"loss": 0.153,
"step": 268
},
{
"epoch": 0.27,
"learning_rate": 0.00017184411197427077,
"loss": 0.1697,
"step": 269
},
{
"epoch": 0.27,
"learning_rate": 0.00017161891679511694,
"loss": 0.1561,
"step": 270
},
{
"epoch": 0.27,
"learning_rate": 0.00017139297345578994,
"loss": 0.1599,
"step": 271
},
{
"epoch": 0.27,
"learning_rate": 0.00017116628431658524,
"loss": 0.1489,
"step": 272
},
{
"epoch": 0.27,
"learning_rate": 0.0001709388517455893,
"loss": 0.1563,
"step": 273
},
{
"epoch": 0.27,
"learning_rate": 0.00017071067811865476,
"loss": 0.1562,
"step": 274
},
{
"epoch": 0.27,
"learning_rate": 0.00017048176581937563,
"loss": 0.1461,
"step": 275
},
{
"epoch": 0.28,
"learning_rate": 0.00017025211723906234,
"loss": 0.1511,
"step": 276
},
{
"epoch": 0.28,
"learning_rate": 0.00017002173477671686,
"loss": 0.1477,
"step": 277
},
{
"epoch": 0.28,
"learning_rate": 0.0001697906208390076,
"loss": 0.1553,
"step": 278
},
{
"epoch": 0.28,
"learning_rate": 0.0001695587778402442,
"loss": 0.16,
"step": 279
},
{
"epoch": 0.28,
"learning_rate": 0.00016932620820235244,
"loss": 0.148,
"step": 280
},
{
"epoch": 0.28,
"learning_rate": 0.0001690929143548488,
"loss": 0.1626,
"step": 281
},
{
"epoch": 0.28,
"learning_rate": 0.00016885889873481514,
"loss": 0.1597,
"step": 282
},
{
"epoch": 0.28,
"learning_rate": 0.0001686241637868734,
"loss": 0.1518,
"step": 283
},
{
"epoch": 0.28,
"learning_rate": 0.00016838871196315963,
"loss": 0.1566,
"step": 284
},
{
"epoch": 0.28,
"learning_rate": 0.00016815254572329896,
"loss": 0.1569,
"step": 285
},
{
"epoch": 0.29,
"learning_rate": 0.00016791566753437932,
"loss": 0.1565,
"step": 286
},
{
"epoch": 0.29,
"learning_rate": 0.00016767807987092621,
"loss": 0.153,
"step": 287
},
{
"epoch": 0.29,
"learning_rate": 0.0001674397852148764,
"loss": 0.153,
"step": 288
},
{
"epoch": 0.29,
"learning_rate": 0.00016720078605555224,
"loss": 0.1569,
"step": 289
},
{
"epoch": 0.29,
"learning_rate": 0.00016696108488963566,
"loss": 0.1431,
"step": 290
},
{
"epoch": 0.29,
"learning_rate": 0.00016672068422114196,
"loss": 0.1563,
"step": 291
},
{
"epoch": 0.29,
"learning_rate": 0.00016647958656139378,
"loss": 0.1587,
"step": 292
},
{
"epoch": 0.29,
"learning_rate": 0.0001662377944289948,
"loss": 0.1606,
"step": 293
},
{
"epoch": 0.29,
"learning_rate": 0.0001659953103498034,
"loss": 0.1536,
"step": 294
},
{
"epoch": 0.29,
"learning_rate": 0.0001657521368569064,
"loss": 0.1541,
"step": 295
},
{
"epoch": 0.3,
"learning_rate": 0.00016550827649059239,
"loss": 0.154,
"step": 296
},
{
"epoch": 0.3,
"learning_rate": 0.00016526373179832546,
"loss": 0.1508,
"step": 297
},
{
"epoch": 0.3,
"learning_rate": 0.00016501850533471836,
"loss": 0.1534,
"step": 298
},
{
"epoch": 0.3,
"learning_rate": 0.00016477259966150588,
"loss": 0.1546,
"step": 299
},
{
"epoch": 0.3,
"learning_rate": 0.00016452601734751824,
"loss": 0.1545,
"step": 300
},
{
"epoch": 0.3,
"learning_rate": 0.00016427876096865394,
"loss": 0.1492,
"step": 301
},
{
"epoch": 0.3,
"learning_rate": 0.00016403083310785317,
"loss": 0.1607,
"step": 302
},
{
"epoch": 0.3,
"learning_rate": 0.0001637822363550706,
"loss": 0.1595,
"step": 303
},
{
"epoch": 0.3,
"learning_rate": 0.0001635329733072485,
"loss": 0.148,
"step": 304
},
{
"epoch": 0.3,
"learning_rate": 0.00016328304656828952,
"loss": 0.1595,
"step": 305
},
{
"epoch": 0.31,
"learning_rate": 0.00016303245874902937,
"loss": 0.1458,
"step": 306
},
{
"epoch": 0.31,
"learning_rate": 0.00016278121246720987,
"loss": 0.1491,
"step": 307
},
{
"epoch": 0.31,
"learning_rate": 0.00016252931034745123,
"loss": 0.1604,
"step": 308
},
{
"epoch": 0.31,
"learning_rate": 0.00016227675502122492,
"loss": 0.1504,
"step": 309
},
{
"epoch": 0.31,
"learning_rate": 0.000162023549126826,
"loss": 0.1547,
"step": 310
},
{
"epoch": 0.31,
"learning_rate": 0.00016176969530934572,
"loss": 0.1576,
"step": 311
},
{
"epoch": 0.31,
"learning_rate": 0.00016151519622064367,
"loss": 0.1542,
"step": 312
},
{
"epoch": 0.31,
"learning_rate": 0.0001612600545193203,
"loss": 0.1559,
"step": 313
},
{
"epoch": 0.31,
"learning_rate": 0.000161004272870689,
"loss": 0.1587,
"step": 314
},
{
"epoch": 0.31,
"learning_rate": 0.00016074785394674837,
"loss": 0.1601,
"step": 315
},
{
"epoch": 0.32,
"learning_rate": 0.00016049080042615415,
"loss": 0.1536,
"step": 316
},
{
"epoch": 0.32,
"learning_rate": 0.0001602331149941915,
"loss": 0.1622,
"step": 317
},
{
"epoch": 0.32,
"learning_rate": 0.00015997480034274651,
"loss": 0.1556,
"step": 318
},
{
"epoch": 0.32,
"learning_rate": 0.00015971585917027862,
"loss": 0.1523,
"step": 319
},
{
"epoch": 0.32,
"learning_rate": 0.00015945629418179203,
"loss": 0.1554,
"step": 320
},
{
"epoch": 0.32,
"learning_rate": 0.0001591961080888076,
"loss": 0.1658,
"step": 321
},
{
"epoch": 0.32,
"learning_rate": 0.00015893530360933447,
"loss": 0.1553,
"step": 322
},
{
"epoch": 0.32,
"learning_rate": 0.0001586738834678418,
"loss": 0.1584,
"step": 323
},
{
"epoch": 0.32,
"learning_rate": 0.00015841185039523002,
"loss": 0.1534,
"step": 324
},
{
"epoch": 0.32,
"learning_rate": 0.00015814920712880267,
"loss": 0.1562,
"step": 325
},
{
"epoch": 0.33,
"learning_rate": 0.00015788595641223752,
"loss": 0.1581,
"step": 326
},
{
"epoch": 0.33,
"learning_rate": 0.00015762210099555803,
"loss": 0.1615,
"step": 327
},
{
"epoch": 0.33,
"learning_rate": 0.0001573576436351046,
"loss": 0.1546,
"step": 328
},
{
"epoch": 0.33,
"learning_rate": 0.00015709258709350582,
"loss": 0.1507,
"step": 329
},
{
"epoch": 0.33,
"learning_rate": 0.0001568269341396495,
"loss": 0.1543,
"step": 330
},
{
"epoch": 0.33,
"learning_rate": 0.00015656068754865387,
"loss": 0.1576,
"step": 331
},
{
"epoch": 0.33,
"learning_rate": 0.0001562938501018385,
"loss": 0.1501,
"step": 332
},
{
"epoch": 0.33,
"learning_rate": 0.00015602642458669528,
"loss": 0.1506,
"step": 333
},
{
"epoch": 0.33,
"learning_rate": 0.00015575841379685928,
"loss": 0.1546,
"step": 334
},
{
"epoch": 0.33,
"learning_rate": 0.0001554898205320797,
"loss": 0.155,
"step": 335
},
{
"epoch": 0.33,
"learning_rate": 0.00015522064759819037,
"loss": 0.1545,
"step": 336
},
{
"epoch": 0.34,
"learning_rate": 0.0001549508978070806,
"loss": 0.1452,
"step": 337
},
{
"epoch": 0.34,
"learning_rate": 0.00015468057397666592,
"loss": 0.1498,
"step": 338
},
{
"epoch": 0.34,
"learning_rate": 0.00015440967893085828,
"loss": 0.1556,
"step": 339
},
{
"epoch": 0.34,
"learning_rate": 0.00015413821549953698,
"loss": 0.1462,
"step": 340
},
{
"epoch": 0.34,
"learning_rate": 0.0001538661865185188,
"loss": 0.16,
"step": 341
},
{
"epoch": 0.34,
"learning_rate": 0.00015359359482952848,
"loss": 0.155,
"step": 342
},
{
"epoch": 0.34,
"learning_rate": 0.00015332044328016914,
"loss": 0.1449,
"step": 343
},
{
"epoch": 0.34,
"learning_rate": 0.00015304673472389227,
"loss": 0.1534,
"step": 344
},
{
"epoch": 0.34,
"learning_rate": 0.0001527724720199682,
"loss": 0.1462,
"step": 345
},
{
"epoch": 0.34,
"learning_rate": 0.000152497658033456,
"loss": 0.1499,
"step": 346
},
{
"epoch": 0.35,
"learning_rate": 0.00015222229563517385,
"loss": 0.1451,
"step": 347
},
{
"epoch": 0.35,
"learning_rate": 0.00015194638770166865,
"loss": 0.1444,
"step": 348
},
{
"epoch": 0.35,
"learning_rate": 0.00015166993711518631,
"loss": 0.1437,
"step": 349
},
{
"epoch": 0.35,
"learning_rate": 0.00015139294676364147,
"loss": 0.1553,
"step": 350
},
{
"epoch": 0.35,
"learning_rate": 0.00015111541954058734,
"loss": 0.1555,
"step": 351
},
{
"epoch": 0.35,
"learning_rate": 0.00015083735834518555,
"loss": 0.1553,
"step": 352
},
{
"epoch": 0.35,
"learning_rate": 0.0001505587660821759,
"loss": 0.1545,
"step": 353
},
{
"epoch": 0.35,
"learning_rate": 0.00015027964566184578,
"loss": 0.1532,
"step": 354
},
{
"epoch": 0.35,
"learning_rate": 0.00015000000000000001,
"loss": 0.1528,
"step": 355
},
{
"epoch": 0.35,
"learning_rate": 0.00014971983201793028,
"loss": 0.157,
"step": 356
},
{
"epoch": 0.36,
"learning_rate": 0.00014943914464238468,
"loss": 0.1606,
"step": 357
},
{
"epoch": 0.36,
"learning_rate": 0.00014915794080553707,
"loss": 0.1428,
"step": 358
},
{
"epoch": 0.36,
"learning_rate": 0.00014887622344495643,
"loss": 0.1521,
"step": 359
},
{
"epoch": 0.36,
"learning_rate": 0.00014859399550357628,
"loss": 0.1559,
"step": 360
},
{
"epoch": 0.36,
"learning_rate": 0.00014831125992966385,
"loss": 0.1543,
"step": 361
},
{
"epoch": 0.36,
"learning_rate": 0.00014802801967678933,
"loss": 0.1555,
"step": 362
},
{
"epoch": 0.36,
"learning_rate": 0.0001477442777037949,
"loss": 0.1623,
"step": 363
},
{
"epoch": 0.36,
"learning_rate": 0.00014746003697476404,
"loss": 0.1547,
"step": 364
},
{
"epoch": 0.36,
"learning_rate": 0.00014717530045899038,
"loss": 0.1475,
"step": 365
},
{
"epoch": 0.36,
"learning_rate": 0.00014689007113094667,
"loss": 0.1456,
"step": 366
},
{
"epoch": 0.37,
"learning_rate": 0.0001466043519702539,
"loss": 0.1515,
"step": 367
},
{
"epoch": 0.37,
"learning_rate": 0.00014631814596164996,
"loss": 0.1519,
"step": 368
},
{
"epoch": 0.37,
"learning_rate": 0.0001460314560949586,
"loss": 0.1453,
"step": 369
},
{
"epoch": 0.37,
"learning_rate": 0.0001457442853650581,
"loss": 0.1538,
"step": 370
},
{
"epoch": 0.37,
"learning_rate": 0.00014545663677185006,
"loss": 0.1481,
"step": 371
},
{
"epoch": 0.37,
"learning_rate": 0.00014516851332022808,
"loss": 0.164,
"step": 372
},
{
"epoch": 0.37,
"learning_rate": 0.00014487991802004623,
"loss": 0.1489,
"step": 373
},
{
"epoch": 0.37,
"learning_rate": 0.00014459085388608778,
"loss": 0.1544,
"step": 374
},
{
"epoch": 0.37,
"learning_rate": 0.00014430132393803352,
"loss": 0.1647,
"step": 375
},
{
"epoch": 0.37,
"learning_rate": 0.00014401133120043049,
"loss": 0.1475,
"step": 376
},
{
"epoch": 0.38,
"learning_rate": 0.0001437208787026601,
"loss": 0.1563,
"step": 377
},
{
"epoch": 0.38,
"learning_rate": 0.00014342996947890656,
"loss": 0.1566,
"step": 378
},
{
"epoch": 0.38,
"learning_rate": 0.00014313860656812536,
"loss": 0.1622,
"step": 379
},
{
"epoch": 0.38,
"learning_rate": 0.0001428467930140113,
"loss": 0.152,
"step": 380
},
{
"epoch": 0.38,
"learning_rate": 0.00014255453186496673,
"loss": 0.1546,
"step": 381
},
{
"epoch": 0.38,
"learning_rate": 0.00014226182617406996,
"loss": 0.1569,
"step": 382
},
{
"epoch": 0.38,
"learning_rate": 0.0001419686789990429,
"loss": 0.1514,
"step": 383
},
{
"epoch": 0.38,
"learning_rate": 0.0001416750934022196,
"loss": 0.1578,
"step": 384
},
{
"epoch": 0.38,
"learning_rate": 0.00014138107245051392,
"loss": 0.1536,
"step": 385
},
{
"epoch": 0.38,
"learning_rate": 0.00014108661921538766,
"loss": 0.1495,
"step": 386
},
{
"epoch": 0.39,
"learning_rate": 0.00014079173677281837,
"loss": 0.1459,
"step": 387
},
{
"epoch": 0.39,
"learning_rate": 0.00014049642820326735,
"loss": 0.1483,
"step": 388
},
{
"epoch": 0.39,
"learning_rate": 0.0001402006965916474,
"loss": 0.1485,
"step": 389
},
{
"epoch": 0.39,
"learning_rate": 0.0001399045450272905,
"loss": 0.1568,
"step": 390
},
{
"epoch": 0.39,
"learning_rate": 0.0001396079766039157,
"loss": 0.148,
"step": 391
},
{
"epoch": 0.39,
"learning_rate": 0.0001393109944195967,
"loss": 0.1466,
"step": 392
},
{
"epoch": 0.39,
"learning_rate": 0.0001390136015767295,
"loss": 0.1503,
"step": 393
},
{
"epoch": 0.39,
"learning_rate": 0.00013871580118200006,
"loss": 0.1498,
"step": 394
},
{
"epoch": 0.39,
"learning_rate": 0.00013841759634635178,
"loss": 0.1616,
"step": 395
},
{
"epoch": 0.39,
"learning_rate": 0.00013811899018495292,
"loss": 0.1516,
"step": 396
},
{
"epoch": 0.4,
"learning_rate": 0.00013781998581716427,
"loss": 0.1441,
"step": 397
},
{
"epoch": 0.4,
"learning_rate": 0.00013752058636650637,
"loss": 0.1543,
"step": 398
},
{
"epoch": 0.4,
"learning_rate": 0.00013722079496062702,
"loss": 0.1548,
"step": 399
},
{
"epoch": 0.4,
"learning_rate": 0.00013692061473126845,
"loss": 0.1492,
"step": 400
},
{
"epoch": 0.4,
"learning_rate": 0.0001366200488142348,
"loss": 0.154,
"step": 401
},
{
"epoch": 0.4,
"learning_rate": 0.00013631910034935923,
"loss": 0.1481,
"step": 402
},
{
"epoch": 0.4,
"learning_rate": 0.00013601777248047105,
"loss": 0.1513,
"step": 403
},
{
"epoch": 0.4,
"learning_rate": 0.00013571606835536314,
"loss": 0.1408,
"step": 404
},
{
"epoch": 0.4,
"learning_rate": 0.00013541399112575878,
"loss": 0.1517,
"step": 405
},
{
"epoch": 0.4,
"learning_rate": 0.0001351115439472789,
"loss": 0.1501,
"step": 406
},
{
"epoch": 0.41,
"learning_rate": 0.00013480872997940905,
"loss": 0.1521,
"step": 407
},
{
"epoch": 0.41,
"learning_rate": 0.00013450555238546648,
"loss": 0.151,
"step": 408
},
{
"epoch": 0.41,
"learning_rate": 0.00013420201433256689,
"loss": 0.1519,
"step": 409
},
{
"epoch": 0.41,
"learning_rate": 0.0001338981189915916,
"loss": 0.1457,
"step": 410
},
{
"epoch": 0.41,
"learning_rate": 0.00013359386953715421,
"loss": 0.1488,
"step": 411
},
{
"epoch": 0.41,
"learning_rate": 0.00013328926914756765,
"loss": 0.1548,
"step": 412
},
{
"epoch": 0.41,
"learning_rate": 0.00013298432100481079,
"loss": 0.1803,
"step": 413
},
{
"epoch": 0.41,
"learning_rate": 0.00013267902829449524,
"loss": 0.2163,
"step": 414
},
{
"epoch": 0.41,
"learning_rate": 0.00013237339420583212,
"loss": 0.1509,
"step": 415
},
{
"epoch": 0.41,
"learning_rate": 0.00013206742193159877,
"loss": 0.1881,
"step": 416
},
{
"epoch": 0.42,
"learning_rate": 0.00013176111466810532,
"loss": 0.1693,
"step": 417
},
{
"epoch": 0.42,
"learning_rate": 0.00013145447561516138,
"loss": 0.1727,
"step": 418
},
{
"epoch": 0.42,
"learning_rate": 0.00013114750797604247,
"loss": 0.1628,
"step": 419
},
{
"epoch": 0.42,
"learning_rate": 0.00013084021495745674,
"loss": 0.1662,
"step": 420
},
{
"epoch": 0.42,
"learning_rate": 0.00013053259976951133,
"loss": 0.1561,
"step": 421
},
{
"epoch": 0.42,
"learning_rate": 0.00013022466562567894,
"loss": 0.1626,
"step": 422
},
{
"epoch": 0.42,
"learning_rate": 0.00012991641574276418,
"loss": 0.1568,
"step": 423
},
{
"epoch": 0.42,
"learning_rate": 0.00012960785334087,
"loss": 0.1491,
"step": 424
},
{
"epoch": 0.42,
"learning_rate": 0.00012929898164336408,
"loss": 0.1549,
"step": 425
},
{
"epoch": 0.42,
"learning_rate": 0.0001289898038768451,
"loss": 0.1539,
"step": 426
},
{
"epoch": 0.43,
"learning_rate": 0.00012868032327110904,
"loss": 0.1518,
"step": 427
},
{
"epoch": 0.43,
"learning_rate": 0.00012837054305911552,
"loss": 0.1566,
"step": 428
},
{
"epoch": 0.43,
"learning_rate": 0.0001280604664769539,
"loss": 0.154,
"step": 429
},
{
"epoch": 0.43,
"learning_rate": 0.00012775009676380957,
"loss": 0.154,
"step": 430
},
{
"epoch": 0.43,
"learning_rate": 0.00012743943716193016,
"loss": 0.1882,
"step": 431
},
{
"epoch": 0.43,
"learning_rate": 0.00012712849091659148,
"loss": 0.1616,
"step": 432
},
{
"epoch": 0.43,
"learning_rate": 0.00012681726127606376,
"loss": 0.1539,
"step": 433
},
{
"epoch": 0.43,
"learning_rate": 0.00012650575149157772,
"loss": 1.3991,
"step": 434
},
{
"epoch": 0.43,
"learning_rate": 0.0001261939648172906,
"loss": 0.1691,
"step": 435
},
{
"epoch": 0.43,
"learning_rate": 0.00012588190451025207,
"loss": 0.2529,
"step": 436
},
{
"epoch": 0.44,
"learning_rate": 0.0001255695738303704,
"loss": 0.1861,
"step": 437
},
{
"epoch": 0.44,
"learning_rate": 0.00012525697604037807,
"loss": 0.1957,
"step": 438
},
{
"epoch": 0.44,
"learning_rate": 0.00012494411440579814,
"loss": 0.1941,
"step": 439
},
{
"epoch": 0.44,
"learning_rate": 0.00012463099219490976,
"loss": 0.1773,
"step": 440
},
{
"epoch": 0.44,
"learning_rate": 0.00012431761267871417,
"loss": 0.1773,
"step": 441
},
{
"epoch": 0.44,
"learning_rate": 0.0001240039791309006,
"loss": 0.1804,
"step": 442
},
{
"epoch": 0.44,
"learning_rate": 0.00012369009482781192,
"loss": 0.1821,
"step": 443
},
{
"epoch": 0.44,
"learning_rate": 0.00012337596304841053,
"loss": 0.1629,
"step": 444
},
{
"epoch": 0.44,
"learning_rate": 0.00012306158707424403,
"loss": 0.1574,
"step": 445
},
{
"epoch": 0.44,
"learning_rate": 0.000122746970189411,
"loss": 0.1712,
"step": 446
},
{
"epoch": 0.45,
"learning_rate": 0.00012243211568052677,
"loss": 0.1652,
"step": 447
},
{
"epoch": 0.45,
"learning_rate": 0.00012211702683668878,
"loss": 0.1675,
"step": 448
},
{
"epoch": 0.45,
"learning_rate": 0.0001218017069494426,
"loss": 0.1739,
"step": 449
},
{
"epoch": 0.45,
"learning_rate": 0.00012148615931274729,
"loss": 0.1712,
"step": 450
},
{
"epoch": 0.45,
"learning_rate": 0.0001211703872229411,
"loss": 0.166,
"step": 451
},
{
"epoch": 0.45,
"learning_rate": 0.00012085439397870699,
"loss": 0.1694,
"step": 452
},
{
"epoch": 0.45,
"learning_rate": 0.0001205381828810382,
"loss": 0.1615,
"step": 453
},
{
"epoch": 0.45,
"learning_rate": 0.00012022175723320381,
"loss": 0.1512,
"step": 454
},
{
"epoch": 0.45,
"learning_rate": 0.00011990512034071406,
"loss": 0.1643,
"step": 455
},
{
"epoch": 0.45,
"learning_rate": 0.00011958827551128603,
"loss": 0.1576,
"step": 456
},
{
"epoch": 0.46,
"learning_rate": 0.00011927122605480898,
"loss": 0.1517,
"step": 457
},
{
"epoch": 0.46,
"learning_rate": 0.00011895397528330974,
"loss": 0.1676,
"step": 458
},
{
"epoch": 0.46,
"learning_rate": 0.00011863652651091823,
"loss": 0.1611,
"step": 459
},
{
"epoch": 0.46,
"learning_rate": 0.00011831888305383268,
"loss": 0.1559,
"step": 460
},
{
"epoch": 0.46,
"learning_rate": 0.00011800104823028515,
"loss": 0.1595,
"step": 461
},
{
"epoch": 0.46,
"learning_rate": 0.00011768302536050675,
"loss": 0.1662,
"step": 462
},
{
"epoch": 0.46,
"learning_rate": 0.00011736481776669306,
"loss": 0.1584,
"step": 463
},
{
"epoch": 0.46,
"learning_rate": 0.00011704642877296926,
"loss": 0.145,
"step": 464
},
{
"epoch": 0.46,
"learning_rate": 0.00011672786170535552,
"loss": 0.1635,
"step": 465
},
{
"epoch": 0.46,
"learning_rate": 0.00011640911989173242,
"loss": 0.1463,
"step": 466
},
{
"epoch": 0.47,
"learning_rate": 0.00011609020666180575,
"loss": 0.1591,
"step": 467
},
{
"epoch": 0.47,
"learning_rate": 0.00011577112534707213,
"loss": 0.1599,
"step": 468
},
{
"epoch": 0.47,
"learning_rate": 0.00011545187928078406,
"loss": 0.1605,
"step": 469
},
{
"epoch": 0.47,
"learning_rate": 0.00011513247179791505,
"loss": 0.1586,
"step": 470
},
{
"epoch": 0.47,
"learning_rate": 0.0001148129062351249,
"loss": 0.1605,
"step": 471
},
{
"epoch": 0.47,
"learning_rate": 0.00011449318593072466,
"loss": 0.1552,
"step": 472
},
{
"epoch": 0.47,
"learning_rate": 0.00011417331422464205,
"loss": 0.1518,
"step": 473
},
{
"epoch": 0.47,
"learning_rate": 0.00011385329445838622,
"loss": 0.1583,
"step": 474
},
{
"epoch": 0.47,
"learning_rate": 0.00011353312997501313,
"loss": 0.1587,
"step": 475
},
{
"epoch": 0.47,
"learning_rate": 0.00011321282411909045,
"loss": 0.1658,
"step": 476
},
{
"epoch": 0.48,
"learning_rate": 0.00011289238023666266,
"loss": 0.1553,
"step": 477
},
{
"epoch": 0.48,
"learning_rate": 0.00011257180167521629,
"loss": 0.1526,
"step": 478
},
{
"epoch": 0.48,
"learning_rate": 0.00011225109178364455,
"loss": 0.1516,
"step": 479
},
{
"epoch": 0.48,
"learning_rate": 0.00011193025391221274,
"loss": 0.1521,
"step": 480
},
{
"epoch": 0.48,
"learning_rate": 0.00011160929141252303,
"loss": 0.1593,
"step": 481
},
{
"epoch": 0.48,
"learning_rate": 0.00011128820763747948,
"loss": 0.1524,
"step": 482
},
{
"epoch": 0.48,
"learning_rate": 0.00011096700594125318,
"loss": 0.1572,
"step": 483
},
{
"epoch": 0.48,
"learning_rate": 0.00011064568967924683,
"loss": 0.1588,
"step": 484
},
{
"epoch": 0.48,
"learning_rate": 0.00011032426220806018,
"loss": 0.1518,
"step": 485
},
{
"epoch": 0.48,
"learning_rate": 0.0001100027268854545,
"loss": 0.1572,
"step": 486
},
{
"epoch": 0.49,
"learning_rate": 0.00010968108707031792,
"loss": 0.154,
"step": 487
},
{
"epoch": 0.49,
"learning_rate": 0.00010935934612263001,
"loss": 0.1599,
"step": 488
},
{
"epoch": 0.49,
"learning_rate": 0.00010903750740342682,
"loss": 0.1553,
"step": 489
},
{
"epoch": 0.49,
"learning_rate": 0.00010871557427476583,
"loss": 0.1534,
"step": 490
},
{
"epoch": 0.49,
"learning_rate": 0.00010839355009969068,
"loss": 0.1518,
"step": 491
},
{
"epoch": 0.49,
"learning_rate": 0.00010807143824219621,
"loss": 0.1486,
"step": 492
},
{
"epoch": 0.49,
"learning_rate": 0.0001077492420671931,
"loss": 0.154,
"step": 493
},
{
"epoch": 0.49,
"learning_rate": 0.00010742696494047295,
"loss": 0.1508,
"step": 494
},
{
"epoch": 0.49,
"learning_rate": 0.00010710461022867302,
"loss": 0.1602,
"step": 495
},
{
"epoch": 0.49,
"learning_rate": 0.00010678218129924094,
"loss": 0.1467,
"step": 496
},
{
"epoch": 0.5,
"learning_rate": 0.0001064596815203998,
"loss": 0.1619,
"step": 497
},
{
"epoch": 0.5,
"learning_rate": 0.00010613711426111263,
"loss": 0.1487,
"step": 498
},
{
"epoch": 0.5,
"learning_rate": 0.00010581448289104758,
"loss": 0.1583,
"step": 499
},
{
"epoch": 0.5,
"learning_rate": 0.00010549179078054243,
"loss": 0.1454,
"step": 500
},
{
"epoch": 0.5,
"learning_rate": 0.00010516904130056946,
"loss": 0.1522,
"step": 501
},
{
"epoch": 0.5,
"learning_rate": 0.00010484623782270031,
"loss": 0.1488,
"step": 502
},
{
"epoch": 0.5,
"learning_rate": 0.00010452338371907064,
"loss": 0.1423,
"step": 503
},
{
"epoch": 0.5,
"learning_rate": 0.00010420048236234505,
"loss": 0.1435,
"step": 504
},
{
"epoch": 0.5,
"learning_rate": 0.0001038775371256817,
"loss": 0.156,
"step": 505
},
{
"epoch": 0.5,
"learning_rate": 0.0001035545513826971,
"loss": 0.1502,
"step": 506
},
{
"epoch": 0.51,
"learning_rate": 0.00010323152850743107,
"loss": 0.1555,
"step": 507
},
{
"epoch": 0.51,
"learning_rate": 0.00010290847187431113,
"loss": 0.1573,
"step": 508
},
{
"epoch": 0.51,
"learning_rate": 0.00010258538485811765,
"loss": 0.1592,
"step": 509
},
{
"epoch": 0.51,
"learning_rate": 0.0001022622708339481,
"loss": 0.1471,
"step": 510
},
{
"epoch": 0.51,
"learning_rate": 0.00010193913317718244,
"loss": 0.1563,
"step": 511
},
{
"epoch": 0.51,
"learning_rate": 0.00010161597526344728,
"loss": 0.1649,
"step": 512
},
{
"epoch": 0.51,
"learning_rate": 0.00010129280046858086,
"loss": 0.1526,
"step": 513
},
{
"epoch": 0.51,
"learning_rate": 0.00010096961216859787,
"loss": 0.1456,
"step": 514
},
{
"epoch": 0.51,
"learning_rate": 0.00010064641373965393,
"loss": 0.1511,
"step": 515
},
{
"epoch": 0.51,
"learning_rate": 0.00010032320855801063,
"loss": 0.153,
"step": 516
},
{
"epoch": 0.52,
"learning_rate": 0.0001,
"loss": 0.1522,
"step": 517
},
{
"epoch": 0.52,
"learning_rate": 9.967679144198939e-05,
"loss": 0.1512,
"step": 518
},
{
"epoch": 0.52,
"learning_rate": 9.935358626034606e-05,
"loss": 0.1515,
"step": 519
},
{
"epoch": 0.52,
"learning_rate": 9.903038783140216e-05,
"loss": 0.1542,
"step": 520
},
{
"epoch": 0.52,
"learning_rate": 9.870719953141917e-05,
"loss": 0.1498,
"step": 521
},
{
"epoch": 0.52,
"learning_rate": 9.838402473655277e-05,
"loss": 0.1613,
"step": 522
},
{
"epoch": 0.52,
"learning_rate": 9.806086682281758e-05,
"loss": 0.1538,
"step": 523
},
{
"epoch": 0.52,
"learning_rate": 9.773772916605191e-05,
"loss": 0.1464,
"step": 524
},
{
"epoch": 0.52,
"learning_rate": 9.741461514188242e-05,
"loss": 0.1497,
"step": 525
},
{
"epoch": 0.52,
"learning_rate": 9.709152812568886e-05,
"loss": 0.1486,
"step": 526
},
{
"epoch": 0.53,
"learning_rate": 9.676847149256895e-05,
"loss": 0.1526,
"step": 527
},
{
"epoch": 0.53,
"learning_rate": 9.644544861730292e-05,
"loss": 0.1451,
"step": 528
},
{
"epoch": 0.53,
"learning_rate": 9.612246287431831e-05,
"loss": 0.1473,
"step": 529
},
{
"epoch": 0.53,
"learning_rate": 9.579951763765498e-05,
"loss": 0.1442,
"step": 530
},
{
"epoch": 0.53,
"learning_rate": 9.547661628092937e-05,
"loss": 0.138,
"step": 531
},
{
"epoch": 0.53,
"learning_rate": 9.515376217729972e-05,
"loss": 0.1527,
"step": 532
},
{
"epoch": 0.53,
"learning_rate": 9.483095869943055e-05,
"loss": 0.1567,
"step": 533
},
{
"epoch": 0.53,
"learning_rate": 9.450820921945758e-05,
"loss": 0.1508,
"step": 534
},
{
"epoch": 0.53,
"learning_rate": 9.418551710895243e-05,
"loss": 0.1468,
"step": 535
},
{
"epoch": 0.53,
"learning_rate": 9.386288573888736e-05,
"loss": 0.1517,
"step": 536
},
{
"epoch": 0.54,
"learning_rate": 9.354031847960022e-05,
"loss": 0.1537,
"step": 537
},
{
"epoch": 0.54,
"learning_rate": 9.321781870075908e-05,
"loss": 0.1454,
"step": 538
},
{
"epoch": 0.54,
"learning_rate": 9.289538977132703e-05,
"loss": 0.1486,
"step": 539
},
{
"epoch": 0.54,
"learning_rate": 9.257303505952704e-05,
"loss": 0.1567,
"step": 540
},
{
"epoch": 0.54,
"learning_rate": 9.225075793280692e-05,
"loss": 0.144,
"step": 541
},
{
"epoch": 0.54,
"learning_rate": 9.192856175780383e-05,
"loss": 0.1458,
"step": 542
},
{
"epoch": 0.54,
"learning_rate": 9.160644990030931e-05,
"loss": 0.1565,
"step": 543
},
{
"epoch": 0.54,
"learning_rate": 9.128442572523417e-05,
"loss": 0.1516,
"step": 544
},
{
"epoch": 0.54,
"learning_rate": 9.09624925965732e-05,
"loss": 0.1674,
"step": 545
},
{
"epoch": 0.54,
"learning_rate": 9.064065387737004e-05,
"loss": 0.1533,
"step": 546
},
{
"epoch": 0.55,
"learning_rate": 9.03189129296821e-05,
"loss": 0.1557,
"step": 547
},
{
"epoch": 0.55,
"learning_rate": 8.99972731145455e-05,
"loss": 0.1504,
"step": 548
},
{
"epoch": 0.55,
"learning_rate": 8.967573779193987e-05,
"loss": 0.1536,
"step": 549
},
{
"epoch": 0.55,
"learning_rate": 8.935431032075318e-05,
"loss": 0.1492,
"step": 550
},
{
"epoch": 0.55,
"learning_rate": 8.903299405874684e-05,
"loss": 0.155,
"step": 551
},
{
"epoch": 0.55,
"learning_rate": 8.871179236252052e-05,
"loss": 0.1509,
"step": 552
},
{
"epoch": 0.55,
"learning_rate": 8.839070858747697e-05,
"loss": 0.1494,
"step": 553
},
{
"epoch": 0.55,
"learning_rate": 8.806974608778728e-05,
"loss": 0.1511,
"step": 554
},
{
"epoch": 0.55,
"learning_rate": 8.774890821635548e-05,
"loss": 0.1579,
"step": 555
},
{
"epoch": 0.55,
"learning_rate": 8.742819832478375e-05,
"loss": 0.1492,
"step": 556
},
{
"epoch": 0.56,
"learning_rate": 8.710761976333734e-05,
"loss": 0.1487,
"step": 557
},
{
"epoch": 0.56,
"learning_rate": 8.678717588090959e-05,
"loss": 0.1469,
"step": 558
},
{
"epoch": 0.56,
"learning_rate": 8.646687002498692e-05,
"loss": 0.1516,
"step": 559
},
{
"epoch": 0.56,
"learning_rate": 8.614670554161376e-05,
"loss": 0.1505,
"step": 560
},
{
"epoch": 0.56,
"learning_rate": 8.582668577535797e-05,
"loss": 0.1573,
"step": 561
},
{
"epoch": 0.56,
"learning_rate": 8.550681406927535e-05,
"loss": 0.1564,
"step": 562
},
{
"epoch": 0.56,
"learning_rate": 8.518709376487515e-05,
"loss": 0.1544,
"step": 563
},
{
"epoch": 0.56,
"learning_rate": 8.486752820208496e-05,
"loss": 0.1508,
"step": 564
},
{
"epoch": 0.56,
"learning_rate": 8.454812071921596e-05,
"loss": 0.1428,
"step": 565
},
{
"epoch": 0.56,
"learning_rate": 8.422887465292789e-05,
"loss": 0.1572,
"step": 566
},
{
"epoch": 0.57,
"learning_rate": 8.390979333819426e-05,
"loss": 0.1478,
"step": 567
},
{
"epoch": 0.57,
"learning_rate": 8.35908801082676e-05,
"loss": 0.1491,
"step": 568
},
{
"epoch": 0.57,
"learning_rate": 8.327213829464449e-05,
"loss": 0.1497,
"step": 569
},
{
"epoch": 0.57,
"learning_rate": 8.29535712270308e-05,
"loss": 0.1591,
"step": 570
},
{
"epoch": 0.57,
"learning_rate": 8.263518223330697e-05,
"loss": 0.147,
"step": 571
},
{
"epoch": 0.57,
"learning_rate": 8.231697463949326e-05,
"loss": 0.143,
"step": 572
},
{
"epoch": 0.57,
"learning_rate": 8.199895176971488e-05,
"loss": 0.1442,
"step": 573
},
{
"epoch": 0.57,
"learning_rate": 8.168111694616733e-05,
"loss": 0.1507,
"step": 574
},
{
"epoch": 0.57,
"learning_rate": 8.13634734890818e-05,
"loss": 0.1547,
"step": 575
},
{
"epoch": 0.57,
"learning_rate": 8.104602471669029e-05,
"loss": 0.1451,
"step": 576
},
{
"epoch": 0.58,
"learning_rate": 8.072877394519102e-05,
"loss": 0.1454,
"step": 577
},
{
"epoch": 0.58,
"learning_rate": 8.041172448871398e-05,
"loss": 0.1508,
"step": 578
},
{
"epoch": 0.58,
"learning_rate": 8.009487965928596e-05,
"loss": 0.1415,
"step": 579
},
{
"epoch": 0.58,
"learning_rate": 7.977824276679623e-05,
"loss": 0.1513,
"step": 580
},
{
"epoch": 0.58,
"learning_rate": 7.94618171189618e-05,
"loss": 0.1543,
"step": 581
},
{
"epoch": 0.58,
"learning_rate": 7.914560602129305e-05,
"loss": 0.1467,
"step": 582
},
{
"epoch": 0.58,
"learning_rate": 7.882961277705895e-05,
"loss": 0.148,
"step": 583
},
{
"epoch": 0.58,
"learning_rate": 7.851384068725272e-05,
"loss": 0.1489,
"step": 584
},
{
"epoch": 0.58,
"learning_rate": 7.819829305055743e-05,
"loss": 0.1537,
"step": 585
},
{
"epoch": 0.58,
"learning_rate": 7.788297316331123e-05,
"loss": 0.1404,
"step": 586
},
{
"epoch": 0.59,
"learning_rate": 7.756788431947326e-05,
"loss": 0.1509,
"step": 587
},
{
"epoch": 0.59,
"learning_rate": 7.725302981058899e-05,
"loss": 0.1435,
"step": 588
},
{
"epoch": 0.59,
"learning_rate": 7.693841292575598e-05,
"loss": 0.1435,
"step": 589
},
{
"epoch": 0.59,
"learning_rate": 7.662403695158952e-05,
"loss": 0.1614,
"step": 590
},
{
"epoch": 0.59,
"learning_rate": 7.630990517218808e-05,
"loss": 0.1499,
"step": 591
},
{
"epoch": 0.59,
"learning_rate": 7.599602086909943e-05,
"loss": 0.148,
"step": 592
},
{
"epoch": 0.59,
"learning_rate": 7.568238732128585e-05,
"loss": 0.1507,
"step": 593
},
{
"epoch": 0.59,
"learning_rate": 7.536900780509029e-05,
"loss": 0.1421,
"step": 594
},
{
"epoch": 0.59,
"learning_rate": 7.505588559420189e-05,
"loss": 0.1509,
"step": 595
},
{
"epoch": 0.59,
"learning_rate": 7.474302395962195e-05,
"loss": 0.1461,
"step": 596
},
{
"epoch": 0.6,
"learning_rate": 7.443042616962965e-05,
"loss": 0.1542,
"step": 597
},
{
"epoch": 0.6,
"learning_rate": 7.411809548974792e-05,
"loss": 0.1439,
"step": 598
},
{
"epoch": 0.6,
"learning_rate": 7.380603518270941e-05,
"loss": 0.1448,
"step": 599
},
{
"epoch": 0.6,
"learning_rate": 7.349424850842232e-05,
"loss": 0.1485,
"step": 600
},
{
"epoch": 0.6,
"learning_rate": 7.318273872393625e-05,
"loss": 0.1398,
"step": 601
},
{
"epoch": 0.6,
"learning_rate": 7.287150908340857e-05,
"loss": 0.1552,
"step": 602
},
{
"epoch": 0.6,
"learning_rate": 7.256056283806986e-05,
"loss": 0.1516,
"step": 603
},
{
"epoch": 0.6,
"learning_rate": 7.224990323619044e-05,
"loss": 0.1511,
"step": 604
},
{
"epoch": 0.6,
"learning_rate": 7.193953352304613e-05,
"loss": 0.1428,
"step": 605
},
{
"epoch": 0.6,
"learning_rate": 7.16294569408845e-05,
"loss": 0.152,
"step": 606
},
{
"epoch": 0.61,
"learning_rate": 7.131967672889101e-05,
"loss": 0.1514,
"step": 607
},
{
"epoch": 0.61,
"learning_rate": 7.101019612315492e-05,
"loss": 0.1418,
"step": 608
},
{
"epoch": 0.61,
"learning_rate": 7.070101835663594e-05,
"loss": 0.1485,
"step": 609
},
{
"epoch": 0.61,
"learning_rate": 7.039214665913003e-05,
"loss": 0.1551,
"step": 610
},
{
"epoch": 0.61,
"learning_rate": 7.008358425723585e-05,
"loss": 0.1491,
"step": 611
},
{
"epoch": 0.61,
"learning_rate": 6.97753343743211e-05,
"loss": 0.1434,
"step": 612
},
{
"epoch": 0.61,
"learning_rate": 6.94674002304887e-05,
"loss": 0.1546,
"step": 613
},
{
"epoch": 0.61,
"learning_rate": 6.915978504254331e-05,
"loss": 0.1557,
"step": 614
},
{
"epoch": 0.61,
"learning_rate": 6.885249202395754e-05,
"loss": 0.1491,
"step": 615
},
{
"epoch": 0.61,
"learning_rate": 6.854552438483865e-05,
"loss": 0.1538,
"step": 616
},
{
"epoch": 0.62,
"learning_rate": 6.823888533189469e-05,
"loss": 0.146,
"step": 617
},
{
"epoch": 0.62,
"learning_rate": 6.793257806840122e-05,
"loss": 0.1529,
"step": 618
},
{
"epoch": 0.62,
"learning_rate": 6.762660579416791e-05,
"loss": 0.1414,
"step": 619
},
{
"epoch": 0.62,
"learning_rate": 6.73209717055048e-05,
"loss": 0.1434,
"step": 620
},
{
"epoch": 0.62,
"learning_rate": 6.701567899518924e-05,
"loss": 0.1489,
"step": 621
},
{
"epoch": 0.62,
"learning_rate": 6.671073085243233e-05,
"loss": 0.1453,
"step": 622
},
{
"epoch": 0.62,
"learning_rate": 6.640613046284581e-05,
"loss": 0.1463,
"step": 623
},
{
"epoch": 0.62,
"learning_rate": 6.610188100840843e-05,
"loss": 0.1488,
"step": 624
},
{
"epoch": 0.62,
"learning_rate": 6.579798566743314e-05,
"loss": 0.158,
"step": 625
},
{
"epoch": 0.62,
"learning_rate": 6.549444761453355e-05,
"loss": 0.1486,
"step": 626
},
{
"epoch": 0.63,
"learning_rate": 6.519127002059095e-05,
"loss": 0.142,
"step": 627
},
{
"epoch": 0.63,
"learning_rate": 6.488845605272113e-05,
"loss": 0.145,
"step": 628
},
{
"epoch": 0.63,
"learning_rate": 6.458600887424125e-05,
"loss": 0.1533,
"step": 629
},
{
"epoch": 0.63,
"learning_rate": 6.428393164463689e-05,
"loss": 0.1493,
"step": 630
},
{
"epoch": 0.63,
"learning_rate": 6.398222751952899e-05,
"loss": 0.15,
"step": 631
},
{
"epoch": 0.63,
"learning_rate": 6.36808996506408e-05,
"loss": 0.1473,
"step": 632
},
{
"epoch": 0.63,
"learning_rate": 6.337995118576521e-05,
"loss": 0.1539,
"step": 633
},
{
"epoch": 0.63,
"learning_rate": 6.307938526873157e-05,
"loss": 0.1474,
"step": 634
},
{
"epoch": 0.63,
"learning_rate": 6.277920503937303e-05,
"loss": 0.1428,
"step": 635
},
{
"epoch": 0.63,
"learning_rate": 6.247941363349365e-05,
"loss": 0.1395,
"step": 636
},
{
"epoch": 0.64,
"learning_rate": 6.218001418283576e-05,
"loss": 0.1539,
"step": 637
},
{
"epoch": 0.64,
"learning_rate": 6.188100981504712e-05,
"loss": 0.1575,
"step": 638
},
{
"epoch": 0.64,
"learning_rate": 6.158240365364823e-05,
"loss": 0.1565,
"step": 639
},
{
"epoch": 0.64,
"learning_rate": 6.128419881799996e-05,
"loss": 0.1525,
"step": 640
},
{
"epoch": 0.64,
"learning_rate": 6.098639842327052e-05,
"loss": 0.149,
"step": 641
},
{
"epoch": 0.64,
"learning_rate": 6.068900558040332e-05,
"loss": 0.1459,
"step": 642
},
{
"epoch": 0.64,
"learning_rate": 6.039202339608432e-05,
"loss": 0.151,
"step": 643
},
{
"epoch": 0.64,
"learning_rate": 6.009545497270952e-05,
"loss": 0.1443,
"step": 644
},
{
"epoch": 0.64,
"learning_rate": 5.979930340835265e-05,
"loss": 0.1554,
"step": 645
},
{
"epoch": 0.64,
"learning_rate": 5.950357179673264e-05,
"loss": 0.1463,
"step": 646
},
{
"epoch": 0.65,
"learning_rate": 5.920826322718165e-05,
"loss": 0.1571,
"step": 647
},
{
"epoch": 0.65,
"learning_rate": 5.891338078461238e-05,
"loss": 0.146,
"step": 648
},
{
"epoch": 0.65,
"learning_rate": 5.861892754948609e-05,
"loss": 0.1595,
"step": 649
},
{
"epoch": 0.65,
"learning_rate": 5.83249065977804e-05,
"loss": 0.1562,
"step": 650
},
{
"epoch": 0.65,
"learning_rate": 5.80313210009571e-05,
"loss": 0.1407,
"step": 651
},
{
"epoch": 0.65,
"learning_rate": 5.773817382593008e-05,
"loss": 0.1441,
"step": 652
},
{
"epoch": 0.65,
"learning_rate": 5.744546813503328e-05,
"loss": 0.1515,
"step": 653
},
{
"epoch": 0.65,
"learning_rate": 5.715320698598875e-05,
"loss": 0.1491,
"step": 654
},
{
"epoch": 0.65,
"learning_rate": 5.6861393431874675e-05,
"loss": 0.1448,
"step": 655
},
{
"epoch": 0.65,
"learning_rate": 5.657003052109344e-05,
"loss": 0.1446,
"step": 656
},
{
"epoch": 0.66,
"learning_rate": 5.6279121297339924e-05,
"loss": 0.1413,
"step": 657
},
{
"epoch": 0.66,
"learning_rate": 5.5988668799569545e-05,
"loss": 0.1481,
"step": 658
},
{
"epoch": 0.66,
"learning_rate": 5.5698676061966515e-05,
"loss": 0.1497,
"step": 659
},
{
"epoch": 0.66,
"learning_rate": 5.540914611391226e-05,
"loss": 0.1498,
"step": 660
},
{
"epoch": 0.66,
"learning_rate": 5.5120081979953785e-05,
"loss": 0.146,
"step": 661
},
{
"epoch": 0.66,
"learning_rate": 5.483148667977194e-05,
"loss": 0.1534,
"step": 662
},
{
"epoch": 0.66,
"learning_rate": 5.4543363228149946e-05,
"loss": 0.1521,
"step": 663
},
{
"epoch": 0.66,
"learning_rate": 5.4255714634941936e-05,
"loss": 0.1556,
"step": 664
},
{
"epoch": 0.66,
"learning_rate": 5.3968543905041444e-05,
"loss": 0.1584,
"step": 665
},
{
"epoch": 0.66,
"learning_rate": 5.3681854038350046e-05,
"loss": 0.1501,
"step": 666
},
{
"epoch": 0.67,
"learning_rate": 5.339564802974615e-05,
"loss": 0.1462,
"step": 667
},
{
"epoch": 0.67,
"learning_rate": 5.310992886905337e-05,
"loss": 0.1456,
"step": 668
},
{
"epoch": 0.67,
"learning_rate": 5.282469954100968e-05,
"loss": 0.1476,
"step": 669
},
{
"epoch": 0.67,
"learning_rate": 5.253996302523596e-05,
"loss": 0.1588,
"step": 670
},
{
"epoch": 0.67,
"learning_rate": 5.22557222962051e-05,
"loss": 0.1463,
"step": 671
},
{
"epoch": 0.67,
"learning_rate": 5.197198032321069e-05,
"loss": 0.1438,
"step": 672
},
{
"epoch": 0.67,
"learning_rate": 5.168874007033615e-05,
"loss": 0.1521,
"step": 673
},
{
"epoch": 0.67,
"learning_rate": 5.140600449642373e-05,
"loss": 0.152,
"step": 674
},
{
"epoch": 0.67,
"learning_rate": 5.112377655504359e-05,
"loss": 0.1514,
"step": 675
},
{
"epoch": 0.67,
"learning_rate": 5.0842059194462956e-05,
"loss": 0.1435,
"step": 676
},
{
"epoch": 0.67,
"learning_rate": 5.056085535761532e-05,
"loss": 0.1495,
"step": 677
},
{
"epoch": 0.68,
"learning_rate": 5.0280167982069735e-05,
"loss": 0.1368,
"step": 678
},
{
"epoch": 0.68,
"learning_rate": 5.000000000000002e-05,
"loss": 0.1505,
"step": 679
},
{
"epoch": 0.68,
"learning_rate": 4.972035433815422e-05,
"loss": 0.1419,
"step": 680
},
{
"epoch": 0.68,
"learning_rate": 4.9441233917824106e-05,
"loss": 0.1493,
"step": 681
},
{
"epoch": 0.68,
"learning_rate": 4.916264165481448e-05,
"loss": 0.1479,
"step": 682
},
{
"epoch": 0.68,
"learning_rate": 4.888458045941269e-05,
"loss": 0.148,
"step": 683
},
{
"epoch": 0.68,
"learning_rate": 4.860705323635857e-05,
"loss": 0.155,
"step": 684
},
{
"epoch": 0.68,
"learning_rate": 4.833006288481371e-05,
"loss": 0.1471,
"step": 685
},
{
"epoch": 0.68,
"learning_rate": 4.805361229833136e-05,
"loss": 0.1526,
"step": 686
},
{
"epoch": 0.68,
"learning_rate": 4.777770436482617e-05,
"loss": 0.1398,
"step": 687
},
{
"epoch": 0.69,
"learning_rate": 4.7502341966544e-05,
"loss": 0.1465,
"step": 688
},
{
"epoch": 0.69,
"learning_rate": 4.722752798003184e-05,
"loss": 0.1469,
"step": 689
},
{
"epoch": 0.69,
"learning_rate": 4.695326527610773e-05,
"loss": 0.145,
"step": 690
},
{
"epoch": 0.69,
"learning_rate": 4.66795567198309e-05,
"loss": 0.1396,
"step": 691
},
{
"epoch": 0.69,
"learning_rate": 4.640640517047154e-05,
"loss": 0.1475,
"step": 692
},
{
"epoch": 0.69,
"learning_rate": 4.6133813481481246e-05,
"loss": 0.1462,
"step": 693
},
{
"epoch": 0.69,
"learning_rate": 4.586178450046303e-05,
"loss": 0.1413,
"step": 694
},
{
"epoch": 0.69,
"learning_rate": 4.559032106914173e-05,
"loss": 0.1518,
"step": 695
},
{
"epoch": 0.69,
"learning_rate": 4.5319426023334155e-05,
"loss": 0.1481,
"step": 696
},
{
"epoch": 0.69,
"learning_rate": 4.50491021929194e-05,
"loss": 0.1484,
"step": 697
},
{
"epoch": 0.7,
"learning_rate": 4.477935240180966e-05,
"loss": 0.1487,
"step": 698
},
{
"epoch": 0.7,
"learning_rate": 4.451017946792032e-05,
"loss": 0.1447,
"step": 699
},
{
"epoch": 0.7,
"learning_rate": 4.424158620314073e-05,
"loss": 0.1407,
"step": 700
},
{
"epoch": 0.7,
"learning_rate": 4.397357541330476e-05,
"loss": 0.1536,
"step": 701
},
{
"epoch": 0.7,
"learning_rate": 4.370614989816154e-05,
"loss": 0.1611,
"step": 702
},
{
"epoch": 0.7,
"learning_rate": 4.343931245134616e-05,
"loss": 0.1463,
"step": 703
},
{
"epoch": 0.7,
"learning_rate": 4.317306586035048e-05,
"loss": 0.1571,
"step": 704
},
{
"epoch": 0.7,
"learning_rate": 4.2907412906494174e-05,
"loss": 0.1476,
"step": 705
},
{
"epoch": 0.7,
"learning_rate": 4.264235636489542e-05,
"loss": 0.1497,
"step": 706
},
{
"epoch": 0.7,
"learning_rate": 4.2377899004441966e-05,
"loss": 0.1489,
"step": 707
},
{
"epoch": 0.71,
"learning_rate": 4.2114043587762486e-05,
"loss": 0.1485,
"step": 708
},
{
"epoch": 0.71,
"learning_rate": 4.185079287119733e-05,
"loss": 0.1453,
"step": 709
},
{
"epoch": 0.71,
"learning_rate": 4.158814960476999e-05,
"loss": 0.1499,
"step": 710
},
{
"epoch": 0.71,
"learning_rate": 4.132611653215822e-05,
"loss": 0.1543,
"step": 711
},
{
"epoch": 0.71,
"learning_rate": 4.106469639066552e-05,
"loss": 0.1469,
"step": 712
},
{
"epoch": 0.71,
"learning_rate": 4.080389191119241e-05,
"loss": 0.1443,
"step": 713
},
{
"epoch": 0.71,
"learning_rate": 4.054370581820796e-05,
"loss": 0.1476,
"step": 714
},
{
"epoch": 0.71,
"learning_rate": 4.028414082972141e-05,
"loss": 0.1501,
"step": 715
},
{
"epoch": 0.71,
"learning_rate": 4.002519965725352e-05,
"loss": 0.1469,
"step": 716
},
{
"epoch": 0.71,
"learning_rate": 3.9766885005808565e-05,
"loss": 0.1457,
"step": 717
},
{
"epoch": 0.72,
"learning_rate": 3.950919957384582e-05,
"loss": 0.1502,
"step": 718
},
{
"epoch": 0.72,
"learning_rate": 3.9252146053251636e-05,
"loss": 0.1474,
"step": 719
},
{
"epoch": 0.72,
"learning_rate": 3.8995727129311035e-05,
"loss": 0.1447,
"step": 720
},
{
"epoch": 0.72,
"learning_rate": 3.873994548067972e-05,
"loss": 0.1554,
"step": 721
},
{
"epoch": 0.72,
"learning_rate": 3.8484803779356346e-05,
"loss": 0.1469,
"step": 722
},
{
"epoch": 0.72,
"learning_rate": 3.8230304690654304e-05,
"loss": 0.1465,
"step": 723
},
{
"epoch": 0.72,
"learning_rate": 3.7976450873174005e-05,
"loss": 0.1477,
"step": 724
},
{
"epoch": 0.72,
"learning_rate": 3.772324497877511e-05,
"loss": 0.1449,
"step": 725
},
{
"epoch": 0.72,
"learning_rate": 3.74706896525488e-05,
"loss": 0.1527,
"step": 726
},
{
"epoch": 0.72,
"learning_rate": 3.721878753279017e-05,
"loss": 0.1551,
"step": 727
},
{
"epoch": 0.73,
"learning_rate": 3.6967541250970604e-05,
"loss": 0.1564,
"step": 728
},
{
"epoch": 0.73,
"learning_rate": 3.671695343171052e-05,
"loss": 0.1413,
"step": 729
},
{
"epoch": 0.73,
"learning_rate": 3.646702669275151e-05,
"loss": 0.156,
"step": 730
},
{
"epoch": 0.73,
"learning_rate": 3.621776364492939e-05,
"loss": 0.1409,
"step": 731
},
{
"epoch": 0.73,
"learning_rate": 3.5969166892146845e-05,
"loss": 0.1389,
"step": 732
},
{
"epoch": 0.73,
"learning_rate": 3.5721239031346066e-05,
"loss": 0.1507,
"step": 733
},
{
"epoch": 0.73,
"learning_rate": 3.5473982652481775e-05,
"loss": 0.1452,
"step": 734
},
{
"epoch": 0.73,
"learning_rate": 3.522740033849411e-05,
"loss": 0.1425,
"step": 735
},
{
"epoch": 0.73,
"learning_rate": 3.4981494665281666e-05,
"loss": 0.1441,
"step": 736
},
{
"epoch": 0.73,
"learning_rate": 3.4736268201674574e-05,
"loss": 0.1412,
"step": 737
},
{
"epoch": 0.74,
"learning_rate": 3.44917235094076e-05,
"loss": 0.1418,
"step": 738
},
{
"epoch": 0.74,
"learning_rate": 3.424786314309365e-05,
"loss": 0.147,
"step": 739
},
{
"epoch": 0.74,
"learning_rate": 3.400468965019663e-05,
"loss": 0.1451,
"step": 740
},
{
"epoch": 0.74,
"learning_rate": 3.376220557100523e-05,
"loss": 0.1563,
"step": 741
},
{
"epoch": 0.74,
"learning_rate": 3.352041343860621e-05,
"loss": 0.152,
"step": 742
},
{
"epoch": 0.74,
"learning_rate": 3.3279315778858036e-05,
"loss": 0.1517,
"step": 743
},
{
"epoch": 0.74,
"learning_rate": 3.3038915110364375e-05,
"loss": 0.1492,
"step": 744
},
{
"epoch": 0.74,
"learning_rate": 3.279921394444776e-05,
"loss": 0.154,
"step": 745
},
{
"epoch": 0.74,
"learning_rate": 3.256021478512362e-05,
"loss": 0.1509,
"step": 746
},
{
"epoch": 0.74,
"learning_rate": 3.2321920129073816e-05,
"loss": 0.1523,
"step": 747
},
{
"epoch": 0.75,
"learning_rate": 3.2084332465620694e-05,
"loss": 0.1524,
"step": 748
},
{
"epoch": 0.75,
"learning_rate": 3.18474542767011e-05,
"loss": 0.1422,
"step": 749
},
{
"epoch": 0.75,
"learning_rate": 3.161128803684039e-05,
"loss": 0.1525,
"step": 750
},
{
"epoch": 0.75,
"learning_rate": 3.137583621312665e-05,
"loss": 0.1458,
"step": 751
},
{
"epoch": 0.75,
"learning_rate": 3.1141101265184835e-05,
"loss": 0.1444,
"step": 752
},
{
"epoch": 0.75,
"learning_rate": 3.090708564515124e-05,
"loss": 0.146,
"step": 753
},
{
"epoch": 0.75,
"learning_rate": 3.06737917976476e-05,
"loss": 0.1475,
"step": 754
},
{
"epoch": 0.75,
"learning_rate": 3.04412221597558e-05,
"loss": 0.1468,
"step": 755
},
{
"epoch": 0.75,
"learning_rate": 3.0209379160992413e-05,
"loss": 0.1498,
"step": 756
},
{
"epoch": 0.75,
"learning_rate": 2.997826522328315e-05,
"loss": 0.1631,
"step": 757
},
{
"epoch": 0.76,
"learning_rate": 2.9747882760937684e-05,
"loss": 0.1557,
"step": 758
},
{
"epoch": 0.76,
"learning_rate": 2.9518234180624393e-05,
"loss": 0.1527,
"step": 759
},
{
"epoch": 0.76,
"learning_rate": 2.9289321881345254e-05,
"loss": 0.15,
"step": 760
},
{
"epoch": 0.76,
"learning_rate": 2.906114825441072e-05,
"loss": 0.1517,
"step": 761
},
{
"epoch": 0.76,
"learning_rate": 2.8833715683414797e-05,
"loss": 0.1502,
"step": 762
},
{
"epoch": 0.76,
"learning_rate": 2.8607026544210114e-05,
"loss": 0.1404,
"step": 763
},
{
"epoch": 0.76,
"learning_rate": 2.8381083204883096e-05,
"loss": 0.1409,
"step": 764
},
{
"epoch": 0.76,
"learning_rate": 2.8155888025729273e-05,
"loss": 0.1553,
"step": 765
},
{
"epoch": 0.76,
"learning_rate": 2.793144335922854e-05,
"loss": 0.1505,
"step": 766
},
{
"epoch": 0.76,
"learning_rate": 2.770775155002071e-05,
"loss": 0.1524,
"step": 767
},
{
"epoch": 0.77,
"learning_rate": 2.748481493488092e-05,
"loss": 0.154,
"step": 768
},
{
"epoch": 0.77,
"learning_rate": 2.7262635842695127e-05,
"loss": 0.1536,
"step": 769
},
{
"epoch": 0.77,
"learning_rate": 2.7041216594436093e-05,
"loss": 0.1491,
"step": 770
},
{
"epoch": 0.77,
"learning_rate": 2.6820559503138797e-05,
"loss": 0.1483,
"step": 771
},
{
"epoch": 0.77,
"learning_rate": 2.660066687387648e-05,
"loss": 0.1522,
"step": 772
},
{
"epoch": 0.77,
"learning_rate": 2.6381541003736486e-05,
"loss": 0.1468,
"step": 773
},
{
"epoch": 0.77,
"learning_rate": 2.6163184181796286e-05,
"loss": 0.142,
"step": 774
},
{
"epoch": 0.77,
"learning_rate": 2.594559868909956e-05,
"loss": 0.1564,
"step": 775
},
{
"epoch": 0.77,
"learning_rate": 2.572878679863234e-05,
"loss": 0.1518,
"step": 776
},
{
"epoch": 0.77,
"learning_rate": 2.5512750775299432e-05,
"loss": 0.1513,
"step": 777
},
{
"epoch": 0.78,
"learning_rate": 2.529749287590042e-05,
"loss": 0.154,
"step": 778
},
{
"epoch": 0.78,
"learning_rate": 2.50830153491064e-05,
"loss": 0.146,
"step": 779
},
{
"epoch": 0.78,
"learning_rate": 2.4869320435436406e-05,
"loss": 0.1504,
"step": 780
},
{
"epoch": 0.78,
"learning_rate": 2.465641036723393e-05,
"loss": 0.1449,
"step": 781
},
{
"epoch": 0.78,
"learning_rate": 2.444428736864368e-05,
"loss": 0.1412,
"step": 782
},
{
"epoch": 0.78,
"learning_rate": 2.423295365558821e-05,
"loss": 0.1535,
"step": 783
},
{
"epoch": 0.78,
"learning_rate": 2.4022411435745074e-05,
"loss": 0.1444,
"step": 784
},
{
"epoch": 0.78,
"learning_rate": 2.381266290852341e-05,
"loss": 0.1502,
"step": 785
},
{
"epoch": 0.78,
"learning_rate": 2.360371026504118e-05,
"loss": 0.1447,
"step": 786
},
{
"epoch": 0.78,
"learning_rate": 2.339555568810221e-05,
"loss": 0.1599,
"step": 787
},
{
"epoch": 0.79,
"learning_rate": 2.318820135217341e-05,
"loss": 0.1512,
"step": 788
},
{
"epoch": 0.79,
"learning_rate": 2.298164942336205e-05,
"loss": 0.148,
"step": 789
},
{
"epoch": 0.79,
"learning_rate": 2.2775902059393085e-05,
"loss": 0.1467,
"step": 790
},
{
"epoch": 0.79,
"learning_rate": 2.2570961409586754e-05,
"loss": 0.1565,
"step": 791
},
{
"epoch": 0.79,
"learning_rate": 2.2366829614835992e-05,
"loss": 0.1468,
"step": 792
},
{
"epoch": 0.79,
"learning_rate": 2.2163508807583998e-05,
"loss": 0.1486,
"step": 793
},
{
"epoch": 0.79,
"learning_rate": 2.196100111180225e-05,
"loss": 0.1459,
"step": 794
},
{
"epoch": 0.79,
"learning_rate": 2.1759308642968025e-05,
"loss": 0.1463,
"step": 795
},
{
"epoch": 0.79,
"learning_rate": 2.155843350804243e-05,
"loss": 0.1503,
"step": 796
},
{
"epoch": 0.79,
"learning_rate": 2.1358377805448382e-05,
"loss": 0.15,
"step": 797
},
{
"epoch": 0.8,
"learning_rate": 2.115914362504865e-05,
"loss": 0.1539,
"step": 798
},
{
"epoch": 0.8,
"learning_rate": 2.0960733048124083e-05,
"loss": 0.1501,
"step": 799
},
{
"epoch": 0.8,
"learning_rate": 2.0763148147351785e-05,
"loss": 0.1496,
"step": 800
},
{
"epoch": 0.8,
"learning_rate": 2.0566390986783646e-05,
"loss": 0.1446,
"step": 801
},
{
"epoch": 0.8,
"learning_rate": 2.0370463621824444e-05,
"loss": 0.1542,
"step": 802
},
{
"epoch": 0.8,
"learning_rate": 2.01753680992107e-05,
"loss": 0.151,
"step": 803
},
{
"epoch": 0.8,
"learning_rate": 1.9981106456989207e-05,
"loss": 0.1412,
"step": 804
},
{
"epoch": 0.8,
"learning_rate": 1.9787680724495617e-05,
"loss": 0.1479,
"step": 805
},
{
"epoch": 0.8,
"learning_rate": 1.9595092922333414e-05,
"loss": 0.1558,
"step": 806
},
{
"epoch": 0.8,
"learning_rate": 1.9403345062352573e-05,
"loss": 0.1498,
"step": 807
},
{
"epoch": 0.81,
"learning_rate": 1.921243914762889e-05,
"loss": 0.1416,
"step": 808
},
{
"epoch": 0.81,
"learning_rate": 1.9022377172442752e-05,
"loss": 0.1435,
"step": 809
},
{
"epoch": 0.81,
"learning_rate": 1.8833161122258402e-05,
"loss": 0.1394,
"step": 810
},
{
"epoch": 0.81,
"learning_rate": 1.864479297370325e-05,
"loss": 0.1488,
"step": 811
},
{
"epoch": 0.81,
"learning_rate": 1.845727469454718e-05,
"loss": 0.1463,
"step": 812
},
{
"epoch": 0.81,
"learning_rate": 1.8270608243681953e-05,
"loss": 0.1525,
"step": 813
},
{
"epoch": 0.81,
"learning_rate": 1.808479557110081e-05,
"loss": 0.1495,
"step": 814
},
{
"epoch": 0.81,
"learning_rate": 1.7899838617878163e-05,
"loss": 0.1455,
"step": 815
},
{
"epoch": 0.81,
"learning_rate": 1.771573931614908e-05,
"loss": 0.1524,
"step": 816
},
{
"epoch": 0.81,
"learning_rate": 1.7532499589089323e-05,
"loss": 0.1468,
"step": 817
},
{
"epoch": 0.82,
"learning_rate": 1.7350121350895253e-05,
"loss": 0.1441,
"step": 818
},
{
"epoch": 0.82,
"learning_rate": 1.7168606506763695e-05,
"loss": 0.1501,
"step": 819
},
{
"epoch": 0.82,
"learning_rate": 1.698795695287212e-05,
"loss": 0.1419,
"step": 820
},
{
"epoch": 0.82,
"learning_rate": 1.6808174576358848e-05,
"loss": 0.1546,
"step": 821
},
{
"epoch": 0.82,
"learning_rate": 1.6629261255303286e-05,
"loss": 0.1511,
"step": 822
},
{
"epoch": 0.82,
"learning_rate": 1.6451218858706374e-05,
"loss": 0.1416,
"step": 823
},
{
"epoch": 0.82,
"learning_rate": 1.6274049246470956e-05,
"loss": 0.1481,
"step": 824
},
{
"epoch": 0.82,
"learning_rate": 1.6097754269382536e-05,
"loss": 0.143,
"step": 825
},
{
"epoch": 0.82,
"learning_rate": 1.59223357690897e-05,
"loss": 0.1588,
"step": 826
},
{
"epoch": 0.82,
"learning_rate": 1.5747795578085046e-05,
"loss": 0.1503,
"step": 827
},
{
"epoch": 0.83,
"learning_rate": 1.557413551968604e-05,
"loss": 0.1467,
"step": 828
},
{
"epoch": 0.83,
"learning_rate": 1.5401357408015893e-05,
"loss": 0.1471,
"step": 829
},
{
"epoch": 0.83,
"learning_rate": 1.522946304798465e-05,
"loss": 0.1608,
"step": 830
},
{
"epoch": 0.83,
"learning_rate": 1.505845423527027e-05,
"loss": 0.1489,
"step": 831
},
{
"epoch": 0.83,
"learning_rate": 1.4888332756300027e-05,
"loss": 0.1523,
"step": 832
},
{
"epoch": 0.83,
"learning_rate": 1.47191003882317e-05,
"loss": 0.1383,
"step": 833
},
{
"epoch": 0.83,
"learning_rate": 1.4550758898935047e-05,
"loss": 0.1401,
"step": 834
},
{
"epoch": 0.83,
"learning_rate": 1.4383310046973365e-05,
"loss": 0.1483,
"step": 835
},
{
"epoch": 0.83,
"learning_rate": 1.4216755581585072e-05,
"loss": 0.1473,
"step": 836
},
{
"epoch": 0.83,
"learning_rate": 1.40510972426655e-05,
"loss": 0.1515,
"step": 837
},
{
"epoch": 0.84,
"learning_rate": 1.388633676074862e-05,
"loss": 0.14,
"step": 838
},
{
"epoch": 0.84,
"learning_rate": 1.3722475856989158e-05,
"loss": 0.1452,
"step": 839
},
{
"epoch": 0.84,
"learning_rate": 1.3559516243144321e-05,
"loss": 0.1431,
"step": 840
},
{
"epoch": 0.84,
"learning_rate": 1.339745962155613e-05,
"loss": 0.1437,
"step": 841
},
{
"epoch": 0.84,
"learning_rate": 1.3236307685133675e-05,
"loss": 0.1491,
"step": 842
},
{
"epoch": 0.84,
"learning_rate": 1.307606211733522e-05,
"loss": 0.1522,
"step": 843
},
{
"epoch": 0.84,
"learning_rate": 1.2916724592150798e-05,
"loss": 0.1566,
"step": 844
},
{
"epoch": 0.84,
"learning_rate": 1.2758296774084632e-05,
"loss": 0.1512,
"step": 845
},
{
"epoch": 0.84,
"learning_rate": 1.2600780318137806e-05,
"loss": 0.1453,
"step": 846
},
{
"epoch": 0.84,
"learning_rate": 1.2444176869790925e-05,
"loss": 0.144,
"step": 847
},
{
"epoch": 0.85,
"learning_rate": 1.2288488064986948e-05,
"loss": 0.1535,
"step": 848
},
{
"epoch": 0.85,
"learning_rate": 1.213371553011411e-05,
"loss": 0.1446,
"step": 849
},
{
"epoch": 0.85,
"learning_rate": 1.1979860881988902e-05,
"loss": 0.1507,
"step": 850
},
{
"epoch": 0.85,
"learning_rate": 1.18269257278392e-05,
"loss": 0.1482,
"step": 851
},
{
"epoch": 0.85,
"learning_rate": 1.1674911665287513e-05,
"loss": 0.1547,
"step": 852
},
{
"epoch": 0.85,
"learning_rate": 1.1523820282334219e-05,
"loss": 0.1425,
"step": 853
},
{
"epoch": 0.85,
"learning_rate": 1.1373653157341036e-05,
"loss": 0.1489,
"step": 854
},
{
"epoch": 0.85,
"learning_rate": 1.1224411859014417e-05,
"loss": 0.1488,
"step": 855
},
{
"epoch": 0.85,
"learning_rate": 1.1076097946389396e-05,
"loss": 0.1485,
"step": 856
},
{
"epoch": 0.85,
"learning_rate": 1.092871296881307e-05,
"loss": 0.1444,
"step": 857
},
{
"epoch": 0.86,
"learning_rate": 1.078225846592854e-05,
"loss": 0.1406,
"step": 858
},
{
"epoch": 0.86,
"learning_rate": 1.0636735967658784e-05,
"loss": 0.1517,
"step": 859
},
{
"epoch": 0.86,
"learning_rate": 1.0492146994190666e-05,
"loss": 0.1444,
"step": 860
},
{
"epoch": 0.86,
"learning_rate": 1.0348493055959062e-05,
"loss": 0.1489,
"step": 861
},
{
"epoch": 0.86,
"learning_rate": 1.0205775653631178e-05,
"loss": 0.1573,
"step": 862
},
{
"epoch": 0.86,
"learning_rate": 1.0063996278090704e-05,
"loss": 0.1425,
"step": 863
},
{
"epoch": 0.86,
"learning_rate": 9.923156410422318e-06,
"loss": 0.1391,
"step": 864
},
{
"epoch": 0.86,
"learning_rate": 9.783257521896227e-06,
"loss": 0.1402,
"step": 865
},
{
"epoch": 0.86,
"learning_rate": 9.644301073952866e-06,
"loss": 0.1475,
"step": 866
},
{
"epoch": 0.86,
"learning_rate": 9.506288518187467e-06,
"loss": 0.14,
"step": 867
},
{
"epoch": 0.87,
"learning_rate": 9.369221296335006e-06,
"loss": 0.1547,
"step": 868
},
{
"epoch": 0.87,
"learning_rate": 9.233100840255127e-06,
"loss": 0.1544,
"step": 869
},
{
"epoch": 0.87,
"learning_rate": 9.097928571917181e-06,
"loss": 0.1459,
"step": 870
},
{
"epoch": 0.87,
"learning_rate": 8.963705903385345e-06,
"loss": 0.149,
"step": 871
},
{
"epoch": 0.87,
"learning_rate": 8.830434236803909e-06,
"loss": 0.1445,
"step": 872
},
{
"epoch": 0.87,
"learning_rate": 8.698114964382598e-06,
"loss": 0.1443,
"step": 873
},
{
"epoch": 0.87,
"learning_rate": 8.566749468382074e-06,
"loss": 0.1502,
"step": 874
},
{
"epoch": 0.87,
"learning_rate": 8.436339121099412e-06,
"loss": 0.1427,
"step": 875
},
{
"epoch": 0.87,
"learning_rate": 8.30688528485387e-06,
"loss": 0.1506,
"step": 876
},
{
"epoch": 0.87,
"learning_rate": 8.178389311972612e-06,
"loss": 0.1498,
"step": 877
},
{
"epoch": 0.88,
"learning_rate": 8.050852544776489e-06,
"loss": 0.1386,
"step": 878
},
{
"epoch": 0.88,
"learning_rate": 7.92427631556617e-06,
"loss": 0.1413,
"step": 879
},
{
"epoch": 0.88,
"learning_rate": 7.798661946608166e-06,
"loss": 0.1555,
"step": 880
},
{
"epoch": 0.88,
"learning_rate": 7.674010750120964e-06,
"loss": 0.1508,
"step": 881
},
{
"epoch": 0.88,
"learning_rate": 7.550324028261391e-06,
"loss": 0.1514,
"step": 882
},
{
"epoch": 0.88,
"learning_rate": 7.427603073110967e-06,
"loss": 0.1485,
"step": 883
},
{
"epoch": 0.88,
"learning_rate": 7.305849166662427e-06,
"loss": 0.1509,
"step": 884
},
{
"epoch": 0.88,
"learning_rate": 7.185063580806317e-06,
"loss": 0.1499,
"step": 885
},
{
"epoch": 0.88,
"learning_rate": 7.0652475773177464e-06,
"loss": 0.1492,
"step": 886
},
{
"epoch": 0.88,
"learning_rate": 6.946402407843155e-06,
"loss": 0.1466,
"step": 887
},
{
"epoch": 0.89,
"learning_rate": 6.828529313887211e-06,
"loss": 0.1425,
"step": 888
},
{
"epoch": 0.89,
"learning_rate": 6.7116295267999455e-06,
"loss": 0.1384,
"step": 889
},
{
"epoch": 0.89,
"learning_rate": 6.595704267763847e-06,
"loss": 0.1482,
"step": 890
},
{
"epoch": 0.89,
"learning_rate": 6.480754747781037e-06,
"loss": 0.1417,
"step": 891
},
{
"epoch": 0.89,
"learning_rate": 6.36678216766069e-06,
"loss": 0.1486,
"step": 892
},
{
"epoch": 0.89,
"learning_rate": 6.253787718006498e-06,
"loss": 0.1549,
"step": 893
},
{
"epoch": 0.89,
"learning_rate": 6.141772579204175e-06,
"loss": 0.1505,
"step": 894
},
{
"epoch": 0.89,
"learning_rate": 6.030737921409169e-06,
"loss": 0.1537,
"step": 895
},
{
"epoch": 0.89,
"learning_rate": 5.920684904534424e-06,
"loss": 0.1563,
"step": 896
},
{
"epoch": 0.89,
"learning_rate": 5.811614678238275e-06,
"loss": 0.1424,
"step": 897
},
{
"epoch": 0.9,
"learning_rate": 5.7035283819124155e-06,
"loss": 0.1563,
"step": 898
},
{
"epoch": 0.9,
"learning_rate": 5.596427144670002e-06,
"loss": 0.1443,
"step": 899
},
{
"epoch": 0.9,
"learning_rate": 5.490312085333904e-06,
"loss": 0.1478,
"step": 900
},
{
"epoch": 0.9,
"learning_rate": 5.385184312424974e-06,
"loss": 0.1556,
"step": 901
},
{
"epoch": 0.9,
"learning_rate": 5.281044924150391e-06,
"loss": 0.1469,
"step": 902
},
{
"epoch": 0.9,
"learning_rate": 5.177895008392353e-06,
"loss": 0.1505,
"step": 903
},
{
"epoch": 0.9,
"learning_rate": 5.075735642696611e-06,
"loss": 0.1488,
"step": 904
},
{
"epoch": 0.9,
"learning_rate": 4.974567894261217e-06,
"loss": 0.1516,
"step": 905
},
{
"epoch": 0.9,
"learning_rate": 4.874392819925388e-06,
"loss": 0.1416,
"step": 906
},
{
"epoch": 0.9,
"learning_rate": 4.775211466158469e-06,
"loss": 0.1462,
"step": 907
},
{
"epoch": 0.91,
"learning_rate": 4.677024869049007e-06,
"loss": 0.157,
"step": 908
},
{
"epoch": 0.91,
"learning_rate": 4.5798340542938855e-06,
"loss": 0.15,
"step": 909
},
{
"epoch": 0.91,
"learning_rate": 4.4836400371876974e-06,
"loss": 0.1398,
"step": 910
},
{
"epoch": 0.91,
"learning_rate": 4.3884438226120424e-06,
"loss": 0.1518,
"step": 911
},
{
"epoch": 0.91,
"learning_rate": 4.2942464050250705e-06,
"loss": 0.146,
"step": 912
},
{
"epoch": 0.91,
"learning_rate": 4.20104876845111e-06,
"loss": 0.1481,
"step": 913
},
{
"epoch": 0.91,
"learning_rate": 4.108851886470389e-06,
"loss": 0.1511,
"step": 914
},
{
"epoch": 0.91,
"learning_rate": 4.017656722208807e-06,
"loss": 0.1528,
"step": 915
},
{
"epoch": 0.91,
"learning_rate": 3.927464228327949e-06,
"loss": 0.1398,
"step": 916
},
{
"epoch": 0.91,
"learning_rate": 3.8382753470150854e-06,
"loss": 0.1418,
"step": 917
},
{
"epoch": 0.92,
"learning_rate": 3.7500910099733444e-06,
"loss": 0.1556,
"step": 918
},
{
"epoch": 0.92,
"learning_rate": 3.662912138411967e-06,
"loss": 0.1479,
"step": 919
},
{
"epoch": 0.92,
"learning_rate": 3.5767396430367306e-06,
"loss": 0.1486,
"step": 920
},
{
"epoch": 0.92,
"learning_rate": 3.4915744240403558e-06,
"loss": 0.1493,
"step": 921
},
{
"epoch": 0.92,
"learning_rate": 3.40741737109318e-06,
"loss": 0.1504,
"step": 922
},
{
"epoch": 0.92,
"learning_rate": 3.3242693633337983e-06,
"loss": 0.1531,
"step": 923
},
{
"epoch": 0.92,
"learning_rate": 3.2421312693599603e-06,
"loss": 0.1507,
"step": 924
},
{
"epoch": 0.92,
"learning_rate": 3.161003947219421e-06,
"loss": 0.1485,
"step": 925
},
{
"epoch": 0.92,
"learning_rate": 3.080888244400959e-06,
"loss": 0.1437,
"step": 926
},
{
"epoch": 0.92,
"learning_rate": 3.0017849978256516e-06,
"loss": 0.1447,
"step": 927
},
{
"epoch": 0.93,
"learning_rate": 2.9236950338380033e-06,
"loss": 0.1458,
"step": 928
},
{
"epoch": 0.93,
"learning_rate": 2.8466191681973507e-06,
"loss": 0.1507,
"step": 929
},
{
"epoch": 0.93,
"learning_rate": 2.7705582060693623e-06,
"loss": 0.1406,
"step": 930
},
{
"epoch": 0.93,
"learning_rate": 2.6955129420176196e-06,
"loss": 0.153,
"step": 931
},
{
"epoch": 0.93,
"learning_rate": 2.6214841599953023e-06,
"loss": 0.1513,
"step": 932
},
{
"epoch": 0.93,
"learning_rate": 2.548472633337007e-06,
"loss": 0.1489,
"step": 933
},
{
"epoch": 0.93,
"learning_rate": 2.476479124750697e-06,
"loss": 0.1408,
"step": 934
},
{
"epoch": 0.93,
"learning_rate": 2.4055043863096428e-06,
"loss": 0.146,
"step": 935
},
{
"epoch": 0.93,
"learning_rate": 2.3355491594447055e-06,
"loss": 0.1507,
"step": 936
},
{
"epoch": 0.93,
"learning_rate": 2.266614174936443e-06,
"loss": 0.1501,
"step": 937
},
{
"epoch": 0.94,
"learning_rate": 2.198700152907629e-06,
"loss": 0.1535,
"step": 938
},
{
"epoch": 0.94,
"learning_rate": 2.1318078028155888e-06,
"loss": 0.1439,
"step": 939
},
{
"epoch": 0.94,
"learning_rate": 2.0659378234448525e-06,
"loss": 0.1509,
"step": 940
},
{
"epoch": 0.94,
"learning_rate": 2.0010909028998827e-06,
"loss": 0.1416,
"step": 941
},
{
"epoch": 0.94,
"learning_rate": 1.937267718597846e-06,
"loss": 0.1559,
"step": 942
},
{
"epoch": 0.94,
"learning_rate": 1.874468937261531e-06,
"loss": 0.1449,
"step": 943
},
{
"epoch": 0.94,
"learning_rate": 1.8126952149124188e-06,
"loss": 0.1534,
"step": 944
},
{
"epoch": 0.94,
"learning_rate": 1.751947196863779e-06,
"loss": 0.1543,
"step": 945
},
{
"epoch": 0.94,
"learning_rate": 1.692225517714008e-06,
"loss": 0.1501,
"step": 946
},
{
"epoch": 0.94,
"learning_rate": 1.6335308013398886e-06,
"loss": 0.1448,
"step": 947
},
{
"epoch": 0.95,
"learning_rate": 1.5758636608901979e-06,
"loss": 0.1408,
"step": 948
},
{
"epoch": 0.95,
"learning_rate": 1.5192246987791981e-06,
"loss": 0.1539,
"step": 949
},
{
"epoch": 0.95,
"learning_rate": 1.4636145066803885e-06,
"loss": 0.1539,
"step": 950
},
{
"epoch": 0.95,
"learning_rate": 1.409033665520354e-06,
"loss": 0.1443,
"step": 951
},
{
"epoch": 0.95,
"learning_rate": 1.3554827454726139e-06,
"loss": 0.1466,
"step": 952
},
{
"epoch": 0.95,
"learning_rate": 1.3029623059517493e-06,
"loss": 0.15,
"step": 953
},
{
"epoch": 0.95,
"learning_rate": 1.2514728956075084e-06,
"loss": 0.1396,
"step": 954
},
{
"epoch": 0.95,
"learning_rate": 1.201015052319099e-06,
"loss": 0.1511,
"step": 955
},
{
"epoch": 0.95,
"learning_rate": 1.1515893031895818e-06,
"loss": 0.1448,
"step": 956
},
{
"epoch": 0.95,
"learning_rate": 1.1031961645402877e-06,
"loss": 0.1449,
"step": 957
},
{
"epoch": 0.96,
"learning_rate": 1.055836141905553e-06,
"loss": 0.1438,
"step": 958
},
{
"epoch": 0.96,
"learning_rate": 1.0095097300273026e-06,
"loss": 0.1494,
"step": 959
},
{
"epoch": 0.96,
"learning_rate": 9.642174128499881e-07,
"loss": 0.1503,
"step": 960
},
{
"epoch": 0.96,
"learning_rate": 9.199596635154683e-07,
"loss": 0.1485,
"step": 961
},
{
"epoch": 0.96,
"learning_rate": 8.767369443581031e-07,
"loss": 0.1494,
"step": 962
},
{
"epoch": 0.96,
"learning_rate": 8.345497068998897e-07,
"loss": 0.1532,
"step": 963
},
{
"epoch": 0.96,
"learning_rate": 7.933983918457677e-07,
"loss": 0.1473,
"step": 964
},
{
"epoch": 0.96,
"learning_rate": 7.532834290790436e-07,
"loss": 0.1454,
"step": 965
},
{
"epoch": 0.96,
"learning_rate": 7.142052376568175e-07,
"loss": 0.1433,
"step": 966
},
{
"epoch": 0.96,
"learning_rate": 6.761642258056978e-07,
"loss": 0.1439,
"step": 967
},
{
"epoch": 0.97,
"learning_rate": 6.391607909174812e-07,
"loss": 0.1531,
"step": 968
},
{
"epoch": 0.97,
"learning_rate": 6.031953195450135e-07,
"loss": 0.1461,
"step": 969
},
{
"epoch": 0.97,
"learning_rate": 5.682681873981577e-07,
"loss": 0.1585,
"step": 970
},
{
"epoch": 0.97,
"learning_rate": 5.343797593398536e-07,
"loss": 0.1505,
"step": 971
},
{
"epoch": 0.97,
"learning_rate": 5.01530389382343e-07,
"loss": 0.1406,
"step": 972
},
{
"epoch": 0.97,
"learning_rate": 4.6972042068341714e-07,
"loss": 0.1428,
"step": 973
},
{
"epoch": 0.97,
"learning_rate": 4.389501855428857e-07,
"loss": 0.1501,
"step": 974
},
{
"epoch": 0.97,
"learning_rate": 4.092200053990691e-07,
"loss": 0.1431,
"step": 975
},
{
"epoch": 0.97,
"learning_rate": 3.805301908254455e-07,
"loss": 0.15,
"step": 976
},
{
"epoch": 0.97,
"learning_rate": 3.5288104152743083e-07,
"loss": 0.1437,
"step": 977
},
{
"epoch": 0.98,
"learning_rate": 3.262728463392262e-07,
"loss": 0.15,
"step": 978
},
{
"epoch": 0.98,
"learning_rate": 3.007058832207976e-07,
"loss": 0.1552,
"step": 979
},
{
"epoch": 0.98,
"learning_rate": 2.7618041925496773e-07,
"loss": 0.1581,
"step": 980
},
{
"epoch": 0.98,
"learning_rate": 2.5269671064467313e-07,
"loss": 0.1496,
"step": 981
},
{
"epoch": 0.98,
"learning_rate": 2.3025500271023347e-07,
"loss": 0.1487,
"step": 982
},
{
"epoch": 0.98,
"learning_rate": 2.088555298867978e-07,
"loss": 0.154,
"step": 983
},
{
"epoch": 0.98,
"learning_rate": 1.8849851572191325e-07,
"loss": 0.1494,
"step": 984
},
{
"epoch": 0.98,
"learning_rate": 1.6918417287318245e-07,
"loss": 0.1442,
"step": 985
},
{
"epoch": 0.98,
"learning_rate": 1.509127031060431e-07,
"loss": 0.1495,
"step": 986
},
{
"epoch": 0.98,
"learning_rate": 1.3368429729168076e-07,
"loss": 0.1453,
"step": 987
},
{
"epoch": 0.99,
"learning_rate": 1.1749913540496371e-07,
"loss": 0.1512,
"step": 988
},
{
"epoch": 0.99,
"learning_rate": 1.023573865226446e-07,
"loss": 0.1422,
"step": 989
},
{
"epoch": 0.99,
"learning_rate": 8.825920882157279e-08,
"loss": 0.1552,
"step": 990
},
{
"epoch": 0.99,
"learning_rate": 7.520474957699586e-08,
"loss": 0.1466,
"step": 991
},
{
"epoch": 0.99,
"learning_rate": 6.31941451610829e-08,
"loss": 0.1509,
"step": 992
},
{
"epoch": 0.99,
"learning_rate": 5.2227521041470216e-08,
"loss": 0.1438,
"step": 993
},
{
"epoch": 0.99,
"learning_rate": 4.230499177994007e-08,
"loss": 0.1527,
"step": 994
},
{
"epoch": 0.99,
"learning_rate": 3.3426661031255026e-08,
"loss": 0.1427,
"step": 995
},
{
"epoch": 0.99,
"learning_rate": 2.5592621542047668e-08,
"loss": 0.1426,
"step": 996
},
{
"epoch": 0.99,
"learning_rate": 1.8802955149865852e-08,
"loss": 0.1545,
"step": 997
},
{
"epoch": 1.0,
"learning_rate": 1.3057732782295606e-08,
"loss": 0.1507,
"step": 998
},
{
"epoch": 1.0,
"learning_rate": 8.357014456272794e-09,
"loss": 0.1472,
"step": 999
},
{
"epoch": 1.0,
"learning_rate": 4.700849277383679e-09,
"loss": 0.1467,
"step": 1000
},
{
"epoch": 1.0,
"learning_rate": 2.0892754394208346e-09,
"loss": 0.1552,
"step": 1001
},
{
"epoch": 1.0,
"learning_rate": 5.223202239501568e-10,
"loss": 0.1489,
"step": 1002
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.1462,
"step": 1003
},
{
"epoch": 1.0,
"step": 1003,
"total_flos": 1949480219836416.0,
"train_loss": 0.16290030455363474,
"train_runtime": 38379.7661,
"train_samples_per_second": 1.045,
"train_steps_per_second": 0.026
}
],
"logging_steps": 1.0,
"max_steps": 1003,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 50000,
"total_flos": 1949480219836416.0,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}