DeQA-Score-LoRA-Mix3 / trainer_state.json
zhiyuanyou
init commit
3c1258b
raw
history blame
77.9 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 3.0,
"eval_steps": 500,
"global_step": 645,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.4037,
"step": 1
},
{
"epoch": 0.01,
"learning_rate": 2.0000000000000003e-06,
"loss": 0.4032,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 3e-06,
"loss": 0.4071,
"step": 3
},
{
"epoch": 0.02,
"learning_rate": 4.000000000000001e-06,
"loss": 0.406,
"step": 4
},
{
"epoch": 0.02,
"learning_rate": 5e-06,
"loss": 0.4054,
"step": 5
},
{
"epoch": 0.03,
"learning_rate": 6e-06,
"loss": 0.4099,
"step": 6
},
{
"epoch": 0.03,
"learning_rate": 7e-06,
"loss": 0.385,
"step": 7
},
{
"epoch": 0.04,
"learning_rate": 8.000000000000001e-06,
"loss": 0.3404,
"step": 8
},
{
"epoch": 0.04,
"learning_rate": 9e-06,
"loss": 0.3211,
"step": 9
},
{
"epoch": 0.05,
"learning_rate": 1e-05,
"loss": 0.2757,
"step": 10
},
{
"epoch": 0.05,
"learning_rate": 1.1000000000000001e-05,
"loss": 0.2633,
"step": 11
},
{
"epoch": 0.06,
"learning_rate": 1.2e-05,
"loss": 0.2397,
"step": 12
},
{
"epoch": 0.06,
"learning_rate": 1.3000000000000001e-05,
"loss": 0.178,
"step": 13
},
{
"epoch": 0.07,
"learning_rate": 1.4e-05,
"loss": 0.1417,
"step": 14
},
{
"epoch": 0.07,
"learning_rate": 1.5000000000000002e-05,
"loss": 0.1309,
"step": 15
},
{
"epoch": 0.07,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.109,
"step": 16
},
{
"epoch": 0.08,
"learning_rate": 1.7e-05,
"loss": 0.1101,
"step": 17
},
{
"epoch": 0.08,
"learning_rate": 1.8e-05,
"loss": 0.0791,
"step": 18
},
{
"epoch": 0.09,
"learning_rate": 1.9e-05,
"loss": 0.0903,
"step": 19
},
{
"epoch": 0.09,
"learning_rate": 2e-05,
"loss": 0.0738,
"step": 20
},
{
"epoch": 0.1,
"learning_rate": 1.999987366932966e-05,
"loss": 0.066,
"step": 21
},
{
"epoch": 0.1,
"learning_rate": 1.999949468051052e-05,
"loss": 0.0583,
"step": 22
},
{
"epoch": 0.11,
"learning_rate": 1.9998863043118163e-05,
"loss": 0.0541,
"step": 23
},
{
"epoch": 0.11,
"learning_rate": 1.999797877311163e-05,
"loss": 0.0733,
"step": 24
},
{
"epoch": 0.12,
"learning_rate": 1.9996841892833e-05,
"loss": 0.0517,
"step": 25
},
{
"epoch": 0.12,
"learning_rate": 1.9995452431006844e-05,
"loss": 0.0579,
"step": 26
},
{
"epoch": 0.13,
"learning_rate": 1.9993810422739496e-05,
"loss": 0.065,
"step": 27
},
{
"epoch": 0.13,
"learning_rate": 1.9991915909518146e-05,
"loss": 0.0462,
"step": 28
},
{
"epoch": 0.13,
"learning_rate": 1.9989768939209826e-05,
"loss": 0.0468,
"step": 29
},
{
"epoch": 0.14,
"learning_rate": 1.998736956606018e-05,
"loss": 0.0531,
"step": 30
},
{
"epoch": 0.14,
"learning_rate": 1.998471785069208e-05,
"loss": 0.053,
"step": 31
},
{
"epoch": 0.15,
"learning_rate": 1.998181386010413e-05,
"loss": 0.052,
"step": 32
},
{
"epoch": 0.15,
"learning_rate": 1.9978657667668945e-05,
"loss": 0.046,
"step": 33
},
{
"epoch": 0.16,
"learning_rate": 1.9975249353131304e-05,
"loss": 0.0524,
"step": 34
},
{
"epoch": 0.16,
"learning_rate": 1.997158900260614e-05,
"loss": 0.055,
"step": 35
},
{
"epoch": 0.17,
"learning_rate": 1.9967676708576362e-05,
"loss": 0.0445,
"step": 36
},
{
"epoch": 0.17,
"learning_rate": 1.9963512569890512e-05,
"loss": 0.0405,
"step": 37
},
{
"epoch": 0.18,
"learning_rate": 1.9959096691760284e-05,
"loss": 0.0434,
"step": 38
},
{
"epoch": 0.18,
"learning_rate": 1.9954429185757835e-05,
"loss": 0.0559,
"step": 39
},
{
"epoch": 0.19,
"learning_rate": 1.9949510169813006e-05,
"loss": 0.0446,
"step": 40
},
{
"epoch": 0.19,
"learning_rate": 1.994433976821031e-05,
"loss": 0.0374,
"step": 41
},
{
"epoch": 0.2,
"learning_rate": 1.9938918111585805e-05,
"loss": 0.037,
"step": 42
},
{
"epoch": 0.2,
"learning_rate": 1.9933245336923798e-05,
"loss": 0.0359,
"step": 43
},
{
"epoch": 0.2,
"learning_rate": 1.9927321587553378e-05,
"loss": 0.0359,
"step": 44
},
{
"epoch": 0.21,
"learning_rate": 1.9921147013144782e-05,
"loss": 0.0359,
"step": 45
},
{
"epoch": 0.21,
"learning_rate": 1.9914721769705637e-05,
"loss": 0.0396,
"step": 46
},
{
"epoch": 0.22,
"learning_rate": 1.990804601957701e-05,
"loss": 0.0348,
"step": 47
},
{
"epoch": 0.22,
"learning_rate": 1.9901119931429294e-05,
"loss": 0.0345,
"step": 48
},
{
"epoch": 0.23,
"learning_rate": 1.9893943680257964e-05,
"loss": 0.0331,
"step": 49
},
{
"epoch": 0.23,
"learning_rate": 1.988651744737914e-05,
"loss": 0.04,
"step": 50
},
{
"epoch": 0.24,
"learning_rate": 1.9878841420425023e-05,
"loss": 0.0335,
"step": 51
},
{
"epoch": 0.24,
"learning_rate": 1.9870915793339137e-05,
"loss": 0.034,
"step": 52
},
{
"epoch": 0.25,
"learning_rate": 1.9862740766371434e-05,
"loss": 0.0345,
"step": 53
},
{
"epoch": 0.25,
"learning_rate": 1.985431654607325e-05,
"loss": 0.0329,
"step": 54
},
{
"epoch": 0.26,
"learning_rate": 1.9845643345292055e-05,
"loss": 0.0334,
"step": 55
},
{
"epoch": 0.26,
"learning_rate": 1.983672138316611e-05,
"loss": 0.037,
"step": 56
},
{
"epoch": 0.27,
"learning_rate": 1.9827550885118902e-05,
"loss": 0.0337,
"step": 57
},
{
"epoch": 0.27,
"learning_rate": 1.9818132082853466e-05,
"loss": 0.0354,
"step": 58
},
{
"epoch": 0.27,
"learning_rate": 1.9808465214346525e-05,
"loss": 0.0333,
"step": 59
},
{
"epoch": 0.28,
"learning_rate": 1.979855052384247e-05,
"loss": 0.0375,
"step": 60
},
{
"epoch": 0.28,
"learning_rate": 1.9788388261847204e-05,
"loss": 0.0417,
"step": 61
},
{
"epoch": 0.29,
"learning_rate": 1.97779786851218e-05,
"loss": 0.0378,
"step": 62
},
{
"epoch": 0.29,
"learning_rate": 1.9767322056676018e-05,
"loss": 0.0371,
"step": 63
},
{
"epoch": 0.3,
"learning_rate": 1.975641864576166e-05,
"loss": 0.0352,
"step": 64
},
{
"epoch": 0.3,
"learning_rate": 1.9745268727865774e-05,
"loss": 0.0326,
"step": 65
},
{
"epoch": 0.31,
"learning_rate": 1.9733872584703673e-05,
"loss": 0.0333,
"step": 66
},
{
"epoch": 0.31,
"learning_rate": 1.9722230504211843e-05,
"loss": 0.0279,
"step": 67
},
{
"epoch": 0.32,
"learning_rate": 1.971034278054065e-05,
"loss": 0.0345,
"step": 68
},
{
"epoch": 0.32,
"learning_rate": 1.969820971404691e-05,
"loss": 0.0321,
"step": 69
},
{
"epoch": 0.33,
"learning_rate": 1.9685831611286312e-05,
"loss": 0.0341,
"step": 70
},
{
"epoch": 0.33,
"learning_rate": 1.9673208785005658e-05,
"loss": 0.0335,
"step": 71
},
{
"epoch": 0.33,
"learning_rate": 1.9660341554134972e-05,
"loss": 0.0372,
"step": 72
},
{
"epoch": 0.34,
"learning_rate": 1.9647230243779432e-05,
"loss": 0.0333,
"step": 73
},
{
"epoch": 0.34,
"learning_rate": 1.963387518521116e-05,
"loss": 0.0422,
"step": 74
},
{
"epoch": 0.35,
"learning_rate": 1.962027671586086e-05,
"loss": 0.0412,
"step": 75
},
{
"epoch": 0.35,
"learning_rate": 1.9606435179309284e-05,
"loss": 0.0314,
"step": 76
},
{
"epoch": 0.36,
"learning_rate": 1.9592350925278546e-05,
"loss": 0.0475,
"step": 77
},
{
"epoch": 0.36,
"learning_rate": 1.9578024309623296e-05,
"loss": 0.0524,
"step": 78
},
{
"epoch": 0.37,
"learning_rate": 1.956345569432173e-05,
"loss": 0.031,
"step": 79
},
{
"epoch": 0.37,
"learning_rate": 1.9548645447466433e-05,
"loss": 0.037,
"step": 80
},
{
"epoch": 0.38,
"learning_rate": 1.9533593943255087e-05,
"loss": 0.0382,
"step": 81
},
{
"epoch": 0.38,
"learning_rate": 1.9518301561981016e-05,
"loss": 0.0323,
"step": 82
},
{
"epoch": 0.39,
"learning_rate": 1.9502768690023574e-05,
"loss": 0.0349,
"step": 83
},
{
"epoch": 0.39,
"learning_rate": 1.9486995719838392e-05,
"loss": 0.0417,
"step": 84
},
{
"epoch": 0.4,
"learning_rate": 1.9470983049947446e-05,
"loss": 0.0389,
"step": 85
},
{
"epoch": 0.4,
"learning_rate": 1.9454731084928995e-05,
"loss": 0.0354,
"step": 86
},
{
"epoch": 0.4,
"learning_rate": 1.9438240235407375e-05,
"loss": 0.0366,
"step": 87
},
{
"epoch": 0.41,
"learning_rate": 1.9421510918042593e-05,
"loss": 0.0388,
"step": 88
},
{
"epoch": 0.41,
"learning_rate": 1.940454355551983e-05,
"loss": 0.0312,
"step": 89
},
{
"epoch": 0.42,
"learning_rate": 1.9387338576538743e-05,
"loss": 0.0354,
"step": 90
},
{
"epoch": 0.42,
"learning_rate": 1.936989641580263e-05,
"loss": 0.0383,
"step": 91
},
{
"epoch": 0.43,
"learning_rate": 1.935221751400747e-05,
"loss": 0.0331,
"step": 92
},
{
"epoch": 0.43,
"learning_rate": 1.9334302317830764e-05,
"loss": 0.0297,
"step": 93
},
{
"epoch": 0.44,
"learning_rate": 1.931615127992026e-05,
"loss": 0.0312,
"step": 94
},
{
"epoch": 0.44,
"learning_rate": 1.9297764858882516e-05,
"loss": 0.0378,
"step": 95
},
{
"epoch": 0.45,
"learning_rate": 1.927914351927131e-05,
"loss": 0.0266,
"step": 96
},
{
"epoch": 0.45,
"learning_rate": 1.9260287731575902e-05,
"loss": 0.0295,
"step": 97
},
{
"epoch": 0.46,
"learning_rate": 1.9241197972209157e-05,
"loss": 0.0371,
"step": 98
},
{
"epoch": 0.46,
"learning_rate": 1.9221874723495494e-05,
"loss": 0.0307,
"step": 99
},
{
"epoch": 0.47,
"learning_rate": 1.9202318473658707e-05,
"loss": 0.0294,
"step": 100
},
{
"epoch": 0.47,
"learning_rate": 1.9182529716809618e-05,
"loss": 0.0311,
"step": 101
},
{
"epoch": 0.47,
"learning_rate": 1.916250895293362e-05,
"loss": 0.025,
"step": 102
},
{
"epoch": 0.48,
"learning_rate": 1.9142256687878012e-05,
"loss": 0.0294,
"step": 103
},
{
"epoch": 0.48,
"learning_rate": 1.912177343333924e-05,
"loss": 0.0336,
"step": 104
},
{
"epoch": 0.49,
"learning_rate": 1.9101059706849957e-05,
"loss": 0.035,
"step": 105
},
{
"epoch": 0.49,
"learning_rate": 1.908011603176596e-05,
"loss": 0.0303,
"step": 106
},
{
"epoch": 0.5,
"learning_rate": 1.9058942937252943e-05,
"loss": 0.0324,
"step": 107
},
{
"epoch": 0.5,
"learning_rate": 1.903754095827316e-05,
"loss": 0.0387,
"step": 108
},
{
"epoch": 0.51,
"learning_rate": 1.901591063557187e-05,
"loss": 0.0323,
"step": 109
},
{
"epoch": 0.51,
"learning_rate": 1.899405251566371e-05,
"loss": 0.0289,
"step": 110
},
{
"epoch": 0.52,
"learning_rate": 1.897196715081888e-05,
"loss": 0.0286,
"step": 111
},
{
"epoch": 0.52,
"learning_rate": 1.894965509904915e-05,
"loss": 0.0318,
"step": 112
},
{
"epoch": 0.53,
"learning_rate": 1.8927116924093824e-05,
"loss": 0.0261,
"step": 113
},
{
"epoch": 0.53,
"learning_rate": 1.890435319540545e-05,
"loss": 0.0307,
"step": 114
},
{
"epoch": 0.53,
"learning_rate": 1.8881364488135448e-05,
"loss": 0.0287,
"step": 115
},
{
"epoch": 0.54,
"learning_rate": 1.8858151383119576e-05,
"loss": 0.0274,
"step": 116
},
{
"epoch": 0.54,
"learning_rate": 1.883471446686326e-05,
"loss": 0.0274,
"step": 117
},
{
"epoch": 0.55,
"learning_rate": 1.881105433152677e-05,
"loss": 0.0286,
"step": 118
},
{
"epoch": 0.55,
"learning_rate": 1.878717157491025e-05,
"loss": 0.0274,
"step": 119
},
{
"epoch": 0.56,
"learning_rate": 1.8763066800438638e-05,
"loss": 0.0214,
"step": 120
},
{
"epoch": 0.56,
"learning_rate": 1.8738740617146396e-05,
"loss": 0.0217,
"step": 121
},
{
"epoch": 0.57,
"learning_rate": 1.871419363966213e-05,
"loss": 0.0335,
"step": 122
},
{
"epoch": 0.57,
"learning_rate": 1.8689426488193066e-05,
"loss": 0.0233,
"step": 123
},
{
"epoch": 0.58,
"learning_rate": 1.866443978850937e-05,
"loss": 0.0325,
"step": 124
},
{
"epoch": 0.58,
"learning_rate": 1.8639234171928355e-05,
"loss": 0.0242,
"step": 125
},
{
"epoch": 0.59,
"learning_rate": 1.86138102752985e-05,
"loss": 0.0296,
"step": 126
},
{
"epoch": 0.59,
"learning_rate": 1.858816874098339e-05,
"loss": 0.0275,
"step": 127
},
{
"epoch": 0.6,
"learning_rate": 1.8562310216845463e-05,
"loss": 0.0294,
"step": 128
},
{
"epoch": 0.6,
"learning_rate": 1.8536235356229667e-05,
"loss": 0.0295,
"step": 129
},
{
"epoch": 0.6,
"learning_rate": 1.850994481794692e-05,
"loss": 0.0259,
"step": 130
},
{
"epoch": 0.61,
"learning_rate": 1.8483439266257485e-05,
"loss": 0.0308,
"step": 131
},
{
"epoch": 0.61,
"learning_rate": 1.845671937085419e-05,
"loss": 0.0299,
"step": 132
},
{
"epoch": 0.62,
"learning_rate": 1.842978580684549e-05,
"loss": 0.024,
"step": 133
},
{
"epoch": 0.62,
"learning_rate": 1.8402639254738422e-05,
"loss": 0.0266,
"step": 134
},
{
"epoch": 0.63,
"learning_rate": 1.837528040042142e-05,
"loss": 0.0241,
"step": 135
},
{
"epoch": 0.63,
"learning_rate": 1.8347709935146958e-05,
"loss": 0.0273,
"step": 136
},
{
"epoch": 0.64,
"learning_rate": 1.8319928555514108e-05,
"loss": 0.028,
"step": 137
},
{
"epoch": 0.64,
"learning_rate": 1.8291936963450933e-05,
"loss": 0.0277,
"step": 138
},
{
"epoch": 0.65,
"learning_rate": 1.8263735866196758e-05,
"loss": 0.0314,
"step": 139
},
{
"epoch": 0.65,
"learning_rate": 1.8235325976284276e-05,
"loss": 0.0273,
"step": 140
},
{
"epoch": 0.66,
"learning_rate": 1.820670801152158e-05,
"loss": 0.0242,
"step": 141
},
{
"epoch": 0.66,
"learning_rate": 1.8177882694974008e-05,
"loss": 0.0265,
"step": 142
},
{
"epoch": 0.67,
"learning_rate": 1.8148850754945865e-05,
"loss": 0.0282,
"step": 143
},
{
"epoch": 0.67,
"learning_rate": 1.8119612924962043e-05,
"loss": 0.0244,
"step": 144
},
{
"epoch": 0.67,
"learning_rate": 1.8090169943749477e-05,
"loss": 0.0249,
"step": 145
},
{
"epoch": 0.68,
"learning_rate": 1.806052255521847e-05,
"loss": 0.0246,
"step": 146
},
{
"epoch": 0.68,
"learning_rate": 1.8030671508443928e-05,
"loss": 0.028,
"step": 147
},
{
"epoch": 0.69,
"learning_rate": 1.8000617557646392e-05,
"loss": 0.0225,
"step": 148
},
{
"epoch": 0.69,
"learning_rate": 1.797036146217301e-05,
"loss": 0.0257,
"step": 149
},
{
"epoch": 0.7,
"learning_rate": 1.7939903986478354e-05,
"loss": 0.0283,
"step": 150
},
{
"epoch": 0.7,
"learning_rate": 1.7909245900105085e-05,
"loss": 0.0264,
"step": 151
},
{
"epoch": 0.71,
"learning_rate": 1.7878387977664522e-05,
"loss": 0.0277,
"step": 152
},
{
"epoch": 0.71,
"learning_rate": 1.784733099881707e-05,
"loss": 0.0244,
"step": 153
},
{
"epoch": 0.72,
"learning_rate": 1.7816075748252526e-05,
"loss": 0.0246,
"step": 154
},
{
"epoch": 0.72,
"learning_rate": 1.7784623015670237e-05,
"loss": 0.0306,
"step": 155
},
{
"epoch": 0.73,
"learning_rate": 1.775297359575916e-05,
"loss": 0.0234,
"step": 156
},
{
"epoch": 0.73,
"learning_rate": 1.7721128288177782e-05,
"loss": 0.0267,
"step": 157
},
{
"epoch": 0.73,
"learning_rate": 1.7689087897533916e-05,
"loss": 0.0272,
"step": 158
},
{
"epoch": 0.74,
"learning_rate": 1.765685323336437e-05,
"loss": 0.0271,
"step": 159
},
{
"epoch": 0.74,
"learning_rate": 1.762442511011448e-05,
"loss": 0.0269,
"step": 160
},
{
"epoch": 0.75,
"learning_rate": 1.759180434711757e-05,
"loss": 0.0258,
"step": 161
},
{
"epoch": 0.75,
"learning_rate": 1.7558991768574197e-05,
"loss": 0.0254,
"step": 162
},
{
"epoch": 0.76,
"learning_rate": 1.752598820353138e-05,
"loss": 0.029,
"step": 163
},
{
"epoch": 0.76,
"learning_rate": 1.749279448586162e-05,
"loss": 0.0262,
"step": 164
},
{
"epoch": 0.77,
"learning_rate": 1.7459411454241822e-05,
"loss": 0.0328,
"step": 165
},
{
"epoch": 0.77,
"learning_rate": 1.7425839952132157e-05,
"loss": 0.0284,
"step": 166
},
{
"epoch": 0.78,
"learning_rate": 1.739208082775469e-05,
"loss": 0.0306,
"step": 167
},
{
"epoch": 0.78,
"learning_rate": 1.7358134934071978e-05,
"loss": 0.0297,
"step": 168
},
{
"epoch": 0.79,
"learning_rate": 1.7324003128765536e-05,
"loss": 0.0253,
"step": 169
},
{
"epoch": 0.79,
"learning_rate": 1.7289686274214116e-05,
"loss": 0.0273,
"step": 170
},
{
"epoch": 0.8,
"learning_rate": 1.7255185237471978e-05,
"loss": 0.0268,
"step": 171
},
{
"epoch": 0.8,
"learning_rate": 1.7220500890246944e-05,
"loss": 0.0313,
"step": 172
},
{
"epoch": 0.8,
"learning_rate": 1.7185634108878367e-05,
"loss": 0.0319,
"step": 173
},
{
"epoch": 0.81,
"learning_rate": 1.715058577431503e-05,
"loss": 0.0274,
"step": 174
},
{
"epoch": 0.81,
"learning_rate": 1.7115356772092858e-05,
"loss": 0.0268,
"step": 175
},
{
"epoch": 0.82,
"learning_rate": 1.707994799231253e-05,
"loss": 0.0255,
"step": 176
},
{
"epoch": 0.82,
"learning_rate": 1.704436032961703e-05,
"loss": 0.0273,
"step": 177
},
{
"epoch": 0.83,
"learning_rate": 1.7008594683169018e-05,
"loss": 0.0261,
"step": 178
},
{
"epoch": 0.83,
"learning_rate": 1.6972651956628108e-05,
"loss": 0.0269,
"step": 179
},
{
"epoch": 0.84,
"learning_rate": 1.693653305812805e-05,
"loss": 0.0259,
"step": 180
},
{
"epoch": 0.84,
"learning_rate": 1.6900238900253777e-05,
"loss": 0.0248,
"step": 181
},
{
"epoch": 0.85,
"learning_rate": 1.6863770400018344e-05,
"loss": 0.0256,
"step": 182
},
{
"epoch": 0.85,
"learning_rate": 1.6827128478839767e-05,
"loss": 0.0242,
"step": 183
},
{
"epoch": 0.86,
"learning_rate": 1.679031406251774e-05,
"loss": 0.0262,
"step": 184
},
{
"epoch": 0.86,
"learning_rate": 1.6753328081210244e-05,
"loss": 0.0285,
"step": 185
},
{
"epoch": 0.87,
"learning_rate": 1.6716171469410042e-05,
"loss": 0.0324,
"step": 186
},
{
"epoch": 0.87,
"learning_rate": 1.6678845165921066e-05,
"loss": 0.0236,
"step": 187
},
{
"epoch": 0.87,
"learning_rate": 1.6641350113834705e-05,
"loss": 0.0247,
"step": 188
},
{
"epoch": 0.88,
"learning_rate": 1.660368726050597e-05,
"loss": 0.0247,
"step": 189
},
{
"epoch": 0.88,
"learning_rate": 1.6565857557529567e-05,
"loss": 0.0273,
"step": 190
},
{
"epoch": 0.89,
"learning_rate": 1.652786196071584e-05,
"loss": 0.0275,
"step": 191
},
{
"epoch": 0.89,
"learning_rate": 1.6489701430066632e-05,
"loss": 0.0244,
"step": 192
},
{
"epoch": 0.9,
"learning_rate": 1.6451376929751028e-05,
"loss": 0.0259,
"step": 193
},
{
"epoch": 0.9,
"learning_rate": 1.6412889428080992e-05,
"loss": 0.0281,
"step": 194
},
{
"epoch": 0.91,
"learning_rate": 1.63742398974869e-05,
"loss": 0.0255,
"step": 195
},
{
"epoch": 0.91,
"learning_rate": 1.633542931449297e-05,
"loss": 0.0259,
"step": 196
},
{
"epoch": 0.92,
"learning_rate": 1.62964586596926e-05,
"loss": 0.034,
"step": 197
},
{
"epoch": 0.92,
"learning_rate": 1.625732891772358e-05,
"loss": 0.0243,
"step": 198
},
{
"epoch": 0.93,
"learning_rate": 1.6218041077243213e-05,
"loss": 0.0229,
"step": 199
},
{
"epoch": 0.93,
"learning_rate": 1.6178596130903345e-05,
"loss": 0.024,
"step": 200
},
{
"epoch": 0.93,
"learning_rate": 1.6138995075325277e-05,
"loss": 0.0233,
"step": 201
},
{
"epoch": 0.94,
"learning_rate": 1.609923891107459e-05,
"loss": 0.0277,
"step": 202
},
{
"epoch": 0.94,
"learning_rate": 1.6059328642635864e-05,
"loss": 0.0315,
"step": 203
},
{
"epoch": 0.95,
"learning_rate": 1.6019265278387287e-05,
"loss": 0.027,
"step": 204
},
{
"epoch": 0.95,
"learning_rate": 1.597904983057519e-05,
"loss": 0.0285,
"step": 205
},
{
"epoch": 0.96,
"learning_rate": 1.5938683315288472e-05,
"loss": 0.0237,
"step": 206
},
{
"epoch": 0.96,
"learning_rate": 1.589816675243292e-05,
"loss": 0.0254,
"step": 207
},
{
"epoch": 0.97,
"learning_rate": 1.5857501165705443e-05,
"loss": 0.0281,
"step": 208
},
{
"epoch": 0.97,
"learning_rate": 1.581668758256821e-05,
"loss": 0.03,
"step": 209
},
{
"epoch": 0.98,
"learning_rate": 1.5775727034222675e-05,
"loss": 0.0257,
"step": 210
},
{
"epoch": 0.98,
"learning_rate": 1.5734620555583555e-05,
"loss": 0.0262,
"step": 211
},
{
"epoch": 0.99,
"learning_rate": 1.5693369185252648e-05,
"loss": 0.0238,
"step": 212
},
{
"epoch": 0.99,
"learning_rate": 1.56519739654926e-05,
"loss": 0.0275,
"step": 213
},
{
"epoch": 1.0,
"learning_rate": 1.561043594220059e-05,
"loss": 0.0221,
"step": 214
},
{
"epoch": 1.0,
"learning_rate": 1.556875616488188e-05,
"loss": 0.0291,
"step": 215
},
{
"epoch": 1.0,
"learning_rate": 1.5526935686623316e-05,
"loss": 0.0223,
"step": 216
},
{
"epoch": 1.01,
"learning_rate": 1.5484975564066704e-05,
"loss": 0.0208,
"step": 217
},
{
"epoch": 1.01,
"learning_rate": 1.544287685738213e-05,
"loss": 0.0292,
"step": 218
},
{
"epoch": 1.02,
"learning_rate": 1.540064063024116e-05,
"loss": 0.023,
"step": 219
},
{
"epoch": 1.02,
"learning_rate": 1.5358267949789968e-05,
"loss": 0.0252,
"step": 220
},
{
"epoch": 1.03,
"learning_rate": 1.531575988662238e-05,
"loss": 0.0253,
"step": 221
},
{
"epoch": 1.03,
"learning_rate": 1.5273117514752826e-05,
"loss": 0.029,
"step": 222
},
{
"epoch": 1.04,
"learning_rate": 1.5230341911589183e-05,
"loss": 0.02,
"step": 223
},
{
"epoch": 1.04,
"learning_rate": 1.5187434157905575e-05,
"loss": 0.0212,
"step": 224
},
{
"epoch": 1.05,
"learning_rate": 1.5144395337815066e-05,
"loss": 0.0204,
"step": 225
},
{
"epoch": 1.05,
"learning_rate": 1.5101226538742248e-05,
"loss": 0.0211,
"step": 226
},
{
"epoch": 1.06,
"learning_rate": 1.505792885139579e-05,
"loss": 0.0252,
"step": 227
},
{
"epoch": 1.06,
"learning_rate": 1.5014503369740866e-05,
"loss": 0.0241,
"step": 228
},
{
"epoch": 1.07,
"learning_rate": 1.4970951190971512e-05,
"loss": 0.0271,
"step": 229
},
{
"epoch": 1.07,
"learning_rate": 1.4927273415482916e-05,
"loss": 0.0291,
"step": 230
},
{
"epoch": 1.07,
"learning_rate": 1.4883471146843617e-05,
"loss": 0.0285,
"step": 231
},
{
"epoch": 1.08,
"learning_rate": 1.4839545491767599e-05,
"loss": 0.0256,
"step": 232
},
{
"epoch": 1.08,
"learning_rate": 1.4795497560086358e-05,
"loss": 0.0265,
"step": 233
},
{
"epoch": 1.09,
"learning_rate": 1.4751328464720842e-05,
"loss": 0.0229,
"step": 234
},
{
"epoch": 1.09,
"learning_rate": 1.470703932165333e-05,
"loss": 0.0257,
"step": 235
},
{
"epoch": 1.1,
"learning_rate": 1.4662631249899248e-05,
"loss": 0.0257,
"step": 236
},
{
"epoch": 1.1,
"learning_rate": 1.4618105371478896e-05,
"loss": 0.0235,
"step": 237
},
{
"epoch": 1.11,
"learning_rate": 1.4573462811389087e-05,
"loss": 0.0199,
"step": 238
},
{
"epoch": 1.11,
"learning_rate": 1.4528704697574729e-05,
"loss": 0.0201,
"step": 239
},
{
"epoch": 1.12,
"learning_rate": 1.4483832160900326e-05,
"loss": 0.0229,
"step": 240
},
{
"epoch": 1.12,
"learning_rate": 1.4438846335121402e-05,
"loss": 0.0275,
"step": 241
},
{
"epoch": 1.13,
"learning_rate": 1.4393748356855865e-05,
"loss": 0.0195,
"step": 242
},
{
"epoch": 1.13,
"learning_rate": 1.4348539365555283e-05,
"loss": 0.0273,
"step": 243
},
{
"epoch": 1.13,
"learning_rate": 1.430322050347609e-05,
"loss": 0.0304,
"step": 244
},
{
"epoch": 1.14,
"learning_rate": 1.4257792915650728e-05,
"loss": 0.0228,
"step": 245
},
{
"epoch": 1.14,
"learning_rate": 1.4212257749858727e-05,
"loss": 0.022,
"step": 246
},
{
"epoch": 1.15,
"learning_rate": 1.416661615659768e-05,
"loss": 0.0266,
"step": 247
},
{
"epoch": 1.15,
"learning_rate": 1.412086928905421e-05,
"loss": 0.0254,
"step": 248
},
{
"epoch": 1.16,
"learning_rate": 1.4075018303074808e-05,
"loss": 0.0229,
"step": 249
},
{
"epoch": 1.16,
"learning_rate": 1.4029064357136628e-05,
"loss": 0.0246,
"step": 250
},
{
"epoch": 1.17,
"learning_rate": 1.398300861231823e-05,
"loss": 0.0221,
"step": 251
},
{
"epoch": 1.17,
"learning_rate": 1.3936852232270236e-05,
"loss": 0.0213,
"step": 252
},
{
"epoch": 1.18,
"learning_rate": 1.3890596383185934e-05,
"loss": 0.0266,
"step": 253
},
{
"epoch": 1.18,
"learning_rate": 1.384424223377181e-05,
"loss": 0.0241,
"step": 254
},
{
"epoch": 1.19,
"learning_rate": 1.3797790955218014e-05,
"loss": 0.0231,
"step": 255
},
{
"epoch": 1.19,
"learning_rate": 1.3751243721168778e-05,
"loss": 0.0272,
"step": 256
},
{
"epoch": 1.2,
"learning_rate": 1.3704601707692762e-05,
"loss": 0.023,
"step": 257
},
{
"epoch": 1.2,
"learning_rate": 1.3657866093253327e-05,
"loss": 0.0194,
"step": 258
},
{
"epoch": 1.2,
"learning_rate": 1.3611038058678776e-05,
"loss": 0.0205,
"step": 259
},
{
"epoch": 1.21,
"learning_rate": 1.3564118787132507e-05,
"loss": 0.0233,
"step": 260
},
{
"epoch": 1.21,
"learning_rate": 1.3517109464083129e-05,
"loss": 0.0238,
"step": 261
},
{
"epoch": 1.22,
"learning_rate": 1.3470011277274497e-05,
"loss": 0.017,
"step": 262
},
{
"epoch": 1.22,
"learning_rate": 1.3422825416695713e-05,
"loss": 0.0233,
"step": 263
},
{
"epoch": 1.23,
"learning_rate": 1.337555307455106e-05,
"loss": 0.0201,
"step": 264
},
{
"epoch": 1.23,
"learning_rate": 1.3328195445229869e-05,
"loss": 0.0242,
"step": 265
},
{
"epoch": 1.24,
"learning_rate": 1.3280753725276352e-05,
"loss": 0.0255,
"step": 266
},
{
"epoch": 1.24,
"learning_rate": 1.3233229113359368e-05,
"loss": 0.0239,
"step": 267
},
{
"epoch": 1.25,
"learning_rate": 1.3185622810242129e-05,
"loss": 0.0215,
"step": 268
},
{
"epoch": 1.25,
"learning_rate": 1.3137936018751876e-05,
"loss": 0.0225,
"step": 269
},
{
"epoch": 1.26,
"learning_rate": 1.3090169943749475e-05,
"loss": 0.0212,
"step": 270
},
{
"epoch": 1.26,
"learning_rate": 1.3042325792098982e-05,
"loss": 0.027,
"step": 271
},
{
"epoch": 1.27,
"learning_rate": 1.2994404772637145e-05,
"loss": 0.0278,
"step": 272
},
{
"epoch": 1.27,
"learning_rate": 1.2946408096142866e-05,
"loss": 0.0243,
"step": 273
},
{
"epoch": 1.27,
"learning_rate": 1.289833697530661e-05,
"loss": 0.0221,
"step": 274
},
{
"epoch": 1.28,
"learning_rate": 1.2850192624699762e-05,
"loss": 0.0212,
"step": 275
},
{
"epoch": 1.28,
"learning_rate": 1.2801976260743937e-05,
"loss": 0.0241,
"step": 276
},
{
"epoch": 1.29,
"learning_rate": 1.2753689101680252e-05,
"loss": 0.0207,
"step": 277
},
{
"epoch": 1.29,
"learning_rate": 1.2705332367538539e-05,
"loss": 0.0191,
"step": 278
},
{
"epoch": 1.3,
"learning_rate": 1.2656907280106528e-05,
"loss": 0.0193,
"step": 279
},
{
"epoch": 1.3,
"learning_rate": 1.2608415062898971e-05,
"loss": 0.0237,
"step": 280
},
{
"epoch": 1.31,
"learning_rate": 1.255985694112673e-05,
"loss": 0.0272,
"step": 281
},
{
"epoch": 1.31,
"learning_rate": 1.2511234141665816e-05,
"loss": 0.0201,
"step": 282
},
{
"epoch": 1.32,
"learning_rate": 1.2462547893026403e-05,
"loss": 0.0216,
"step": 283
},
{
"epoch": 1.32,
"learning_rate": 1.2413799425321774e-05,
"loss": 0.0225,
"step": 284
},
{
"epoch": 1.33,
"learning_rate": 1.236498997023725e-05,
"loss": 0.0259,
"step": 285
},
{
"epoch": 1.33,
"learning_rate": 1.2316120760999066e-05,
"loss": 0.0283,
"step": 286
},
{
"epoch": 1.33,
"learning_rate": 1.2267193032343219e-05,
"loss": 0.0201,
"step": 287
},
{
"epoch": 1.34,
"learning_rate": 1.2218208020484255e-05,
"loss": 0.0199,
"step": 288
},
{
"epoch": 1.34,
"learning_rate": 1.2169166963084056e-05,
"loss": 0.0221,
"step": 289
},
{
"epoch": 1.35,
"learning_rate": 1.212007109922055e-05,
"loss": 0.0274,
"step": 290
},
{
"epoch": 1.35,
"learning_rate": 1.2070921669356415e-05,
"loss": 0.0299,
"step": 291
},
{
"epoch": 1.36,
"learning_rate": 1.2021719915307737e-05,
"loss": 0.0263,
"step": 292
},
{
"epoch": 1.36,
"learning_rate": 1.197246708021263e-05,
"loss": 0.023,
"step": 293
},
{
"epoch": 1.37,
"learning_rate": 1.192316440849983e-05,
"loss": 0.0228,
"step": 294
},
{
"epoch": 1.37,
"learning_rate": 1.187381314585725e-05,
"loss": 0.0232,
"step": 295
},
{
"epoch": 1.38,
"learning_rate": 1.1824414539200505e-05,
"loss": 0.0179,
"step": 296
},
{
"epoch": 1.38,
"learning_rate": 1.1774969836641417e-05,
"loss": 0.0189,
"step": 297
},
{
"epoch": 1.39,
"learning_rate": 1.1725480287456467e-05,
"loss": 0.0222,
"step": 298
},
{
"epoch": 1.39,
"learning_rate": 1.1675947142055241e-05,
"loss": 0.0204,
"step": 299
},
{
"epoch": 1.4,
"learning_rate": 1.1626371651948839e-05,
"loss": 0.0228,
"step": 300
},
{
"epoch": 1.4,
"learning_rate": 1.1576755069718229e-05,
"loss": 0.0253,
"step": 301
},
{
"epoch": 1.4,
"learning_rate": 1.1527098648982634e-05,
"loss": 0.0224,
"step": 302
},
{
"epoch": 1.41,
"learning_rate": 1.1477403644367839e-05,
"loss": 0.0199,
"step": 303
},
{
"epoch": 1.41,
"learning_rate": 1.1427671311474489e-05,
"loss": 0.0217,
"step": 304
},
{
"epoch": 1.42,
"learning_rate": 1.137790290684638e-05,
"loss": 0.0226,
"step": 305
},
{
"epoch": 1.42,
"learning_rate": 1.1328099687938696e-05,
"loss": 0.0212,
"step": 306
},
{
"epoch": 1.43,
"learning_rate": 1.1278262913086238e-05,
"loss": 0.0224,
"step": 307
},
{
"epoch": 1.43,
"learning_rate": 1.1228393841471644e-05,
"loss": 0.021,
"step": 308
},
{
"epoch": 1.44,
"learning_rate": 1.117849373309356e-05,
"loss": 0.0266,
"step": 309
},
{
"epoch": 1.44,
"learning_rate": 1.1128563848734817e-05,
"loss": 0.0221,
"step": 310
},
{
"epoch": 1.45,
"learning_rate": 1.1078605449930569e-05,
"loss": 0.0231,
"step": 311
},
{
"epoch": 1.45,
"learning_rate": 1.1028619798936418e-05,
"loss": 0.0226,
"step": 312
},
{
"epoch": 1.46,
"learning_rate": 1.0978608158696517e-05,
"loss": 0.0164,
"step": 313
},
{
"epoch": 1.46,
"learning_rate": 1.092857179281168e-05,
"loss": 0.0204,
"step": 314
},
{
"epoch": 1.47,
"learning_rate": 1.0878511965507435e-05,
"loss": 0.0228,
"step": 315
},
{
"epoch": 1.47,
"learning_rate": 1.0828429941602082e-05,
"loss": 0.023,
"step": 316
},
{
"epoch": 1.47,
"learning_rate": 1.0778326986474765e-05,
"loss": 0.0208,
"step": 317
},
{
"epoch": 1.48,
"learning_rate": 1.072820436603346e-05,
"loss": 0.018,
"step": 318
},
{
"epoch": 1.48,
"learning_rate": 1.067806334668301e-05,
"loss": 0.0199,
"step": 319
},
{
"epoch": 1.49,
"learning_rate": 1.0627905195293135e-05,
"loss": 0.0218,
"step": 320
},
{
"epoch": 1.49,
"learning_rate": 1.0577731179166415e-05,
"loss": 0.0275,
"step": 321
},
{
"epoch": 1.5,
"learning_rate": 1.0527542566006266e-05,
"loss": 0.0201,
"step": 322
},
{
"epoch": 1.5,
"learning_rate": 1.0477340623884917e-05,
"loss": 0.0199,
"step": 323
},
{
"epoch": 1.51,
"learning_rate": 1.0427126621211372e-05,
"loss": 0.0243,
"step": 324
},
{
"epoch": 1.51,
"learning_rate": 1.0376901826699349e-05,
"loss": 0.0207,
"step": 325
},
{
"epoch": 1.52,
"learning_rate": 1.032666750933524e-05,
"loss": 0.0202,
"step": 326
},
{
"epoch": 1.52,
"learning_rate": 1.0276424938346044e-05,
"loss": 0.0209,
"step": 327
},
{
"epoch": 1.53,
"learning_rate": 1.0226175383167299e-05,
"loss": 0.0222,
"step": 328
},
{
"epoch": 1.53,
"learning_rate": 1.0175920113410998e-05,
"loss": 0.0191,
"step": 329
},
{
"epoch": 1.53,
"learning_rate": 1.0125660398833528e-05,
"loss": 0.0197,
"step": 330
},
{
"epoch": 1.54,
"learning_rate": 1.0075397509303573e-05,
"loss": 0.0258,
"step": 331
},
{
"epoch": 1.54,
"learning_rate": 1.0025132714770041e-05,
"loss": 0.02,
"step": 332
},
{
"epoch": 1.55,
"learning_rate": 9.974867285229962e-06,
"loss": 0.0224,
"step": 333
},
{
"epoch": 1.55,
"learning_rate": 9.92460249069643e-06,
"loss": 0.0236,
"step": 334
},
{
"epoch": 1.56,
"learning_rate": 9.874339601166474e-06,
"loss": 0.0242,
"step": 335
},
{
"epoch": 1.56,
"learning_rate": 9.824079886589004e-06,
"loss": 0.0219,
"step": 336
},
{
"epoch": 1.57,
"learning_rate": 9.773824616832706e-06,
"loss": 0.0169,
"step": 337
},
{
"epoch": 1.57,
"learning_rate": 9.723575061653957e-06,
"loss": 0.0302,
"step": 338
},
{
"epoch": 1.58,
"learning_rate": 9.673332490664765e-06,
"loss": 0.0225,
"step": 339
},
{
"epoch": 1.58,
"learning_rate": 9.623098173300655e-06,
"loss": 0.0224,
"step": 340
},
{
"epoch": 1.59,
"learning_rate": 9.572873378788633e-06,
"loss": 0.0199,
"step": 341
},
{
"epoch": 1.59,
"learning_rate": 9.522659376115086e-06,
"loss": 0.0271,
"step": 342
},
{
"epoch": 1.6,
"learning_rate": 9.472457433993736e-06,
"loss": 0.0248,
"step": 343
},
{
"epoch": 1.6,
"learning_rate": 9.422268820833588e-06,
"loss": 0.0208,
"step": 344
},
{
"epoch": 1.6,
"learning_rate": 9.372094804706867e-06,
"loss": 0.0227,
"step": 345
},
{
"epoch": 1.61,
"learning_rate": 9.321936653316995e-06,
"loss": 0.0203,
"step": 346
},
{
"epoch": 1.61,
"learning_rate": 9.271795633966544e-06,
"loss": 0.0236,
"step": 347
},
{
"epoch": 1.62,
"learning_rate": 9.221673013525235e-06,
"loss": 0.0219,
"step": 348
},
{
"epoch": 1.62,
"learning_rate": 9.171570058397919e-06,
"loss": 0.021,
"step": 349
},
{
"epoch": 1.63,
"learning_rate": 9.121488034492569e-06,
"loss": 0.0222,
"step": 350
},
{
"epoch": 1.63,
"learning_rate": 9.071428207188324e-06,
"loss": 0.0208,
"step": 351
},
{
"epoch": 1.64,
"learning_rate": 9.021391841303484e-06,
"loss": 0.0224,
"step": 352
},
{
"epoch": 1.64,
"learning_rate": 8.971380201063586e-06,
"loss": 0.0211,
"step": 353
},
{
"epoch": 1.65,
"learning_rate": 8.921394550069434e-06,
"loss": 0.022,
"step": 354
},
{
"epoch": 1.65,
"learning_rate": 8.871436151265183e-06,
"loss": 0.0192,
"step": 355
},
{
"epoch": 1.66,
"learning_rate": 8.821506266906444e-06,
"loss": 0.0202,
"step": 356
},
{
"epoch": 1.66,
"learning_rate": 8.77160615852836e-06,
"loss": 0.0217,
"step": 357
},
{
"epoch": 1.67,
"learning_rate": 8.721737086913766e-06,
"loss": 0.0193,
"step": 358
},
{
"epoch": 1.67,
"learning_rate": 8.671900312061306e-06,
"loss": 0.0167,
"step": 359
},
{
"epoch": 1.67,
"learning_rate": 8.62209709315362e-06,
"loss": 0.0234,
"step": 360
},
{
"epoch": 1.68,
"learning_rate": 8.572328688525513e-06,
"loss": 0.0233,
"step": 361
},
{
"epoch": 1.68,
"learning_rate": 8.522596355632164e-06,
"loss": 0.0247,
"step": 362
},
{
"epoch": 1.69,
"learning_rate": 8.472901351017371e-06,
"loss": 0.0209,
"step": 363
},
{
"epoch": 1.69,
"learning_rate": 8.423244930281774e-06,
"loss": 0.0231,
"step": 364
},
{
"epoch": 1.7,
"learning_rate": 8.373628348051165e-06,
"loss": 0.0196,
"step": 365
},
{
"epoch": 1.7,
"learning_rate": 8.32405285794476e-06,
"loss": 0.0188,
"step": 366
},
{
"epoch": 1.71,
"learning_rate": 8.274519712543535e-06,
"loss": 0.02,
"step": 367
},
{
"epoch": 1.71,
"learning_rate": 8.22503016335859e-06,
"loss": 0.0184,
"step": 368
},
{
"epoch": 1.72,
"learning_rate": 8.175585460799499e-06,
"loss": 0.0192,
"step": 369
},
{
"epoch": 1.72,
"learning_rate": 8.126186854142752e-06,
"loss": 0.0195,
"step": 370
},
{
"epoch": 1.73,
"learning_rate": 8.076835591500172e-06,
"loss": 0.0219,
"step": 371
},
{
"epoch": 1.73,
"learning_rate": 8.027532919787372e-06,
"loss": 0.0208,
"step": 372
},
{
"epoch": 1.73,
"learning_rate": 7.978280084692268e-06,
"loss": 0.0194,
"step": 373
},
{
"epoch": 1.74,
"learning_rate": 7.92907833064359e-06,
"loss": 0.0242,
"step": 374
},
{
"epoch": 1.74,
"learning_rate": 7.879928900779457e-06,
"loss": 0.0203,
"step": 375
},
{
"epoch": 1.75,
"learning_rate": 7.830833036915947e-06,
"loss": 0.0212,
"step": 376
},
{
"epoch": 1.75,
"learning_rate": 7.781791979515749e-06,
"loss": 0.0225,
"step": 377
},
{
"epoch": 1.76,
"learning_rate": 7.732806967656785e-06,
"loss": 0.0263,
"step": 378
},
{
"epoch": 1.76,
"learning_rate": 7.683879239000936e-06,
"loss": 0.0184,
"step": 379
},
{
"epoch": 1.77,
"learning_rate": 7.635010029762755e-06,
"loss": 0.0166,
"step": 380
},
{
"epoch": 1.77,
"learning_rate": 7.586200574678231e-06,
"loss": 0.0218,
"step": 381
},
{
"epoch": 1.78,
"learning_rate": 7.5374521069736e-06,
"loss": 0.0199,
"step": 382
},
{
"epoch": 1.78,
"learning_rate": 7.488765858334188e-06,
"loss": 0.0219,
"step": 383
},
{
"epoch": 1.79,
"learning_rate": 7.440143058873275e-06,
"loss": 0.0231,
"step": 384
},
{
"epoch": 1.79,
"learning_rate": 7.391584937101034e-06,
"loss": 0.0207,
"step": 385
},
{
"epoch": 1.8,
"learning_rate": 7.3430927198934745e-06,
"loss": 0.0189,
"step": 386
},
{
"epoch": 1.8,
"learning_rate": 7.294667632461463e-06,
"loss": 0.0255,
"step": 387
},
{
"epoch": 1.8,
"learning_rate": 7.246310898319753e-06,
"loss": 0.0229,
"step": 388
},
{
"epoch": 1.81,
"learning_rate": 7.198023739256066e-06,
"loss": 0.0194,
"step": 389
},
{
"epoch": 1.81,
"learning_rate": 7.149807375300239e-06,
"loss": 0.0204,
"step": 390
},
{
"epoch": 1.82,
"learning_rate": 7.10166302469339e-06,
"loss": 0.0213,
"step": 391
},
{
"epoch": 1.82,
"learning_rate": 7.053591903857138e-06,
"loss": 0.0206,
"step": 392
},
{
"epoch": 1.83,
"learning_rate": 7.005595227362858e-06,
"loss": 0.0239,
"step": 393
},
{
"epoch": 1.83,
"learning_rate": 6.957674207901018e-06,
"loss": 0.0214,
"step": 394
},
{
"epoch": 1.84,
"learning_rate": 6.909830056250527e-06,
"loss": 0.0198,
"step": 395
},
{
"epoch": 1.84,
"learning_rate": 6.862063981248126e-06,
"loss": 0.0221,
"step": 396
},
{
"epoch": 1.85,
"learning_rate": 6.814377189757876e-06,
"loss": 0.0245,
"step": 397
},
{
"epoch": 1.85,
"learning_rate": 6.766770886640637e-06,
"loss": 0.0219,
"step": 398
},
{
"epoch": 1.86,
"learning_rate": 6.71924627472365e-06,
"loss": 0.0206,
"step": 399
},
{
"epoch": 1.86,
"learning_rate": 6.671804554770135e-06,
"loss": 0.0215,
"step": 400
},
{
"epoch": 1.87,
"learning_rate": 6.624446925448944e-06,
"loss": 0.026,
"step": 401
},
{
"epoch": 1.87,
"learning_rate": 6.577174583304289e-06,
"loss": 0.0198,
"step": 402
},
{
"epoch": 1.87,
"learning_rate": 6.529988722725506e-06,
"loss": 0.0232,
"step": 403
},
{
"epoch": 1.88,
"learning_rate": 6.482890535916876e-06,
"loss": 0.0222,
"step": 404
},
{
"epoch": 1.88,
"learning_rate": 6.435881212867494e-06,
"loss": 0.0203,
"step": 405
},
{
"epoch": 1.89,
"learning_rate": 6.388961941321225e-06,
"loss": 0.0218,
"step": 406
},
{
"epoch": 1.89,
"learning_rate": 6.342133906746676e-06,
"loss": 0.0222,
"step": 407
},
{
"epoch": 1.9,
"learning_rate": 6.295398292307242e-06,
"loss": 0.0218,
"step": 408
},
{
"epoch": 1.9,
"learning_rate": 6.2487562788312275e-06,
"loss": 0.0242,
"step": 409
},
{
"epoch": 1.91,
"learning_rate": 6.202209044781991e-06,
"loss": 0.017,
"step": 410
},
{
"epoch": 1.91,
"learning_rate": 6.155757766228192e-06,
"loss": 0.0198,
"step": 411
},
{
"epoch": 1.92,
"learning_rate": 6.109403616814069e-06,
"loss": 0.0164,
"step": 412
},
{
"epoch": 1.92,
"learning_rate": 6.063147767729764e-06,
"loss": 0.0222,
"step": 413
},
{
"epoch": 1.93,
"learning_rate": 6.016991387681775e-06,
"loss": 0.0194,
"step": 414
},
{
"epoch": 1.93,
"learning_rate": 5.970935642863375e-06,
"loss": 0.0216,
"step": 415
},
{
"epoch": 1.93,
"learning_rate": 5.924981696925192e-06,
"loss": 0.0212,
"step": 416
},
{
"epoch": 1.94,
"learning_rate": 5.879130710945791e-06,
"loss": 0.018,
"step": 417
},
{
"epoch": 1.94,
"learning_rate": 5.83338384340232e-06,
"loss": 0.0201,
"step": 418
},
{
"epoch": 1.95,
"learning_rate": 5.787742250141279e-06,
"loss": 0.0185,
"step": 419
},
{
"epoch": 1.95,
"learning_rate": 5.742207084349274e-06,
"loss": 0.0213,
"step": 420
},
{
"epoch": 1.96,
"learning_rate": 5.696779496523913e-06,
"loss": 0.0219,
"step": 421
},
{
"epoch": 1.96,
"learning_rate": 5.651460634444716e-06,
"loss": 0.0176,
"step": 422
},
{
"epoch": 1.97,
"learning_rate": 5.606251643144136e-06,
"loss": 0.0191,
"step": 423
},
{
"epoch": 1.97,
"learning_rate": 5.561153664878603e-06,
"loss": 0.0213,
"step": 424
},
{
"epoch": 1.98,
"learning_rate": 5.516167839099679e-06,
"loss": 0.0181,
"step": 425
},
{
"epoch": 1.98,
"learning_rate": 5.471295302425277e-06,
"loss": 0.0229,
"step": 426
},
{
"epoch": 1.99,
"learning_rate": 5.426537188610916e-06,
"loss": 0.0214,
"step": 427
},
{
"epoch": 1.99,
"learning_rate": 5.3818946285211045e-06,
"loss": 0.0262,
"step": 428
},
{
"epoch": 2.0,
"learning_rate": 5.337368750100756e-06,
"loss": 0.0171,
"step": 429
},
{
"epoch": 2.0,
"learning_rate": 5.292960678346674e-06,
"loss": 0.0246,
"step": 430
},
{
"epoch": 2.0,
"learning_rate": 5.248671535279164e-06,
"loss": 0.0192,
"step": 431
},
{
"epoch": 2.01,
"learning_rate": 5.204502439913641e-06,
"loss": 0.018,
"step": 432
},
{
"epoch": 2.01,
"learning_rate": 5.160454508232398e-06,
"loss": 0.0168,
"step": 433
},
{
"epoch": 2.02,
"learning_rate": 5.116528853156384e-06,
"loss": 0.0208,
"step": 434
},
{
"epoch": 2.02,
"learning_rate": 5.072726584517086e-06,
"loss": 0.0207,
"step": 435
},
{
"epoch": 2.03,
"learning_rate": 5.029048809028496e-06,
"loss": 0.0172,
"step": 436
},
{
"epoch": 2.03,
"learning_rate": 4.98549663025914e-06,
"loss": 0.0198,
"step": 437
},
{
"epoch": 2.04,
"learning_rate": 4.942071148604215e-06,
"loss": 0.0199,
"step": 438
},
{
"epoch": 2.04,
"learning_rate": 4.8987734612577544e-06,
"loss": 0.0167,
"step": 439
},
{
"epoch": 2.05,
"learning_rate": 4.855604662184935e-06,
"loss": 0.02,
"step": 440
},
{
"epoch": 2.05,
"learning_rate": 4.812565842094426e-06,
"loss": 0.0175,
"step": 441
},
{
"epoch": 2.06,
"learning_rate": 4.769658088410819e-06,
"loss": 0.019,
"step": 442
},
{
"epoch": 2.06,
"learning_rate": 4.726882485247177e-06,
"loss": 0.0151,
"step": 443
},
{
"epoch": 2.07,
"learning_rate": 4.684240113377619e-06,
"loss": 0.0171,
"step": 444
},
{
"epoch": 2.07,
"learning_rate": 4.641732050210032e-06,
"loss": 0.018,
"step": 445
},
{
"epoch": 2.07,
"learning_rate": 4.599359369758842e-06,
"loss": 0.0175,
"step": 446
},
{
"epoch": 2.08,
"learning_rate": 4.557123142617873e-06,
"loss": 0.018,
"step": 447
},
{
"epoch": 2.08,
"learning_rate": 4.5150244359333e-06,
"loss": 0.0153,
"step": 448
},
{
"epoch": 2.09,
"learning_rate": 4.4730643133766874e-06,
"loss": 0.0216,
"step": 449
},
{
"epoch": 2.09,
"learning_rate": 4.4312438351181246e-06,
"loss": 0.0207,
"step": 450
},
{
"epoch": 2.1,
"learning_rate": 4.389564057799413e-06,
"loss": 0.0165,
"step": 451
},
{
"epoch": 2.1,
"learning_rate": 4.348026034507402e-06,
"loss": 0.0153,
"step": 452
},
{
"epoch": 2.11,
"learning_rate": 4.306630814747358e-06,
"loss": 0.0166,
"step": 453
},
{
"epoch": 2.11,
"learning_rate": 4.265379444416446e-06,
"loss": 0.017,
"step": 454
},
{
"epoch": 2.12,
"learning_rate": 4.224272965777326e-06,
"loss": 0.0275,
"step": 455
},
{
"epoch": 2.12,
"learning_rate": 4.183312417431793e-06,
"loss": 0.0152,
"step": 456
},
{
"epoch": 2.13,
"learning_rate": 4.142498834294555e-06,
"loss": 0.0235,
"step": 457
},
{
"epoch": 2.13,
"learning_rate": 4.1018332475670795e-06,
"loss": 0.021,
"step": 458
},
{
"epoch": 2.13,
"learning_rate": 4.06131668471153e-06,
"loss": 0.0194,
"step": 459
},
{
"epoch": 2.14,
"learning_rate": 4.020950169424815e-06,
"loss": 0.0193,
"step": 460
},
{
"epoch": 2.14,
"learning_rate": 3.980734721612718e-06,
"loss": 0.0181,
"step": 461
},
{
"epoch": 2.15,
"learning_rate": 3.940671357364137e-06,
"loss": 0.022,
"step": 462
},
{
"epoch": 2.15,
"learning_rate": 3.900761088925411e-06,
"loss": 0.0134,
"step": 463
},
{
"epoch": 2.16,
"learning_rate": 3.861004924674724e-06,
"loss": 0.0183,
"step": 464
},
{
"epoch": 2.16,
"learning_rate": 3.821403869096658e-06,
"loss": 0.0191,
"step": 465
},
{
"epoch": 2.17,
"learning_rate": 3.7819589227567886e-06,
"loss": 0.0175,
"step": 466
},
{
"epoch": 2.17,
"learning_rate": 3.7426710822764234e-06,
"loss": 0.0168,
"step": 467
},
{
"epoch": 2.18,
"learning_rate": 3.7035413403073995e-06,
"loss": 0.0169,
"step": 468
},
{
"epoch": 2.18,
"learning_rate": 3.6645706855070296e-06,
"loss": 0.0175,
"step": 469
},
{
"epoch": 2.19,
"learning_rate": 3.625760102513103e-06,
"loss": 0.0175,
"step": 470
},
{
"epoch": 2.19,
"learning_rate": 3.5871105719190115e-06,
"loss": 0.0173,
"step": 471
},
{
"epoch": 2.2,
"learning_rate": 3.5486230702489767e-06,
"loss": 0.0215,
"step": 472
},
{
"epoch": 2.2,
"learning_rate": 3.5102985699333713e-06,
"loss": 0.0201,
"step": 473
},
{
"epoch": 2.2,
"learning_rate": 3.4721380392841632e-06,
"loss": 0.0233,
"step": 474
},
{
"epoch": 2.21,
"learning_rate": 3.4341424424704373e-06,
"loss": 0.0213,
"step": 475
},
{
"epoch": 2.21,
"learning_rate": 3.396312739494032e-06,
"loss": 0.0279,
"step": 476
},
{
"epoch": 2.22,
"learning_rate": 3.3586498861653004e-06,
"loss": 0.0195,
"step": 477
},
{
"epoch": 2.22,
"learning_rate": 3.3211548340789366e-06,
"loss": 0.0187,
"step": 478
},
{
"epoch": 2.23,
"learning_rate": 3.2838285305899586e-06,
"loss": 0.0167,
"step": 479
},
{
"epoch": 2.23,
"learning_rate": 3.2466719187897555e-06,
"loss": 0.0178,
"step": 480
},
{
"epoch": 2.24,
"learning_rate": 3.2096859374822587e-06,
"loss": 0.018,
"step": 481
},
{
"epoch": 2.24,
"learning_rate": 3.172871521160236e-06,
"loss": 0.018,
"step": 482
},
{
"epoch": 2.25,
"learning_rate": 3.1362295999816617e-06,
"loss": 0.0172,
"step": 483
},
{
"epoch": 2.25,
"learning_rate": 3.0997610997462292e-06,
"loss": 0.0158,
"step": 484
},
{
"epoch": 2.26,
"learning_rate": 3.063466941871952e-06,
"loss": 0.0186,
"step": 485
},
{
"epoch": 2.26,
"learning_rate": 3.027348043371893e-06,
"loss": 0.0178,
"step": 486
},
{
"epoch": 2.27,
"learning_rate": 2.991405316830985e-06,
"loss": 0.0155,
"step": 487
},
{
"epoch": 2.27,
"learning_rate": 2.9556396703829705e-06,
"loss": 0.0144,
"step": 488
},
{
"epoch": 2.27,
"learning_rate": 2.920052007687475e-06,
"loss": 0.0181,
"step": 489
},
{
"epoch": 2.28,
"learning_rate": 2.884643227907147e-06,
"loss": 0.0132,
"step": 490
},
{
"epoch": 2.28,
"learning_rate": 2.8494142256849676e-06,
"loss": 0.0182,
"step": 491
},
{
"epoch": 2.29,
"learning_rate": 2.814365891121634e-06,
"loss": 0.0168,
"step": 492
},
{
"epoch": 2.29,
"learning_rate": 2.77949910975306e-06,
"loss": 0.0153,
"step": 493
},
{
"epoch": 2.3,
"learning_rate": 2.744814762528022e-06,
"loss": 0.0155,
"step": 494
},
{
"epoch": 2.3,
"learning_rate": 2.7103137257858867e-06,
"loss": 0.0192,
"step": 495
},
{
"epoch": 2.31,
"learning_rate": 2.67599687123447e-06,
"loss": 0.0208,
"step": 496
},
{
"epoch": 2.31,
"learning_rate": 2.6418650659280253e-06,
"loss": 0.021,
"step": 497
},
{
"epoch": 2.32,
"learning_rate": 2.6079191722453157e-06,
"loss": 0.0201,
"step": 498
},
{
"epoch": 2.32,
"learning_rate": 2.5741600478678474e-06,
"loss": 0.0199,
"step": 499
},
{
"epoch": 2.33,
"learning_rate": 2.5405885457581793e-06,
"loss": 0.0165,
"step": 500
},
{
"epoch": 2.33,
"learning_rate": 2.5072055141383876e-06,
"loss": 0.0199,
"step": 501
},
{
"epoch": 2.33,
"learning_rate": 2.4740117964686218e-06,
"loss": 0.0201,
"step": 502
},
{
"epoch": 2.34,
"learning_rate": 2.4410082314258034e-06,
"loss": 0.0174,
"step": 503
},
{
"epoch": 2.34,
"learning_rate": 2.408195652882436e-06,
"loss": 0.0189,
"step": 504
},
{
"epoch": 2.35,
"learning_rate": 2.37557488988552e-06,
"loss": 0.0148,
"step": 505
},
{
"epoch": 2.35,
"learning_rate": 2.3431467666356356e-06,
"loss": 0.017,
"step": 506
},
{
"epoch": 2.36,
"learning_rate": 2.3109121024660873e-06,
"loss": 0.0171,
"step": 507
},
{
"epoch": 2.36,
"learning_rate": 2.2788717118222203e-06,
"loss": 0.021,
"step": 508
},
{
"epoch": 2.37,
"learning_rate": 2.2470264042408452e-06,
"loss": 0.0183,
"step": 509
},
{
"epoch": 2.37,
"learning_rate": 2.215376984329767e-06,
"loss": 0.0144,
"step": 510
},
{
"epoch": 2.38,
"learning_rate": 2.183924251747477e-06,
"loss": 0.0179,
"step": 511
},
{
"epoch": 2.38,
"learning_rate": 2.1526690011829297e-06,
"loss": 0.0197,
"step": 512
},
{
"epoch": 2.39,
"learning_rate": 2.1216120223354817e-06,
"loss": 0.0158,
"step": 513
},
{
"epoch": 2.39,
"learning_rate": 2.090754099894917e-06,
"loss": 0.0185,
"step": 514
},
{
"epoch": 2.4,
"learning_rate": 2.0600960135216463e-06,
"loss": 0.0138,
"step": 515
},
{
"epoch": 2.4,
"learning_rate": 2.02963853782699e-06,
"loss": 0.0156,
"step": 516
},
{
"epoch": 2.4,
"learning_rate": 1.99938244235361e-06,
"loss": 0.022,
"step": 517
},
{
"epoch": 2.41,
"learning_rate": 1.969328491556074e-06,
"loss": 0.0201,
"step": 518
},
{
"epoch": 2.41,
"learning_rate": 1.93947744478153e-06,
"loss": 0.0171,
"step": 519
},
{
"epoch": 2.42,
"learning_rate": 1.9098300562505266e-06,
"loss": 0.0167,
"step": 520
},
{
"epoch": 2.42,
"learning_rate": 1.8803870750379605e-06,
"loss": 0.0173,
"step": 521
},
{
"epoch": 2.43,
"learning_rate": 1.8511492450541379e-06,
"loss": 0.0201,
"step": 522
},
{
"epoch": 2.43,
"learning_rate": 1.8221173050259976e-06,
"loss": 0.0171,
"step": 523
},
{
"epoch": 2.44,
"learning_rate": 1.7932919884784216e-06,
"loss": 0.0172,
"step": 524
},
{
"epoch": 2.44,
"learning_rate": 1.7646740237157256e-06,
"loss": 0.0215,
"step": 525
},
{
"epoch": 2.45,
"learning_rate": 1.736264133803246e-06,
"loss": 0.0174,
"step": 526
},
{
"epoch": 2.45,
"learning_rate": 1.7080630365490658e-06,
"loss": 0.0181,
"step": 527
},
{
"epoch": 2.46,
"learning_rate": 1.6800714444858945e-06,
"loss": 0.0224,
"step": 528
},
{
"epoch": 2.46,
"learning_rate": 1.652290064853045e-06,
"loss": 0.0203,
"step": 529
},
{
"epoch": 2.47,
"learning_rate": 1.6247195995785836e-06,
"loss": 0.0197,
"step": 530
},
{
"epoch": 2.47,
"learning_rate": 1.5973607452615802e-06,
"loss": 0.0167,
"step": 531
},
{
"epoch": 2.47,
"learning_rate": 1.5702141931545144e-06,
"loss": 0.0212,
"step": 532
},
{
"epoch": 2.48,
"learning_rate": 1.5432806291458146e-06,
"loss": 0.0196,
"step": 533
},
{
"epoch": 2.48,
"learning_rate": 1.5165607337425158e-06,
"loss": 0.0205,
"step": 534
},
{
"epoch": 2.49,
"learning_rate": 1.490055182053083e-06,
"loss": 0.0186,
"step": 535
},
{
"epoch": 2.49,
"learning_rate": 1.4637646437703335e-06,
"loss": 0.0177,
"step": 536
},
{
"epoch": 2.5,
"learning_rate": 1.4376897831545345e-06,
"loss": 0.0202,
"step": 537
},
{
"epoch": 2.5,
"learning_rate": 1.4118312590166128e-06,
"loss": 0.0176,
"step": 538
},
{
"epoch": 2.51,
"learning_rate": 1.386189724701501e-06,
"loss": 0.0166,
"step": 539
},
{
"epoch": 2.51,
"learning_rate": 1.3607658280716474e-06,
"loss": 0.0246,
"step": 540
},
{
"epoch": 2.52,
"learning_rate": 1.3355602114906285e-06,
"loss": 0.0159,
"step": 541
},
{
"epoch": 2.52,
"learning_rate": 1.3105735118069373e-06,
"loss": 0.0186,
"step": 542
},
{
"epoch": 2.53,
"learning_rate": 1.2858063603378724e-06,
"loss": 0.0165,
"step": 543
},
{
"epoch": 2.53,
"learning_rate": 1.261259382853608e-06,
"loss": 0.0188,
"step": 544
},
{
"epoch": 2.53,
"learning_rate": 1.2369331995613664e-06,
"loss": 0.0148,
"step": 545
},
{
"epoch": 2.54,
"learning_rate": 1.2128284250897526e-06,
"loss": 0.0222,
"step": 546
},
{
"epoch": 2.54,
"learning_rate": 1.1889456684732358e-06,
"loss": 0.0206,
"step": 547
},
{
"epoch": 2.55,
"learning_rate": 1.165285533136743e-06,
"loss": 0.0162,
"step": 548
},
{
"epoch": 2.55,
"learning_rate": 1.1418486168804255e-06,
"loss": 0.019,
"step": 549
},
{
"epoch": 2.56,
"learning_rate": 1.1186355118645552e-06,
"loss": 0.0214,
"step": 550
},
{
"epoch": 2.56,
"learning_rate": 1.0956468045945512e-06,
"loss": 0.0155,
"step": 551
},
{
"epoch": 2.57,
"learning_rate": 1.0728830759061781e-06,
"loss": 0.0186,
"step": 552
},
{
"epoch": 2.57,
"learning_rate": 1.0503449009508516e-06,
"loss": 0.0178,
"step": 553
},
{
"epoch": 2.58,
"learning_rate": 1.0280328491811265e-06,
"loss": 0.0153,
"step": 554
},
{
"epoch": 2.58,
"learning_rate": 1.0059474843362893e-06,
"loss": 0.0171,
"step": 555
},
{
"epoch": 2.59,
"learning_rate": 9.840893644281347e-07,
"loss": 0.0173,
"step": 556
},
{
"epoch": 2.59,
"learning_rate": 9.624590417268475e-07,
"loss": 0.0155,
"step": 557
},
{
"epoch": 2.6,
"learning_rate": 9.410570627470594e-07,
"loss": 0.0158,
"step": 558
},
{
"epoch": 2.6,
"learning_rate": 9.198839682340432e-07,
"loss": 0.0174,
"step": 559
},
{
"epoch": 2.6,
"learning_rate": 8.989402931500434e-07,
"loss": 0.0202,
"step": 560
},
{
"epoch": 2.61,
"learning_rate": 8.782265666607614e-07,
"loss": 0.0204,
"step": 561
},
{
"epoch": 2.61,
"learning_rate": 8.577433121219903e-07,
"loss": 0.024,
"step": 562
},
{
"epoch": 2.62,
"learning_rate": 8.374910470663821e-07,
"loss": 0.0172,
"step": 563
},
{
"epoch": 2.62,
"learning_rate": 8.174702831903836e-07,
"loss": 0.0147,
"step": 564
},
{
"epoch": 2.63,
"learning_rate": 7.976815263412963e-07,
"loss": 0.0205,
"step": 565
},
{
"epoch": 2.63,
"learning_rate": 7.781252765045078e-07,
"loss": 0.0201,
"step": 566
},
{
"epoch": 2.64,
"learning_rate": 7.588020277908426e-07,
"loss": 0.018,
"step": 567
},
{
"epoch": 2.64,
"learning_rate": 7.397122684240998e-07,
"loss": 0.0195,
"step": 568
},
{
"epoch": 2.65,
"learning_rate": 7.208564807286945e-07,
"loss": 0.0183,
"step": 569
},
{
"epoch": 2.65,
"learning_rate": 7.022351411174866e-07,
"loss": 0.0202,
"step": 570
},
{
"epoch": 2.66,
"learning_rate": 6.838487200797406e-07,
"loss": 0.0158,
"step": 571
},
{
"epoch": 2.66,
"learning_rate": 6.656976821692374e-07,
"loss": 0.0176,
"step": 572
},
{
"epoch": 2.67,
"learning_rate": 6.477824859925297e-07,
"loss": 0.018,
"step": 573
},
{
"epoch": 2.67,
"learning_rate": 6.301035841973702e-07,
"loss": 0.0239,
"step": 574
},
{
"epoch": 2.67,
"learning_rate": 6.126614234612593e-07,
"loss": 0.0159,
"step": 575
},
{
"epoch": 2.68,
"learning_rate": 5.954564444801703e-07,
"loss": 0.0203,
"step": 576
},
{
"epoch": 2.68,
"learning_rate": 5.784890819574074e-07,
"loss": 0.0207,
"step": 577
},
{
"epoch": 2.69,
"learning_rate": 5.617597645926265e-07,
"loss": 0.0215,
"step": 578
},
{
"epoch": 2.69,
"learning_rate": 5.452689150710055e-07,
"loss": 0.0161,
"step": 579
},
{
"epoch": 2.7,
"learning_rate": 5.290169500525577e-07,
"loss": 0.0204,
"step": 580
},
{
"epoch": 2.7,
"learning_rate": 5.130042801616109e-07,
"loss": 0.0196,
"step": 581
},
{
"epoch": 2.71,
"learning_rate": 4.97231309976427e-07,
"loss": 0.0194,
"step": 582
},
{
"epoch": 2.71,
"learning_rate": 4.81698438018986e-07,
"loss": 0.018,
"step": 583
},
{
"epoch": 2.72,
"learning_rate": 4.664060567449169e-07,
"loss": 0.0177,
"step": 584
},
{
"epoch": 2.72,
"learning_rate": 4.5135455253357053e-07,
"loss": 0.0197,
"step": 585
},
{
"epoch": 2.73,
"learning_rate": 4.3654430567827367e-07,
"loss": 0.0208,
"step": 586
},
{
"epoch": 2.73,
"learning_rate": 4.2197569037670584e-07,
"loss": 0.0151,
"step": 587
},
{
"epoch": 2.73,
"learning_rate": 4.0764907472145677e-07,
"loss": 0.0188,
"step": 588
},
{
"epoch": 2.74,
"learning_rate": 3.935648206907194e-07,
"loss": 0.0181,
"step": 589
},
{
"epoch": 2.74,
"learning_rate": 3.7972328413914074e-07,
"loss": 0.0148,
"step": 590
},
{
"epoch": 2.75,
"learning_rate": 3.661248147888419e-07,
"loss": 0.0173,
"step": 591
},
{
"epoch": 2.75,
"learning_rate": 3.527697562205723e-07,
"loss": 0.0174,
"step": 592
},
{
"epoch": 2.76,
"learning_rate": 3.3965844586503095e-07,
"loss": 0.0155,
"step": 593
},
{
"epoch": 2.76,
"learning_rate": 3.267912149943431e-07,
"loss": 0.0176,
"step": 594
},
{
"epoch": 2.77,
"learning_rate": 3.1416838871368925e-07,
"loss": 0.0157,
"step": 595
},
{
"epoch": 2.77,
"learning_rate": 3.017902859530919e-07,
"loss": 0.0166,
"step": 596
},
{
"epoch": 2.78,
"learning_rate": 2.896572194593528e-07,
"loss": 0.0173,
"step": 597
},
{
"epoch": 2.78,
"learning_rate": 2.777694957881594e-07,
"loss": 0.0173,
"step": 598
},
{
"epoch": 2.79,
"learning_rate": 2.661274152963278e-07,
"loss": 0.0179,
"step": 599
},
{
"epoch": 2.79,
"learning_rate": 2.547312721342277e-07,
"loss": 0.0221,
"step": 600
},
{
"epoch": 2.8,
"learning_rate": 2.435813542383403e-07,
"loss": 0.0207,
"step": 601
},
{
"epoch": 2.8,
"learning_rate": 2.3267794332398454e-07,
"loss": 0.0208,
"step": 602
},
{
"epoch": 2.8,
"learning_rate": 2.2202131487820244e-07,
"loss": 0.0157,
"step": 603
},
{
"epoch": 2.81,
"learning_rate": 2.1161173815279824e-07,
"loss": 0.0195,
"step": 604
},
{
"epoch": 2.81,
"learning_rate": 2.014494761575314e-07,
"loss": 0.0129,
"step": 605
},
{
"epoch": 2.82,
"learning_rate": 1.9153478565347773e-07,
"loss": 0.0157,
"step": 606
},
{
"epoch": 2.82,
"learning_rate": 1.8186791714653428e-07,
"loss": 0.0168,
"step": 607
},
{
"epoch": 2.83,
"learning_rate": 1.7244911488109916e-07,
"loss": 0.0188,
"step": 608
},
{
"epoch": 2.83,
"learning_rate": 1.6327861683389179e-07,
"loss": 0.0182,
"step": 609
},
{
"epoch": 2.84,
"learning_rate": 1.543566547079467e-07,
"loss": 0.0207,
"step": 610
},
{
"epoch": 2.84,
"learning_rate": 1.4568345392675375e-07,
"loss": 0.0173,
"step": 611
},
{
"epoch": 2.85,
"learning_rate": 1.3725923362856607e-07,
"loss": 0.0157,
"step": 612
},
{
"epoch": 2.85,
"learning_rate": 1.290842066608655e-07,
"loss": 0.0188,
"step": 613
},
{
"epoch": 2.86,
"learning_rate": 1.2115857957497813e-07,
"loss": 0.0199,
"step": 614
},
{
"epoch": 2.86,
"learning_rate": 1.134825526208605e-07,
"loss": 0.0182,
"step": 615
},
{
"epoch": 2.87,
"learning_rate": 1.0605631974203834e-07,
"loss": 0.0173,
"step": 616
},
{
"epoch": 2.87,
"learning_rate": 9.888006857070698e-08,
"loss": 0.016,
"step": 617
},
{
"epoch": 2.87,
"learning_rate": 9.195398042299298e-08,
"loss": 0.0161,
"step": 618
},
{
"epoch": 2.88,
"learning_rate": 8.527823029436444e-08,
"loss": 0.0178,
"step": 619
},
{
"epoch": 2.88,
"learning_rate": 7.885298685522235e-08,
"loss": 0.0172,
"step": 620
},
{
"epoch": 2.89,
"learning_rate": 7.267841244662622e-08,
"loss": 0.0173,
"step": 621
},
{
"epoch": 2.89,
"learning_rate": 6.675466307620282e-08,
"loss": 0.0171,
"step": 622
},
{
"epoch": 2.9,
"learning_rate": 6.108188841419727e-08,
"loss": 0.0175,
"step": 623
},
{
"epoch": 2.9,
"learning_rate": 5.566023178969482e-08,
"loss": 0.0269,
"step": 624
},
{
"epoch": 2.91,
"learning_rate": 5.048983018699827e-08,
"loss": 0.0169,
"step": 625
},
{
"epoch": 2.91,
"learning_rate": 4.557081424216847e-08,
"loss": 0.0157,
"step": 626
},
{
"epoch": 2.92,
"learning_rate": 4.090330823972033e-08,
"loss": 0.0157,
"step": 627
},
{
"epoch": 2.92,
"learning_rate": 3.648743010948863e-08,
"loss": 0.0159,
"step": 628
},
{
"epoch": 2.93,
"learning_rate": 3.232329142363933e-08,
"loss": 0.0178,
"step": 629
},
{
"epoch": 2.93,
"learning_rate": 2.8410997393860663e-08,
"loss": 0.0164,
"step": 630
},
{
"epoch": 2.93,
"learning_rate": 2.4750646868697546e-08,
"loss": 0.0187,
"step": 631
},
{
"epoch": 2.94,
"learning_rate": 2.134233233105798e-08,
"loss": 0.0184,
"step": 632
},
{
"epoch": 2.94,
"learning_rate": 1.81861398958727e-08,
"loss": 0.0148,
"step": 633
},
{
"epoch": 2.95,
"learning_rate": 1.5282149307922488e-08,
"loss": 0.018,
"step": 634
},
{
"epoch": 2.95,
"learning_rate": 1.2630433939825326e-08,
"loss": 0.0162,
"step": 635
},
{
"epoch": 2.96,
"learning_rate": 1.0231060790174552e-08,
"loss": 0.0152,
"step": 636
},
{
"epoch": 2.96,
"learning_rate": 8.084090481855766e-09,
"loss": 0.0169,
"step": 637
},
{
"epoch": 2.97,
"learning_rate": 6.189577260508062e-09,
"loss": 0.0168,
"step": 638
},
{
"epoch": 2.97,
"learning_rate": 4.5475689931551155e-09,
"loss": 0.0194,
"step": 639
},
{
"epoch": 2.98,
"learning_rate": 3.1581071670006013e-09,
"loss": 0.0199,
"step": 640
},
{
"epoch": 2.98,
"learning_rate": 2.0212268883701424e-09,
"loss": 0.0169,
"step": 641
},
{
"epoch": 2.99,
"learning_rate": 1.1369568818364596e-09,
"loss": 0.0161,
"step": 642
},
{
"epoch": 2.99,
"learning_rate": 5.053194894821811e-10,
"loss": 0.015,
"step": 643
},
{
"epoch": 3.0,
"learning_rate": 1.2633067034251156e-10,
"loss": 0.0139,
"step": 644
},
{
"epoch": 3.0,
"learning_rate": 0.0,
"loss": 0.0186,
"step": 645
},
{
"epoch": 3.0,
"step": 645,
"total_flos": 0.0,
"train_loss": 0.031338086711286114,
"train_runtime": 10862.5621,
"train_samples_per_second": 6.647,
"train_steps_per_second": 0.059
}
],
"logging_steps": 1.0,
"max_steps": 645,
"num_input_tokens_seen": 0,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 0.0,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}