Qilin-Med-VL / trainer_state.json
williamliu's picture
Upload Qilin-Med-VL model weights
1272da4
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9645628014258755,
"global_step": 1150,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.555555555555555e-07,
"loss": 2.258,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.111111111111111e-06,
"loss": 2.2773,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 1.6666666666666667e-06,
"loss": 2.2713,
"step": 3
},
{
"epoch": 0.0,
"learning_rate": 2.222222222222222e-06,
"loss": 2.2116,
"step": 4
},
{
"epoch": 0.0,
"learning_rate": 2.7777777777777783e-06,
"loss": 2.0276,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 3.3333333333333333e-06,
"loss": 1.71,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 3.88888888888889e-06,
"loss": 1.5804,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 4.444444444444444e-06,
"loss": 1.4892,
"step": 8
},
{
"epoch": 0.01,
"learning_rate": 5e-06,
"loss": 1.5167,
"step": 9
},
{
"epoch": 0.01,
"learning_rate": 5.555555555555557e-06,
"loss": 1.5724,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 6.111111111111112e-06,
"loss": 1.4348,
"step": 11
},
{
"epoch": 0.01,
"learning_rate": 6.666666666666667e-06,
"loss": 1.5742,
"step": 12
},
{
"epoch": 0.01,
"learning_rate": 7.222222222222223e-06,
"loss": 1.5096,
"step": 13
},
{
"epoch": 0.01,
"learning_rate": 7.77777777777778e-06,
"loss": 1.4846,
"step": 14
},
{
"epoch": 0.01,
"learning_rate": 8.333333333333334e-06,
"loss": 1.5117,
"step": 15
},
{
"epoch": 0.01,
"learning_rate": 8.888888888888888e-06,
"loss": 1.5584,
"step": 16
},
{
"epoch": 0.01,
"learning_rate": 9.444444444444445e-06,
"loss": 1.5195,
"step": 17
},
{
"epoch": 0.02,
"learning_rate": 1e-05,
"loss": 1.6049,
"step": 18
},
{
"epoch": 0.02,
"learning_rate": 1.0555555555555557e-05,
"loss": 1.3504,
"step": 19
},
{
"epoch": 0.02,
"learning_rate": 1.1111111111111113e-05,
"loss": 1.5413,
"step": 20
},
{
"epoch": 0.02,
"learning_rate": 1.1666666666666668e-05,
"loss": 1.4658,
"step": 21
},
{
"epoch": 0.02,
"learning_rate": 1.2222222222222224e-05,
"loss": 1.4348,
"step": 22
},
{
"epoch": 0.02,
"learning_rate": 1.2777777777777777e-05,
"loss": 1.5483,
"step": 23
},
{
"epoch": 0.02,
"learning_rate": 1.3333333333333333e-05,
"loss": 1.3411,
"step": 24
},
{
"epoch": 0.02,
"learning_rate": 1.388888888888889e-05,
"loss": 1.4523,
"step": 25
},
{
"epoch": 0.02,
"learning_rate": 1.4444444444444446e-05,
"loss": 1.3985,
"step": 26
},
{
"epoch": 0.02,
"learning_rate": 1.5000000000000002e-05,
"loss": 1.3115,
"step": 27
},
{
"epoch": 0.02,
"learning_rate": 1.555555555555556e-05,
"loss": 1.3854,
"step": 28
},
{
"epoch": 0.02,
"learning_rate": 1.6111111111111115e-05,
"loss": 1.4561,
"step": 29
},
{
"epoch": 0.03,
"learning_rate": 1.6666666666666667e-05,
"loss": 1.3643,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 1.7222222222222224e-05,
"loss": 1.4077,
"step": 31
},
{
"epoch": 0.03,
"learning_rate": 1.7777777777777777e-05,
"loss": 1.5276,
"step": 32
},
{
"epoch": 0.03,
"learning_rate": 1.8333333333333333e-05,
"loss": 1.4767,
"step": 33
},
{
"epoch": 0.03,
"learning_rate": 1.888888888888889e-05,
"loss": 1.4163,
"step": 34
},
{
"epoch": 0.03,
"learning_rate": 1.9444444444444445e-05,
"loss": 1.3612,
"step": 35
},
{
"epoch": 0.03,
"learning_rate": 2e-05,
"loss": 1.4327,
"step": 36
},
{
"epoch": 0.03,
"learning_rate": 1.9999963072167756e-05,
"loss": 1.3824,
"step": 37
},
{
"epoch": 0.03,
"learning_rate": 1.9999852288943748e-05,
"loss": 1.335,
"step": 38
},
{
"epoch": 0.03,
"learning_rate": 1.9999667651146174e-05,
"loss": 1.3785,
"step": 39
},
{
"epoch": 0.03,
"learning_rate": 1.9999409160138695e-05,
"loss": 1.4464,
"step": 40
},
{
"epoch": 0.03,
"learning_rate": 1.9999076817830406e-05,
"loss": 1.2467,
"step": 41
},
{
"epoch": 0.04,
"learning_rate": 1.9998670626675845e-05,
"loss": 1.3647,
"step": 42
},
{
"epoch": 0.04,
"learning_rate": 1.999819058967497e-05,
"loss": 1.3731,
"step": 43
},
{
"epoch": 0.04,
"learning_rate": 1.9997636710373117e-05,
"loss": 1.367,
"step": 44
},
{
"epoch": 0.04,
"learning_rate": 1.9997008992861006e-05,
"loss": 1.4122,
"step": 45
},
{
"epoch": 0.04,
"learning_rate": 1.9996307441774685e-05,
"loss": 1.1763,
"step": 46
},
{
"epoch": 0.04,
"learning_rate": 1.99955320622955e-05,
"loss": 1.3545,
"step": 47
},
{
"epoch": 0.04,
"learning_rate": 1.9994682860150073e-05,
"loss": 1.3445,
"step": 48
},
{
"epoch": 0.04,
"learning_rate": 1.9993759841610242e-05,
"loss": 1.3004,
"step": 49
},
{
"epoch": 0.04,
"learning_rate": 1.9992763013493023e-05,
"loss": 1.0944,
"step": 50
},
{
"epoch": 0.04,
"learning_rate": 1.9991692383160553e-05,
"loss": 1.3616,
"step": 51
},
{
"epoch": 0.04,
"learning_rate": 1.9990547958520047e-05,
"loss": 1.2798,
"step": 52
},
{
"epoch": 0.04,
"learning_rate": 1.9989329748023728e-05,
"loss": 1.4324,
"step": 53
},
{
"epoch": 0.05,
"learning_rate": 1.9988037760668767e-05,
"loss": 1.324,
"step": 54
},
{
"epoch": 0.05,
"learning_rate": 1.9986672005997227e-05,
"loss": 1.2707,
"step": 55
},
{
"epoch": 0.05,
"learning_rate": 1.9985232494095982e-05,
"loss": 1.2232,
"step": 56
},
{
"epoch": 0.05,
"learning_rate": 1.9983719235596632e-05,
"loss": 1.3018,
"step": 57
},
{
"epoch": 0.05,
"learning_rate": 1.998213224167546e-05,
"loss": 1.3571,
"step": 58
},
{
"epoch": 0.05,
"learning_rate": 1.9980471524053315e-05,
"loss": 1.2873,
"step": 59
},
{
"epoch": 0.05,
"learning_rate": 1.9978737094995525e-05,
"loss": 1.3495,
"step": 60
},
{
"epoch": 0.05,
"learning_rate": 1.9976928967311844e-05,
"loss": 1.3412,
"step": 61
},
{
"epoch": 0.05,
"learning_rate": 1.997504715435631e-05,
"loss": 1.379,
"step": 62
},
{
"epoch": 0.05,
"learning_rate": 1.9973091670027184e-05,
"loss": 1.3084,
"step": 63
},
{
"epoch": 0.05,
"learning_rate": 1.9971062528766825e-05,
"loss": 1.2442,
"step": 64
},
{
"epoch": 0.05,
"learning_rate": 1.9968959745561588e-05,
"loss": 1.3888,
"step": 65
},
{
"epoch": 0.06,
"learning_rate": 1.9966783335941717e-05,
"loss": 1.3991,
"step": 66
},
{
"epoch": 0.06,
"learning_rate": 1.9964533315981235e-05,
"loss": 1.3098,
"step": 67
},
{
"epoch": 0.06,
"learning_rate": 1.9962209702297807e-05,
"loss": 1.2983,
"step": 68
},
{
"epoch": 0.06,
"learning_rate": 1.9959812512052644e-05,
"loss": 1.1825,
"step": 69
},
{
"epoch": 0.06,
"learning_rate": 1.9957341762950346e-05,
"loss": 1.2862,
"step": 70
},
{
"epoch": 0.06,
"learning_rate": 1.99547974732388e-05,
"loss": 1.2423,
"step": 71
},
{
"epoch": 0.06,
"learning_rate": 1.9952179661709028e-05,
"loss": 1.2596,
"step": 72
},
{
"epoch": 0.06,
"learning_rate": 1.9949488347695044e-05,
"loss": 1.2378,
"step": 73
},
{
"epoch": 0.06,
"learning_rate": 1.9946723551073733e-05,
"loss": 1.2155,
"step": 74
},
{
"epoch": 0.06,
"learning_rate": 1.994388529226468e-05,
"loss": 1.4262,
"step": 75
},
{
"epoch": 0.06,
"learning_rate": 1.9940973592230038e-05,
"loss": 1.3928,
"step": 76
},
{
"epoch": 0.06,
"learning_rate": 1.993798847247436e-05,
"loss": 1.2511,
"step": 77
},
{
"epoch": 0.07,
"learning_rate": 1.993492995504444e-05,
"loss": 1.3199,
"step": 78
},
{
"epoch": 0.07,
"learning_rate": 1.9931798062529174e-05,
"loss": 1.1963,
"step": 79
},
{
"epoch": 0.07,
"learning_rate": 1.992859281805935e-05,
"loss": 1.2605,
"step": 80
},
{
"epoch": 0.07,
"learning_rate": 1.9925314245307523e-05,
"loss": 1.09,
"step": 81
},
{
"epoch": 0.07,
"learning_rate": 1.9921962368487805e-05,
"loss": 1.19,
"step": 82
},
{
"epoch": 0.07,
"learning_rate": 1.9918537212355704e-05,
"loss": 1.2609,
"step": 83
},
{
"epoch": 0.07,
"learning_rate": 1.9915038802207946e-05,
"loss": 1.4035,
"step": 84
},
{
"epoch": 0.07,
"learning_rate": 1.9911467163882258e-05,
"loss": 1.2619,
"step": 85
},
{
"epoch": 0.07,
"learning_rate": 1.9907822323757225e-05,
"loss": 1.4575,
"step": 86
},
{
"epoch": 0.07,
"learning_rate": 1.9904104308752053e-05,
"loss": 1.3853,
"step": 87
},
{
"epoch": 0.07,
"learning_rate": 1.9900313146326384e-05,
"loss": 1.2737,
"step": 88
},
{
"epoch": 0.07,
"learning_rate": 1.9896448864480104e-05,
"loss": 1.2943,
"step": 89
},
{
"epoch": 0.08,
"learning_rate": 1.9892511491753126e-05,
"loss": 1.4215,
"step": 90
},
{
"epoch": 0.08,
"learning_rate": 1.988850105722517e-05,
"loss": 1.2784,
"step": 91
},
{
"epoch": 0.08,
"learning_rate": 1.9884417590515574e-05,
"loss": 1.3207,
"step": 92
},
{
"epoch": 0.08,
"learning_rate": 1.9880261121783046e-05,
"loss": 1.3498,
"step": 93
},
{
"epoch": 0.08,
"learning_rate": 1.9876031681725468e-05,
"loss": 1.2851,
"step": 94
},
{
"epoch": 0.08,
"learning_rate": 1.9871729301579648e-05,
"loss": 1.2137,
"step": 95
},
{
"epoch": 0.08,
"learning_rate": 1.98673540131211e-05,
"loss": 1.177,
"step": 96
},
{
"epoch": 0.08,
"learning_rate": 1.986290584866381e-05,
"loss": 1.1539,
"step": 97
},
{
"epoch": 0.08,
"learning_rate": 1.985838484105999e-05,
"loss": 1.1612,
"step": 98
},
{
"epoch": 0.08,
"learning_rate": 1.985379102369984e-05,
"loss": 1.2737,
"step": 99
},
{
"epoch": 0.08,
"learning_rate": 1.984912443051131e-05,
"loss": 1.3064,
"step": 100
},
{
"epoch": 0.08,
"learning_rate": 1.9844385095959832e-05,
"loss": 1.2523,
"step": 101
},
{
"epoch": 0.09,
"learning_rate": 1.983957305504807e-05,
"loss": 1.4239,
"step": 102
},
{
"epoch": 0.09,
"learning_rate": 1.983468834331568e-05,
"loss": 1.3386,
"step": 103
},
{
"epoch": 0.09,
"learning_rate": 1.982973099683902e-05,
"loss": 1.3065,
"step": 104
},
{
"epoch": 0.09,
"learning_rate": 1.98247010522309e-05,
"loss": 1.2748,
"step": 105
},
{
"epoch": 0.09,
"learning_rate": 1.981959854664032e-05,
"loss": 1.2802,
"step": 106
},
{
"epoch": 0.09,
"learning_rate": 1.9814423517752163e-05,
"loss": 1.1896,
"step": 107
},
{
"epoch": 0.09,
"learning_rate": 1.9809176003786953e-05,
"loss": 1.2393,
"step": 108
},
{
"epoch": 0.09,
"learning_rate": 1.9803856043500554e-05,
"loss": 1.1552,
"step": 109
},
{
"epoch": 0.09,
"learning_rate": 1.9798463676183887e-05,
"loss": 1.1761,
"step": 110
},
{
"epoch": 0.09,
"learning_rate": 1.9792998941662638e-05,
"loss": 1.1529,
"step": 111
},
{
"epoch": 0.09,
"learning_rate": 1.9787461880296964e-05,
"loss": 1.1921,
"step": 112
},
{
"epoch": 0.09,
"learning_rate": 1.9781852532981204e-05,
"loss": 1.2738,
"step": 113
},
{
"epoch": 0.1,
"learning_rate": 1.9776170941143563e-05,
"loss": 1.1381,
"step": 114
},
{
"epoch": 0.1,
"learning_rate": 1.9770417146745818e-05,
"loss": 1.2991,
"step": 115
},
{
"epoch": 0.1,
"learning_rate": 1.9764591192282994e-05,
"loss": 1.322,
"step": 116
},
{
"epoch": 0.1,
"learning_rate": 1.975869312078307e-05,
"loss": 1.3013,
"step": 117
},
{
"epoch": 0.1,
"learning_rate": 1.9752722975806643e-05,
"loss": 1.3464,
"step": 118
},
{
"epoch": 0.1,
"learning_rate": 1.9746680801446617e-05,
"loss": 1.223,
"step": 119
},
{
"epoch": 0.1,
"learning_rate": 1.9740566642327868e-05,
"loss": 1.2459,
"step": 120
},
{
"epoch": 0.1,
"learning_rate": 1.9734380543606932e-05,
"loss": 1.1452,
"step": 121
},
{
"epoch": 0.1,
"learning_rate": 1.972812255097164e-05,
"loss": 1.2957,
"step": 122
},
{
"epoch": 0.1,
"learning_rate": 1.972179271064083e-05,
"loss": 1.0999,
"step": 123
},
{
"epoch": 0.1,
"learning_rate": 1.971539106936394e-05,
"loss": 1.2046,
"step": 124
},
{
"epoch": 0.1,
"learning_rate": 1.970891767442073e-05,
"loss": 1.3477,
"step": 125
},
{
"epoch": 0.11,
"learning_rate": 1.9702372573620883e-05,
"loss": 1.1943,
"step": 126
},
{
"epoch": 0.11,
"learning_rate": 1.9695755815303678e-05,
"loss": 1.2248,
"step": 127
},
{
"epoch": 0.11,
"learning_rate": 1.968906744833762e-05,
"loss": 1.1426,
"step": 128
},
{
"epoch": 0.11,
"learning_rate": 1.968230752212009e-05,
"loss": 1.2379,
"step": 129
},
{
"epoch": 0.11,
"learning_rate": 1.967547608657697e-05,
"loss": 1.0793,
"step": 130
},
{
"epoch": 0.11,
"learning_rate": 1.966857319216229e-05,
"loss": 1.1698,
"step": 131
},
{
"epoch": 0.11,
"learning_rate": 1.966159888985782e-05,
"loss": 1.1846,
"step": 132
},
{
"epoch": 0.11,
"learning_rate": 1.9654553231172748e-05,
"loss": 1.2352,
"step": 133
},
{
"epoch": 0.11,
"learning_rate": 1.9647436268143248e-05,
"loss": 1.2779,
"step": 134
},
{
"epoch": 0.11,
"learning_rate": 1.9640248053332122e-05,
"loss": 1.2378,
"step": 135
},
{
"epoch": 0.11,
"learning_rate": 1.9632988639828407e-05,
"loss": 1.2137,
"step": 136
},
{
"epoch": 0.11,
"learning_rate": 1.962565808124699e-05,
"loss": 1.1914,
"step": 137
},
{
"epoch": 0.12,
"learning_rate": 1.961825643172819e-05,
"loss": 1.2478,
"step": 138
},
{
"epoch": 0.12,
"learning_rate": 1.9610783745937392e-05,
"loss": 1.1782,
"step": 139
},
{
"epoch": 0.12,
"learning_rate": 1.9603240079064605e-05,
"loss": 1.1752,
"step": 140
},
{
"epoch": 0.12,
"learning_rate": 1.9595625486824086e-05,
"loss": 1.2017,
"step": 141
},
{
"epoch": 0.12,
"learning_rate": 1.9587940025453908e-05,
"loss": 1.1926,
"step": 142
},
{
"epoch": 0.12,
"learning_rate": 1.9580183751715563e-05,
"loss": 1.3638,
"step": 143
},
{
"epoch": 0.12,
"learning_rate": 1.957235672289352e-05,
"loss": 1.1866,
"step": 144
},
{
"epoch": 0.12,
"learning_rate": 1.956445899679482e-05,
"loss": 1.1528,
"step": 145
},
{
"epoch": 0.12,
"learning_rate": 1.9556490631748652e-05,
"loss": 1.172,
"step": 146
},
{
"epoch": 0.12,
"learning_rate": 1.95484516866059e-05,
"loss": 1.0545,
"step": 147
},
{
"epoch": 0.12,
"learning_rate": 1.9540342220738726e-05,
"loss": 1.3409,
"step": 148
},
{
"epoch": 0.12,
"learning_rate": 1.953216229404013e-05,
"loss": 1.1431,
"step": 149
},
{
"epoch": 0.13,
"learning_rate": 1.9523911966923506e-05,
"loss": 1.2732,
"step": 150
},
{
"epoch": 0.13,
"learning_rate": 1.9515591300322195e-05,
"loss": 1.2827,
"step": 151
},
{
"epoch": 0.13,
"learning_rate": 1.9507200355689028e-05,
"loss": 1.246,
"step": 152
},
{
"epoch": 0.13,
"learning_rate": 1.9498739194995885e-05,
"loss": 1.2449,
"step": 153
},
{
"epoch": 0.13,
"learning_rate": 1.9490207880733235e-05,
"loss": 1.346,
"step": 154
},
{
"epoch": 0.13,
"learning_rate": 1.948160647590966e-05,
"loss": 1.2723,
"step": 155
},
{
"epoch": 0.13,
"learning_rate": 1.9472935044051412e-05,
"loss": 1.1313,
"step": 156
},
{
"epoch": 0.13,
"learning_rate": 1.9464193649201924e-05,
"loss": 1.3009,
"step": 157
},
{
"epoch": 0.13,
"learning_rate": 1.945538235592135e-05,
"loss": 1.182,
"step": 158
},
{
"epoch": 0.13,
"learning_rate": 1.9446501229286086e-05,
"loss": 1.1277,
"step": 159
},
{
"epoch": 0.13,
"learning_rate": 1.9437550334888277e-05,
"loss": 1.2124,
"step": 160
},
{
"epoch": 0.14,
"learning_rate": 1.942852973883535e-05,
"loss": 1.2483,
"step": 161
},
{
"epoch": 0.14,
"learning_rate": 1.9419439507749515e-05,
"loss": 1.2316,
"step": 162
},
{
"epoch": 0.14,
"learning_rate": 1.9410279708767282e-05,
"loss": 1.1734,
"step": 163
},
{
"epoch": 0.14,
"learning_rate": 1.9401050409538952e-05,
"loss": 1.2339,
"step": 164
},
{
"epoch": 0.14,
"learning_rate": 1.939175167822813e-05,
"loss": 1.2428,
"step": 165
},
{
"epoch": 0.14,
"learning_rate": 1.938238358351121e-05,
"loss": 1.2136,
"step": 166
},
{
"epoch": 0.14,
"learning_rate": 1.937294619457688e-05,
"loss": 1.3297,
"step": 167
},
{
"epoch": 0.14,
"learning_rate": 1.9363439581125603e-05,
"loss": 1.0625,
"step": 168
},
{
"epoch": 0.14,
"learning_rate": 1.9353863813369103e-05,
"loss": 1.1671,
"step": 169
},
{
"epoch": 0.14,
"learning_rate": 1.9344218962029856e-05,
"loss": 1.1805,
"step": 170
},
{
"epoch": 0.14,
"learning_rate": 1.933450509834055e-05,
"loss": 1.2241,
"step": 171
},
{
"epoch": 0.14,
"learning_rate": 1.932472229404356e-05,
"loss": 1.1198,
"step": 172
},
{
"epoch": 0.15,
"learning_rate": 1.931487062139045e-05,
"loss": 1.2854,
"step": 173
},
{
"epoch": 0.15,
"learning_rate": 1.9304950153141394e-05,
"loss": 1.1165,
"step": 174
},
{
"epoch": 0.15,
"learning_rate": 1.9294960962564678e-05,
"loss": 1.125,
"step": 175
},
{
"epoch": 0.15,
"learning_rate": 1.9284903123436126e-05,
"loss": 1.2975,
"step": 176
},
{
"epoch": 0.15,
"learning_rate": 1.9274776710038583e-05,
"loss": 1.2759,
"step": 177
},
{
"epoch": 0.15,
"learning_rate": 1.9264581797161345e-05,
"loss": 1.3289,
"step": 178
},
{
"epoch": 0.15,
"learning_rate": 1.9254318460099615e-05,
"loss": 1.1113,
"step": 179
},
{
"epoch": 0.15,
"learning_rate": 1.9243986774653954e-05,
"loss": 1.1399,
"step": 180
},
{
"epoch": 0.15,
"learning_rate": 1.9233586817129716e-05,
"loss": 1.2984,
"step": 181
},
{
"epoch": 0.15,
"learning_rate": 1.922311866433647e-05,
"loss": 1.2251,
"step": 182
},
{
"epoch": 0.15,
"learning_rate": 1.921258239358746e-05,
"loss": 1.2213,
"step": 183
},
{
"epoch": 0.15,
"learning_rate": 1.9201978082699008e-05,
"loss": 1.2384,
"step": 184
},
{
"epoch": 0.16,
"learning_rate": 1.919130580998996e-05,
"loss": 1.1378,
"step": 185
},
{
"epoch": 0.16,
"learning_rate": 1.91805656542811e-05,
"loss": 1.2728,
"step": 186
},
{
"epoch": 0.16,
"learning_rate": 1.9169757694894557e-05,
"loss": 1.1214,
"step": 187
},
{
"epoch": 0.16,
"learning_rate": 1.9158882011653233e-05,
"loss": 1.115,
"step": 188
},
{
"epoch": 0.16,
"learning_rate": 1.9147938684880213e-05,
"loss": 1.273,
"step": 189
},
{
"epoch": 0.16,
"learning_rate": 1.9136927795398158e-05,
"loss": 1.1804,
"step": 190
},
{
"epoch": 0.16,
"learning_rate": 1.9125849424528725e-05,
"loss": 1.18,
"step": 191
},
{
"epoch": 0.16,
"learning_rate": 1.911470365409196e-05,
"loss": 1.2424,
"step": 192
},
{
"epoch": 0.16,
"learning_rate": 1.9103490566405694e-05,
"loss": 1.2372,
"step": 193
},
{
"epoch": 0.16,
"learning_rate": 1.909221024428493e-05,
"loss": 1.0896,
"step": 194
},
{
"epoch": 0.16,
"learning_rate": 1.9080862771041232e-05,
"loss": 1.1715,
"step": 195
},
{
"epoch": 0.16,
"learning_rate": 1.906944823048212e-05,
"loss": 1.1772,
"step": 196
},
{
"epoch": 0.17,
"learning_rate": 1.905796670691044e-05,
"loss": 1.1724,
"step": 197
},
{
"epoch": 0.17,
"learning_rate": 1.9046418285123755e-05,
"loss": 1.2572,
"step": 198
},
{
"epoch": 0.17,
"learning_rate": 1.9034803050413698e-05,
"loss": 1.2469,
"step": 199
},
{
"epoch": 0.17,
"learning_rate": 1.9023121088565353e-05,
"loss": 1.2741,
"step": 200
},
{
"epoch": 0.17,
"learning_rate": 1.9011372485856624e-05,
"loss": 1.2415,
"step": 201
},
{
"epoch": 0.17,
"learning_rate": 1.8999557329057605e-05,
"loss": 1.1112,
"step": 202
},
{
"epoch": 0.17,
"learning_rate": 1.8987675705429916e-05,
"loss": 1.181,
"step": 203
},
{
"epoch": 0.17,
"learning_rate": 1.8975727702726078e-05,
"loss": 1.1483,
"step": 204
},
{
"epoch": 0.17,
"learning_rate": 1.896371340918886e-05,
"loss": 1.1569,
"step": 205
},
{
"epoch": 0.17,
"learning_rate": 1.8951632913550625e-05,
"loss": 1.1242,
"step": 206
},
{
"epoch": 0.17,
"learning_rate": 1.8939486305032677e-05,
"loss": 1.0974,
"step": 207
},
{
"epoch": 0.17,
"learning_rate": 1.89272736733446e-05,
"loss": 1.3355,
"step": 208
},
{
"epoch": 0.18,
"learning_rate": 1.8914995108683598e-05,
"loss": 1.2029,
"step": 209
},
{
"epoch": 0.18,
"learning_rate": 1.890265070173382e-05,
"loss": 1.1613,
"step": 210
},
{
"epoch": 0.18,
"learning_rate": 1.8890240543665716e-05,
"loss": 1.2549,
"step": 211
},
{
"epoch": 0.18,
"learning_rate": 1.887776472613532e-05,
"loss": 1.2789,
"step": 212
},
{
"epoch": 0.18,
"learning_rate": 1.8865223341283618e-05,
"loss": 1.2193,
"step": 213
},
{
"epoch": 0.18,
"learning_rate": 1.885261648173584e-05,
"loss": 1.2323,
"step": 214
},
{
"epoch": 0.18,
"learning_rate": 1.883994424060079e-05,
"loss": 1.3269,
"step": 215
},
{
"epoch": 0.18,
"learning_rate": 1.882720671147014e-05,
"loss": 1.2591,
"step": 216
},
{
"epoch": 0.18,
"learning_rate": 1.8814403988417758e-05,
"loss": 1.1464,
"step": 217
},
{
"epoch": 0.18,
"learning_rate": 1.8801536165999008e-05,
"loss": 1.126,
"step": 218
},
{
"epoch": 0.18,
"learning_rate": 1.8788603339250048e-05,
"loss": 1.2736,
"step": 219
},
{
"epoch": 0.18,
"learning_rate": 1.8775605603687128e-05,
"loss": 1.1508,
"step": 220
},
{
"epoch": 0.19,
"learning_rate": 1.8762543055305888e-05,
"loss": 1.1655,
"step": 221
},
{
"epoch": 0.19,
"learning_rate": 1.874941579058065e-05,
"loss": 1.2109,
"step": 222
},
{
"epoch": 0.19,
"learning_rate": 1.8736223906463698e-05,
"loss": 1.1558,
"step": 223
},
{
"epoch": 0.19,
"learning_rate": 1.8722967500384564e-05,
"loss": 1.2051,
"step": 224
},
{
"epoch": 0.19,
"learning_rate": 1.870964667024932e-05,
"loss": 1.1103,
"step": 225
},
{
"epoch": 0.19,
"learning_rate": 1.869626151443985e-05,
"loss": 1.2582,
"step": 226
},
{
"epoch": 0.19,
"learning_rate": 1.86828121318131e-05,
"loss": 1.221,
"step": 227
},
{
"epoch": 0.19,
"learning_rate": 1.866929862170038e-05,
"loss": 1.2196,
"step": 228
},
{
"epoch": 0.19,
"learning_rate": 1.865572108390662e-05,
"loss": 1.2237,
"step": 229
},
{
"epoch": 0.19,
"learning_rate": 1.8642079618709627e-05,
"loss": 1.2242,
"step": 230
},
{
"epoch": 0.19,
"learning_rate": 1.862837432685935e-05,
"loss": 1.1561,
"step": 231
},
{
"epoch": 0.19,
"learning_rate": 1.8614605309577135e-05,
"loss": 1.2589,
"step": 232
},
{
"epoch": 0.2,
"learning_rate": 1.860077266855497e-05,
"loss": 1.1399,
"step": 233
},
{
"epoch": 0.2,
"learning_rate": 1.8586876505954742e-05,
"loss": 1.1725,
"step": 234
},
{
"epoch": 0.2,
"learning_rate": 1.857291692440749e-05,
"loss": 1.1633,
"step": 235
},
{
"epoch": 0.2,
"learning_rate": 1.8558894027012626e-05,
"loss": 1.1776,
"step": 236
},
{
"epoch": 0.2,
"learning_rate": 1.854480791733719e-05,
"loss": 1.2233,
"step": 237
},
{
"epoch": 0.2,
"learning_rate": 1.8530658699415088e-05,
"loss": 1.1943,
"step": 238
},
{
"epoch": 0.2,
"learning_rate": 1.8516446477746305e-05,
"loss": 1.1747,
"step": 239
},
{
"epoch": 0.2,
"learning_rate": 1.8502171357296144e-05,
"loss": 1.1233,
"step": 240
},
{
"epoch": 0.2,
"learning_rate": 1.848783344349446e-05,
"loss": 1.1065,
"step": 241
},
{
"epoch": 0.2,
"learning_rate": 1.8473432842234867e-05,
"loss": 1.0629,
"step": 242
},
{
"epoch": 0.2,
"learning_rate": 1.8458969659873966e-05,
"loss": 1.1395,
"step": 243
},
{
"epoch": 0.2,
"learning_rate": 1.844444400323055e-05,
"loss": 1.1783,
"step": 244
},
{
"epoch": 0.21,
"learning_rate": 1.8429855979584816e-05,
"loss": 1.1482,
"step": 245
},
{
"epoch": 0.21,
"learning_rate": 1.841520569667759e-05,
"loss": 1.1096,
"step": 246
},
{
"epoch": 0.21,
"learning_rate": 1.8400493262709503e-05,
"loss": 1.2644,
"step": 247
},
{
"epoch": 0.21,
"learning_rate": 1.8385718786340216e-05,
"loss": 1.2022,
"step": 248
},
{
"epoch": 0.21,
"learning_rate": 1.8370882376687608e-05,
"loss": 1.2342,
"step": 249
},
{
"epoch": 0.21,
"learning_rate": 1.8355984143326968e-05,
"loss": 1.0691,
"step": 250
},
{
"epoch": 0.21,
"learning_rate": 1.834102419629019e-05,
"loss": 1.1144,
"step": 251
},
{
"epoch": 0.21,
"learning_rate": 1.8326002646064947e-05,
"loss": 1.169,
"step": 252
},
{
"epoch": 0.21,
"learning_rate": 1.831091960359391e-05,
"loss": 1.2056,
"step": 253
},
{
"epoch": 0.21,
"learning_rate": 1.8295775180273883e-05,
"loss": 1.1919,
"step": 254
},
{
"epoch": 0.21,
"learning_rate": 1.8280569487955014e-05,
"loss": 1.1419,
"step": 255
},
{
"epoch": 0.21,
"learning_rate": 1.826530263893995e-05,
"loss": 1.1101,
"step": 256
},
{
"epoch": 0.22,
"learning_rate": 1.8249974745983023e-05,
"loss": 1.1707,
"step": 257
},
{
"epoch": 0.22,
"learning_rate": 1.8234585922289408e-05,
"loss": 1.2138,
"step": 258
},
{
"epoch": 0.22,
"learning_rate": 1.8219136281514275e-05,
"loss": 1.2311,
"step": 259
},
{
"epoch": 0.22,
"learning_rate": 1.820362593776198e-05,
"loss": 1.2867,
"step": 260
},
{
"epoch": 0.22,
"learning_rate": 1.81880550055852e-05,
"loss": 1.0333,
"step": 261
},
{
"epoch": 0.22,
"learning_rate": 1.817242359998408e-05,
"loss": 1.213,
"step": 262
},
{
"epoch": 0.22,
"learning_rate": 1.815673183640541e-05,
"loss": 1.1154,
"step": 263
},
{
"epoch": 0.22,
"learning_rate": 1.8140979830741753e-05,
"loss": 1.2085,
"step": 264
},
{
"epoch": 0.22,
"learning_rate": 1.8125167699330598e-05,
"loss": 1.2162,
"step": 265
},
{
"epoch": 0.22,
"learning_rate": 1.810929555895348e-05,
"loss": 1.1229,
"step": 266
},
{
"epoch": 0.22,
"learning_rate": 1.8093363526835158e-05,
"loss": 1.1621,
"step": 267
},
{
"epoch": 0.22,
"learning_rate": 1.807737172064271e-05,
"loss": 1.1808,
"step": 268
},
{
"epoch": 0.23,
"learning_rate": 1.8061320258484678e-05,
"loss": 1.1368,
"step": 269
},
{
"epoch": 0.23,
"learning_rate": 1.804520925891021e-05,
"loss": 1.142,
"step": 270
},
{
"epoch": 0.23,
"learning_rate": 1.8029038840908164e-05,
"loss": 1.1052,
"step": 271
},
{
"epoch": 0.23,
"learning_rate": 1.801280912390623e-05,
"loss": 1.1695,
"step": 272
},
{
"epoch": 0.23,
"learning_rate": 1.7996520227770067e-05,
"loss": 1.1363,
"step": 273
},
{
"epoch": 0.23,
"learning_rate": 1.7980172272802398e-05,
"loss": 1.2725,
"step": 274
},
{
"epoch": 0.23,
"learning_rate": 1.796376537974213e-05,
"loss": 1.2461,
"step": 275
},
{
"epoch": 0.23,
"learning_rate": 1.794729966976346e-05,
"loss": 1.1654,
"step": 276
},
{
"epoch": 0.23,
"learning_rate": 1.793077526447499e-05,
"loss": 1.2284,
"step": 277
},
{
"epoch": 0.23,
"learning_rate": 1.7914192285918807e-05,
"loss": 1.3043,
"step": 278
},
{
"epoch": 0.23,
"learning_rate": 1.7897550856569603e-05,
"loss": 1.0824,
"step": 279
},
{
"epoch": 0.23,
"learning_rate": 1.7880851099333762e-05,
"loss": 1.1501,
"step": 280
},
{
"epoch": 0.24,
"learning_rate": 1.786409313754845e-05,
"loss": 1.1308,
"step": 281
},
{
"epoch": 0.24,
"learning_rate": 1.7847277094980703e-05,
"loss": 1.1703,
"step": 282
},
{
"epoch": 0.24,
"learning_rate": 1.7830403095826527e-05,
"loss": 1.1675,
"step": 283
},
{
"epoch": 0.24,
"learning_rate": 1.7813471264709958e-05,
"loss": 1.098,
"step": 284
},
{
"epoch": 0.24,
"learning_rate": 1.7796481726682165e-05,
"loss": 1.2146,
"step": 285
},
{
"epoch": 0.24,
"learning_rate": 1.7779434607220513e-05,
"loss": 1.1281,
"step": 286
},
{
"epoch": 0.24,
"learning_rate": 1.7762330032227625e-05,
"loss": 1.1142,
"step": 287
},
{
"epoch": 0.24,
"learning_rate": 1.7745168128030483e-05,
"loss": 1.2072,
"step": 288
},
{
"epoch": 0.24,
"learning_rate": 1.772794902137947e-05,
"loss": 1.1925,
"step": 289
},
{
"epoch": 0.24,
"learning_rate": 1.7710672839447442e-05,
"loss": 1.3229,
"step": 290
},
{
"epoch": 0.24,
"learning_rate": 1.769333970982879e-05,
"loss": 1.1819,
"step": 291
},
{
"epoch": 0.24,
"learning_rate": 1.7675949760538495e-05,
"loss": 1.162,
"step": 292
},
{
"epoch": 0.25,
"learning_rate": 1.7658503120011177e-05,
"loss": 1.1212,
"step": 293
},
{
"epoch": 0.25,
"learning_rate": 1.7640999917100166e-05,
"loss": 1.0532,
"step": 294
},
{
"epoch": 0.25,
"learning_rate": 1.7623440281076527e-05,
"loss": 1.162,
"step": 295
},
{
"epoch": 0.25,
"learning_rate": 1.760582434162812e-05,
"loss": 1.201,
"step": 296
},
{
"epoch": 0.25,
"learning_rate": 1.7588152228858633e-05,
"loss": 1.1066,
"step": 297
},
{
"epoch": 0.25,
"learning_rate": 1.7570424073286635e-05,
"loss": 1.2092,
"step": 298
},
{
"epoch": 0.25,
"learning_rate": 1.7552640005844593e-05,
"loss": 1.0971,
"step": 299
},
{
"epoch": 0.25,
"learning_rate": 1.753480015787792e-05,
"loss": 1.104,
"step": 300
},
{
"epoch": 0.25,
"learning_rate": 1.7516904661143994e-05,
"loss": 1.0912,
"step": 301
},
{
"epoch": 0.25,
"learning_rate": 1.7498953647811202e-05,
"loss": 1.1938,
"step": 302
},
{
"epoch": 0.25,
"learning_rate": 1.748094725045794e-05,
"loss": 1.1397,
"step": 303
},
{
"epoch": 0.25,
"learning_rate": 1.7462885602071663e-05,
"loss": 1.166,
"step": 304
},
{
"epoch": 0.26,
"learning_rate": 1.7444768836047864e-05,
"loss": 1.1489,
"step": 305
},
{
"epoch": 0.26,
"learning_rate": 1.7426597086189128e-05,
"loss": 1.0993,
"step": 306
},
{
"epoch": 0.26,
"learning_rate": 1.7408370486704114e-05,
"loss": 1.0746,
"step": 307
},
{
"epoch": 0.26,
"learning_rate": 1.7390089172206594e-05,
"loss": 1.1138,
"step": 308
},
{
"epoch": 0.26,
"learning_rate": 1.7371753277714422e-05,
"loss": 1.0426,
"step": 309
},
{
"epoch": 0.26,
"learning_rate": 1.735336293864857e-05,
"loss": 1.0842,
"step": 310
},
{
"epoch": 0.26,
"learning_rate": 1.733491829083211e-05,
"loss": 1.0861,
"step": 311
},
{
"epoch": 0.26,
"learning_rate": 1.731641947048921e-05,
"loss": 1.1035,
"step": 312
},
{
"epoch": 0.26,
"learning_rate": 1.7297866614244142e-05,
"loss": 1.1717,
"step": 313
},
{
"epoch": 0.26,
"learning_rate": 1.7279259859120253e-05,
"loss": 1.164,
"step": 314
},
{
"epoch": 0.26,
"learning_rate": 1.726059934253897e-05,
"loss": 1.1451,
"step": 315
},
{
"epoch": 0.27,
"learning_rate": 1.7241885202318787e-05,
"loss": 1.1866,
"step": 316
},
{
"epoch": 0.27,
"learning_rate": 1.722311757667422e-05,
"loss": 1.0418,
"step": 317
},
{
"epoch": 0.27,
"learning_rate": 1.7204296604214818e-05,
"loss": 1.0387,
"step": 318
},
{
"epoch": 0.27,
"learning_rate": 1.7185422423944124e-05,
"loss": 1.2088,
"step": 319
},
{
"epoch": 0.27,
"learning_rate": 1.7166495175258654e-05,
"loss": 1.1499,
"step": 320
},
{
"epoch": 0.27,
"learning_rate": 1.7147514997946856e-05,
"loss": 1.2202,
"step": 321
},
{
"epoch": 0.27,
"learning_rate": 1.7128482032188093e-05,
"loss": 1.1988,
"step": 322
},
{
"epoch": 0.27,
"learning_rate": 1.71093964185516e-05,
"loss": 1.2098,
"step": 323
},
{
"epoch": 0.27,
"learning_rate": 1.709025829799544e-05,
"loss": 1.0199,
"step": 324
},
{
"epoch": 0.27,
"learning_rate": 1.7071067811865477e-05,
"loss": 1.173,
"step": 325
},
{
"epoch": 0.27,
"learning_rate": 1.7051825101894323e-05,
"loss": 1.1412,
"step": 326
},
{
"epoch": 0.27,
"learning_rate": 1.7032530310200286e-05,
"loss": 1.0721,
"step": 327
},
{
"epoch": 0.28,
"learning_rate": 1.701318357928634e-05,
"loss": 1.2049,
"step": 328
},
{
"epoch": 0.28,
"learning_rate": 1.6993785052039043e-05,
"loss": 1.1303,
"step": 329
},
{
"epoch": 0.28,
"learning_rate": 1.697433487172752e-05,
"loss": 1.1077,
"step": 330
},
{
"epoch": 0.28,
"learning_rate": 1.695483318200235e-05,
"loss": 1.1597,
"step": 331
},
{
"epoch": 0.28,
"learning_rate": 1.6935280126894577e-05,
"loss": 1.1776,
"step": 332
},
{
"epoch": 0.28,
"learning_rate": 1.691567585081458e-05,
"loss": 1.1358,
"step": 333
},
{
"epoch": 0.28,
"learning_rate": 1.6896020498551047e-05,
"loss": 1.0625,
"step": 334
},
{
"epoch": 0.28,
"learning_rate": 1.687631421526988e-05,
"loss": 1.1508,
"step": 335
},
{
"epoch": 0.28,
"learning_rate": 1.685655714651316e-05,
"loss": 1.1583,
"step": 336
},
{
"epoch": 0.28,
"learning_rate": 1.683674943819801e-05,
"loss": 1.1441,
"step": 337
},
{
"epoch": 0.28,
"learning_rate": 1.6816891236615588e-05,
"loss": 1.1538,
"step": 338
},
{
"epoch": 0.28,
"learning_rate": 1.679698268842996e-05,
"loss": 1.2693,
"step": 339
},
{
"epoch": 0.29,
"learning_rate": 1.6777023940677036e-05,
"loss": 1.0847,
"step": 340
},
{
"epoch": 0.29,
"learning_rate": 1.6757015140763466e-05,
"loss": 1.1827,
"step": 341
},
{
"epoch": 0.29,
"learning_rate": 1.6736956436465573e-05,
"loss": 1.2952,
"step": 342
},
{
"epoch": 0.29,
"learning_rate": 1.6716847975928256e-05,
"loss": 1.1737,
"step": 343
},
{
"epoch": 0.29,
"learning_rate": 1.669668990766388e-05,
"loss": 1.1821,
"step": 344
},
{
"epoch": 0.29,
"learning_rate": 1.6676482380551198e-05,
"loss": 1.1288,
"step": 345
},
{
"epoch": 0.29,
"learning_rate": 1.6656225543834244e-05,
"loss": 1.162,
"step": 346
},
{
"epoch": 0.29,
"learning_rate": 1.663591954712124e-05,
"loss": 1.0906,
"step": 347
},
{
"epoch": 0.29,
"learning_rate": 1.6615564540383465e-05,
"loss": 1.0483,
"step": 348
},
{
"epoch": 0.29,
"learning_rate": 1.659516067395418e-05,
"loss": 1.1827,
"step": 349
},
{
"epoch": 0.29,
"learning_rate": 1.657470809852749e-05,
"loss": 1.0504,
"step": 350
},
{
"epoch": 0.29,
"learning_rate": 1.6554206965157253e-05,
"loss": 1.2106,
"step": 351
},
{
"epoch": 0.3,
"learning_rate": 1.6533657425255954e-05,
"loss": 1.3281,
"step": 352
},
{
"epoch": 0.3,
"learning_rate": 1.651305963059358e-05,
"loss": 1.0249,
"step": 353
},
{
"epoch": 0.3,
"learning_rate": 1.6492413733296518e-05,
"loss": 1.2419,
"step": 354
},
{
"epoch": 0.3,
"learning_rate": 1.647171988584641e-05,
"loss": 1.1585,
"step": 355
},
{
"epoch": 0.3,
"learning_rate": 1.6450978241079047e-05,
"loss": 1.0698,
"step": 356
},
{
"epoch": 0.3,
"learning_rate": 1.643018895218322e-05,
"loss": 1.059,
"step": 357
},
{
"epoch": 0.3,
"learning_rate": 1.64093521726996e-05,
"loss": 1.2459,
"step": 358
},
{
"epoch": 0.3,
"learning_rate": 1.638846805651961e-05,
"loss": 1.1886,
"step": 359
},
{
"epoch": 0.3,
"learning_rate": 1.6367536757884285e-05,
"loss": 1.1925,
"step": 360
},
{
"epoch": 0.3,
"learning_rate": 1.6346558431383118e-05,
"loss": 1.2149,
"step": 361
},
{
"epoch": 0.3,
"learning_rate": 1.6325533231952927e-05,
"loss": 1.1256,
"step": 362
},
{
"epoch": 0.3,
"learning_rate": 1.6304461314876722e-05,
"loss": 1.1315,
"step": 363
},
{
"epoch": 0.31,
"learning_rate": 1.628334283578255e-05,
"loss": 1.1721,
"step": 364
},
{
"epoch": 0.31,
"learning_rate": 1.6262177950642338e-05,
"loss": 1.2249,
"step": 365
},
{
"epoch": 0.31,
"learning_rate": 1.6240966815770756e-05,
"loss": 1.1995,
"step": 366
},
{
"epoch": 0.31,
"learning_rate": 1.6219709587824048e-05,
"loss": 1.0177,
"step": 367
},
{
"epoch": 0.31,
"learning_rate": 1.619840642379888e-05,
"loss": 1.1286,
"step": 368
},
{
"epoch": 0.31,
"learning_rate": 1.6177057481031197e-05,
"loss": 1.0459,
"step": 369
},
{
"epoch": 0.31,
"learning_rate": 1.615566291719502e-05,
"loss": 1.2175,
"step": 370
},
{
"epoch": 0.31,
"learning_rate": 1.6134222890301327e-05,
"loss": 1.1376,
"step": 371
},
{
"epoch": 0.31,
"learning_rate": 1.6112737558696864e-05,
"loss": 0.9767,
"step": 372
},
{
"epoch": 0.31,
"learning_rate": 1.6091207081062973e-05,
"loss": 1.0415,
"step": 373
},
{
"epoch": 0.31,
"learning_rate": 1.606963161641443e-05,
"loss": 1.1501,
"step": 374
},
{
"epoch": 0.31,
"learning_rate": 1.6048011324098254e-05,
"loss": 1.2156,
"step": 375
},
{
"epoch": 0.32,
"learning_rate": 1.6026346363792565e-05,
"loss": 0.9929,
"step": 376
},
{
"epoch": 0.32,
"learning_rate": 1.6004636895505357e-05,
"loss": 1.1557,
"step": 377
},
{
"epoch": 0.32,
"learning_rate": 1.5982883079573354e-05,
"loss": 1.0847,
"step": 378
},
{
"epoch": 0.32,
"learning_rate": 1.5961085076660806e-05,
"loss": 1.0762,
"step": 379
},
{
"epoch": 0.32,
"learning_rate": 1.5939243047758312e-05,
"loss": 1.1354,
"step": 380
},
{
"epoch": 0.32,
"learning_rate": 1.5917357154181633e-05,
"loss": 1.1752,
"step": 381
},
{
"epoch": 0.32,
"learning_rate": 1.5895427557570487e-05,
"loss": 1.1353,
"step": 382
},
{
"epoch": 0.32,
"learning_rate": 1.5873454419887365e-05,
"loss": 1.1302,
"step": 383
},
{
"epoch": 0.32,
"learning_rate": 1.585143790341634e-05,
"loss": 1.0786,
"step": 384
},
{
"epoch": 0.32,
"learning_rate": 1.5829378170761853e-05,
"loss": 1.0211,
"step": 385
},
{
"epoch": 0.32,
"learning_rate": 1.5807275384847528e-05,
"loss": 1.1838,
"step": 386
},
{
"epoch": 0.32,
"learning_rate": 1.5785129708914957e-05,
"loss": 1.0755,
"step": 387
},
{
"epoch": 0.33,
"learning_rate": 1.5762941306522504e-05,
"loss": 1.1503,
"step": 388
},
{
"epoch": 0.33,
"learning_rate": 1.5740710341544088e-05,
"loss": 1.1327,
"step": 389
},
{
"epoch": 0.33,
"learning_rate": 1.5718436978167976e-05,
"loss": 1.2214,
"step": 390
},
{
"epoch": 0.33,
"learning_rate": 1.5696121380895576e-05,
"loss": 1.176,
"step": 391
},
{
"epoch": 0.33,
"learning_rate": 1.5673763714540214e-05,
"loss": 1.1124,
"step": 392
},
{
"epoch": 0.33,
"learning_rate": 1.565136414422592e-05,
"loss": 1.1212,
"step": 393
},
{
"epoch": 0.33,
"learning_rate": 1.562892283538621e-05,
"loss": 1.0597,
"step": 394
},
{
"epoch": 0.33,
"learning_rate": 1.560643995376286e-05,
"loss": 1.1578,
"step": 395
},
{
"epoch": 0.33,
"learning_rate": 1.5583915665404688e-05,
"loss": 1.0946,
"step": 396
},
{
"epoch": 0.33,
"learning_rate": 1.5561350136666316e-05,
"loss": 1.0713,
"step": 397
},
{
"epoch": 0.33,
"learning_rate": 1.5538743534206968e-05,
"loss": 1.0381,
"step": 398
},
{
"epoch": 0.33,
"learning_rate": 1.55160960249892e-05,
"loss": 1.1601,
"step": 399
},
{
"epoch": 0.34,
"learning_rate": 1.54934077762777e-05,
"loss": 1.0874,
"step": 400
},
{
"epoch": 0.34,
"learning_rate": 1.5470678955638032e-05,
"loss": 1.1824,
"step": 401
},
{
"epoch": 0.34,
"learning_rate": 1.5447909730935413e-05,
"loss": 1.135,
"step": 402
},
{
"epoch": 0.34,
"learning_rate": 1.542510027033347e-05,
"loss": 1.2259,
"step": 403
},
{
"epoch": 0.34,
"learning_rate": 1.5402250742292983e-05,
"loss": 1.1228,
"step": 404
},
{
"epoch": 0.34,
"learning_rate": 1.5379361315570663e-05,
"loss": 1.0783,
"step": 405
},
{
"epoch": 0.34,
"learning_rate": 1.5356432159217893e-05,
"loss": 1.1195,
"step": 406
},
{
"epoch": 0.34,
"learning_rate": 1.533346344257948e-05,
"loss": 1.056,
"step": 407
},
{
"epoch": 0.34,
"learning_rate": 1.5310455335292404e-05,
"loss": 1.0922,
"step": 408
},
{
"epoch": 0.34,
"learning_rate": 1.5287408007284575e-05,
"loss": 1.0628,
"step": 409
},
{
"epoch": 0.34,
"learning_rate": 1.526432162877356e-05,
"loss": 1.09,
"step": 410
},
{
"epoch": 0.34,
"learning_rate": 1.5241196370265348e-05,
"loss": 1.11,
"step": 411
},
{
"epoch": 0.35,
"learning_rate": 1.5218032402553067e-05,
"loss": 1.0801,
"step": 412
},
{
"epoch": 0.35,
"learning_rate": 1.5194829896715741e-05,
"loss": 1.1495,
"step": 413
},
{
"epoch": 0.35,
"learning_rate": 1.5171589024117022e-05,
"loss": 1.121,
"step": 414
},
{
"epoch": 0.35,
"learning_rate": 1.5148309956403915e-05,
"loss": 1.1426,
"step": 415
},
{
"epoch": 0.35,
"learning_rate": 1.5124992865505523e-05,
"loss": 1.0894,
"step": 416
},
{
"epoch": 0.35,
"learning_rate": 1.510163792363177e-05,
"loss": 1.0876,
"step": 417
},
{
"epoch": 0.35,
"learning_rate": 1.5078245303272133e-05,
"loss": 1.165,
"step": 418
},
{
"epoch": 0.35,
"learning_rate": 1.5054815177194362e-05,
"loss": 1.2857,
"step": 419
},
{
"epoch": 0.35,
"learning_rate": 1.5031347718443212e-05,
"loss": 1.1519,
"step": 420
},
{
"epoch": 0.35,
"learning_rate": 1.5007843100339155e-05,
"loss": 1.0477,
"step": 421
},
{
"epoch": 0.35,
"learning_rate": 1.4984301496477111e-05,
"loss": 1.0129,
"step": 422
},
{
"epoch": 0.35,
"learning_rate": 1.4960723080725164e-05,
"loss": 1.0625,
"step": 423
},
{
"epoch": 0.36,
"learning_rate": 1.4937108027223266e-05,
"loss": 1.0532,
"step": 424
},
{
"epoch": 0.36,
"learning_rate": 1.4913456510381966e-05,
"loss": 1.1655,
"step": 425
},
{
"epoch": 0.36,
"learning_rate": 1.4889768704881112e-05,
"loss": 1.0847,
"step": 426
},
{
"epoch": 0.36,
"learning_rate": 1.4866044785668563e-05,
"loss": 0.9805,
"step": 427
},
{
"epoch": 0.36,
"learning_rate": 1.4842284927958908e-05,
"loss": 1.0598,
"step": 428
},
{
"epoch": 0.36,
"learning_rate": 1.4818489307232149e-05,
"loss": 1.0831,
"step": 429
},
{
"epoch": 0.36,
"learning_rate": 1.4794658099232426e-05,
"loss": 1.0979,
"step": 430
},
{
"epoch": 0.36,
"learning_rate": 1.4770791479966708e-05,
"loss": 0.9962,
"step": 431
},
{
"epoch": 0.36,
"learning_rate": 1.4746889625703502e-05,
"loss": 1.0091,
"step": 432
},
{
"epoch": 0.36,
"learning_rate": 1.4722952712971535e-05,
"loss": 1.0671,
"step": 433
},
{
"epoch": 0.36,
"learning_rate": 1.4698980918558467e-05,
"loss": 1.0411,
"step": 434
},
{
"epoch": 0.36,
"learning_rate": 1.4674974419509582e-05,
"loss": 1.1817,
"step": 435
},
{
"epoch": 0.37,
"learning_rate": 1.4650933393126476e-05,
"loss": 1.267,
"step": 436
},
{
"epoch": 0.37,
"learning_rate": 1.4626858016965742e-05,
"loss": 1.0557,
"step": 437
},
{
"epoch": 0.37,
"learning_rate": 1.4602748468837669e-05,
"loss": 1.071,
"step": 438
},
{
"epoch": 0.37,
"learning_rate": 1.4578604926804932e-05,
"loss": 1.1764,
"step": 439
},
{
"epoch": 0.37,
"learning_rate": 1.455442756918126e-05,
"loss": 1.1305,
"step": 440
},
{
"epoch": 0.37,
"learning_rate": 1.453021657453014e-05,
"loss": 1.2061,
"step": 441
},
{
"epoch": 0.37,
"learning_rate": 1.4505972121663479e-05,
"loss": 1.1132,
"step": 442
},
{
"epoch": 0.37,
"learning_rate": 1.4481694389640291e-05,
"loss": 1.0902,
"step": 443
},
{
"epoch": 0.37,
"learning_rate": 1.4457383557765385e-05,
"loss": 1.2242,
"step": 444
},
{
"epoch": 0.37,
"learning_rate": 1.443303980558802e-05,
"loss": 1.1325,
"step": 445
},
{
"epoch": 0.37,
"learning_rate": 1.4408663312900602e-05,
"loss": 1.0879,
"step": 446
},
{
"epoch": 0.37,
"learning_rate": 1.4384254259737332e-05,
"loss": 1.1637,
"step": 447
},
{
"epoch": 0.38,
"learning_rate": 1.4359812826372894e-05,
"loss": 1.1082,
"step": 448
},
{
"epoch": 0.38,
"learning_rate": 1.4335339193321123e-05,
"loss": 1.0338,
"step": 449
},
{
"epoch": 0.38,
"learning_rate": 1.4310833541333658e-05,
"loss": 1.0382,
"step": 450
},
{
"epoch": 0.38,
"learning_rate": 1.428629605139862e-05,
"loss": 1.1959,
"step": 451
},
{
"epoch": 0.38,
"learning_rate": 1.4261726904739273e-05,
"loss": 1.0936,
"step": 452
},
{
"epoch": 0.38,
"learning_rate": 1.4237126282812684e-05,
"loss": 1.0443,
"step": 453
},
{
"epoch": 0.38,
"learning_rate": 1.4212494367308374e-05,
"loss": 1.1352,
"step": 454
},
{
"epoch": 0.38,
"learning_rate": 1.4187831340147e-05,
"loss": 1.1318,
"step": 455
},
{
"epoch": 0.38,
"learning_rate": 1.4163137383478984e-05,
"loss": 0.9967,
"step": 456
},
{
"epoch": 0.38,
"learning_rate": 1.4138412679683187e-05,
"loss": 1.0163,
"step": 457
},
{
"epoch": 0.38,
"learning_rate": 1.4113657411365546e-05,
"loss": 1.1117,
"step": 458
},
{
"epoch": 0.38,
"learning_rate": 1.4088871761357745e-05,
"loss": 1.0453,
"step": 459
},
{
"epoch": 0.39,
"learning_rate": 1.4064055912715846e-05,
"loss": 1.2473,
"step": 460
},
{
"epoch": 0.39,
"learning_rate": 1.403921004871895e-05,
"loss": 1.1516,
"step": 461
},
{
"epoch": 0.39,
"learning_rate": 1.401433435286784e-05,
"loss": 1.0157,
"step": 462
},
{
"epoch": 0.39,
"learning_rate": 1.3989429008883613e-05,
"loss": 1.1479,
"step": 463
},
{
"epoch": 0.39,
"learning_rate": 1.3964494200706344e-05,
"loss": 1.1181,
"step": 464
},
{
"epoch": 0.39,
"learning_rate": 1.393953011249372e-05,
"loss": 1.2199,
"step": 465
},
{
"epoch": 0.39,
"learning_rate": 1.3914536928619671e-05,
"loss": 1.0544,
"step": 466
},
{
"epoch": 0.39,
"learning_rate": 1.3889514833673016e-05,
"loss": 1.1422,
"step": 467
},
{
"epoch": 0.39,
"learning_rate": 1.3864464012456103e-05,
"loss": 1.1584,
"step": 468
},
{
"epoch": 0.39,
"learning_rate": 1.3839384649983432e-05,
"loss": 1.1335,
"step": 469
},
{
"epoch": 0.39,
"learning_rate": 1.3814276931480308e-05,
"loss": 1.1361,
"step": 470
},
{
"epoch": 0.4,
"learning_rate": 1.3789141042381446e-05,
"loss": 1.0482,
"step": 471
},
{
"epoch": 0.4,
"learning_rate": 1.3763977168329632e-05,
"loss": 1.1425,
"step": 472
},
{
"epoch": 0.4,
"learning_rate": 1.3738785495174325e-05,
"loss": 1.1768,
"step": 473
},
{
"epoch": 0.4,
"learning_rate": 1.3713566208970303e-05,
"loss": 1.0069,
"step": 474
},
{
"epoch": 0.4,
"learning_rate": 1.3688319495976283e-05,
"loss": 1.1906,
"step": 475
},
{
"epoch": 0.4,
"learning_rate": 1.3663045542653539e-05,
"loss": 1.0321,
"step": 476
},
{
"epoch": 0.4,
"learning_rate": 1.3637744535664532e-05,
"loss": 1.0541,
"step": 477
},
{
"epoch": 0.4,
"learning_rate": 1.3612416661871532e-05,
"loss": 1.1265,
"step": 478
},
{
"epoch": 0.4,
"learning_rate": 1.3587062108335231e-05,
"loss": 1.2049,
"step": 479
},
{
"epoch": 0.4,
"learning_rate": 1.356168106231337e-05,
"loss": 1.1211,
"step": 480
},
{
"epoch": 0.4,
"learning_rate": 1.3536273711259353e-05,
"loss": 1.0433,
"step": 481
},
{
"epoch": 0.4,
"learning_rate": 1.351084024282086e-05,
"loss": 1.1116,
"step": 482
},
{
"epoch": 0.41,
"learning_rate": 1.3485380844838461e-05,
"loss": 1.0837,
"step": 483
},
{
"epoch": 0.41,
"learning_rate": 1.3459895705344228e-05,
"loss": 1.1204,
"step": 484
},
{
"epoch": 0.41,
"learning_rate": 1.3434385012560355e-05,
"loss": 1.1462,
"step": 485
},
{
"epoch": 0.41,
"learning_rate": 1.3408848954897757e-05,
"loss": 0.9902,
"step": 486
},
{
"epoch": 0.41,
"learning_rate": 1.3383287720954692e-05,
"loss": 1.1093,
"step": 487
},
{
"epoch": 0.41,
"learning_rate": 1.3357701499515345e-05,
"loss": 1.0436,
"step": 488
},
{
"epoch": 0.41,
"learning_rate": 1.3332090479548457e-05,
"loss": 1.0496,
"step": 489
},
{
"epoch": 0.41,
"learning_rate": 1.3306454850205914e-05,
"loss": 1.1103,
"step": 490
},
{
"epoch": 0.41,
"learning_rate": 1.3280794800821366e-05,
"loss": 1.0282,
"step": 491
},
{
"epoch": 0.41,
"learning_rate": 1.3255110520908807e-05,
"loss": 1.1134,
"step": 492
},
{
"epoch": 0.41,
"learning_rate": 1.3229402200161197e-05,
"loss": 1.2148,
"step": 493
},
{
"epoch": 0.41,
"learning_rate": 1.3203670028449045e-05,
"loss": 1.076,
"step": 494
},
{
"epoch": 0.42,
"learning_rate": 1.3177914195819018e-05,
"loss": 1.0989,
"step": 495
},
{
"epoch": 0.42,
"learning_rate": 1.3152134892492525e-05,
"loss": 1.0703,
"step": 496
},
{
"epoch": 0.42,
"learning_rate": 1.3126332308864328e-05,
"loss": 1.1575,
"step": 497
},
{
"epoch": 0.42,
"learning_rate": 1.310050663550112e-05,
"loss": 1.1597,
"step": 498
},
{
"epoch": 0.42,
"learning_rate": 1.307465806314013e-05,
"loss": 1.1396,
"step": 499
},
{
"epoch": 0.42,
"learning_rate": 1.3048786782687706e-05,
"loss": 1.2354,
"step": 500
},
{
"epoch": 0.42,
"learning_rate": 1.3022892985217906e-05,
"loss": 1.0416,
"step": 501
},
{
"epoch": 0.42,
"learning_rate": 1.2996976861971095e-05,
"loss": 1.0604,
"step": 502
},
{
"epoch": 0.42,
"learning_rate": 1.2971038604352521e-05,
"loss": 1.228,
"step": 503
},
{
"epoch": 0.42,
"learning_rate": 1.2945078403930916e-05,
"loss": 1.0783,
"step": 504
},
{
"epoch": 0.42,
"learning_rate": 1.2919096452437057e-05,
"loss": 1.1292,
"step": 505
},
{
"epoch": 0.42,
"learning_rate": 1.2893092941762375e-05,
"loss": 1.1129,
"step": 506
},
{
"epoch": 0.43,
"learning_rate": 1.2867068063957527e-05,
"loss": 1.1042,
"step": 507
},
{
"epoch": 0.43,
"learning_rate": 1.284102201123098e-05,
"loss": 1.1973,
"step": 508
},
{
"epoch": 0.43,
"learning_rate": 1.2814954975947585e-05,
"loss": 1.0227,
"step": 509
},
{
"epoch": 0.43,
"learning_rate": 1.2788867150627163e-05,
"loss": 1.135,
"step": 510
},
{
"epoch": 0.43,
"learning_rate": 1.276275872794308e-05,
"loss": 1.1289,
"step": 511
},
{
"epoch": 0.43,
"learning_rate": 1.2736629900720832e-05,
"loss": 1.1767,
"step": 512
},
{
"epoch": 0.43,
"learning_rate": 1.27104808619366e-05,
"loss": 1.0454,
"step": 513
},
{
"epoch": 0.43,
"learning_rate": 1.2684311804715854e-05,
"loss": 1.075,
"step": 514
},
{
"epoch": 0.43,
"learning_rate": 1.2658122922331906e-05,
"loss": 1.0904,
"step": 515
},
{
"epoch": 0.43,
"learning_rate": 1.2631914408204481e-05,
"loss": 1.1208,
"step": 516
},
{
"epoch": 0.43,
"learning_rate": 1.2605686455898306e-05,
"loss": 1.0028,
"step": 517
},
{
"epoch": 0.43,
"learning_rate": 1.2579439259121665e-05,
"loss": 1.1063,
"step": 518
},
{
"epoch": 0.44,
"learning_rate": 1.2553173011724972e-05,
"loss": 1.0617,
"step": 519
},
{
"epoch": 0.44,
"learning_rate": 1.2526887907699349e-05,
"loss": 1.0826,
"step": 520
},
{
"epoch": 0.44,
"learning_rate": 1.2500584141175169e-05,
"loss": 1.0823,
"step": 521
},
{
"epoch": 0.44,
"learning_rate": 1.2474261906420655e-05,
"loss": 1.2205,
"step": 522
},
{
"epoch": 0.44,
"learning_rate": 1.2447921397840417e-05,
"loss": 1.0679,
"step": 523
},
{
"epoch": 0.44,
"learning_rate": 1.2421562809974032e-05,
"loss": 0.9495,
"step": 524
},
{
"epoch": 0.44,
"learning_rate": 1.2395186337494602e-05,
"loss": 1.1252,
"step": 525
},
{
"epoch": 0.44,
"learning_rate": 1.2368792175207317e-05,
"loss": 0.9648,
"step": 526
},
{
"epoch": 0.44,
"learning_rate": 1.2342380518048018e-05,
"loss": 1.1101,
"step": 527
},
{
"epoch": 0.44,
"learning_rate": 1.2315951561081754e-05,
"loss": 1.171,
"step": 528
},
{
"epoch": 0.44,
"learning_rate": 1.2289505499501341e-05,
"loss": 1.0684,
"step": 529
},
{
"epoch": 0.44,
"learning_rate": 1.2263042528625928e-05,
"loss": 1.0531,
"step": 530
},
{
"epoch": 0.45,
"learning_rate": 1.223656284389954e-05,
"loss": 1.1063,
"step": 531
},
{
"epoch": 0.45,
"learning_rate": 1.2210066640889648e-05,
"loss": 1.1428,
"step": 532
},
{
"epoch": 0.45,
"learning_rate": 1.2183554115285726e-05,
"loss": 1.0169,
"step": 533
},
{
"epoch": 0.45,
"learning_rate": 1.2157025462897789e-05,
"loss": 1.0298,
"step": 534
},
{
"epoch": 0.45,
"learning_rate": 1.213048087965496e-05,
"loss": 1.1656,
"step": 535
},
{
"epoch": 0.45,
"learning_rate": 1.2103920561604027e-05,
"loss": 0.9962,
"step": 536
},
{
"epoch": 0.45,
"learning_rate": 1.2077344704907984e-05,
"loss": 1.1055,
"step": 537
},
{
"epoch": 0.45,
"learning_rate": 1.2050753505844581e-05,
"loss": 1.105,
"step": 538
},
{
"epoch": 0.45,
"learning_rate": 1.2024147160804891e-05,
"loss": 1.0703,
"step": 539
},
{
"epoch": 0.45,
"learning_rate": 1.1997525866291842e-05,
"loss": 0.9513,
"step": 540
},
{
"epoch": 0.45,
"learning_rate": 1.1970889818918774e-05,
"loss": 1.0886,
"step": 541
},
{
"epoch": 0.45,
"learning_rate": 1.1944239215407981e-05,
"loss": 1.0455,
"step": 542
},
{
"epoch": 0.46,
"learning_rate": 1.191757425258927e-05,
"loss": 1.1027,
"step": 543
},
{
"epoch": 0.46,
"learning_rate": 1.1890895127398497e-05,
"loss": 1.166,
"step": 544
},
{
"epoch": 0.46,
"learning_rate": 1.186420203687611e-05,
"loss": 0.9527,
"step": 545
},
{
"epoch": 0.46,
"learning_rate": 1.1837495178165706e-05,
"loss": 1.0732,
"step": 546
},
{
"epoch": 0.46,
"learning_rate": 1.1810774748512559e-05,
"loss": 1.0596,
"step": 547
},
{
"epoch": 0.46,
"learning_rate": 1.1784040945262185e-05,
"loss": 1.0608,
"step": 548
},
{
"epoch": 0.46,
"learning_rate": 1.1757293965858862e-05,
"loss": 1.059,
"step": 549
},
{
"epoch": 0.46,
"learning_rate": 1.1730534007844186e-05,
"loss": 0.9504,
"step": 550
},
{
"epoch": 0.46,
"learning_rate": 1.1703761268855596e-05,
"loss": 1.0897,
"step": 551
},
{
"epoch": 0.46,
"learning_rate": 1.1676975946624945e-05,
"loss": 1.0546,
"step": 552
},
{
"epoch": 0.46,
"learning_rate": 1.1650178238977004e-05,
"loss": 1.0778,
"step": 553
},
{
"epoch": 0.46,
"learning_rate": 1.1623368343828028e-05,
"loss": 1.0734,
"step": 554
},
{
"epoch": 0.47,
"learning_rate": 1.1596546459184274e-05,
"loss": 1.0447,
"step": 555
},
{
"epoch": 0.47,
"learning_rate": 1.1569712783140559e-05,
"loss": 1.0512,
"step": 556
},
{
"epoch": 0.47,
"learning_rate": 1.1542867513878774e-05,
"loss": 1.0295,
"step": 557
},
{
"epoch": 0.47,
"learning_rate": 1.1516010849666446e-05,
"loss": 1.1233,
"step": 558
},
{
"epoch": 0.47,
"learning_rate": 1.1489142988855248e-05,
"loss": 1.1078,
"step": 559
},
{
"epoch": 0.47,
"learning_rate": 1.1462264129879555e-05,
"loss": 1.0433,
"step": 560
},
{
"epoch": 0.47,
"learning_rate": 1.1435374471254964e-05,
"loss": 0.9486,
"step": 561
},
{
"epoch": 0.47,
"learning_rate": 1.1408474211576834e-05,
"loss": 1.0185,
"step": 562
},
{
"epoch": 0.47,
"learning_rate": 1.1381563549518823e-05,
"loss": 1.0445,
"step": 563
},
{
"epoch": 0.47,
"learning_rate": 1.1354642683831414e-05,
"loss": 1.0032,
"step": 564
},
{
"epoch": 0.47,
"learning_rate": 1.1327711813340447e-05,
"loss": 1.0455,
"step": 565
},
{
"epoch": 0.47,
"learning_rate": 1.1300771136945659e-05,
"loss": 1.0844,
"step": 566
},
{
"epoch": 0.48,
"learning_rate": 1.1273820853619201e-05,
"loss": 1.1341,
"step": 567
},
{
"epoch": 0.48,
"learning_rate": 1.1246861162404184e-05,
"loss": 1.102,
"step": 568
},
{
"epoch": 0.48,
"learning_rate": 1.12198922624132e-05,
"loss": 1.1376,
"step": 569
},
{
"epoch": 0.48,
"learning_rate": 1.1192914352826849e-05,
"loss": 0.9603,
"step": 570
},
{
"epoch": 0.48,
"learning_rate": 1.1165927632892281e-05,
"loss": 1.0845,
"step": 571
},
{
"epoch": 0.48,
"learning_rate": 1.1138932301921703e-05,
"loss": 1.0532,
"step": 572
},
{
"epoch": 0.48,
"learning_rate": 1.1111928559290928e-05,
"loss": 1.0438,
"step": 573
},
{
"epoch": 0.48,
"learning_rate": 1.108491660443789e-05,
"loss": 1.1825,
"step": 574
},
{
"epoch": 0.48,
"learning_rate": 1.1057896636861181e-05,
"loss": 1.1446,
"step": 575
},
{
"epoch": 0.48,
"learning_rate": 1.103086885611856e-05,
"loss": 1.1086,
"step": 576
},
{
"epoch": 0.48,
"learning_rate": 1.1003833461825504e-05,
"loss": 1.0804,
"step": 577
},
{
"epoch": 0.48,
"learning_rate": 1.097679065365371e-05,
"loss": 1.0641,
"step": 578
},
{
"epoch": 0.49,
"learning_rate": 1.0949740631329636e-05,
"loss": 1.1698,
"step": 579
},
{
"epoch": 0.49,
"learning_rate": 1.092268359463302e-05,
"loss": 1.113,
"step": 580
},
{
"epoch": 0.49,
"learning_rate": 1.0895619743395405e-05,
"loss": 1.0982,
"step": 581
},
{
"epoch": 0.49,
"learning_rate": 1.0868549277498661e-05,
"loss": 1.0553,
"step": 582
},
{
"epoch": 0.49,
"learning_rate": 1.0841472396873516e-05,
"loss": 1.0902,
"step": 583
},
{
"epoch": 0.49,
"learning_rate": 1.0814389301498067e-05,
"loss": 0.9517,
"step": 584
},
{
"epoch": 0.49,
"learning_rate": 1.0787300191396314e-05,
"loss": 0.9944,
"step": 585
},
{
"epoch": 0.49,
"learning_rate": 1.0760205266636686e-05,
"loss": 1.0683,
"step": 586
},
{
"epoch": 0.49,
"learning_rate": 1.0733104727330544e-05,
"loss": 1.0865,
"step": 587
},
{
"epoch": 0.49,
"learning_rate": 1.0705998773630722e-05,
"loss": 1.0567,
"step": 588
},
{
"epoch": 0.49,
"learning_rate": 1.0678887605730043e-05,
"loss": 1.0276,
"step": 589
},
{
"epoch": 0.49,
"learning_rate": 1.0651771423859845e-05,
"loss": 1.0733,
"step": 590
},
{
"epoch": 0.5,
"learning_rate": 1.0624650428288486e-05,
"loss": 1.0938,
"step": 591
},
{
"epoch": 0.5,
"learning_rate": 1.059752481931988e-05,
"loss": 0.9869,
"step": 592
},
{
"epoch": 0.5,
"learning_rate": 1.0570394797292015e-05,
"loss": 1.1595,
"step": 593
},
{
"epoch": 0.5,
"learning_rate": 1.054326056257547e-05,
"loss": 1.1582,
"step": 594
},
{
"epoch": 0.5,
"learning_rate": 1.0516122315571944e-05,
"loss": 1.0947,
"step": 595
},
{
"epoch": 0.5,
"learning_rate": 1.0488980256712762e-05,
"loss": 1.0521,
"step": 596
},
{
"epoch": 0.5,
"learning_rate": 1.0461834586457398e-05,
"loss": 1.074,
"step": 597
},
{
"epoch": 0.5,
"learning_rate": 1.0434685505292008e-05,
"loss": 1.1291,
"step": 598
},
{
"epoch": 0.5,
"learning_rate": 1.0407533213727934e-05,
"loss": 1.0596,
"step": 599
},
{
"epoch": 0.5,
"learning_rate": 1.0380377912300231e-05,
"loss": 1.0019,
"step": 600
},
{
"epoch": 0.5,
"learning_rate": 1.0353219801566178e-05,
"loss": 1.2203,
"step": 601
},
{
"epoch": 0.5,
"learning_rate": 1.0326059082103812e-05,
"loss": 1.0156,
"step": 602
},
{
"epoch": 0.51,
"learning_rate": 1.0298895954510426e-05,
"loss": 1.0873,
"step": 603
},
{
"epoch": 0.51,
"learning_rate": 1.0271730619401112e-05,
"loss": 1.1523,
"step": 604
},
{
"epoch": 0.51,
"learning_rate": 1.0244563277407246e-05,
"loss": 0.9182,
"step": 605
},
{
"epoch": 0.51,
"learning_rate": 1.0217394129175047e-05,
"loss": 0.9912,
"step": 606
},
{
"epoch": 0.51,
"learning_rate": 1.019022337536406e-05,
"loss": 1.0356,
"step": 607
},
{
"epoch": 0.51,
"learning_rate": 1.0163051216645693e-05,
"loss": 1.0776,
"step": 608
},
{
"epoch": 0.51,
"learning_rate": 1.0135877853701734e-05,
"loss": 1.0743,
"step": 609
},
{
"epoch": 0.51,
"learning_rate": 1.0108703487222855e-05,
"loss": 1.022,
"step": 610
},
{
"epoch": 0.51,
"learning_rate": 1.0081528317907147e-05,
"loss": 0.992,
"step": 611
},
{
"epoch": 0.51,
"learning_rate": 1.0054352546458634e-05,
"loss": 1.0643,
"step": 612
},
{
"epoch": 0.51,
"learning_rate": 1.0027176373585774e-05,
"loss": 1.0449,
"step": 613
},
{
"epoch": 0.51,
"learning_rate": 1e-05,
"loss": 1.0523,
"step": 614
},
{
"epoch": 0.52,
"learning_rate": 9.972823626414231e-06,
"loss": 1.0675,
"step": 615
},
{
"epoch": 0.52,
"learning_rate": 9.94564745354137e-06,
"loss": 0.9872,
"step": 616
},
{
"epoch": 0.52,
"learning_rate": 9.918471682092855e-06,
"loss": 0.9549,
"step": 617
},
{
"epoch": 0.52,
"learning_rate": 9.891296512777145e-06,
"loss": 1.0482,
"step": 618
},
{
"epoch": 0.52,
"learning_rate": 9.86412214629827e-06,
"loss": 1.0516,
"step": 619
},
{
"epoch": 0.52,
"learning_rate": 9.836948783354308e-06,
"loss": 1.0875,
"step": 620
},
{
"epoch": 0.52,
"learning_rate": 9.809776624635941e-06,
"loss": 1.1198,
"step": 621
},
{
"epoch": 0.52,
"learning_rate": 9.782605870824956e-06,
"loss": 1.024,
"step": 622
},
{
"epoch": 0.52,
"learning_rate": 9.755436722592757e-06,
"loss": 1.223,
"step": 623
},
{
"epoch": 0.52,
"learning_rate": 9.728269380598891e-06,
"loss": 1.1825,
"step": 624
},
{
"epoch": 0.52,
"learning_rate": 9.701104045489576e-06,
"loss": 1.2434,
"step": 625
},
{
"epoch": 0.53,
"learning_rate": 9.67394091789619e-06,
"loss": 1.0738,
"step": 626
},
{
"epoch": 0.53,
"learning_rate": 9.646780198433823e-06,
"loss": 1.0535,
"step": 627
},
{
"epoch": 0.53,
"learning_rate": 9.619622087699774e-06,
"loss": 1.0694,
"step": 628
},
{
"epoch": 0.53,
"learning_rate": 9.59246678627207e-06,
"loss": 1.0112,
"step": 629
},
{
"epoch": 0.53,
"learning_rate": 9.565314494707995e-06,
"loss": 1.055,
"step": 630
},
{
"epoch": 0.53,
"learning_rate": 9.538165413542607e-06,
"loss": 1.0697,
"step": 631
},
{
"epoch": 0.53,
"learning_rate": 9.511019743287242e-06,
"loss": 1.0657,
"step": 632
},
{
"epoch": 0.53,
"learning_rate": 9.483877684428059e-06,
"loss": 1.0955,
"step": 633
},
{
"epoch": 0.53,
"learning_rate": 9.45673943742453e-06,
"loss": 1.1197,
"step": 634
},
{
"epoch": 0.53,
"learning_rate": 9.429605202707989e-06,
"loss": 1.0304,
"step": 635
},
{
"epoch": 0.53,
"learning_rate": 9.402475180680125e-06,
"loss": 0.9521,
"step": 636
},
{
"epoch": 0.53,
"learning_rate": 9.375349571711517e-06,
"loss": 0.9929,
"step": 637
},
{
"epoch": 0.54,
"learning_rate": 9.348228576140159e-06,
"loss": 1.0895,
"step": 638
},
{
"epoch": 0.54,
"learning_rate": 9.321112394269955e-06,
"loss": 1.0061,
"step": 639
},
{
"epoch": 0.54,
"learning_rate": 9.294001226369281e-06,
"loss": 1.0651,
"step": 640
},
{
"epoch": 0.54,
"learning_rate": 9.266895272669461e-06,
"loss": 0.9835,
"step": 641
},
{
"epoch": 0.54,
"learning_rate": 9.239794733363316e-06,
"loss": 1.0125,
"step": 642
},
{
"epoch": 0.54,
"learning_rate": 9.212699808603687e-06,
"loss": 1.1509,
"step": 643
},
{
"epoch": 0.54,
"learning_rate": 9.185610698501938e-06,
"loss": 1.0558,
"step": 644
},
{
"epoch": 0.54,
"learning_rate": 9.158527603126486e-06,
"loss": 0.9462,
"step": 645
},
{
"epoch": 0.54,
"learning_rate": 9.13145072250134e-06,
"loss": 1.0415,
"step": 646
},
{
"epoch": 0.54,
"learning_rate": 9.104380256604595e-06,
"loss": 0.9683,
"step": 647
},
{
"epoch": 0.54,
"learning_rate": 9.07731640536698e-06,
"loss": 0.9727,
"step": 648
},
{
"epoch": 0.54,
"learning_rate": 9.050259368670367e-06,
"loss": 1.1236,
"step": 649
},
{
"epoch": 0.55,
"learning_rate": 9.023209346346293e-06,
"loss": 1.142,
"step": 650
},
{
"epoch": 0.55,
"learning_rate": 8.9961665381745e-06,
"loss": 1.0868,
"step": 651
},
{
"epoch": 0.55,
"learning_rate": 8.969131143881445e-06,
"loss": 0.9866,
"step": 652
},
{
"epoch": 0.55,
"learning_rate": 8.942103363138824e-06,
"loss": 1.0948,
"step": 653
},
{
"epoch": 0.55,
"learning_rate": 8.915083395562114e-06,
"loss": 1.0301,
"step": 654
},
{
"epoch": 0.55,
"learning_rate": 8.888071440709074e-06,
"loss": 1.0685,
"step": 655
},
{
"epoch": 0.55,
"learning_rate": 8.8610676980783e-06,
"loss": 1.0353,
"step": 656
},
{
"epoch": 0.55,
"learning_rate": 8.834072367107724e-06,
"loss": 1.1193,
"step": 657
},
{
"epoch": 0.55,
"learning_rate": 8.807085647173151e-06,
"loss": 1.017,
"step": 658
},
{
"epoch": 0.55,
"learning_rate": 8.780107737586804e-06,
"loss": 1.1112,
"step": 659
},
{
"epoch": 0.55,
"learning_rate": 8.753138837595818e-06,
"loss": 0.9645,
"step": 660
},
{
"epoch": 0.55,
"learning_rate": 8.7261791463808e-06,
"loss": 1.051,
"step": 661
},
{
"epoch": 0.56,
"learning_rate": 8.699228863054345e-06,
"loss": 1.0719,
"step": 662
},
{
"epoch": 0.56,
"learning_rate": 8.672288186659555e-06,
"loss": 0.9992,
"step": 663
},
{
"epoch": 0.56,
"learning_rate": 8.64535731616859e-06,
"loss": 1.1188,
"step": 664
},
{
"epoch": 0.56,
"learning_rate": 8.618436450481182e-06,
"loss": 0.9611,
"step": 665
},
{
"epoch": 0.56,
"learning_rate": 8.591525788423167e-06,
"loss": 1.0776,
"step": 666
},
{
"epoch": 0.56,
"learning_rate": 8.564625528745041e-06,
"loss": 1.1213,
"step": 667
},
{
"epoch": 0.56,
"learning_rate": 8.537735870120447e-06,
"loss": 1.0828,
"step": 668
},
{
"epoch": 0.56,
"learning_rate": 8.510857011144754e-06,
"loss": 1.1018,
"step": 669
},
{
"epoch": 0.56,
"learning_rate": 8.483989150333556e-06,
"loss": 1.0427,
"step": 670
},
{
"epoch": 0.56,
"learning_rate": 8.457132486121228e-06,
"loss": 1.0854,
"step": 671
},
{
"epoch": 0.56,
"learning_rate": 8.430287216859444e-06,
"loss": 0.9942,
"step": 672
},
{
"epoch": 0.56,
"learning_rate": 8.403453540815729e-06,
"loss": 1.0068,
"step": 673
},
{
"epoch": 0.57,
"learning_rate": 8.376631656171974e-06,
"loss": 1.071,
"step": 674
},
{
"epoch": 0.57,
"learning_rate": 8.349821761023e-06,
"loss": 0.9944,
"step": 675
},
{
"epoch": 0.57,
"learning_rate": 8.323024053375057e-06,
"loss": 1.0718,
"step": 676
},
{
"epoch": 0.57,
"learning_rate": 8.296238731144406e-06,
"loss": 1.1308,
"step": 677
},
{
"epoch": 0.57,
"learning_rate": 8.26946599215582e-06,
"loss": 1.05,
"step": 678
},
{
"epoch": 0.57,
"learning_rate": 8.24270603414114e-06,
"loss": 0.9964,
"step": 679
},
{
"epoch": 0.57,
"learning_rate": 8.215959054737817e-06,
"loss": 1.0898,
"step": 680
},
{
"epoch": 0.57,
"learning_rate": 8.189225251487441e-06,
"loss": 0.9617,
"step": 681
},
{
"epoch": 0.57,
"learning_rate": 8.162504821834296e-06,
"loss": 1.0859,
"step": 682
},
{
"epoch": 0.57,
"learning_rate": 8.135797963123894e-06,
"loss": 1.1574,
"step": 683
},
{
"epoch": 0.57,
"learning_rate": 8.109104872601505e-06,
"loss": 1.1689,
"step": 684
},
{
"epoch": 0.57,
"learning_rate": 8.082425747410732e-06,
"loss": 1.1252,
"step": 685
},
{
"epoch": 0.58,
"learning_rate": 8.055760784592024e-06,
"loss": 1.0719,
"step": 686
},
{
"epoch": 0.58,
"learning_rate": 8.02911018108123e-06,
"loss": 1.1428,
"step": 687
},
{
"epoch": 0.58,
"learning_rate": 8.002474133708163e-06,
"loss": 0.9778,
"step": 688
},
{
"epoch": 0.58,
"learning_rate": 7.97585283919511e-06,
"loss": 1.0778,
"step": 689
},
{
"epoch": 0.58,
"learning_rate": 7.94924649415542e-06,
"loss": 1.0562,
"step": 690
},
{
"epoch": 0.58,
"learning_rate": 7.92265529509202e-06,
"loss": 1.0858,
"step": 691
},
{
"epoch": 0.58,
"learning_rate": 7.896079438395974e-06,
"loss": 0.9493,
"step": 692
},
{
"epoch": 0.58,
"learning_rate": 7.869519120345042e-06,
"loss": 1.1294,
"step": 693
},
{
"epoch": 0.58,
"learning_rate": 7.842974537102218e-06,
"loss": 1.0639,
"step": 694
},
{
"epoch": 0.58,
"learning_rate": 7.816445884714277e-06,
"loss": 0.9792,
"step": 695
},
{
"epoch": 0.58,
"learning_rate": 7.789933359110355e-06,
"loss": 1.1122,
"step": 696
},
{
"epoch": 0.58,
"learning_rate": 7.763437156100461e-06,
"loss": 0.9382,
"step": 697
},
{
"epoch": 0.59,
"learning_rate": 7.736957471374075e-06,
"loss": 1.011,
"step": 698
},
{
"epoch": 0.59,
"learning_rate": 7.710494500498662e-06,
"loss": 1.1003,
"step": 699
},
{
"epoch": 0.59,
"learning_rate": 7.684048438918247e-06,
"loss": 1.0887,
"step": 700
},
{
"epoch": 0.59,
"learning_rate": 7.657619481951985e-06,
"loss": 1.067,
"step": 701
},
{
"epoch": 0.59,
"learning_rate": 7.631207824792688e-06,
"loss": 1.0497,
"step": 702
},
{
"epoch": 0.59,
"learning_rate": 7.6048136625054e-06,
"loss": 1.0982,
"step": 703
},
{
"epoch": 0.59,
"learning_rate": 7.578437190025972e-06,
"loss": 1.0662,
"step": 704
},
{
"epoch": 0.59,
"learning_rate": 7.5520786021595846e-06,
"loss": 1.015,
"step": 705
},
{
"epoch": 0.59,
"learning_rate": 7.525738093579349e-06,
"loss": 0.9984,
"step": 706
},
{
"epoch": 0.59,
"learning_rate": 7.499415858824834e-06,
"loss": 0.9834,
"step": 707
},
{
"epoch": 0.59,
"learning_rate": 7.473112092300654e-06,
"loss": 1.0425,
"step": 708
},
{
"epoch": 0.59,
"learning_rate": 7.446826988275029e-06,
"loss": 0.9609,
"step": 709
},
{
"epoch": 0.6,
"learning_rate": 7.420560740878335e-06,
"loss": 1.1199,
"step": 710
},
{
"epoch": 0.6,
"learning_rate": 7.394313544101697e-06,
"loss": 1.1281,
"step": 711
},
{
"epoch": 0.6,
"learning_rate": 7.368085591795522e-06,
"loss": 1.0895,
"step": 712
},
{
"epoch": 0.6,
"learning_rate": 7.341877077668098e-06,
"loss": 1.0241,
"step": 713
},
{
"epoch": 0.6,
"learning_rate": 7.315688195284147e-06,
"loss": 1.0198,
"step": 714
},
{
"epoch": 0.6,
"learning_rate": 7.289519138063403e-06,
"loss": 1.0346,
"step": 715
},
{
"epoch": 0.6,
"learning_rate": 7.263370099279173e-06,
"loss": 1.075,
"step": 716
},
{
"epoch": 0.6,
"learning_rate": 7.237241272056923e-06,
"loss": 1.0962,
"step": 717
},
{
"epoch": 0.6,
"learning_rate": 7.211132849372838e-06,
"loss": 1.0543,
"step": 718
},
{
"epoch": 0.6,
"learning_rate": 7.1850450240524175e-06,
"loss": 1.0309,
"step": 719
},
{
"epoch": 0.6,
"learning_rate": 7.1589779887690235e-06,
"loss": 1.0215,
"step": 720
},
{
"epoch": 0.6,
"learning_rate": 7.132931936042473e-06,
"loss": 0.9455,
"step": 721
},
{
"epoch": 0.61,
"learning_rate": 7.106907058237626e-06,
"loss": 1.0395,
"step": 722
},
{
"epoch": 0.61,
"learning_rate": 7.080903547562949e-06,
"loss": 0.9929,
"step": 723
},
{
"epoch": 0.61,
"learning_rate": 7.0549215960690865e-06,
"loss": 1.0952,
"step": 724
},
{
"epoch": 0.61,
"learning_rate": 7.02896139564748e-06,
"loss": 0.9785,
"step": 725
},
{
"epoch": 0.61,
"learning_rate": 7.003023138028906e-06,
"loss": 1.0301,
"step": 726
},
{
"epoch": 0.61,
"learning_rate": 6.977107014782098e-06,
"loss": 0.9947,
"step": 727
},
{
"epoch": 0.61,
"learning_rate": 6.951213217312301e-06,
"loss": 0.9629,
"step": 728
},
{
"epoch": 0.61,
"learning_rate": 6.925341936859873e-06,
"loss": 1.0344,
"step": 729
},
{
"epoch": 0.61,
"learning_rate": 6.899493364498884e-06,
"loss": 1.0306,
"step": 730
},
{
"epoch": 0.61,
"learning_rate": 6.873667691135672e-06,
"loss": 1.014,
"step": 731
},
{
"epoch": 0.61,
"learning_rate": 6.847865107507478e-06,
"loss": 1.053,
"step": 732
},
{
"epoch": 0.61,
"learning_rate": 6.822085804180985e-06,
"loss": 0.8946,
"step": 733
},
{
"epoch": 0.62,
"learning_rate": 6.796329971550957e-06,
"loss": 1.1176,
"step": 734
},
{
"epoch": 0.62,
"learning_rate": 6.770597799838805e-06,
"loss": 1.0073,
"step": 735
},
{
"epoch": 0.62,
"learning_rate": 6.744889479091197e-06,
"loss": 1.1047,
"step": 736
},
{
"epoch": 0.62,
"learning_rate": 6.719205199178637e-06,
"loss": 1.0084,
"step": 737
},
{
"epoch": 0.62,
"learning_rate": 6.69354514979409e-06,
"loss": 1.0155,
"step": 738
},
{
"epoch": 0.62,
"learning_rate": 6.667909520451546e-06,
"loss": 0.9673,
"step": 739
},
{
"epoch": 0.62,
"learning_rate": 6.642298500484657e-06,
"loss": 0.935,
"step": 740
},
{
"epoch": 0.62,
"learning_rate": 6.616712279045312e-06,
"loss": 1.0023,
"step": 741
},
{
"epoch": 0.62,
"learning_rate": 6.591151045102242e-06,
"loss": 1.0854,
"step": 742
},
{
"epoch": 0.62,
"learning_rate": 6.565614987439648e-06,
"loss": 0.9026,
"step": 743
},
{
"epoch": 0.62,
"learning_rate": 6.540104294655778e-06,
"loss": 0.938,
"step": 744
},
{
"epoch": 0.62,
"learning_rate": 6.514619155161543e-06,
"loss": 1.0195,
"step": 745
},
{
"epoch": 0.63,
"learning_rate": 6.489159757179144e-06,
"loss": 0.9787,
"step": 746
},
{
"epoch": 0.63,
"learning_rate": 6.463726288740646e-06,
"loss": 0.9446,
"step": 747
},
{
"epoch": 0.63,
"learning_rate": 6.438318937686631e-06,
"loss": 1.1951,
"step": 748
},
{
"epoch": 0.63,
"learning_rate": 6.412937891664775e-06,
"loss": 1.0241,
"step": 749
},
{
"epoch": 0.63,
"learning_rate": 6.387583338128471e-06,
"loss": 0.9006,
"step": 750
},
{
"epoch": 0.63,
"learning_rate": 6.362255464335472e-06,
"loss": 0.9233,
"step": 751
},
{
"epoch": 0.63,
"learning_rate": 6.336954457346463e-06,
"loss": 1.081,
"step": 752
},
{
"epoch": 0.63,
"learning_rate": 6.311680504023718e-06,
"loss": 1.0389,
"step": 753
},
{
"epoch": 0.63,
"learning_rate": 6.286433791029698e-06,
"loss": 1.127,
"step": 754
},
{
"epoch": 0.63,
"learning_rate": 6.2612145048256746e-06,
"loss": 1.1039,
"step": 755
},
{
"epoch": 0.63,
"learning_rate": 6.236022831670372e-06,
"loss": 0.9573,
"step": 756
},
{
"epoch": 0.63,
"learning_rate": 6.210858957618557e-06,
"loss": 1.118,
"step": 757
},
{
"epoch": 0.64,
"learning_rate": 6.1857230685196955e-06,
"loss": 1.0596,
"step": 758
},
{
"epoch": 0.64,
"learning_rate": 6.16061535001657e-06,
"loss": 0.9847,
"step": 759
},
{
"epoch": 0.64,
"learning_rate": 6.1355359875438995e-06,
"loss": 0.9658,
"step": 760
},
{
"epoch": 0.64,
"learning_rate": 6.110485166326986e-06,
"loss": 1.1449,
"step": 761
},
{
"epoch": 0.64,
"learning_rate": 6.085463071380333e-06,
"loss": 0.9484,
"step": 762
},
{
"epoch": 0.64,
"learning_rate": 6.060469887506282e-06,
"loss": 1.1344,
"step": 763
},
{
"epoch": 0.64,
"learning_rate": 6.0355057992936575e-06,
"loss": 1.1659,
"step": 764
},
{
"epoch": 0.64,
"learning_rate": 6.0105709911163925e-06,
"loss": 1.0415,
"step": 765
},
{
"epoch": 0.64,
"learning_rate": 5.985665647132164e-06,
"loss": 1.0571,
"step": 766
},
{
"epoch": 0.64,
"learning_rate": 5.960789951281052e-06,
"loss": 1.1086,
"step": 767
},
{
"epoch": 0.64,
"learning_rate": 5.935944087284155e-06,
"loss": 0.9445,
"step": 768
},
{
"epoch": 0.64,
"learning_rate": 5.911128238642259e-06,
"loss": 1.0085,
"step": 769
},
{
"epoch": 0.65,
"learning_rate": 5.886342588634458e-06,
"loss": 0.941,
"step": 770
},
{
"epoch": 0.65,
"learning_rate": 5.861587320316815e-06,
"loss": 1.1245,
"step": 771
},
{
"epoch": 0.65,
"learning_rate": 5.836862616521017e-06,
"loss": 1.1156,
"step": 772
},
{
"epoch": 0.65,
"learning_rate": 5.812168659852998e-06,
"loss": 1.1118,
"step": 773
},
{
"epoch": 0.65,
"learning_rate": 5.7875056326916254e-06,
"loss": 0.9864,
"step": 774
},
{
"epoch": 0.65,
"learning_rate": 5.76287371718732e-06,
"loss": 1.0627,
"step": 775
},
{
"epoch": 0.65,
"learning_rate": 5.738273095260728e-06,
"loss": 1.0345,
"step": 776
},
{
"epoch": 0.65,
"learning_rate": 5.713703948601384e-06,
"loss": 1.0092,
"step": 777
},
{
"epoch": 0.65,
"learning_rate": 5.689166458666348e-06,
"loss": 0.8705,
"step": 778
},
{
"epoch": 0.65,
"learning_rate": 5.66466080667888e-06,
"loss": 1.0813,
"step": 779
},
{
"epoch": 0.65,
"learning_rate": 5.64018717362711e-06,
"loss": 1.0446,
"step": 780
},
{
"epoch": 0.66,
"learning_rate": 5.615745740262671e-06,
"loss": 0.9739,
"step": 781
},
{
"epoch": 0.66,
"learning_rate": 5.591336687099403e-06,
"loss": 1.0484,
"step": 782
},
{
"epoch": 0.66,
"learning_rate": 5.566960194411984e-06,
"loss": 0.973,
"step": 783
},
{
"epoch": 0.66,
"learning_rate": 5.542616442234618e-06,
"loss": 1.093,
"step": 784
},
{
"epoch": 0.66,
"learning_rate": 5.518305610359713e-06,
"loss": 1.0987,
"step": 785
},
{
"epoch": 0.66,
"learning_rate": 5.494027878336529e-06,
"loss": 1.0304,
"step": 786
},
{
"epoch": 0.66,
"learning_rate": 5.469783425469864e-06,
"loss": 0.8761,
"step": 787
},
{
"epoch": 0.66,
"learning_rate": 5.445572430818744e-06,
"loss": 0.9181,
"step": 788
},
{
"epoch": 0.66,
"learning_rate": 5.421395073195072e-06,
"loss": 1.0648,
"step": 789
},
{
"epoch": 0.66,
"learning_rate": 5.397251531162332e-06,
"loss": 1.1156,
"step": 790
},
{
"epoch": 0.66,
"learning_rate": 5.373141983034262e-06,
"loss": 1.1206,
"step": 791
},
{
"epoch": 0.66,
"learning_rate": 5.349066606873525e-06,
"loss": 1.1167,
"step": 792
},
{
"epoch": 0.67,
"learning_rate": 5.3250255804904176e-06,
"loss": 0.9647,
"step": 793
},
{
"epoch": 0.67,
"learning_rate": 5.301019081441532e-06,
"loss": 0.9224,
"step": 794
},
{
"epoch": 0.67,
"learning_rate": 5.277047287028468e-06,
"loss": 1.1204,
"step": 795
},
{
"epoch": 0.67,
"learning_rate": 5.253110374296502e-06,
"loss": 1.0848,
"step": 796
},
{
"epoch": 0.67,
"learning_rate": 5.229208520033291e-06,
"loss": 0.9863,
"step": 797
},
{
"epoch": 0.67,
"learning_rate": 5.205341900767575e-06,
"loss": 1.0207,
"step": 798
},
{
"epoch": 0.67,
"learning_rate": 5.1815106927678545e-06,
"loss": 1.0449,
"step": 799
},
{
"epoch": 0.67,
"learning_rate": 5.157715072041094e-06,
"loss": 1.0775,
"step": 800
},
{
"epoch": 0.67,
"learning_rate": 5.133955214331439e-06,
"loss": 1.0521,
"step": 801
},
{
"epoch": 0.67,
"learning_rate": 5.110231295118891e-06,
"loss": 0.9102,
"step": 802
},
{
"epoch": 0.67,
"learning_rate": 5.0865434896180385e-06,
"loss": 1.2522,
"step": 803
},
{
"epoch": 0.67,
"learning_rate": 5.062891972776739e-06,
"loss": 0.9721,
"step": 804
},
{
"epoch": 0.68,
"learning_rate": 5.039276919274839e-06,
"loss": 1.0027,
"step": 805
},
{
"epoch": 0.68,
"learning_rate": 5.015698503522892e-06,
"loss": 1.116,
"step": 806
},
{
"epoch": 0.68,
"learning_rate": 4.992156899660852e-06,
"loss": 0.9207,
"step": 807
},
{
"epoch": 0.68,
"learning_rate": 4.968652281556794e-06,
"loss": 1.1083,
"step": 808
},
{
"epoch": 0.68,
"learning_rate": 4.945184822805643e-06,
"loss": 0.9862,
"step": 809
},
{
"epoch": 0.68,
"learning_rate": 4.92175469672787e-06,
"loss": 1.061,
"step": 810
},
{
"epoch": 0.68,
"learning_rate": 4.898362076368229e-06,
"loss": 1.1168,
"step": 811
},
{
"epoch": 0.68,
"learning_rate": 4.87500713449448e-06,
"loss": 1.1573,
"step": 812
},
{
"epoch": 0.68,
"learning_rate": 4.851690043596086e-06,
"loss": 0.9872,
"step": 813
},
{
"epoch": 0.68,
"learning_rate": 4.82841097588298e-06,
"loss": 0.9923,
"step": 814
},
{
"epoch": 0.68,
"learning_rate": 4.805170103284258e-06,
"loss": 0.9865,
"step": 815
},
{
"epoch": 0.68,
"learning_rate": 4.781967597446936e-06,
"loss": 1.0246,
"step": 816
},
{
"epoch": 0.69,
"learning_rate": 4.7588036297346575e-06,
"loss": 1.0087,
"step": 817
},
{
"epoch": 0.69,
"learning_rate": 4.7356783712264405e-06,
"loss": 1.0542,
"step": 818
},
{
"epoch": 0.69,
"learning_rate": 4.71259199271543e-06,
"loss": 0.9015,
"step": 819
},
{
"epoch": 0.69,
"learning_rate": 4.6895446647076e-06,
"loss": 0.9257,
"step": 820
},
{
"epoch": 0.69,
"learning_rate": 4.6665365574205236e-06,
"loss": 0.9308,
"step": 821
},
{
"epoch": 0.69,
"learning_rate": 4.6435678407821105e-06,
"loss": 1.0453,
"step": 822
},
{
"epoch": 0.69,
"learning_rate": 4.620638684429337e-06,
"loss": 1.0321,
"step": 823
},
{
"epoch": 0.69,
"learning_rate": 4.5977492577070196e-06,
"loss": 0.8717,
"step": 824
},
{
"epoch": 0.69,
"learning_rate": 4.574899729666535e-06,
"loss": 1.1597,
"step": 825
},
{
"epoch": 0.69,
"learning_rate": 4.552090269064587e-06,
"loss": 0.9731,
"step": 826
},
{
"epoch": 0.69,
"learning_rate": 4.5293210443619716e-06,
"loss": 0.9214,
"step": 827
},
{
"epoch": 0.69,
"learning_rate": 4.506592223722306e-06,
"loss": 1.0562,
"step": 828
},
{
"epoch": 0.7,
"learning_rate": 4.4839039750108015e-06,
"loss": 1.1395,
"step": 829
},
{
"epoch": 0.7,
"learning_rate": 4.461256465793033e-06,
"loss": 0.979,
"step": 830
},
{
"epoch": 0.7,
"learning_rate": 4.438649863333684e-06,
"loss": 1.0556,
"step": 831
},
{
"epoch": 0.7,
"learning_rate": 4.416084334595314e-06,
"loss": 0.9651,
"step": 832
},
{
"epoch": 0.7,
"learning_rate": 4.393560046237143e-06,
"loss": 0.9939,
"step": 833
},
{
"epoch": 0.7,
"learning_rate": 4.371077164613791e-06,
"loss": 1.0127,
"step": 834
},
{
"epoch": 0.7,
"learning_rate": 4.348635855774082e-06,
"loss": 1.0357,
"step": 835
},
{
"epoch": 0.7,
"learning_rate": 4.326236285459789e-06,
"loss": 1.1271,
"step": 836
},
{
"epoch": 0.7,
"learning_rate": 4.303878619104425e-06,
"loss": 1.0574,
"step": 837
},
{
"epoch": 0.7,
"learning_rate": 4.281563021832027e-06,
"loss": 0.8575,
"step": 838
},
{
"epoch": 0.7,
"learning_rate": 4.259289658455914e-06,
"loss": 1.0136,
"step": 839
},
{
"epoch": 0.7,
"learning_rate": 4.237058693477499e-06,
"loss": 0.951,
"step": 840
},
{
"epoch": 0.71,
"learning_rate": 4.214870291085047e-06,
"loss": 1.0942,
"step": 841
},
{
"epoch": 0.71,
"learning_rate": 4.192724615152475e-06,
"loss": 1.0084,
"step": 842
},
{
"epoch": 0.71,
"learning_rate": 4.170621829238152e-06,
"loss": 1.0073,
"step": 843
},
{
"epoch": 0.71,
"learning_rate": 4.1485620965836626e-06,
"loss": 1.0049,
"step": 844
},
{
"epoch": 0.71,
"learning_rate": 4.1265455801126376e-06,
"loss": 1.0503,
"step": 845
},
{
"epoch": 0.71,
"learning_rate": 4.104572442429519e-06,
"loss": 0.9951,
"step": 846
},
{
"epoch": 0.71,
"learning_rate": 4.08264284581837e-06,
"loss": 0.9679,
"step": 847
},
{
"epoch": 0.71,
"learning_rate": 4.060756952241691e-06,
"loss": 1.1065,
"step": 848
},
{
"epoch": 0.71,
"learning_rate": 4.038914923339201e-06,
"loss": 0.9822,
"step": 849
},
{
"epoch": 0.71,
"learning_rate": 4.017116920426652e-06,
"loss": 1.0682,
"step": 850
},
{
"epoch": 0.71,
"learning_rate": 3.995363104494644e-06,
"loss": 1.0281,
"step": 851
},
{
"epoch": 0.71,
"learning_rate": 3.973653636207437e-06,
"loss": 1.0002,
"step": 852
},
{
"epoch": 0.72,
"learning_rate": 3.951988675901744e-06,
"loss": 1.0596,
"step": 853
},
{
"epoch": 0.72,
"learning_rate": 3.930368383585574e-06,
"loss": 1.1168,
"step": 854
},
{
"epoch": 0.72,
"learning_rate": 3.908792918937027e-06,
"loss": 1.1252,
"step": 855
},
{
"epoch": 0.72,
"learning_rate": 3.887262441303139e-06,
"loss": 0.9952,
"step": 856
},
{
"epoch": 0.72,
"learning_rate": 3.865777109698677e-06,
"loss": 1.0558,
"step": 857
},
{
"epoch": 0.72,
"learning_rate": 3.844337082804984e-06,
"loss": 1.0051,
"step": 858
},
{
"epoch": 0.72,
"learning_rate": 3.822942518968809e-06,
"loss": 0.9312,
"step": 859
},
{
"epoch": 0.72,
"learning_rate": 3.801593576201118e-06,
"loss": 1.0311,
"step": 860
},
{
"epoch": 0.72,
"learning_rate": 3.780290412175954e-06,
"loss": 1.032,
"step": 861
},
{
"epoch": 0.72,
"learning_rate": 3.759033184229247e-06,
"loss": 1.149,
"step": 862
},
{
"epoch": 0.72,
"learning_rate": 3.737822049357662e-06,
"loss": 0.9795,
"step": 863
},
{
"epoch": 0.72,
"learning_rate": 3.716657164217454e-06,
"loss": 0.8837,
"step": 864
},
{
"epoch": 0.73,
"learning_rate": 3.6955386851232798e-06,
"loss": 1.0688,
"step": 865
},
{
"epoch": 0.73,
"learning_rate": 3.674466768047078e-06,
"loss": 1.1097,
"step": 866
},
{
"epoch": 0.73,
"learning_rate": 3.653441568616888e-06,
"loss": 0.9773,
"step": 867
},
{
"epoch": 0.73,
"learning_rate": 3.6324632421157147e-06,
"loss": 1.1088,
"step": 868
},
{
"epoch": 0.73,
"learning_rate": 3.6115319434803897e-06,
"loss": 1.016,
"step": 869
},
{
"epoch": 0.73,
"learning_rate": 3.5906478273004053e-06,
"loss": 1.0481,
"step": 870
},
{
"epoch": 0.73,
"learning_rate": 3.569811047816786e-06,
"loss": 1.0176,
"step": 871
},
{
"epoch": 0.73,
"learning_rate": 3.5490217589209554e-06,
"loss": 1.037,
"step": 872
},
{
"epoch": 0.73,
"learning_rate": 3.5282801141535915e-06,
"loss": 0.9981,
"step": 873
},
{
"epoch": 0.73,
"learning_rate": 3.507586266703482e-06,
"loss": 1.003,
"step": 874
},
{
"epoch": 0.73,
"learning_rate": 3.486940369406422e-06,
"loss": 1.0723,
"step": 875
},
{
"epoch": 0.73,
"learning_rate": 3.466342574744047e-06,
"loss": 0.9298,
"step": 876
},
{
"epoch": 0.74,
"learning_rate": 3.4457930348427483e-06,
"loss": 0.9549,
"step": 877
},
{
"epoch": 0.74,
"learning_rate": 3.4252919014725137e-06,
"loss": 1.0966,
"step": 878
},
{
"epoch": 0.74,
"learning_rate": 3.4048393260458225e-06,
"loss": 1.0261,
"step": 879
},
{
"epoch": 0.74,
"learning_rate": 3.3844354596165364e-06,
"loss": 0.9912,
"step": 880
},
{
"epoch": 0.74,
"learning_rate": 3.3640804528787606e-06,
"loss": 0.9125,
"step": 881
},
{
"epoch": 0.74,
"learning_rate": 3.343774456165756e-06,
"loss": 0.9502,
"step": 882
},
{
"epoch": 0.74,
"learning_rate": 3.3235176194488073e-06,
"loss": 1.037,
"step": 883
},
{
"epoch": 0.74,
"learning_rate": 3.303310092336125e-06,
"loss": 0.9492,
"step": 884
},
{
"epoch": 0.74,
"learning_rate": 3.2831520240717494e-06,
"loss": 1.0116,
"step": 885
},
{
"epoch": 0.74,
"learning_rate": 3.2630435635344283e-06,
"loss": 0.9723,
"step": 886
},
{
"epoch": 0.74,
"learning_rate": 3.2429848592365377e-06,
"loss": 1.068,
"step": 887
},
{
"epoch": 0.74,
"learning_rate": 3.2229760593229686e-06,
"loss": 1.0179,
"step": 888
},
{
"epoch": 0.75,
"learning_rate": 3.203017311570039e-06,
"loss": 0.9453,
"step": 889
},
{
"epoch": 0.75,
"learning_rate": 3.183108763384415e-06,
"loss": 0.9429,
"step": 890
},
{
"epoch": 0.75,
"learning_rate": 3.163250561801993e-06,
"loss": 1.0132,
"step": 891
},
{
"epoch": 0.75,
"learning_rate": 3.1434428534868468e-06,
"loss": 1.0406,
"step": 892
},
{
"epoch": 0.75,
"learning_rate": 3.123685784730118e-06,
"loss": 1.1853,
"step": 893
},
{
"epoch": 0.75,
"learning_rate": 3.103979501448956e-06,
"loss": 0.9048,
"step": 894
},
{
"epoch": 0.75,
"learning_rate": 3.0843241491854194e-06,
"loss": 1.0461,
"step": 895
},
{
"epoch": 0.75,
"learning_rate": 3.064719873105424e-06,
"loss": 0.9589,
"step": 896
},
{
"epoch": 0.75,
"learning_rate": 3.045166817997648e-06,
"loss": 0.957,
"step": 897
},
{
"epoch": 0.75,
"learning_rate": 3.0256651282724857e-06,
"loss": 1.0895,
"step": 898
},
{
"epoch": 0.75,
"learning_rate": 3.0062149479609583e-06,
"loss": 1.084,
"step": 899
},
{
"epoch": 0.75,
"learning_rate": 2.986816420713662e-06,
"loss": 1.0945,
"step": 900
},
{
"epoch": 0.76,
"learning_rate": 2.9674696897997156e-06,
"loss": 0.9155,
"step": 901
},
{
"epoch": 0.76,
"learning_rate": 2.948174898105679e-06,
"loss": 1.0602,
"step": 902
},
{
"epoch": 0.76,
"learning_rate": 2.9289321881345257e-06,
"loss": 0.9653,
"step": 903
},
{
"epoch": 0.76,
"learning_rate": 2.909741702004565e-06,
"loss": 1.0404,
"step": 904
},
{
"epoch": 0.76,
"learning_rate": 2.8906035814484037e-06,
"loss": 0.9577,
"step": 905
},
{
"epoch": 0.76,
"learning_rate": 2.87151796781191e-06,
"loss": 1.0365,
"step": 906
},
{
"epoch": 0.76,
"learning_rate": 2.8524850020531448e-06,
"loss": 0.9499,
"step": 907
},
{
"epoch": 0.76,
"learning_rate": 2.833504824741349e-06,
"loss": 1.0653,
"step": 908
},
{
"epoch": 0.76,
"learning_rate": 2.814577576055878e-06,
"loss": 0.9587,
"step": 909
},
{
"epoch": 0.76,
"learning_rate": 2.795703395785184e-06,
"loss": 1.1562,
"step": 910
},
{
"epoch": 0.76,
"learning_rate": 2.776882423325784e-06,
"loss": 1.1083,
"step": 911
},
{
"epoch": 0.76,
"learning_rate": 2.758114797681215e-06,
"loss": 0.9218,
"step": 912
},
{
"epoch": 0.77,
"learning_rate": 2.73940065746103e-06,
"loss": 1.0869,
"step": 913
},
{
"epoch": 0.77,
"learning_rate": 2.7207401408797483e-06,
"loss": 0.9998,
"step": 914
},
{
"epoch": 0.77,
"learning_rate": 2.702133385755862e-06,
"loss": 0.9218,
"step": 915
},
{
"epoch": 0.77,
"learning_rate": 2.6835805295107897e-06,
"loss": 1.0468,
"step": 916
},
{
"epoch": 0.77,
"learning_rate": 2.6650817091678927e-06,
"loss": 0.9748,
"step": 917
},
{
"epoch": 0.77,
"learning_rate": 2.646637061351429e-06,
"loss": 1.0311,
"step": 918
},
{
"epoch": 0.77,
"learning_rate": 2.62824672228558e-06,
"loss": 1.0616,
"step": 919
},
{
"epoch": 0.77,
"learning_rate": 2.6099108277934105e-06,
"loss": 0.9472,
"step": 920
},
{
"epoch": 0.77,
"learning_rate": 2.591629513295887e-06,
"loss": 1.0391,
"step": 921
},
{
"epoch": 0.77,
"learning_rate": 2.5734029138108774e-06,
"loss": 1.0123,
"step": 922
},
{
"epoch": 0.77,
"learning_rate": 2.5552311639521376e-06,
"loss": 1.0171,
"step": 923
},
{
"epoch": 0.78,
"learning_rate": 2.5371143979283395e-06,
"loss": 0.9545,
"step": 924
},
{
"epoch": 0.78,
"learning_rate": 2.5190527495420593e-06,
"loss": 1.014,
"step": 925
},
{
"epoch": 0.78,
"learning_rate": 2.5010463521887997e-06,
"loss": 0.9456,
"step": 926
},
{
"epoch": 0.78,
"learning_rate": 2.4830953388560096e-06,
"loss": 0.9799,
"step": 927
},
{
"epoch": 0.78,
"learning_rate": 2.4651998421220847e-06,
"loss": 0.8907,
"step": 928
},
{
"epoch": 0.78,
"learning_rate": 2.4473599941554115e-06,
"loss": 1.135,
"step": 929
},
{
"epoch": 0.78,
"learning_rate": 2.42957592671337e-06,
"loss": 0.9295,
"step": 930
},
{
"epoch": 0.78,
"learning_rate": 2.411847771141369e-06,
"loss": 0.9934,
"step": 931
},
{
"epoch": 0.78,
"learning_rate": 2.3941756583718846e-06,
"loss": 0.9745,
"step": 932
},
{
"epoch": 0.78,
"learning_rate": 2.3765597189234756e-06,
"loss": 0.9394,
"step": 933
},
{
"epoch": 0.78,
"learning_rate": 2.3590000828998373e-06,
"loss": 0.8665,
"step": 934
},
{
"epoch": 0.78,
"learning_rate": 2.341496879988824e-06,
"loss": 0.9889,
"step": 935
},
{
"epoch": 0.79,
"learning_rate": 2.324050239461507e-06,
"loss": 1.0121,
"step": 936
},
{
"epoch": 0.79,
"learning_rate": 2.306660290171211e-06,
"loss": 1.0178,
"step": 937
},
{
"epoch": 0.79,
"learning_rate": 2.289327160552559e-06,
"loss": 1.1345,
"step": 938
},
{
"epoch": 0.79,
"learning_rate": 2.27205097862053e-06,
"loss": 1.0549,
"step": 939
},
{
"epoch": 0.79,
"learning_rate": 2.2548318719695182e-06,
"loss": 1.0208,
"step": 940
},
{
"epoch": 0.79,
"learning_rate": 2.2376699677723778e-06,
"loss": 0.9378,
"step": 941
},
{
"epoch": 0.79,
"learning_rate": 2.2205653927794903e-06,
"loss": 1.0551,
"step": 942
},
{
"epoch": 0.79,
"learning_rate": 2.203518273317835e-06,
"loss": 0.9825,
"step": 943
},
{
"epoch": 0.79,
"learning_rate": 2.186528735290041e-06,
"loss": 0.8767,
"step": 944
},
{
"epoch": 0.79,
"learning_rate": 2.1695969041734755e-06,
"loss": 1.1121,
"step": 945
},
{
"epoch": 0.79,
"learning_rate": 2.1527229050192998e-06,
"loss": 0.9899,
"step": 946
},
{
"epoch": 0.79,
"learning_rate": 2.135906862451553e-06,
"loss": 1.0901,
"step": 947
},
{
"epoch": 0.8,
"learning_rate": 2.1191489006662415e-06,
"loss": 0.9815,
"step": 948
},
{
"epoch": 0.8,
"learning_rate": 2.1024491434303974e-06,
"loss": 0.9499,
"step": 949
},
{
"epoch": 0.8,
"learning_rate": 2.085807714081195e-06,
"loss": 0.9434,
"step": 950
},
{
"epoch": 0.8,
"learning_rate": 2.069224735525014e-06,
"loss": 0.9371,
"step": 951
},
{
"epoch": 0.8,
"learning_rate": 2.052700330236541e-06,
"loss": 1.0445,
"step": 952
},
{
"epoch": 0.8,
"learning_rate": 2.0362346202578753e-06,
"loss": 0.9939,
"step": 953
},
{
"epoch": 0.8,
"learning_rate": 2.019827727197605e-06,
"loss": 0.9905,
"step": 954
},
{
"epoch": 0.8,
"learning_rate": 2.003479772229937e-06,
"loss": 1.0823,
"step": 955
},
{
"epoch": 0.8,
"learning_rate": 1.9871908760937718e-06,
"loss": 0.8509,
"step": 956
},
{
"epoch": 0.8,
"learning_rate": 1.9709611590918375e-06,
"loss": 0.923,
"step": 957
},
{
"epoch": 0.8,
"learning_rate": 1.9547907410897902e-06,
"loss": 0.9731,
"step": 958
},
{
"epoch": 0.8,
"learning_rate": 1.9386797415153237e-06,
"loss": 0.8809,
"step": 959
},
{
"epoch": 0.81,
"learning_rate": 1.9226282793572927e-06,
"loss": 0.9575,
"step": 960
},
{
"epoch": 0.81,
"learning_rate": 1.9066364731648446e-06,
"loss": 1.0167,
"step": 961
},
{
"epoch": 0.81,
"learning_rate": 1.8907044410465225e-06,
"loss": 1.072,
"step": 962
},
{
"epoch": 0.81,
"learning_rate": 1.8748323006694058e-06,
"loss": 1.0137,
"step": 963
},
{
"epoch": 0.81,
"learning_rate": 1.8590201692582478e-06,
"loss": 0.8644,
"step": 964
},
{
"epoch": 0.81,
"learning_rate": 1.8432681635945905e-06,
"loss": 0.9449,
"step": 965
},
{
"epoch": 0.81,
"learning_rate": 1.8275764000159223e-06,
"loss": 1.0449,
"step": 966
},
{
"epoch": 0.81,
"learning_rate": 1.8119449944148049e-06,
"loss": 1.0895,
"step": 967
},
{
"epoch": 0.81,
"learning_rate": 1.7963740622380199e-06,
"loss": 0.9953,
"step": 968
},
{
"epoch": 0.81,
"learning_rate": 1.7808637184857269e-06,
"loss": 0.976,
"step": 969
},
{
"epoch": 0.81,
"learning_rate": 1.7654140777105954e-06,
"loss": 0.9967,
"step": 970
},
{
"epoch": 0.81,
"learning_rate": 1.7500252540169782e-06,
"loss": 1.0752,
"step": 971
},
{
"epoch": 0.82,
"learning_rate": 1.7346973610600527e-06,
"loss": 0.9982,
"step": 972
},
{
"epoch": 0.82,
"learning_rate": 1.7194305120449895e-06,
"loss": 1.1195,
"step": 973
},
{
"epoch": 0.82,
"learning_rate": 1.7042248197261201e-06,
"loss": 1.0732,
"step": 974
},
{
"epoch": 0.82,
"learning_rate": 1.689080396406092e-06,
"loss": 0.9155,
"step": 975
},
{
"epoch": 0.82,
"learning_rate": 1.673997353935054e-06,
"loss": 1.0223,
"step": 976
},
{
"epoch": 0.82,
"learning_rate": 1.6589758037098135e-06,
"loss": 0.9531,
"step": 977
},
{
"epoch": 0.82,
"learning_rate": 1.6440158566730314e-06,
"loss": 0.966,
"step": 978
},
{
"epoch": 0.82,
"learning_rate": 1.6291176233123918e-06,
"loss": 0.8838,
"step": 979
},
{
"epoch": 0.82,
"learning_rate": 1.6142812136597852e-06,
"loss": 1.0138,
"step": 980
},
{
"epoch": 0.82,
"learning_rate": 1.5995067372904983e-06,
"loss": 1.0851,
"step": 981
},
{
"epoch": 0.82,
"learning_rate": 1.5847943033224134e-06,
"loss": 1.0452,
"step": 982
},
{
"epoch": 0.82,
"learning_rate": 1.5701440204151864e-06,
"loss": 0.9344,
"step": 983
},
{
"epoch": 0.83,
"learning_rate": 1.5555559967694522e-06,
"loss": 0.9664,
"step": 984
},
{
"epoch": 0.83,
"learning_rate": 1.5410303401260352e-06,
"loss": 0.9267,
"step": 985
},
{
"epoch": 0.83,
"learning_rate": 1.526567157765132e-06,
"loss": 0.9689,
"step": 986
},
{
"epoch": 0.83,
"learning_rate": 1.5121665565055422e-06,
"loss": 1.0024,
"step": 987
},
{
"epoch": 0.83,
"learning_rate": 1.4978286427038602e-06,
"loss": 0.978,
"step": 988
},
{
"epoch": 0.83,
"learning_rate": 1.483553522253699e-06,
"loss": 0.9922,
"step": 989
},
{
"epoch": 0.83,
"learning_rate": 1.4693413005849143e-06,
"loss": 1.0123,
"step": 990
},
{
"epoch": 0.83,
"learning_rate": 1.4551920826628119e-06,
"loss": 0.9587,
"step": 991
},
{
"epoch": 0.83,
"learning_rate": 1.4411059729873767e-06,
"loss": 0.9426,
"step": 992
},
{
"epoch": 0.83,
"learning_rate": 1.4270830755925148e-06,
"loss": 0.937,
"step": 993
},
{
"epoch": 0.83,
"learning_rate": 1.41312349404526e-06,
"loss": 1.0221,
"step": 994
},
{
"epoch": 0.83,
"learning_rate": 1.3992273314450345e-06,
"loss": 1.182,
"step": 995
},
{
"epoch": 0.84,
"learning_rate": 1.3853946904228678e-06,
"loss": 0.9292,
"step": 996
},
{
"epoch": 0.84,
"learning_rate": 1.3716256731406485e-06,
"loss": 0.9613,
"step": 997
},
{
"epoch": 0.84,
"learning_rate": 1.357920381290374e-06,
"loss": 1.095,
"step": 998
},
{
"epoch": 0.84,
"learning_rate": 1.344278916093381e-06,
"loss": 0.9372,
"step": 999
},
{
"epoch": 0.84,
"learning_rate": 1.3307013782996237e-06,
"loss": 1.037,
"step": 1000
},
{
"epoch": 0.84,
"learning_rate": 1.3171878681869055e-06,
"loss": 0.8803,
"step": 1001
},
{
"epoch": 0.84,
"learning_rate": 1.3037384855601531e-06,
"loss": 1.0021,
"step": 1002
},
{
"epoch": 0.84,
"learning_rate": 1.2903533297506787e-06,
"loss": 0.9836,
"step": 1003
},
{
"epoch": 0.84,
"learning_rate": 1.277032499615438e-06,
"loss": 1.0138,
"step": 1004
},
{
"epoch": 0.84,
"learning_rate": 1.2637760935363053e-06,
"loss": 1.0184,
"step": 1005
},
{
"epoch": 0.84,
"learning_rate": 1.2505842094193522e-06,
"loss": 0.9866,
"step": 1006
},
{
"epoch": 0.84,
"learning_rate": 1.237456944694111e-06,
"loss": 1.0235,
"step": 1007
},
{
"epoch": 0.85,
"learning_rate": 1.2243943963128735e-06,
"loss": 0.8899,
"step": 1008
},
{
"epoch": 0.85,
"learning_rate": 1.211396660749955e-06,
"loss": 1.0218,
"step": 1009
},
{
"epoch": 0.85,
"learning_rate": 1.1984638340009935e-06,
"loss": 0.9961,
"step": 1010
},
{
"epoch": 0.85,
"learning_rate": 1.1855960115822451e-06,
"loss": 0.9893,
"step": 1011
},
{
"epoch": 0.85,
"learning_rate": 1.1727932885298655e-06,
"loss": 0.9737,
"step": 1012
},
{
"epoch": 0.85,
"learning_rate": 1.1600557593992135e-06,
"loss": 0.9805,
"step": 1013
},
{
"epoch": 0.85,
"learning_rate": 1.14738351826416e-06,
"loss": 1.0718,
"step": 1014
},
{
"epoch": 0.85,
"learning_rate": 1.1347766587163834e-06,
"loss": 1.0577,
"step": 1015
},
{
"epoch": 0.85,
"learning_rate": 1.1222352738646825e-06,
"loss": 0.9835,
"step": 1016
},
{
"epoch": 0.85,
"learning_rate": 1.1097594563342873e-06,
"loss": 1.1131,
"step": 1017
},
{
"epoch": 0.85,
"learning_rate": 1.0973492982661792e-06,
"loss": 0.9929,
"step": 1018
},
{
"epoch": 0.85,
"learning_rate": 1.0850048913164058e-06,
"loss": 1.0831,
"step": 1019
},
{
"epoch": 0.86,
"learning_rate": 1.0727263266554012e-06,
"loss": 0.9017,
"step": 1020
},
{
"epoch": 0.86,
"learning_rate": 1.0605136949673245e-06,
"loss": 0.9709,
"step": 1021
},
{
"epoch": 0.86,
"learning_rate": 1.0483670864493777e-06,
"loss": 1.0269,
"step": 1022
},
{
"epoch": 0.86,
"learning_rate": 1.0362865908111418e-06,
"loss": 0.9862,
"step": 1023
},
{
"epoch": 0.86,
"learning_rate": 1.024272297273925e-06,
"loss": 1.0975,
"step": 1024
},
{
"epoch": 0.86,
"learning_rate": 1.0123242945700884e-06,
"loss": 0.9475,
"step": 1025
},
{
"epoch": 0.86,
"learning_rate": 1.0004426709423976e-06,
"loss": 0.9136,
"step": 1026
},
{
"epoch": 0.86,
"learning_rate": 9.886275141433777e-07,
"loss": 0.9978,
"step": 1027
},
{
"epoch": 0.86,
"learning_rate": 9.7687891143465e-07,
"loss": 0.9688,
"step": 1028
},
{
"epoch": 0.86,
"learning_rate": 9.651969495863055e-07,
"loss": 0.855,
"step": 1029
},
{
"epoch": 0.86,
"learning_rate": 9.535817148762461e-07,
"loss": 1.0286,
"step": 1030
},
{
"epoch": 0.86,
"learning_rate": 9.420332930895582e-07,
"loss": 0.8879,
"step": 1031
},
{
"epoch": 0.87,
"learning_rate": 9.305517695178833e-07,
"loss": 0.9423,
"step": 1032
},
{
"epoch": 0.87,
"learning_rate": 9.19137228958773e-07,
"loss": 1.1645,
"step": 1033
},
{
"epoch": 0.87,
"learning_rate": 9.07789755715075e-07,
"loss": 1.0885,
"step": 1034
},
{
"epoch": 0.87,
"learning_rate": 8.965094335943092e-07,
"loss": 0.9917,
"step": 1035
},
{
"epoch": 0.87,
"learning_rate": 8.85296345908041e-07,
"loss": 0.9441,
"step": 1036
},
{
"epoch": 0.87,
"learning_rate": 8.741505754712786e-07,
"loss": 0.939,
"step": 1037
},
{
"epoch": 0.87,
"learning_rate": 8.630722046018458e-07,
"loss": 0.9883,
"step": 1038
},
{
"epoch": 0.87,
"learning_rate": 8.520613151197899e-07,
"loss": 1.0169,
"step": 1039
},
{
"epoch": 0.87,
"learning_rate": 8.411179883467668e-07,
"loss": 0.9508,
"step": 1040
},
{
"epoch": 0.87,
"learning_rate": 8.30242305105442e-07,
"loss": 0.9238,
"step": 1041
},
{
"epoch": 0.87,
"learning_rate": 8.194343457188991e-07,
"loss": 0.8847,
"step": 1042
},
{
"epoch": 0.87,
"learning_rate": 8.086941900100387e-07,
"loss": 0.9735,
"step": 1043
},
{
"epoch": 0.88,
"learning_rate": 7.980219173009929e-07,
"loss": 0.9614,
"step": 1044
},
{
"epoch": 0.88,
"learning_rate": 7.87417606412545e-07,
"loss": 0.9477,
"step": 1045
},
{
"epoch": 0.88,
"learning_rate": 7.76881335663534e-07,
"loss": 0.9325,
"step": 1046
},
{
"epoch": 0.88,
"learning_rate": 7.664131828702881e-07,
"loss": 0.9876,
"step": 1047
},
{
"epoch": 0.88,
"learning_rate": 7.560132253460484e-07,
"loss": 1.0217,
"step": 1048
},
{
"epoch": 0.88,
"learning_rate": 7.45681539900388e-07,
"loss": 0.8251,
"step": 1049
},
{
"epoch": 0.88,
"learning_rate": 7.354182028386591e-07,
"loss": 0.9359,
"step": 1050
},
{
"epoch": 0.88,
"learning_rate": 7.252232899614198e-07,
"loss": 0.9102,
"step": 1051
},
{
"epoch": 0.88,
"learning_rate": 7.150968765638744e-07,
"loss": 0.9019,
"step": 1052
},
{
"epoch": 0.88,
"learning_rate": 7.050390374353244e-07,
"loss": 0.9941,
"step": 1053
},
{
"epoch": 0.88,
"learning_rate": 6.950498468586075e-07,
"loss": 0.9471,
"step": 1054
},
{
"epoch": 0.88,
"learning_rate": 6.851293786095537e-07,
"loss": 1.0083,
"step": 1055
},
{
"epoch": 0.89,
"learning_rate": 6.752777059564431e-07,
"loss": 0.8577,
"step": 1056
},
{
"epoch": 0.89,
"learning_rate": 6.654949016594559e-07,
"loss": 0.9752,
"step": 1057
},
{
"epoch": 0.89,
"learning_rate": 6.557810379701446e-07,
"loss": 0.9655,
"step": 1058
},
{
"epoch": 0.89,
"learning_rate": 6.461361866308957e-07,
"loss": 0.9066,
"step": 1059
},
{
"epoch": 0.89,
"learning_rate": 6.365604188743979e-07,
"loss": 1.0098,
"step": 1060
},
{
"epoch": 0.89,
"learning_rate": 6.270538054231235e-07,
"loss": 1.0067,
"step": 1061
},
{
"epoch": 0.89,
"learning_rate": 6.176164164887932e-07,
"loss": 1.0514,
"step": 1062
},
{
"epoch": 0.89,
"learning_rate": 6.082483217718737e-07,
"loss": 0.9625,
"step": 1063
},
{
"epoch": 0.89,
"learning_rate": 5.989495904610509e-07,
"loss": 1.0184,
"step": 1064
},
{
"epoch": 0.89,
"learning_rate": 5.897202912327193e-07,
"loss": 1.0366,
"step": 1065
},
{
"epoch": 0.89,
"learning_rate": 5.805604922504859e-07,
"loss": 1.0617,
"step": 1066
},
{
"epoch": 0.89,
"learning_rate": 5.714702611646528e-07,
"loss": 0.9393,
"step": 1067
},
{
"epoch": 0.9,
"learning_rate": 5.624496651117251e-07,
"loss": 1.1327,
"step": 1068
},
{
"epoch": 0.9,
"learning_rate": 5.534987707139161e-07,
"loss": 1.0187,
"step": 1069
},
{
"epoch": 0.9,
"learning_rate": 5.446176440786488e-07,
"loss": 0.9388,
"step": 1070
},
{
"epoch": 0.9,
"learning_rate": 5.358063507980771e-07,
"loss": 0.93,
"step": 1071
},
{
"epoch": 0.9,
"learning_rate": 5.270649559485908e-07,
"loss": 0.9258,
"step": 1072
},
{
"epoch": 0.9,
"learning_rate": 5.183935240903415e-07,
"loss": 1.0145,
"step": 1073
},
{
"epoch": 0.9,
"learning_rate": 5.097921192667688e-07,
"loss": 1.0403,
"step": 1074
},
{
"epoch": 0.9,
"learning_rate": 5.012608050041179e-07,
"loss": 0.9184,
"step": 1075
},
{
"epoch": 0.9,
"learning_rate": 4.927996443109751e-07,
"loss": 1.0223,
"step": 1076
},
{
"epoch": 0.9,
"learning_rate": 4.844086996778085e-07,
"loss": 1.036,
"step": 1077
},
{
"epoch": 0.9,
"learning_rate": 4.760880330764939e-07,
"loss": 0.926,
"step": 1078
},
{
"epoch": 0.91,
"learning_rate": 4.678377059598693e-07,
"loss": 0.9898,
"step": 1079
},
{
"epoch": 0.91,
"learning_rate": 4.5965777926127554e-07,
"loss": 1.0136,
"step": 1080
},
{
"epoch": 0.91,
"learning_rate": 4.515483133941012e-07,
"loss": 0.9358,
"step": 1081
},
{
"epoch": 0.91,
"learning_rate": 4.4350936825134804e-07,
"loss": 0.9285,
"step": 1082
},
{
"epoch": 0.91,
"learning_rate": 4.3554100320517767e-07,
"loss": 0.9766,
"step": 1083
},
{
"epoch": 0.91,
"learning_rate": 4.27643277106482e-07,
"loss": 0.9281,
"step": 1084
},
{
"epoch": 0.91,
"learning_rate": 4.1981624828443923e-07,
"loss": 0.9481,
"step": 1085
},
{
"epoch": 0.91,
"learning_rate": 4.120599745460918e-07,
"loss": 0.9562,
"step": 1086
},
{
"epoch": 0.91,
"learning_rate": 4.043745131759169e-07,
"loss": 0.9552,
"step": 1087
},
{
"epoch": 0.91,
"learning_rate": 3.9675992093539674e-07,
"loss": 0.9501,
"step": 1088
},
{
"epoch": 0.91,
"learning_rate": 3.892162540626099e-07,
"loss": 0.9911,
"step": 1089
},
{
"epoch": 0.91,
"learning_rate": 3.817435682718096e-07,
"loss": 0.9654,
"step": 1090
},
{
"epoch": 0.92,
"learning_rate": 3.7434191875301284e-07,
"loss": 0.9835,
"step": 1091
},
{
"epoch": 0.92,
"learning_rate": 3.670113601715941e-07,
"loss": 1.0431,
"step": 1092
},
{
"epoch": 0.92,
"learning_rate": 3.5975194666788224e-07,
"loss": 1.0309,
"step": 1093
},
{
"epoch": 0.92,
"learning_rate": 3.525637318567554e-07,
"loss": 0.9364,
"step": 1094
},
{
"epoch": 0.92,
"learning_rate": 3.4544676882725334e-07,
"loss": 0.9987,
"step": 1095
},
{
"epoch": 0.92,
"learning_rate": 3.3840111014218027e-07,
"loss": 1.07,
"step": 1096
},
{
"epoch": 0.92,
"learning_rate": 3.3142680783771363e-07,
"loss": 1.0139,
"step": 1097
},
{
"epoch": 0.92,
"learning_rate": 3.245239134230305e-07,
"loss": 1.0289,
"step": 1098
},
{
"epoch": 0.92,
"learning_rate": 3.1769247787991177e-07,
"loss": 1.019,
"step": 1099
},
{
"epoch": 0.92,
"learning_rate": 3.109325516623818e-07,
"loss": 0.8893,
"step": 1100
},
{
"epoch": 0.92,
"learning_rate": 3.042441846963251e-07,
"loss": 0.9983,
"step": 1101
},
{
"epoch": 0.92,
"learning_rate": 2.976274263791179e-07,
"loss": 0.9806,
"step": 1102
},
{
"epoch": 0.93,
"learning_rate": 2.9108232557927164e-07,
"loss": 1.0431,
"step": 1103
},
{
"epoch": 0.93,
"learning_rate": 2.8460893063606e-07,
"loss": 0.9457,
"step": 1104
},
{
"epoch": 0.93,
"learning_rate": 2.7820728935917364e-07,
"loss": 1.0079,
"step": 1105
},
{
"epoch": 0.93,
"learning_rate": 2.7187744902835933e-07,
"loss": 0.9633,
"step": 1106
},
{
"epoch": 0.93,
"learning_rate": 2.656194563930714e-07,
"loss": 0.971,
"step": 1107
},
{
"epoch": 0.93,
"learning_rate": 2.594333576721331e-07,
"loss": 1.0131,
"step": 1108
},
{
"epoch": 0.93,
"learning_rate": 2.5331919855338674e-07,
"loss": 0.9247,
"step": 1109
},
{
"epoch": 0.93,
"learning_rate": 2.4727702419335865e-07,
"loss": 0.9996,
"step": 1110
},
{
"epoch": 0.93,
"learning_rate": 2.4130687921693264e-07,
"loss": 0.945,
"step": 1111
},
{
"epoch": 0.93,
"learning_rate": 2.3540880771700802e-07,
"loss": 1.0117,
"step": 1112
},
{
"epoch": 0.93,
"learning_rate": 2.295828532541855e-07,
"loss": 1.0389,
"step": 1113
},
{
"epoch": 0.93,
"learning_rate": 2.2382905885643846e-07,
"loss": 0.9601,
"step": 1114
},
{
"epoch": 0.94,
"learning_rate": 2.1814746701879775e-07,
"loss": 0.9107,
"step": 1115
},
{
"epoch": 0.94,
"learning_rate": 2.125381197030374e-07,
"loss": 0.9632,
"step": 1116
},
{
"epoch": 0.94,
"learning_rate": 2.0700105833736495e-07,
"loss": 0.9383,
"step": 1117
},
{
"epoch": 0.94,
"learning_rate": 2.01536323816115e-07,
"loss": 0.9638,
"step": 1118
},
{
"epoch": 0.94,
"learning_rate": 1.9614395649944605e-07,
"loss": 0.9566,
"step": 1119
},
{
"epoch": 0.94,
"learning_rate": 1.9082399621304758e-07,
"loss": 1.1037,
"step": 1120
},
{
"epoch": 0.94,
"learning_rate": 1.8557648224783898e-07,
"loss": 0.983,
"step": 1121
},
{
"epoch": 0.94,
"learning_rate": 1.8040145335968317e-07,
"loss": 1.0316,
"step": 1122
},
{
"epoch": 0.94,
"learning_rate": 1.7529894776909917e-07,
"loss": 1.0653,
"step": 1123
},
{
"epoch": 0.94,
"learning_rate": 1.7026900316098217e-07,
"loss": 1.0403,
"step": 1124
},
{
"epoch": 0.94,
"learning_rate": 1.653116566843227e-07,
"loss": 0.9818,
"step": 1125
},
{
"epoch": 0.94,
"learning_rate": 1.6042694495193024e-07,
"loss": 0.9345,
"step": 1126
},
{
"epoch": 0.95,
"learning_rate": 1.5561490404016998e-07,
"loss": 0.9756,
"step": 1127
},
{
"epoch": 0.95,
"learning_rate": 1.5087556948868876e-07,
"loss": 0.9944,
"step": 1128
},
{
"epoch": 0.95,
"learning_rate": 1.462089763001584e-07,
"loss": 0.9678,
"step": 1129
},
{
"epoch": 0.95,
"learning_rate": 1.4161515894001165e-07,
"loss": 0.9121,
"step": 1130
},
{
"epoch": 0.95,
"learning_rate": 1.3709415133619232e-07,
"loss": 0.928,
"step": 1131
},
{
"epoch": 0.95,
"learning_rate": 1.3264598687890207e-07,
"loss": 0.9385,
"step": 1132
},
{
"epoch": 0.95,
"learning_rate": 1.2827069842035412e-07,
"loss": 1.0498,
"step": 1133
},
{
"epoch": 0.95,
"learning_rate": 1.2396831827453437e-07,
"loss": 0.9499,
"step": 1134
},
{
"epoch": 0.95,
"learning_rate": 1.197388782169573e-07,
"loss": 0.919,
"step": 1135
},
{
"epoch": 0.95,
"learning_rate": 1.1558240948443045e-07,
"loss": 0.9192,
"step": 1136
},
{
"epoch": 0.95,
"learning_rate": 1.1149894277483253e-07,
"loss": 0.9782,
"step": 1137
},
{
"epoch": 0.95,
"learning_rate": 1.0748850824687795e-07,
"loss": 0.964,
"step": 1138
},
{
"epoch": 0.96,
"learning_rate": 1.0355113551989593e-07,
"loss": 1.0389,
"step": 1139
},
{
"epoch": 0.96,
"learning_rate": 9.96868536736162e-08,
"loss": 0.9545,
"step": 1140
},
{
"epoch": 0.96,
"learning_rate": 9.589569124794918e-08,
"loss": 0.9763,
"step": 1141
},
{
"epoch": 0.96,
"learning_rate": 9.217767624277396e-08,
"loss": 0.9615,
"step": 1142
},
{
"epoch": 0.96,
"learning_rate": 8.853283611774177e-08,
"loss": 0.975,
"step": 1143
},
{
"epoch": 0.96,
"learning_rate": 8.496119779205724e-08,
"loss": 0.9582,
"step": 1144
},
{
"epoch": 0.96,
"learning_rate": 8.146278764429638e-08,
"loss": 1.0936,
"step": 1145
},
{
"epoch": 0.96,
"learning_rate": 7.803763151219779e-08,
"loss": 0.9458,
"step": 1146
},
{
"epoch": 0.96,
"learning_rate": 7.468575469247841e-08,
"loss": 1.0109,
"step": 1147
},
{
"epoch": 0.96,
"learning_rate": 7.140718194065033e-08,
"loss": 0.9568,
"step": 1148
},
{
"epoch": 0.96,
"learning_rate": 6.82019374708287e-08,
"loss": 0.9067,
"step": 1149
},
{
"epoch": 0.96,
"learning_rate": 6.507004495555968e-08,
"loss": 0.9071,
"step": 1150
}
],
"max_steps": 1192,
"num_train_epochs": 1,
"total_flos": 478109396893696.0,
"trial_name": null,
"trial_params": null
}