llama-2-7b-chat-lyre-lm / trainer_state.json
evanfrick's picture
added model
802a40f
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 4.996907854050711,
"eval_steps": 500,
"global_step": 505,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 1.25e-06,
"loss": 1.9805,
"step": 1
},
{
"epoch": 0.02,
"learning_rate": 2.5e-06,
"loss": 1.9742,
"step": 2
},
{
"epoch": 0.03,
"learning_rate": 3.7500000000000005e-06,
"loss": 1.9398,
"step": 3
},
{
"epoch": 0.04,
"learning_rate": 5e-06,
"loss": 1.8268,
"step": 4
},
{
"epoch": 0.05,
"learning_rate": 6.25e-06,
"loss": 1.9396,
"step": 5
},
{
"epoch": 0.06,
"learning_rate": 7.500000000000001e-06,
"loss": 1.8366,
"step": 6
},
{
"epoch": 0.07,
"learning_rate": 8.750000000000001e-06,
"loss": 1.7488,
"step": 7
},
{
"epoch": 0.08,
"learning_rate": 1e-05,
"loss": 1.8458,
"step": 8
},
{
"epoch": 0.09,
"learning_rate": 1.125e-05,
"loss": 1.818,
"step": 9
},
{
"epoch": 0.1,
"learning_rate": 1.25e-05,
"loss": 1.7825,
"step": 10
},
{
"epoch": 0.11,
"learning_rate": 1.375e-05,
"loss": 1.8167,
"step": 11
},
{
"epoch": 0.12,
"learning_rate": 1.5000000000000002e-05,
"loss": 1.6685,
"step": 12
},
{
"epoch": 0.13,
"learning_rate": 1.6250000000000002e-05,
"loss": 1.6598,
"step": 13
},
{
"epoch": 0.14,
"learning_rate": 1.7500000000000002e-05,
"loss": 1.6377,
"step": 14
},
{
"epoch": 0.15,
"learning_rate": 1.8750000000000002e-05,
"loss": 1.7032,
"step": 15
},
{
"epoch": 0.16,
"learning_rate": 2e-05,
"loss": 1.7164,
"step": 16
},
{
"epoch": 0.17,
"learning_rate": 1.9999793628111833e-05,
"loss": 1.6622,
"step": 17
},
{
"epoch": 0.18,
"learning_rate": 1.9999174520965194e-05,
"loss": 1.6692,
"step": 18
},
{
"epoch": 0.19,
"learning_rate": 1.999814270411335e-05,
"loss": 1.6883,
"step": 19
},
{
"epoch": 0.2,
"learning_rate": 1.99966982201439e-05,
"loss": 1.698,
"step": 20
},
{
"epoch": 0.21,
"learning_rate": 1.999484112867702e-05,
"loss": 1.7371,
"step": 21
},
{
"epoch": 0.22,
"learning_rate": 1.9992571506362997e-05,
"loss": 1.6785,
"step": 22
},
{
"epoch": 0.23,
"learning_rate": 1.9989889446879092e-05,
"loss": 1.6484,
"step": 23
},
{
"epoch": 0.24,
"learning_rate": 1.9986795060925636e-05,
"loss": 1.6336,
"step": 24
},
{
"epoch": 0.25,
"learning_rate": 1.9983288476221482e-05,
"loss": 1.5657,
"step": 25
},
{
"epoch": 0.26,
"learning_rate": 1.997936983749873e-05,
"loss": 1.6443,
"step": 26
},
{
"epoch": 0.27,
"learning_rate": 1.997503930649676e-05,
"loss": 1.6247,
"step": 27
},
{
"epoch": 0.28,
"learning_rate": 1.9970297061955533e-05,
"loss": 1.5329,
"step": 28
},
{
"epoch": 0.29,
"learning_rate": 1.9965143299608253e-05,
"loss": 1.6595,
"step": 29
},
{
"epoch": 0.3,
"learning_rate": 1.995957823217325e-05,
"loss": 1.4952,
"step": 30
},
{
"epoch": 0.31,
"learning_rate": 1.9953602089345215e-05,
"loss": 1.5722,
"step": 31
},
{
"epoch": 0.32,
"learning_rate": 1.9947215117785727e-05,
"loss": 1.5964,
"step": 32
},
{
"epoch": 0.33,
"learning_rate": 1.9940417581113062e-05,
"loss": 1.6671,
"step": 33
},
{
"epoch": 0.34,
"learning_rate": 1.9933209759891318e-05,
"loss": 1.6731,
"step": 34
},
{
"epoch": 0.35,
"learning_rate": 1.9925591951618822e-05,
"loss": 1.5952,
"step": 35
},
{
"epoch": 0.36,
"learning_rate": 1.9917564470715876e-05,
"loss": 1.6384,
"step": 36
},
{
"epoch": 0.37,
"learning_rate": 1.9909127648511758e-05,
"loss": 1.5503,
"step": 37
},
{
"epoch": 0.38,
"learning_rate": 1.990028183323105e-05,
"loss": 1.731,
"step": 38
},
{
"epoch": 0.39,
"learning_rate": 1.989102738997928e-05,
"loss": 1.68,
"step": 39
},
{
"epoch": 0.4,
"learning_rate": 1.9881364700727827e-05,
"loss": 1.6194,
"step": 40
},
{
"epoch": 0.41,
"learning_rate": 1.9871294164298175e-05,
"loss": 1.6532,
"step": 41
},
{
"epoch": 0.42,
"learning_rate": 1.986081619634545e-05,
"loss": 1.7531,
"step": 42
},
{
"epoch": 0.43,
"learning_rate": 1.9849931229341258e-05,
"loss": 1.6211,
"step": 43
},
{
"epoch": 0.44,
"learning_rate": 1.9838639712555842e-05,
"loss": 1.6807,
"step": 44
},
{
"epoch": 0.45,
"learning_rate": 1.982694211203952e-05,
"loss": 1.7175,
"step": 45
},
{
"epoch": 0.46,
"learning_rate": 1.981483891060348e-05,
"loss": 1.6584,
"step": 46
},
{
"epoch": 0.47,
"learning_rate": 1.9802330607799832e-05,
"loss": 1.5696,
"step": 47
},
{
"epoch": 0.47,
"learning_rate": 1.978941771990098e-05,
"loss": 1.6854,
"step": 48
},
{
"epoch": 0.48,
"learning_rate": 1.9776100779878344e-05,
"loss": 1.5761,
"step": 49
},
{
"epoch": 0.49,
"learning_rate": 1.976238033738033e-05,
"loss": 1.6526,
"step": 50
},
{
"epoch": 0.5,
"learning_rate": 1.9748256958709666e-05,
"loss": 1.5919,
"step": 51
},
{
"epoch": 0.51,
"learning_rate": 1.9733731226800016e-05,
"loss": 1.5532,
"step": 52
},
{
"epoch": 0.52,
"learning_rate": 1.9718803741191918e-05,
"loss": 1.6093,
"step": 53
},
{
"epoch": 0.53,
"learning_rate": 1.970347511800806e-05,
"loss": 1.6669,
"step": 54
},
{
"epoch": 0.54,
"learning_rate": 1.9687745989927823e-05,
"loss": 1.594,
"step": 55
},
{
"epoch": 0.55,
"learning_rate": 1.967161700616117e-05,
"loss": 1.5944,
"step": 56
},
{
"epoch": 0.56,
"learning_rate": 1.965508883242188e-05,
"loss": 1.6027,
"step": 57
},
{
"epoch": 0.57,
"learning_rate": 1.9638162150900028e-05,
"loss": 1.5561,
"step": 58
},
{
"epoch": 0.58,
"learning_rate": 1.9620837660233866e-05,
"loss": 1.6239,
"step": 59
},
{
"epoch": 0.59,
"learning_rate": 1.960311607548096e-05,
"loss": 1.6061,
"step": 60
},
{
"epoch": 0.6,
"learning_rate": 1.9584998128088686e-05,
"loss": 1.5782,
"step": 61
},
{
"epoch": 0.61,
"learning_rate": 1.9566484565864056e-05,
"loss": 1.7248,
"step": 62
},
{
"epoch": 0.62,
"learning_rate": 1.954757615294283e-05,
"loss": 1.5613,
"step": 63
},
{
"epoch": 0.63,
"learning_rate": 1.9528273669757974e-05,
"loss": 1.7137,
"step": 64
},
{
"epoch": 0.64,
"learning_rate": 1.9508577913007475e-05,
"loss": 1.5516,
"step": 65
},
{
"epoch": 0.65,
"learning_rate": 1.9488489695621432e-05,
"loss": 1.5342,
"step": 66
},
{
"epoch": 0.66,
"learning_rate": 1.9468009846728515e-05,
"loss": 1.519,
"step": 67
},
{
"epoch": 0.67,
"learning_rate": 1.944713921162174e-05,
"loss": 1.5515,
"step": 68
},
{
"epoch": 0.68,
"learning_rate": 1.942587865172359e-05,
"loss": 1.5703,
"step": 69
},
{
"epoch": 0.69,
"learning_rate": 1.9404229044550432e-05,
"loss": 1.5526,
"step": 70
},
{
"epoch": 0.7,
"learning_rate": 1.9382191283676336e-05,
"loss": 1.5859,
"step": 71
},
{
"epoch": 0.71,
"learning_rate": 1.9359766278696165e-05,
"loss": 1.5213,
"step": 72
},
{
"epoch": 0.72,
"learning_rate": 1.9336954955188042e-05,
"loss": 1.6142,
"step": 73
},
{
"epoch": 0.73,
"learning_rate": 1.9313758254675143e-05,
"loss": 1.5944,
"step": 74
},
{
"epoch": 0.74,
"learning_rate": 1.929017713458685e-05,
"loss": 1.4959,
"step": 75
},
{
"epoch": 0.75,
"learning_rate": 1.9266212568219223e-05,
"loss": 1.494,
"step": 76
},
{
"epoch": 0.76,
"learning_rate": 1.9241865544694817e-05,
"loss": 1.5576,
"step": 77
},
{
"epoch": 0.77,
"learning_rate": 1.9217137068921875e-05,
"loss": 1.576,
"step": 78
},
{
"epoch": 0.78,
"learning_rate": 1.9192028161552848e-05,
"loss": 1.6083,
"step": 79
},
{
"epoch": 0.79,
"learning_rate": 1.9166539858942258e-05,
"loss": 1.5894,
"step": 80
},
{
"epoch": 0.8,
"learning_rate": 1.9140673213103932e-05,
"loss": 1.678,
"step": 81
},
{
"epoch": 0.81,
"learning_rate": 1.9114429291667583e-05,
"loss": 1.5729,
"step": 82
},
{
"epoch": 0.82,
"learning_rate": 1.908780917783473e-05,
"loss": 1.6561,
"step": 83
},
{
"epoch": 0.83,
"learning_rate": 1.906081397033401e-05,
"loss": 1.618,
"step": 84
},
{
"epoch": 0.84,
"learning_rate": 1.9033444783375806e-05,
"loss": 1.5488,
"step": 85
},
{
"epoch": 0.85,
"learning_rate": 1.9005702746606274e-05,
"loss": 1.5514,
"step": 86
},
{
"epoch": 0.86,
"learning_rate": 1.8977589005060723e-05,
"loss": 1.47,
"step": 87
},
{
"epoch": 0.87,
"learning_rate": 1.8949104719116334e-05,
"loss": 1.5237,
"step": 88
},
{
"epoch": 0.88,
"learning_rate": 1.8920251064444284e-05,
"loss": 1.5974,
"step": 89
},
{
"epoch": 0.89,
"learning_rate": 1.8891029231961208e-05,
"loss": 1.5083,
"step": 90
},
{
"epoch": 0.9,
"learning_rate": 1.886144042778006e-05,
"loss": 1.6104,
"step": 91
},
{
"epoch": 0.91,
"learning_rate": 1.8831485873160312e-05,
"loss": 1.5195,
"step": 92
},
{
"epoch": 0.92,
"learning_rate": 1.880116680445757e-05,
"loss": 1.619,
"step": 93
},
{
"epoch": 0.93,
"learning_rate": 1.8770484473072518e-05,
"loss": 1.5151,
"step": 94
},
{
"epoch": 0.94,
"learning_rate": 1.8739440145399295e-05,
"loss": 1.6652,
"step": 95
},
{
"epoch": 0.95,
"learning_rate": 1.8708035102773198e-05,
"loss": 1.5574,
"step": 96
},
{
"epoch": 0.96,
"learning_rate": 1.8676270641417824e-05,
"loss": 1.6027,
"step": 97
},
{
"epoch": 0.97,
"learning_rate": 1.864414807239154e-05,
"loss": 1.6076,
"step": 98
},
{
"epoch": 0.98,
"learning_rate": 1.861166872153339e-05,
"loss": 1.65,
"step": 99
},
{
"epoch": 0.99,
"learning_rate": 1.857883392940837e-05,
"loss": 1.5535,
"step": 100
},
{
"epoch": 1.0,
"learning_rate": 1.8545645051252094e-05,
"loss": 1.4654,
"step": 101
},
{
"epoch": 1.01,
"learning_rate": 1.851210345691484e-05,
"loss": 1.6053,
"step": 102
},
{
"epoch": 1.02,
"learning_rate": 1.847821053080505e-05,
"loss": 1.5897,
"step": 103
},
{
"epoch": 1.03,
"learning_rate": 1.844396767183215e-05,
"loss": 1.5679,
"step": 104
},
{
"epoch": 1.04,
"learning_rate": 1.8409376293348836e-05,
"loss": 1.4711,
"step": 105
},
{
"epoch": 1.05,
"learning_rate": 1.8374437823092726e-05,
"loss": 1.6241,
"step": 106
},
{
"epoch": 1.06,
"learning_rate": 1.833915370312743e-05,
"loss": 1.5392,
"step": 107
},
{
"epoch": 1.07,
"learning_rate": 1.8303525389783045e-05,
"loss": 1.456,
"step": 108
},
{
"epoch": 1.08,
"learning_rate": 1.8267554353596027e-05,
"loss": 1.5794,
"step": 109
},
{
"epoch": 1.09,
"learning_rate": 1.8231242079248512e-05,
"loss": 1.5491,
"step": 110
},
{
"epoch": 1.1,
"learning_rate": 1.819459006550702e-05,
"loss": 1.5011,
"step": 111
},
{
"epoch": 1.11,
"learning_rate": 1.815759982516061e-05,
"loss": 1.5362,
"step": 112
},
{
"epoch": 1.12,
"learning_rate": 1.812027288495843e-05,
"loss": 1.363,
"step": 113
},
{
"epoch": 1.13,
"learning_rate": 1.808261078554671e-05,
"loss": 1.3702,
"step": 114
},
{
"epoch": 1.14,
"learning_rate": 1.8044615081405153e-05,
"loss": 1.3961,
"step": 115
},
{
"epoch": 1.15,
"learning_rate": 1.8006287340782807e-05,
"loss": 1.4011,
"step": 116
},
{
"epoch": 1.16,
"learning_rate": 1.7967629145633312e-05,
"loss": 1.5261,
"step": 117
},
{
"epoch": 1.17,
"learning_rate": 1.7928642091549616e-05,
"loss": 1.3632,
"step": 118
},
{
"epoch": 1.18,
"learning_rate": 1.7889327787698105e-05,
"loss": 1.4158,
"step": 119
},
{
"epoch": 1.19,
"learning_rate": 1.784968785675221e-05,
"loss": 1.4353,
"step": 120
},
{
"epoch": 1.2,
"learning_rate": 1.7809723934825405e-05,
"loss": 1.4338,
"step": 121
},
{
"epoch": 1.21,
"learning_rate": 1.77694376714037e-05,
"loss": 1.4254,
"step": 122
},
{
"epoch": 1.22,
"learning_rate": 1.772883072927754e-05,
"loss": 1.4691,
"step": 123
},
{
"epoch": 1.23,
"learning_rate": 1.768790478447319e-05,
"loss": 1.3882,
"step": 124
},
{
"epoch": 1.24,
"learning_rate": 1.764666152618355e-05,
"loss": 1.388,
"step": 125
},
{
"epoch": 1.25,
"learning_rate": 1.7605102656698444e-05,
"loss": 1.3245,
"step": 126
},
{
"epoch": 1.26,
"learning_rate": 1.756322989133434e-05,
"loss": 1.4139,
"step": 127
},
{
"epoch": 1.27,
"learning_rate": 1.7521044958363567e-05,
"loss": 1.3521,
"step": 128
},
{
"epoch": 1.28,
"learning_rate": 1.7478549598942983e-05,
"loss": 1.2939,
"step": 129
},
{
"epoch": 1.29,
"learning_rate": 1.7435745567042096e-05,
"loss": 1.3955,
"step": 130
},
{
"epoch": 1.3,
"learning_rate": 1.7392634629370684e-05,
"loss": 1.2642,
"step": 131
},
{
"epoch": 1.31,
"learning_rate": 1.734921856530587e-05,
"loss": 1.3465,
"step": 132
},
{
"epoch": 1.32,
"learning_rate": 1.730549916681868e-05,
"loss": 1.3436,
"step": 133
},
{
"epoch": 1.33,
"learning_rate": 1.726147823840007e-05,
"loss": 1.401,
"step": 134
},
{
"epoch": 1.34,
"learning_rate": 1.7217157596986474e-05,
"loss": 1.4414,
"step": 135
},
{
"epoch": 1.35,
"learning_rate": 1.717253907188477e-05,
"loss": 1.3111,
"step": 136
},
{
"epoch": 1.36,
"learning_rate": 1.7127624504696824e-05,
"loss": 1.4064,
"step": 137
},
{
"epoch": 1.37,
"learning_rate": 1.7082415749243436e-05,
"loss": 1.3095,
"step": 138
},
{
"epoch": 1.38,
"learning_rate": 1.7036914671487854e-05,
"loss": 1.4675,
"step": 139
},
{
"epoch": 1.39,
"learning_rate": 1.699112314945874e-05,
"loss": 1.4249,
"step": 140
},
{
"epoch": 1.4,
"learning_rate": 1.694504307317267e-05,
"loss": 1.3799,
"step": 141
},
{
"epoch": 1.41,
"learning_rate": 1.689867634455612e-05,
"loss": 1.4039,
"step": 142
},
{
"epoch": 1.41,
"learning_rate": 1.6852024877366945e-05,
"loss": 1.5129,
"step": 143
},
{
"epoch": 1.42,
"learning_rate": 1.6805090597115424e-05,
"loss": 1.4277,
"step": 144
},
{
"epoch": 1.43,
"learning_rate": 1.675787544098477e-05,
"loss": 1.4467,
"step": 145
},
{
"epoch": 1.44,
"learning_rate": 1.6710381357751155e-05,
"loss": 1.4491,
"step": 146
},
{
"epoch": 1.45,
"learning_rate": 1.6662610307703318e-05,
"loss": 1.4455,
"step": 147
},
{
"epoch": 1.46,
"learning_rate": 1.661456426256161e-05,
"loss": 1.3074,
"step": 148
},
{
"epoch": 1.47,
"learning_rate": 1.6566245205396647e-05,
"loss": 1.4863,
"step": 149
},
{
"epoch": 1.48,
"learning_rate": 1.6517655130547435e-05,
"loss": 1.3705,
"step": 150
},
{
"epoch": 1.49,
"learning_rate": 1.6468796043539082e-05,
"loss": 1.4302,
"step": 151
},
{
"epoch": 1.5,
"learning_rate": 1.641966996099999e-05,
"loss": 1.3508,
"step": 152
},
{
"epoch": 1.51,
"learning_rate": 1.6370278910578644e-05,
"loss": 1.3224,
"step": 153
},
{
"epoch": 1.52,
"learning_rate": 1.6320624930859905e-05,
"loss": 1.3957,
"step": 154
},
{
"epoch": 1.53,
"learning_rate": 1.627071007128089e-05,
"loss": 1.4123,
"step": 155
},
{
"epoch": 1.54,
"learning_rate": 1.6220536392046357e-05,
"loss": 1.3938,
"step": 156
},
{
"epoch": 1.55,
"learning_rate": 1.6170105964043698e-05,
"loss": 1.3844,
"step": 157
},
{
"epoch": 1.56,
"learning_rate": 1.6119420868757433e-05,
"loss": 1.3879,
"step": 158
},
{
"epoch": 1.57,
"learning_rate": 1.606848319818333e-05,
"loss": 1.3161,
"step": 159
},
{
"epoch": 1.58,
"learning_rate": 1.6017295054742045e-05,
"loss": 1.4198,
"step": 160
},
{
"epoch": 1.59,
"learning_rate": 1.596585855119233e-05,
"loss": 1.3897,
"step": 161
},
{
"epoch": 1.6,
"learning_rate": 1.5914175810543868e-05,
"loss": 1.3397,
"step": 162
},
{
"epoch": 1.61,
"learning_rate": 1.5862248965969604e-05,
"loss": 1.4773,
"step": 163
},
{
"epoch": 1.62,
"learning_rate": 1.5810080160717737e-05,
"loss": 1.3474,
"step": 164
},
{
"epoch": 1.63,
"learning_rate": 1.575767154802323e-05,
"loss": 1.5057,
"step": 165
},
{
"epoch": 1.64,
"learning_rate": 1.570502529101896e-05,
"loss": 1.3343,
"step": 166
},
{
"epoch": 1.65,
"learning_rate": 1.5652143562646416e-05,
"loss": 1.3129,
"step": 167
},
{
"epoch": 1.66,
"learning_rate": 1.5599028545566028e-05,
"loss": 1.2845,
"step": 168
},
{
"epoch": 1.67,
"learning_rate": 1.5545682432067068e-05,
"loss": 1.337,
"step": 169
},
{
"epoch": 1.68,
"learning_rate": 1.5492107423977167e-05,
"loss": 1.3302,
"step": 170
},
{
"epoch": 1.69,
"learning_rate": 1.5438305732571445e-05,
"loss": 1.348,
"step": 171
},
{
"epoch": 1.7,
"learning_rate": 1.5384279578481223e-05,
"loss": 1.3215,
"step": 172
},
{
"epoch": 1.71,
"learning_rate": 1.5330031191602395e-05,
"loss": 1.3139,
"step": 173
},
{
"epoch": 1.72,
"learning_rate": 1.5275562811003363e-05,
"loss": 1.3987,
"step": 174
},
{
"epoch": 1.73,
"learning_rate": 1.522087668483264e-05,
"loss": 1.3642,
"step": 175
},
{
"epoch": 1.74,
"learning_rate": 1.5165975070226045e-05,
"loss": 1.3091,
"step": 176
},
{
"epoch": 1.75,
"learning_rate": 1.5110860233213556e-05,
"loss": 1.2608,
"step": 177
},
{
"epoch": 1.76,
"learning_rate": 1.5055534448625766e-05,
"loss": 1.3476,
"step": 178
},
{
"epoch": 1.77,
"learning_rate": 1.5000000000000002e-05,
"loss": 1.3632,
"step": 179
},
{
"epoch": 1.78,
"learning_rate": 1.4944259179486068e-05,
"loss": 1.3465,
"step": 180
},
{
"epoch": 1.79,
"learning_rate": 1.488831428775164e-05,
"loss": 1.3829,
"step": 181
},
{
"epoch": 1.8,
"learning_rate": 1.4832167633887306e-05,
"loss": 1.3986,
"step": 182
},
{
"epoch": 1.81,
"learning_rate": 1.477582153531126e-05,
"loss": 1.3538,
"step": 183
},
{
"epoch": 1.82,
"learning_rate": 1.4719278317673655e-05,
"loss": 1.4173,
"step": 184
},
{
"epoch": 1.83,
"learning_rate": 1.4662540314760608e-05,
"loss": 1.3762,
"step": 185
},
{
"epoch": 1.84,
"learning_rate": 1.4605609868397874e-05,
"loss": 1.3599,
"step": 186
},
{
"epoch": 1.85,
"learning_rate": 1.4548489328354197e-05,
"loss": 1.324,
"step": 187
},
{
"epoch": 1.86,
"learning_rate": 1.4491181052244317e-05,
"loss": 1.2417,
"step": 188
},
{
"epoch": 1.87,
"learning_rate": 1.4433687405431663e-05,
"loss": 1.2887,
"step": 189
},
{
"epoch": 1.88,
"learning_rate": 1.437601076093073e-05,
"loss": 1.3587,
"step": 190
},
{
"epoch": 1.89,
"learning_rate": 1.4318153499309118e-05,
"loss": 1.2952,
"step": 191
},
{
"epoch": 1.9,
"learning_rate": 1.4260118008589294e-05,
"loss": 1.3474,
"step": 192
},
{
"epoch": 1.91,
"learning_rate": 1.420190668415002e-05,
"loss": 1.348,
"step": 193
},
{
"epoch": 1.92,
"learning_rate": 1.4143521928627479e-05,
"loss": 1.3807,
"step": 194
},
{
"epoch": 1.93,
"learning_rate": 1.4084966151816124e-05,
"loss": 1.3044,
"step": 195
},
{
"epoch": 1.94,
"learning_rate": 1.4026241770569198e-05,
"loss": 1.4231,
"step": 196
},
{
"epoch": 1.95,
"learning_rate": 1.3967351208698985e-05,
"loss": 1.3206,
"step": 197
},
{
"epoch": 1.96,
"learning_rate": 1.3908296896876778e-05,
"loss": 1.3744,
"step": 198
},
{
"epoch": 1.97,
"learning_rate": 1.3849081272532545e-05,
"loss": 1.4214,
"step": 199
},
{
"epoch": 1.98,
"learning_rate": 1.3789706779754326e-05,
"loss": 1.4233,
"step": 200
},
{
"epoch": 1.99,
"learning_rate": 1.373017586918736e-05,
"loss": 1.3545,
"step": 201
},
{
"epoch": 2.0,
"learning_rate": 1.3670490997932922e-05,
"loss": 1.2513,
"step": 202
},
{
"epoch": 2.01,
"learning_rate": 1.3610654629446938e-05,
"loss": 1.3838,
"step": 203
},
{
"epoch": 2.02,
"learning_rate": 1.3550669233438271e-05,
"loss": 1.3655,
"step": 204
},
{
"epoch": 2.03,
"learning_rate": 1.3490537285766809e-05,
"loss": 1.3503,
"step": 205
},
{
"epoch": 2.04,
"learning_rate": 1.3430261268341272e-05,
"loss": 1.2948,
"step": 206
},
{
"epoch": 2.05,
"learning_rate": 1.3369843669016757e-05,
"loss": 1.3722,
"step": 207
},
{
"epoch": 2.06,
"learning_rate": 1.3309286981492084e-05,
"loss": 1.3155,
"step": 208
},
{
"epoch": 2.07,
"learning_rate": 1.3248593705206838e-05,
"loss": 1.2492,
"step": 209
},
{
"epoch": 2.08,
"learning_rate": 1.3187766345238222e-05,
"loss": 1.3089,
"step": 210
},
{
"epoch": 2.09,
"learning_rate": 1.3126807412197666e-05,
"loss": 1.3134,
"step": 211
},
{
"epoch": 2.1,
"learning_rate": 1.3065719422127188e-05,
"loss": 1.2995,
"step": 212
},
{
"epoch": 2.11,
"learning_rate": 1.3004504896395564e-05,
"loss": 1.3436,
"step": 213
},
{
"epoch": 2.12,
"learning_rate": 1.2943166361594242e-05,
"loss": 1.1327,
"step": 214
},
{
"epoch": 2.13,
"learning_rate": 1.288170634943307e-05,
"loss": 1.1199,
"step": 215
},
{
"epoch": 2.14,
"learning_rate": 1.2820127396635802e-05,
"loss": 1.1588,
"step": 216
},
{
"epoch": 2.15,
"learning_rate": 1.275843204483539e-05,
"loss": 1.1224,
"step": 217
},
{
"epoch": 2.16,
"learning_rate": 1.2696622840469084e-05,
"loss": 1.3147,
"step": 218
},
{
"epoch": 2.17,
"learning_rate": 1.263470233467332e-05,
"loss": 1.1618,
"step": 219
},
{
"epoch": 2.18,
"learning_rate": 1.2572673083178448e-05,
"loss": 1.195,
"step": 220
},
{
"epoch": 2.19,
"learning_rate": 1.2510537646203209e-05,
"loss": 1.1712,
"step": 221
},
{
"epoch": 2.2,
"learning_rate": 1.2448298588349097e-05,
"loss": 1.1937,
"step": 222
},
{
"epoch": 2.21,
"learning_rate": 1.2385958478494487e-05,
"loss": 1.192,
"step": 223
},
{
"epoch": 2.22,
"learning_rate": 1.2323519889688615e-05,
"loss": 1.2389,
"step": 224
},
{
"epoch": 2.23,
"learning_rate": 1.2260985399045379e-05,
"loss": 1.1535,
"step": 225
},
{
"epoch": 2.24,
"learning_rate": 1.2198357587636958e-05,
"loss": 1.1672,
"step": 226
},
{
"epoch": 2.25,
"learning_rate": 1.2135639040387291e-05,
"loss": 1.1308,
"step": 227
},
{
"epoch": 2.26,
"learning_rate": 1.2072832345965381e-05,
"loss": 1.1642,
"step": 228
},
{
"epoch": 2.27,
"learning_rate": 1.2009940096678451e-05,
"loss": 1.0826,
"step": 229
},
{
"epoch": 2.28,
"learning_rate": 1.1946964888364949e-05,
"loss": 1.0454,
"step": 230
},
{
"epoch": 2.29,
"learning_rate": 1.1883909320287406e-05,
"loss": 1.1272,
"step": 231
},
{
"epoch": 2.3,
"learning_rate": 1.1820775995025147e-05,
"loss": 1.0187,
"step": 232
},
{
"epoch": 2.31,
"learning_rate": 1.1757567518366883e-05,
"loss": 1.1232,
"step": 233
},
{
"epoch": 2.32,
"learning_rate": 1.169428649920315e-05,
"loss": 1.0917,
"step": 234
},
{
"epoch": 2.33,
"learning_rate": 1.1630935549418627e-05,
"loss": 1.1499,
"step": 235
},
{
"epoch": 2.34,
"learning_rate": 1.1567517283784344e-05,
"loss": 1.2013,
"step": 236
},
{
"epoch": 2.35,
"learning_rate": 1.1504034319849741e-05,
"loss": 1.0021,
"step": 237
},
{
"epoch": 2.35,
"learning_rate": 1.1440489277834645e-05,
"loss": 1.145,
"step": 238
},
{
"epoch": 2.36,
"learning_rate": 1.1376884780521117e-05,
"loss": 1.0523,
"step": 239
},
{
"epoch": 2.37,
"learning_rate": 1.1313223453145202e-05,
"loss": 1.199,
"step": 240
},
{
"epoch": 2.38,
"learning_rate": 1.1249507923288563e-05,
"loss": 1.1381,
"step": 241
},
{
"epoch": 2.39,
"learning_rate": 1.1185740820770042e-05,
"loss": 1.1056,
"step": 242
},
{
"epoch": 2.4,
"learning_rate": 1.1121924777537108e-05,
"loss": 1.152,
"step": 243
},
{
"epoch": 2.41,
"learning_rate": 1.105806242755723e-05,
"loss": 1.2463,
"step": 244
},
{
"epoch": 2.42,
"learning_rate": 1.0994156406709155e-05,
"loss": 1.1991,
"step": 245
},
{
"epoch": 2.43,
"learning_rate": 1.0930209352674123e-05,
"loss": 1.2074,
"step": 246
},
{
"epoch": 2.44,
"learning_rate": 1.0866223904826992e-05,
"loss": 1.2137,
"step": 247
},
{
"epoch": 2.45,
"learning_rate": 1.0802202704127293e-05,
"loss": 1.1762,
"step": 248
},
{
"epoch": 2.46,
"learning_rate": 1.0738148393010251e-05,
"loss": 1.1095,
"step": 249
},
{
"epoch": 2.47,
"learning_rate": 1.0674063615277681e-05,
"loss": 1.2061,
"step": 250
},
{
"epoch": 2.48,
"learning_rate": 1.0609951015988907e-05,
"loss": 1.1453,
"step": 251
},
{
"epoch": 2.49,
"learning_rate": 1.054581324135156e-05,
"loss": 1.1776,
"step": 252
},
{
"epoch": 2.5,
"learning_rate": 1.0481652938612374e-05,
"loss": 1.1084,
"step": 253
},
{
"epoch": 2.51,
"learning_rate": 1.0417472755947908e-05,
"loss": 1.1028,
"step": 254
},
{
"epoch": 2.52,
"learning_rate": 1.0353275342355262e-05,
"loss": 1.1285,
"step": 255
},
{
"epoch": 2.53,
"learning_rate": 1.0289063347542727e-05,
"loss": 1.1332,
"step": 256
},
{
"epoch": 2.54,
"learning_rate": 1.0224839421820426e-05,
"loss": 1.182,
"step": 257
},
{
"epoch": 2.55,
"learning_rate": 1.0160606215990922e-05,
"loss": 1.1823,
"step": 258
},
{
"epoch": 2.56,
"learning_rate": 1.0096366381239808e-05,
"loss": 1.14,
"step": 259
},
{
"epoch": 2.57,
"learning_rate": 1.0032122569026284e-05,
"loss": 1.0837,
"step": 260
},
{
"epoch": 2.58,
"learning_rate": 9.967877430973716e-06,
"loss": 1.1733,
"step": 261
},
{
"epoch": 2.59,
"learning_rate": 9.903633618760195e-06,
"loss": 1.128,
"step": 262
},
{
"epoch": 2.6,
"learning_rate": 9.839393784009078e-06,
"loss": 1.1153,
"step": 263
},
{
"epoch": 2.61,
"learning_rate": 9.775160578179575e-06,
"loss": 1.2042,
"step": 264
},
{
"epoch": 2.62,
"learning_rate": 9.710936652457276e-06,
"loss": 1.1457,
"step": 265
},
{
"epoch": 2.63,
"learning_rate": 9.64672465764474e-06,
"loss": 1.2454,
"step": 266
},
{
"epoch": 2.64,
"learning_rate": 9.582527244052095e-06,
"loss": 1.078,
"step": 267
},
{
"epoch": 2.65,
"learning_rate": 9.518347061387629e-06,
"loss": 1.0835,
"step": 268
},
{
"epoch": 2.66,
"learning_rate": 9.454186758648444e-06,
"loss": 1.0497,
"step": 269
},
{
"epoch": 2.67,
"learning_rate": 9.390048984011095e-06,
"loss": 1.0864,
"step": 270
},
{
"epoch": 2.68,
"learning_rate": 9.325936384722322e-06,
"loss": 1.0472,
"step": 271
},
{
"epoch": 2.69,
"learning_rate": 9.261851606989754e-06,
"loss": 1.1284,
"step": 272
},
{
"epoch": 2.7,
"learning_rate": 9.197797295872709e-06,
"loss": 1.0829,
"step": 273
},
{
"epoch": 2.71,
"learning_rate": 9.133776095173015e-06,
"loss": 1.0973,
"step": 274
},
{
"epoch": 2.72,
"learning_rate": 9.069790647325879e-06,
"loss": 1.1343,
"step": 275
},
{
"epoch": 2.73,
"learning_rate": 9.005843593290849e-06,
"loss": 1.1389,
"step": 276
},
{
"epoch": 2.74,
"learning_rate": 8.941937572442773e-06,
"loss": 1.0836,
"step": 277
},
{
"epoch": 2.75,
"learning_rate": 8.878075222462896e-06,
"loss": 1.0652,
"step": 278
},
{
"epoch": 2.76,
"learning_rate": 8.81425917922996e-06,
"loss": 1.0596,
"step": 279
},
{
"epoch": 2.77,
"learning_rate": 8.750492076711439e-06,
"loss": 1.1175,
"step": 280
},
{
"epoch": 2.78,
"learning_rate": 8.6867765468548e-06,
"loss": 1.0994,
"step": 281
},
{
"epoch": 2.79,
"learning_rate": 8.623115219478884e-06,
"loss": 1.1334,
"step": 282
},
{
"epoch": 2.8,
"learning_rate": 8.55951072216536e-06,
"loss": 1.1725,
"step": 283
},
{
"epoch": 2.81,
"learning_rate": 8.49596568015026e-06,
"loss": 1.0953,
"step": 284
},
{
"epoch": 2.82,
"learning_rate": 8.432482716215663e-06,
"loss": 1.1551,
"step": 285
},
{
"epoch": 2.83,
"learning_rate": 8.369064450581374e-06,
"loss": 1.0854,
"step": 286
},
{
"epoch": 2.84,
"learning_rate": 8.305713500796852e-06,
"loss": 1.1419,
"step": 287
},
{
"epoch": 2.85,
"learning_rate": 8.242432481633119e-06,
"loss": 1.0436,
"step": 288
},
{
"epoch": 2.86,
"learning_rate": 8.179224004974857e-06,
"loss": 1.0331,
"step": 289
},
{
"epoch": 2.87,
"learning_rate": 8.116090679712601e-06,
"loss": 1.0594,
"step": 290
},
{
"epoch": 2.88,
"learning_rate": 8.053035111635054e-06,
"loss": 1.0856,
"step": 291
},
{
"epoch": 2.89,
"learning_rate": 7.990059903321554e-06,
"loss": 1.0206,
"step": 292
},
{
"epoch": 2.9,
"learning_rate": 7.927167654034622e-06,
"loss": 1.1111,
"step": 293
},
{
"epoch": 2.91,
"learning_rate": 7.864360959612714e-06,
"loss": 1.1125,
"step": 294
},
{
"epoch": 2.92,
"learning_rate": 7.801642412363042e-06,
"loss": 1.1747,
"step": 295
},
{
"epoch": 2.93,
"learning_rate": 7.739014600954623e-06,
"loss": 1.0466,
"step": 296
},
{
"epoch": 2.94,
"learning_rate": 7.676480110311385e-06,
"loss": 1.1687,
"step": 297
},
{
"epoch": 2.95,
"learning_rate": 7.614041521505517e-06,
"loss": 1.0808,
"step": 298
},
{
"epoch": 2.96,
"learning_rate": 7.5517014116509094e-06,
"loss": 1.1393,
"step": 299
},
{
"epoch": 2.97,
"learning_rate": 7.489462353796792e-06,
"loss": 1.1399,
"step": 300
},
{
"epoch": 2.98,
"learning_rate": 7.427326916821557e-06,
"loss": 1.1977,
"step": 301
},
{
"epoch": 2.99,
"learning_rate": 7.3652976653266785e-06,
"loss": 1.1136,
"step": 302
},
{
"epoch": 3.0,
"learning_rate": 7.303377159530919e-06,
"loss": 1.0574,
"step": 303
},
{
"epoch": 3.01,
"learning_rate": 7.24156795516461e-06,
"loss": 1.1441,
"step": 304
},
{
"epoch": 3.02,
"learning_rate": 7.1798726033642e-06,
"loss": 1.1542,
"step": 305
},
{
"epoch": 3.03,
"learning_rate": 7.118293650566931e-06,
"loss": 1.1092,
"step": 306
},
{
"epoch": 3.04,
"learning_rate": 7.056833638405762e-06,
"loss": 1.064,
"step": 307
},
{
"epoch": 3.05,
"learning_rate": 6.995495103604442e-06,
"loss": 1.1304,
"step": 308
},
{
"epoch": 3.06,
"learning_rate": 6.934280577872814e-06,
"loss": 1.0682,
"step": 309
},
{
"epoch": 3.07,
"learning_rate": 6.87319258780234e-06,
"loss": 1.0223,
"step": 310
},
{
"epoch": 3.08,
"learning_rate": 6.812233654761779e-06,
"loss": 1.0467,
"step": 311
},
{
"epoch": 3.09,
"learning_rate": 6.7514062947931655e-06,
"loss": 1.0722,
"step": 312
},
{
"epoch": 3.1,
"learning_rate": 6.690713018507917e-06,
"loss": 1.0569,
"step": 313
},
{
"epoch": 3.11,
"learning_rate": 6.630156330983244e-06,
"loss": 1.1766,
"step": 314
},
{
"epoch": 3.12,
"learning_rate": 6.569738731658735e-06,
"loss": 0.9828,
"step": 315
},
{
"epoch": 3.13,
"learning_rate": 6.509462714233194e-06,
"loss": 0.9987,
"step": 316
},
{
"epoch": 3.14,
"learning_rate": 6.449330766561735e-06,
"loss": 0.9584,
"step": 317
},
{
"epoch": 3.15,
"learning_rate": 6.389345370553065e-06,
"loss": 0.9931,
"step": 318
},
{
"epoch": 3.16,
"learning_rate": 6.32950900206708e-06,
"loss": 1.1554,
"step": 319
},
{
"epoch": 3.17,
"learning_rate": 6.269824130812645e-06,
"loss": 0.9724,
"step": 320
},
{
"epoch": 3.18,
"learning_rate": 6.210293220245678e-06,
"loss": 1.0328,
"step": 321
},
{
"epoch": 3.19,
"learning_rate": 6.150918727467455e-06,
"loss": 0.9994,
"step": 322
},
{
"epoch": 3.2,
"learning_rate": 6.091703103123223e-06,
"loss": 1.0123,
"step": 323
},
{
"epoch": 3.21,
"learning_rate": 6.032648791301019e-06,
"loss": 1.003,
"step": 324
},
{
"epoch": 3.22,
"learning_rate": 5.973758229430806e-06,
"loss": 1.0226,
"step": 325
},
{
"epoch": 3.23,
"learning_rate": 5.91503384818388e-06,
"loss": 0.9729,
"step": 326
},
{
"epoch": 3.24,
"learning_rate": 5.856478071372521e-06,
"loss": 0.9728,
"step": 327
},
{
"epoch": 3.25,
"learning_rate": 5.798093315849984e-06,
"loss": 0.9271,
"step": 328
},
{
"epoch": 3.26,
"learning_rate": 5.739881991410707e-06,
"loss": 0.962,
"step": 329
},
{
"epoch": 3.27,
"learning_rate": 5.681846500690884e-06,
"loss": 0.8697,
"step": 330
},
{
"epoch": 3.28,
"learning_rate": 5.623989239069275e-06,
"loss": 0.8456,
"step": 331
},
{
"epoch": 3.29,
"learning_rate": 5.56631259456834e-06,
"loss": 0.9068,
"step": 332
},
{
"epoch": 3.29,
"learning_rate": 5.508818947755687e-06,
"loss": 0.8098,
"step": 333
},
{
"epoch": 3.3,
"learning_rate": 5.451510671645806e-06,
"loss": 0.9104,
"step": 334
},
{
"epoch": 3.31,
"learning_rate": 5.394390131602133e-06,
"loss": 0.904,
"step": 335
},
{
"epoch": 3.32,
"learning_rate": 5.337459685239395e-06,
"loss": 0.9783,
"step": 336
},
{
"epoch": 3.33,
"learning_rate": 5.280721682326349e-06,
"loss": 1.0197,
"step": 337
},
{
"epoch": 3.34,
"learning_rate": 5.224178464688742e-06,
"loss": 0.8167,
"step": 338
},
{
"epoch": 3.35,
"learning_rate": 5.167832366112695e-06,
"loss": 0.9123,
"step": 339
},
{
"epoch": 3.36,
"learning_rate": 5.111685712248364e-06,
"loss": 0.8262,
"step": 340
},
{
"epoch": 3.37,
"learning_rate": 5.055740820513932e-06,
"loss": 1.0102,
"step": 341
},
{
"epoch": 3.38,
"learning_rate": 5.000000000000003e-06,
"loss": 0.9307,
"step": 342
},
{
"epoch": 3.39,
"learning_rate": 4.944465551374238e-06,
"loss": 0.8742,
"step": 343
},
{
"epoch": 3.4,
"learning_rate": 4.889139766786447e-06,
"loss": 0.9395,
"step": 344
},
{
"epoch": 3.41,
"learning_rate": 4.834024929773956e-06,
"loss": 1.0364,
"step": 345
},
{
"epoch": 3.42,
"learning_rate": 4.779123315167362e-06,
"loss": 1.0022,
"step": 346
},
{
"epoch": 3.43,
"learning_rate": 4.7244371889966374e-06,
"loss": 0.9957,
"step": 347
},
{
"epoch": 3.44,
"learning_rate": 4.669968808397609e-06,
"loss": 1.027,
"step": 348
},
{
"epoch": 3.45,
"learning_rate": 4.61572042151878e-06,
"loss": 0.9526,
"step": 349
},
{
"epoch": 3.46,
"learning_rate": 4.56169426742856e-06,
"loss": 0.9176,
"step": 350
},
{
"epoch": 3.47,
"learning_rate": 4.507892576022838e-06,
"loss": 1.0136,
"step": 351
},
{
"epoch": 3.48,
"learning_rate": 4.4543175679329345e-06,
"loss": 0.959,
"step": 352
},
{
"epoch": 3.49,
"learning_rate": 4.4009714544339755e-06,
"loss": 0.9847,
"step": 353
},
{
"epoch": 3.5,
"learning_rate": 4.347856437353584e-06,
"loss": 0.8823,
"step": 354
},
{
"epoch": 3.51,
"learning_rate": 4.294974708981041e-06,
"loss": 0.917,
"step": 355
},
{
"epoch": 3.52,
"learning_rate": 4.242328451976774e-06,
"loss": 0.9402,
"step": 356
},
{
"epoch": 3.53,
"learning_rate": 4.189919839282265e-06,
"loss": 0.9244,
"step": 357
},
{
"epoch": 3.54,
"learning_rate": 4.1377510340304e-06,
"loss": 1.0061,
"step": 358
},
{
"epoch": 3.55,
"learning_rate": 4.085824189456136e-06,
"loss": 0.9854,
"step": 359
},
{
"epoch": 3.56,
"learning_rate": 4.03414144880767e-06,
"loss": 0.9865,
"step": 360
},
{
"epoch": 3.57,
"learning_rate": 3.982704945257957e-06,
"loss": 0.9202,
"step": 361
},
{
"epoch": 3.58,
"learning_rate": 3.931516801816668e-06,
"loss": 0.971,
"step": 362
},
{
"epoch": 3.59,
"learning_rate": 3.880579131242567e-06,
"loss": 0.9601,
"step": 363
},
{
"epoch": 3.6,
"learning_rate": 3.829894035956306e-06,
"loss": 0.9382,
"step": 364
},
{
"epoch": 3.61,
"learning_rate": 3.779463607953644e-06,
"loss": 0.9866,
"step": 365
},
{
"epoch": 3.62,
"learning_rate": 3.729289928719113e-06,
"loss": 0.9524,
"step": 366
},
{
"epoch": 3.63,
"learning_rate": 3.6793750691400996e-06,
"loss": 1.0385,
"step": 367
},
{
"epoch": 3.64,
"learning_rate": 3.62972108942136e-06,
"loss": 0.8792,
"step": 368
},
{
"epoch": 3.65,
"learning_rate": 3.580330039000014e-06,
"loss": 0.9042,
"step": 369
},
{
"epoch": 3.66,
"learning_rate": 3.5312039564609203e-06,
"loss": 0.8561,
"step": 370
},
{
"epoch": 3.67,
"learning_rate": 3.482344869452565e-06,
"loss": 0.948,
"step": 371
},
{
"epoch": 3.68,
"learning_rate": 3.4337547946033557e-06,
"loss": 0.8488,
"step": 372
},
{
"epoch": 3.69,
"learning_rate": 3.3854357374383905e-06,
"loss": 0.9748,
"step": 373
},
{
"epoch": 3.7,
"learning_rate": 3.3373896922966863e-06,
"loss": 0.9182,
"step": 374
},
{
"epoch": 3.71,
"learning_rate": 3.2896186422488463e-06,
"loss": 0.9528,
"step": 375
},
{
"epoch": 3.72,
"learning_rate": 3.242124559015234e-06,
"loss": 0.9543,
"step": 376
},
{
"epoch": 3.73,
"learning_rate": 3.194909402884576e-06,
"loss": 0.9576,
"step": 377
},
{
"epoch": 3.74,
"learning_rate": 3.1479751226330567e-06,
"loss": 0.8864,
"step": 378
},
{
"epoch": 3.75,
"learning_rate": 3.101323655443882e-06,
"loss": 0.9,
"step": 379
},
{
"epoch": 3.76,
"learning_rate": 3.0549569268273316e-06,
"loss": 0.9028,
"step": 380
},
{
"epoch": 3.77,
"learning_rate": 3.0088768505412623e-06,
"loss": 0.9111,
"step": 381
},
{
"epoch": 3.78,
"learning_rate": 2.9630853285121506e-06,
"loss": 0.9345,
"step": 382
},
{
"epoch": 3.79,
"learning_rate": 2.91758425075657e-06,
"loss": 0.9312,
"step": 383
},
{
"epoch": 3.8,
"learning_rate": 2.872375495303178e-06,
"loss": 0.9943,
"step": 384
},
{
"epoch": 3.81,
"learning_rate": 2.8274609281152322e-06,
"loss": 0.9443,
"step": 385
},
{
"epoch": 3.82,
"learning_rate": 2.7828424030135305e-06,
"loss": 0.9603,
"step": 386
},
{
"epoch": 3.83,
"learning_rate": 2.7385217615999303e-06,
"loss": 0.9195,
"step": 387
},
{
"epoch": 3.84,
"learning_rate": 2.694500833181323e-06,
"loss": 0.9257,
"step": 388
},
{
"epoch": 3.85,
"learning_rate": 2.65078143469413e-06,
"loss": 0.8675,
"step": 389
},
{
"epoch": 3.86,
"learning_rate": 2.6073653706293202e-06,
"loss": 0.8377,
"step": 390
},
{
"epoch": 3.87,
"learning_rate": 2.5642544329579088e-06,
"loss": 0.8685,
"step": 391
},
{
"epoch": 3.88,
"learning_rate": 2.5214504010570217e-06,
"loss": 0.8838,
"step": 392
},
{
"epoch": 3.89,
"learning_rate": 2.478955041636435e-06,
"loss": 0.8548,
"step": 393
},
{
"epoch": 3.9,
"learning_rate": 2.4367701086656625e-06,
"loss": 0.9218,
"step": 394
},
{
"epoch": 3.91,
"learning_rate": 2.3948973433015564e-06,
"loss": 0.9963,
"step": 395
},
{
"epoch": 3.92,
"learning_rate": 2.353338473816451e-06,
"loss": 0.9655,
"step": 396
},
{
"epoch": 3.93,
"learning_rate": 2.312095215526814e-06,
"loss": 0.8899,
"step": 397
},
{
"epoch": 3.94,
"learning_rate": 2.271169270722464e-06,
"loss": 0.9778,
"step": 398
},
{
"epoch": 3.95,
"learning_rate": 2.230562328596306e-06,
"loss": 0.946,
"step": 399
},
{
"epoch": 3.96,
"learning_rate": 2.190276065174596e-06,
"loss": 0.9608,
"step": 400
},
{
"epoch": 3.97,
"learning_rate": 2.1503121432477936e-06,
"loss": 0.9379,
"step": 401
},
{
"epoch": 3.98,
"learning_rate": 2.1106722123018965e-06,
"loss": 1.0413,
"step": 402
},
{
"epoch": 3.99,
"learning_rate": 2.0713579084503877e-06,
"loss": 0.9567,
"step": 403
},
{
"epoch": 4.0,
"learning_rate": 2.0323708543666888e-06,
"loss": 0.8854,
"step": 404
},
{
"epoch": 4.01,
"learning_rate": 1.993712659217194e-06,
"loss": 0.952,
"step": 405
},
{
"epoch": 4.02,
"learning_rate": 1.9553849185948514e-06,
"loss": 1.0019,
"step": 406
},
{
"epoch": 4.03,
"learning_rate": 1.9173892144532957e-06,
"loss": 0.965,
"step": 407
},
{
"epoch": 4.04,
"learning_rate": 1.8797271150415709e-06,
"loss": 0.8878,
"step": 408
},
{
"epoch": 4.05,
"learning_rate": 1.8424001748393905e-06,
"loss": 0.9816,
"step": 409
},
{
"epoch": 4.06,
"learning_rate": 1.8054099344929833e-06,
"loss": 0.9631,
"step": 410
},
{
"epoch": 4.07,
"learning_rate": 1.7687579207514893e-06,
"loss": 0.8626,
"step": 411
},
{
"epoch": 4.08,
"learning_rate": 1.7324456464039751e-06,
"loss": 0.8982,
"step": 412
},
{
"epoch": 4.09,
"learning_rate": 1.6964746102169582e-06,
"loss": 0.9542,
"step": 413
},
{
"epoch": 4.1,
"learning_rate": 1.6608462968725736e-06,
"loss": 0.8928,
"step": 414
},
{
"epoch": 4.11,
"learning_rate": 1.6255621769072805e-06,
"loss": 1.0105,
"step": 415
},
{
"epoch": 4.12,
"learning_rate": 1.5906237066511643e-06,
"loss": 0.8821,
"step": 416
},
{
"epoch": 4.13,
"learning_rate": 1.5560323281678514e-06,
"loss": 0.8703,
"step": 417
},
{
"epoch": 4.14,
"learning_rate": 1.521789469194952e-06,
"loss": 0.8437,
"step": 418
},
{
"epoch": 4.15,
"learning_rate": 1.4878965430851612e-06,
"loss": 0.8663,
"step": 419
},
{
"epoch": 4.16,
"learning_rate": 1.4543549487479092e-06,
"loss": 0.9994,
"step": 420
},
{
"epoch": 4.17,
"learning_rate": 1.4211660705916286e-06,
"loss": 0.8591,
"step": 421
},
{
"epoch": 4.18,
"learning_rate": 1.3883312784666091e-06,
"loss": 0.8815,
"step": 422
},
{
"epoch": 4.19,
"learning_rate": 1.3558519276084636e-06,
"loss": 0.8615,
"step": 423
},
{
"epoch": 4.2,
"learning_rate": 1.3237293585821786e-06,
"loss": 0.8511,
"step": 424
},
{
"epoch": 4.21,
"learning_rate": 1.291964897226803e-06,
"loss": 0.8524,
"step": 425
},
{
"epoch": 4.22,
"learning_rate": 1.260559854600709e-06,
"loss": 0.8683,
"step": 426
},
{
"epoch": 4.23,
"learning_rate": 1.2295155269274827e-06,
"loss": 0.8299,
"step": 427
},
{
"epoch": 4.24,
"learning_rate": 1.198833195542435e-06,
"loss": 0.8564,
"step": 428
},
{
"epoch": 4.24,
"learning_rate": 1.1685141268396906e-06,
"loss": 0.7924,
"step": 429
},
{
"epoch": 4.25,
"learning_rate": 1.1385595722199438e-06,
"loss": 0.8414,
"step": 430
},
{
"epoch": 4.26,
"learning_rate": 1.1089707680387962e-06,
"loss": 0.7434,
"step": 431
},
{
"epoch": 4.27,
"learning_rate": 1.079748935555719e-06,
"loss": 0.7299,
"step": 432
},
{
"epoch": 4.28,
"learning_rate": 1.0508952808836682e-06,
"loss": 0.7797,
"step": 433
},
{
"epoch": 4.29,
"learning_rate": 1.022410994939279e-06,
"loss": 0.7147,
"step": 434
},
{
"epoch": 4.3,
"learning_rate": 9.942972533937268e-07,
"loss": 0.8071,
"step": 435
},
{
"epoch": 4.31,
"learning_rate": 9.665552166241965e-07,
"loss": 0.7704,
"step": 436
},
{
"epoch": 4.32,
"learning_rate": 9.391860296659916e-07,
"loss": 0.8572,
"step": 437
},
{
"epoch": 4.33,
"learning_rate": 9.121908221652675e-07,
"loss": 0.8579,
"step": 438
},
{
"epoch": 4.34,
"learning_rate": 8.855707083324183e-07,
"loss": 0.7123,
"step": 439
},
{
"epoch": 4.35,
"learning_rate": 8.593267868960675e-07,
"loss": 0.7775,
"step": 440
},
{
"epoch": 4.36,
"learning_rate": 8.334601410577436e-07,
"loss": 0.7357,
"step": 441
},
{
"epoch": 4.37,
"learning_rate": 8.079718384471557e-07,
"loss": 0.867,
"step": 442
},
{
"epoch": 4.38,
"learning_rate": 7.828629310781266e-07,
"loss": 0.821,
"step": 443
},
{
"epoch": 4.39,
"learning_rate": 7.581344553051873e-07,
"loss": 0.7604,
"step": 444
},
{
"epoch": 4.4,
"learning_rate": 7.337874317807803e-07,
"loss": 0.8181,
"step": 445
},
{
"epoch": 4.41,
"learning_rate": 7.098228654131489e-07,
"loss": 0.8672,
"step": 446
},
{
"epoch": 4.42,
"learning_rate": 6.862417453248593e-07,
"loss": 0.871,
"step": 447
},
{
"epoch": 4.43,
"learning_rate": 6.630450448119618e-07,
"loss": 0.8699,
"step": 448
},
{
"epoch": 4.44,
"learning_rate": 6.402337213038379e-07,
"loss": 0.9146,
"step": 449
},
{
"epoch": 4.45,
"learning_rate": 6.178087163236645e-07,
"loss": 0.8129,
"step": 450
},
{
"epoch": 4.46,
"learning_rate": 5.957709554495683e-07,
"loss": 0.7845,
"step": 451
},
{
"epoch": 4.47,
"learning_rate": 5.741213482764118e-07,
"loss": 0.9181,
"step": 452
},
{
"epoch": 4.48,
"learning_rate": 5.528607883782599e-07,
"loss": 0.8529,
"step": 453
},
{
"epoch": 4.49,
"learning_rate": 5.319901532714877e-07,
"loss": 0.8863,
"step": 454
},
{
"epoch": 4.5,
"learning_rate": 5.115103043785718e-07,
"loss": 0.7747,
"step": 455
},
{
"epoch": 4.51,
"learning_rate": 4.91422086992529e-07,
"loss": 0.8225,
"step": 456
},
{
"epoch": 4.52,
"learning_rate": 4.717263302420283e-07,
"loss": 0.8055,
"step": 457
},
{
"epoch": 4.53,
"learning_rate": 4.5242384705717404e-07,
"loss": 0.8073,
"step": 458
},
{
"epoch": 4.54,
"learning_rate": 4.3351543413594263e-07,
"loss": 0.9058,
"step": 459
},
{
"epoch": 4.55,
"learning_rate": 4.150018719113147e-07,
"loss": 0.8856,
"step": 460
},
{
"epoch": 4.56,
"learning_rate": 3.968839245190448e-07,
"loss": 0.8794,
"step": 461
},
{
"epoch": 4.57,
"learning_rate": 3.7916233976613704e-07,
"loss": 0.8192,
"step": 462
},
{
"epoch": 4.58,
"learning_rate": 3.618378490999719e-07,
"loss": 0.8839,
"step": 463
},
{
"epoch": 4.59,
"learning_rate": 3.4491116757812024e-07,
"loss": 0.8596,
"step": 464
},
{
"epoch": 4.6,
"learning_rate": 3.283829938388294e-07,
"loss": 0.8171,
"step": 465
},
{
"epoch": 4.61,
"learning_rate": 3.122540100721794e-07,
"loss": 0.8854,
"step": 466
},
{
"epoch": 4.62,
"learning_rate": 2.965248819919397e-07,
"loss": 0.8724,
"step": 467
},
{
"epoch": 4.63,
"learning_rate": 2.8119625880808185e-07,
"loss": 0.9394,
"step": 468
},
{
"epoch": 4.64,
"learning_rate": 2.66268773199988e-07,
"loss": 0.7804,
"step": 469
},
{
"epoch": 4.65,
"learning_rate": 2.5174304129033655e-07,
"loss": 0.7995,
"step": 470
},
{
"epoch": 4.66,
"learning_rate": 2.3761966261967252e-07,
"loss": 0.7713,
"step": 471
},
{
"epoch": 4.67,
"learning_rate": 2.2389922012165944e-07,
"loss": 0.8227,
"step": 472
},
{
"epoch": 4.68,
"learning_rate": 2.1058228009902094e-07,
"loss": 0.763,
"step": 473
},
{
"epoch": 4.69,
"learning_rate": 1.9766939220017155e-07,
"loss": 0.8846,
"step": 474
},
{
"epoch": 4.7,
"learning_rate": 1.8516108939651945e-07,
"loss": 0.8191,
"step": 475
},
{
"epoch": 4.71,
"learning_rate": 1.7305788796048274e-07,
"loss": 0.8646,
"step": 476
},
{
"epoch": 4.72,
"learning_rate": 1.613602874441622e-07,
"loss": 0.8464,
"step": 477
},
{
"epoch": 4.73,
"learning_rate": 1.5006877065874338e-07,
"loss": 0.8767,
"step": 478
},
{
"epoch": 4.74,
"learning_rate": 1.3918380365455232e-07,
"loss": 0.7908,
"step": 479
},
{
"epoch": 4.75,
"learning_rate": 1.287058357018278e-07,
"loss": 0.8039,
"step": 480
},
{
"epoch": 4.76,
"learning_rate": 1.1863529927217731e-07,
"loss": 0.7987,
"step": 481
},
{
"epoch": 4.77,
"learning_rate": 1.0897261002072223e-07,
"loss": 0.8231,
"step": 482
},
{
"epoch": 4.78,
"learning_rate": 9.971816676894952e-08,
"loss": 0.8304,
"step": 483
},
{
"epoch": 4.79,
"learning_rate": 9.08723514882437e-08,
"loss": 0.846,
"step": 484
},
{
"epoch": 4.8,
"learning_rate": 8.243552928412501e-08,
"loss": 0.8961,
"step": 485
},
{
"epoch": 4.81,
"learning_rate": 7.440804838117932e-08,
"loss": 0.8834,
"step": 486
},
{
"epoch": 4.82,
"learning_rate": 6.679024010868617e-08,
"loss": 0.8625,
"step": 487
},
{
"epoch": 4.83,
"learning_rate": 5.958241888693872e-08,
"loss": 0.8292,
"step": 488
},
{
"epoch": 4.84,
"learning_rate": 5.2784882214274025e-08,
"loss": 0.8513,
"step": 489
},
{
"epoch": 4.85,
"learning_rate": 4.639791065478738e-08,
"loss": 0.7609,
"step": 490
},
{
"epoch": 4.86,
"learning_rate": 4.042176782675267e-08,
"loss": 0.7706,
"step": 491
},
{
"epoch": 4.87,
"learning_rate": 3.485670039174882e-08,
"loss": 0.7774,
"step": 492
},
{
"epoch": 4.88,
"learning_rate": 2.9702938044468e-08,
"loss": 0.8261,
"step": 493
},
{
"epoch": 4.89,
"learning_rate": 2.496069350324537e-08,
"loss": 0.7775,
"step": 494
},
{
"epoch": 4.9,
"learning_rate": 2.063016250127281e-08,
"loss": 0.8115,
"step": 495
},
{
"epoch": 4.91,
"learning_rate": 1.671152377852092e-08,
"loss": 0.9538,
"step": 496
},
{
"epoch": 4.92,
"learning_rate": 1.320493907436604e-08,
"loss": 0.8701,
"step": 497
},
{
"epoch": 4.93,
"learning_rate": 1.0110553120908917e-08,
"loss": 0.8319,
"step": 498
},
{
"epoch": 4.94,
"learning_rate": 7.42849363700282e-09,
"loss": 0.8792,
"step": 499
},
{
"epoch": 4.95,
"learning_rate": 5.158871322984426e-09,
"loss": 0.8577,
"step": 500
},
{
"epoch": 4.96,
"learning_rate": 3.3017798561030268e-09,
"loss": 0.9151,
"step": 501
},
{
"epoch": 4.97,
"learning_rate": 1.8572958866514e-09,
"loss": 0.8622,
"step": 502
},
{
"epoch": 4.98,
"learning_rate": 8.254790348072306e-10,
"loss": 0.9606,
"step": 503
},
{
"epoch": 4.99,
"learning_rate": 2.063718881695209e-10,
"loss": 0.8872,
"step": 504
},
{
"epoch": 5.0,
"learning_rate": 0.0,
"loss": 0.819,
"step": 505
},
{
"epoch": 5.0,
"step": 505,
"total_flos": 2.6761493632096666e+18,
"train_loss": 1.198445222283354,
"train_runtime": 29037.4355,
"train_samples_per_second": 2.227,
"train_steps_per_second": 0.017
}
],
"logging_steps": 1.0,
"max_steps": 505,
"num_train_epochs": 5,
"save_steps": 101,
"total_flos": 2.6761493632096666e+18,
"trial_name": null,
"trial_params": null
}