race_cot_128 / trainer_state.json
brettbbb's picture
End of training
2e2eaef
raw
history blame contribute delete
No virus
77.6 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 20.0,
"eval_steps": 500,
"global_step": 640,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.03,
"learning_rate": 2e-05,
"loss": 2.5069,
"step": 1
},
{
"epoch": 0.06,
"learning_rate": 4e-05,
"loss": 2.1327,
"step": 2
},
{
"epoch": 0.09,
"learning_rate": 6e-05,
"loss": 2.4537,
"step": 3
},
{
"epoch": 0.12,
"learning_rate": 8e-05,
"loss": 2.3612,
"step": 4
},
{
"epoch": 0.16,
"learning_rate": 0.0001,
"loss": 2.3904,
"step": 5
},
{
"epoch": 0.19,
"learning_rate": 9.984251968503937e-05,
"loss": 2.0011,
"step": 6
},
{
"epoch": 0.22,
"learning_rate": 9.968503937007875e-05,
"loss": 2.1121,
"step": 7
},
{
"epoch": 0.25,
"learning_rate": 9.952755905511811e-05,
"loss": 1.918,
"step": 8
},
{
"epoch": 0.28,
"learning_rate": 9.937007874015748e-05,
"loss": 1.8568,
"step": 9
},
{
"epoch": 0.31,
"learning_rate": 9.921259842519686e-05,
"loss": 1.7663,
"step": 10
},
{
"epoch": 0.34,
"learning_rate": 9.905511811023622e-05,
"loss": 1.6871,
"step": 11
},
{
"epoch": 0.38,
"learning_rate": 9.88976377952756e-05,
"loss": 1.5877,
"step": 12
},
{
"epoch": 0.41,
"learning_rate": 9.874015748031497e-05,
"loss": 1.897,
"step": 13
},
{
"epoch": 0.44,
"learning_rate": 9.874015748031497e-05,
"loss": 1.7996,
"step": 14
},
{
"epoch": 0.47,
"learning_rate": 9.858267716535433e-05,
"loss": 1.8437,
"step": 15
},
{
"epoch": 0.5,
"learning_rate": 9.842519685039371e-05,
"loss": 1.7606,
"step": 16
},
{
"epoch": 0.53,
"learning_rate": 9.826771653543308e-05,
"loss": 1.7194,
"step": 17
},
{
"epoch": 0.56,
"learning_rate": 9.811023622047244e-05,
"loss": 1.8213,
"step": 18
},
{
"epoch": 0.59,
"learning_rate": 9.795275590551182e-05,
"loss": 1.6995,
"step": 19
},
{
"epoch": 0.62,
"learning_rate": 9.779527559055119e-05,
"loss": 1.5309,
"step": 20
},
{
"epoch": 0.66,
"learning_rate": 9.763779527559055e-05,
"loss": 1.4977,
"step": 21
},
{
"epoch": 0.69,
"learning_rate": 9.748031496062993e-05,
"loss": 1.6145,
"step": 22
},
{
"epoch": 0.72,
"learning_rate": 9.73228346456693e-05,
"loss": 1.6686,
"step": 23
},
{
"epoch": 0.75,
"learning_rate": 9.716535433070866e-05,
"loss": 1.3646,
"step": 24
},
{
"epoch": 0.78,
"learning_rate": 9.700787401574803e-05,
"loss": 1.5323,
"step": 25
},
{
"epoch": 0.81,
"learning_rate": 9.68503937007874e-05,
"loss": 1.4045,
"step": 26
},
{
"epoch": 0.84,
"learning_rate": 9.669291338582677e-05,
"loss": 1.6301,
"step": 27
},
{
"epoch": 0.88,
"learning_rate": 9.653543307086614e-05,
"loss": 1.4687,
"step": 28
},
{
"epoch": 0.91,
"learning_rate": 9.637795275590552e-05,
"loss": 1.4465,
"step": 29
},
{
"epoch": 0.94,
"learning_rate": 9.622047244094488e-05,
"loss": 1.3959,
"step": 30
},
{
"epoch": 0.97,
"learning_rate": 9.606299212598425e-05,
"loss": 1.4754,
"step": 31
},
{
"epoch": 1.0,
"learning_rate": 9.590551181102363e-05,
"loss": 1.445,
"step": 32
},
{
"epoch": 1.03,
"learning_rate": 9.574803149606299e-05,
"loss": 1.4998,
"step": 33
},
{
"epoch": 1.06,
"learning_rate": 9.559055118110236e-05,
"loss": 1.5282,
"step": 34
},
{
"epoch": 1.09,
"learning_rate": 9.543307086614174e-05,
"loss": 1.1835,
"step": 35
},
{
"epoch": 1.12,
"learning_rate": 9.52755905511811e-05,
"loss": 1.3461,
"step": 36
},
{
"epoch": 1.16,
"learning_rate": 9.511811023622048e-05,
"loss": 1.2879,
"step": 37
},
{
"epoch": 1.19,
"learning_rate": 9.496062992125985e-05,
"loss": 1.293,
"step": 38
},
{
"epoch": 1.22,
"learning_rate": 9.480314960629921e-05,
"loss": 1.2846,
"step": 39
},
{
"epoch": 1.25,
"learning_rate": 9.464566929133859e-05,
"loss": 1.3323,
"step": 40
},
{
"epoch": 1.28,
"learning_rate": 9.448818897637796e-05,
"loss": 1.3537,
"step": 41
},
{
"epoch": 1.31,
"learning_rate": 9.433070866141732e-05,
"loss": 1.3553,
"step": 42
},
{
"epoch": 1.34,
"learning_rate": 9.41732283464567e-05,
"loss": 1.2661,
"step": 43
},
{
"epoch": 1.38,
"learning_rate": 9.401574803149607e-05,
"loss": 1.345,
"step": 44
},
{
"epoch": 1.41,
"learning_rate": 9.385826771653545e-05,
"loss": 1.2593,
"step": 45
},
{
"epoch": 1.44,
"learning_rate": 9.370078740157481e-05,
"loss": 1.1177,
"step": 46
},
{
"epoch": 1.47,
"learning_rate": 9.354330708661418e-05,
"loss": 1.2686,
"step": 47
},
{
"epoch": 1.5,
"learning_rate": 9.338582677165355e-05,
"loss": 1.1689,
"step": 48
},
{
"epoch": 1.53,
"learning_rate": 9.322834645669292e-05,
"loss": 1.145,
"step": 49
},
{
"epoch": 1.56,
"learning_rate": 9.307086614173229e-05,
"loss": 1.5517,
"step": 50
},
{
"epoch": 1.59,
"learning_rate": 9.291338582677166e-05,
"loss": 1.1943,
"step": 51
},
{
"epoch": 1.62,
"learning_rate": 9.275590551181103e-05,
"loss": 1.2215,
"step": 52
},
{
"epoch": 1.66,
"learning_rate": 9.259842519685041e-05,
"loss": 1.1956,
"step": 53
},
{
"epoch": 1.69,
"learning_rate": 9.244094488188977e-05,
"loss": 1.1969,
"step": 54
},
{
"epoch": 1.72,
"learning_rate": 9.228346456692914e-05,
"loss": 1.2588,
"step": 55
},
{
"epoch": 1.75,
"learning_rate": 9.21259842519685e-05,
"loss": 1.2618,
"step": 56
},
{
"epoch": 1.78,
"learning_rate": 9.196850393700787e-05,
"loss": 1.0742,
"step": 57
},
{
"epoch": 1.81,
"learning_rate": 9.181102362204725e-05,
"loss": 1.2993,
"step": 58
},
{
"epoch": 1.84,
"learning_rate": 9.165354330708661e-05,
"loss": 1.229,
"step": 59
},
{
"epoch": 1.88,
"learning_rate": 9.149606299212598e-05,
"loss": 1.5104,
"step": 60
},
{
"epoch": 1.91,
"learning_rate": 9.133858267716536e-05,
"loss": 1.34,
"step": 61
},
{
"epoch": 1.94,
"learning_rate": 9.118110236220472e-05,
"loss": 1.3365,
"step": 62
},
{
"epoch": 1.97,
"learning_rate": 9.102362204724409e-05,
"loss": 1.3477,
"step": 63
},
{
"epoch": 2.0,
"learning_rate": 9.086614173228347e-05,
"loss": 1.1649,
"step": 64
},
{
"epoch": 2.03,
"learning_rate": 9.070866141732283e-05,
"loss": 1.0887,
"step": 65
},
{
"epoch": 2.06,
"learning_rate": 9.05511811023622e-05,
"loss": 1.0539,
"step": 66
},
{
"epoch": 2.09,
"learning_rate": 9.039370078740158e-05,
"loss": 1.2224,
"step": 67
},
{
"epoch": 2.12,
"learning_rate": 9.023622047244094e-05,
"loss": 0.7727,
"step": 68
},
{
"epoch": 2.16,
"learning_rate": 9.007874015748032e-05,
"loss": 0.8996,
"step": 69
},
{
"epoch": 2.19,
"learning_rate": 8.992125984251969e-05,
"loss": 0.8348,
"step": 70
},
{
"epoch": 2.22,
"learning_rate": 8.976377952755905e-05,
"loss": 0.9522,
"step": 71
},
{
"epoch": 2.25,
"learning_rate": 8.960629921259843e-05,
"loss": 0.9277,
"step": 72
},
{
"epoch": 2.28,
"learning_rate": 8.94488188976378e-05,
"loss": 0.9566,
"step": 73
},
{
"epoch": 2.31,
"learning_rate": 8.929133858267716e-05,
"loss": 0.9179,
"step": 74
},
{
"epoch": 2.34,
"learning_rate": 8.913385826771654e-05,
"loss": 0.9606,
"step": 75
},
{
"epoch": 2.38,
"learning_rate": 8.897637795275591e-05,
"loss": 0.8982,
"step": 76
},
{
"epoch": 2.41,
"learning_rate": 8.881889763779529e-05,
"loss": 0.9102,
"step": 77
},
{
"epoch": 2.44,
"learning_rate": 8.866141732283465e-05,
"loss": 0.8879,
"step": 78
},
{
"epoch": 2.47,
"learning_rate": 8.850393700787402e-05,
"loss": 0.6698,
"step": 79
},
{
"epoch": 2.5,
"learning_rate": 8.83464566929134e-05,
"loss": 0.8272,
"step": 80
},
{
"epoch": 2.53,
"learning_rate": 8.818897637795276e-05,
"loss": 1.0421,
"step": 81
},
{
"epoch": 2.56,
"learning_rate": 8.803149606299213e-05,
"loss": 0.9424,
"step": 82
},
{
"epoch": 2.59,
"learning_rate": 8.78740157480315e-05,
"loss": 0.9943,
"step": 83
},
{
"epoch": 2.62,
"learning_rate": 8.771653543307087e-05,
"loss": 1.1647,
"step": 84
},
{
"epoch": 2.66,
"learning_rate": 8.755905511811025e-05,
"loss": 1.0718,
"step": 85
},
{
"epoch": 2.69,
"learning_rate": 8.740157480314962e-05,
"loss": 0.9349,
"step": 86
},
{
"epoch": 2.72,
"learning_rate": 8.724409448818898e-05,
"loss": 1.014,
"step": 87
},
{
"epoch": 2.75,
"learning_rate": 8.708661417322835e-05,
"loss": 1.2487,
"step": 88
},
{
"epoch": 2.78,
"learning_rate": 8.692913385826773e-05,
"loss": 0.8633,
"step": 89
},
{
"epoch": 2.81,
"learning_rate": 8.677165354330709e-05,
"loss": 0.9087,
"step": 90
},
{
"epoch": 2.84,
"learning_rate": 8.661417322834646e-05,
"loss": 1.1013,
"step": 91
},
{
"epoch": 2.88,
"learning_rate": 8.645669291338582e-05,
"loss": 1.0405,
"step": 92
},
{
"epoch": 2.91,
"learning_rate": 8.62992125984252e-05,
"loss": 0.7865,
"step": 93
},
{
"epoch": 2.94,
"learning_rate": 8.614173228346457e-05,
"loss": 0.9032,
"step": 94
},
{
"epoch": 2.97,
"learning_rate": 8.598425196850393e-05,
"loss": 0.9319,
"step": 95
},
{
"epoch": 3.0,
"learning_rate": 8.582677165354331e-05,
"loss": 0.8459,
"step": 96
},
{
"epoch": 3.03,
"learning_rate": 8.566929133858268e-05,
"loss": 0.5466,
"step": 97
},
{
"epoch": 3.06,
"learning_rate": 8.551181102362204e-05,
"loss": 0.4219,
"step": 98
},
{
"epoch": 3.09,
"learning_rate": 8.535433070866142e-05,
"loss": 0.7185,
"step": 99
},
{
"epoch": 3.12,
"learning_rate": 8.519685039370079e-05,
"loss": 0.5795,
"step": 100
},
{
"epoch": 3.16,
"learning_rate": 8.503937007874016e-05,
"loss": 0.5285,
"step": 101
},
{
"epoch": 3.19,
"learning_rate": 8.488188976377953e-05,
"loss": 0.6236,
"step": 102
},
{
"epoch": 3.22,
"learning_rate": 8.47244094488189e-05,
"loss": 0.6287,
"step": 103
},
{
"epoch": 3.25,
"learning_rate": 8.456692913385827e-05,
"loss": 0.4723,
"step": 104
},
{
"epoch": 3.28,
"learning_rate": 8.440944881889764e-05,
"loss": 0.4926,
"step": 105
},
{
"epoch": 3.31,
"learning_rate": 8.4251968503937e-05,
"loss": 0.5282,
"step": 106
},
{
"epoch": 3.34,
"learning_rate": 8.409448818897638e-05,
"loss": 0.4181,
"step": 107
},
{
"epoch": 3.38,
"learning_rate": 8.393700787401575e-05,
"loss": 0.5563,
"step": 108
},
{
"epoch": 3.41,
"learning_rate": 8.377952755905513e-05,
"loss": 0.3991,
"step": 109
},
{
"epoch": 3.44,
"learning_rate": 8.36220472440945e-05,
"loss": 0.4067,
"step": 110
},
{
"epoch": 3.47,
"learning_rate": 8.346456692913386e-05,
"loss": 0.538,
"step": 111
},
{
"epoch": 3.5,
"learning_rate": 8.330708661417324e-05,
"loss": 0.5638,
"step": 112
},
{
"epoch": 3.53,
"learning_rate": 8.31496062992126e-05,
"loss": 0.4859,
"step": 113
},
{
"epoch": 3.56,
"learning_rate": 8.299212598425197e-05,
"loss": 0.4579,
"step": 114
},
{
"epoch": 3.59,
"learning_rate": 8.283464566929135e-05,
"loss": 0.7117,
"step": 115
},
{
"epoch": 3.62,
"learning_rate": 8.267716535433071e-05,
"loss": 0.5797,
"step": 116
},
{
"epoch": 3.66,
"learning_rate": 8.251968503937009e-05,
"loss": 0.5797,
"step": 117
},
{
"epoch": 3.69,
"learning_rate": 8.236220472440946e-05,
"loss": 0.5154,
"step": 118
},
{
"epoch": 3.72,
"learning_rate": 8.220472440944882e-05,
"loss": 0.5505,
"step": 119
},
{
"epoch": 3.75,
"learning_rate": 8.20472440944882e-05,
"loss": 0.6365,
"step": 120
},
{
"epoch": 3.78,
"learning_rate": 8.188976377952757e-05,
"loss": 0.5842,
"step": 121
},
{
"epoch": 3.81,
"learning_rate": 8.173228346456693e-05,
"loss": 0.6596,
"step": 122
},
{
"epoch": 3.84,
"learning_rate": 8.15748031496063e-05,
"loss": 0.544,
"step": 123
},
{
"epoch": 3.88,
"learning_rate": 8.141732283464568e-05,
"loss": 0.6828,
"step": 124
},
{
"epoch": 3.91,
"learning_rate": 8.125984251968504e-05,
"loss": 0.5819,
"step": 125
},
{
"epoch": 3.94,
"learning_rate": 8.110236220472441e-05,
"loss": 0.5688,
"step": 126
},
{
"epoch": 3.97,
"learning_rate": 8.094488188976377e-05,
"loss": 0.5328,
"step": 127
},
{
"epoch": 4.0,
"learning_rate": 8.078740157480315e-05,
"loss": 0.5887,
"step": 128
},
{
"epoch": 4.03,
"learning_rate": 8.062992125984252e-05,
"loss": 0.3647,
"step": 129
},
{
"epoch": 4.06,
"learning_rate": 8.047244094488188e-05,
"loss": 0.3172,
"step": 130
},
{
"epoch": 4.09,
"learning_rate": 8.031496062992126e-05,
"loss": 0.3931,
"step": 131
},
{
"epoch": 4.12,
"learning_rate": 8.015748031496063e-05,
"loss": 0.2472,
"step": 132
},
{
"epoch": 4.16,
"learning_rate": 8e-05,
"loss": 0.1936,
"step": 133
},
{
"epoch": 4.19,
"learning_rate": 7.984251968503937e-05,
"loss": 0.2755,
"step": 134
},
{
"epoch": 4.22,
"learning_rate": 7.968503937007874e-05,
"loss": 0.3037,
"step": 135
},
{
"epoch": 4.25,
"learning_rate": 7.952755905511812e-05,
"loss": 0.2445,
"step": 136
},
{
"epoch": 4.28,
"learning_rate": 7.937007874015748e-05,
"loss": 0.1991,
"step": 137
},
{
"epoch": 4.31,
"learning_rate": 7.921259842519685e-05,
"loss": 0.249,
"step": 138
},
{
"epoch": 4.34,
"learning_rate": 7.905511811023623e-05,
"loss": 0.3154,
"step": 139
},
{
"epoch": 4.38,
"learning_rate": 7.889763779527559e-05,
"loss": 0.218,
"step": 140
},
{
"epoch": 4.41,
"learning_rate": 7.874015748031497e-05,
"loss": 0.3027,
"step": 141
},
{
"epoch": 4.44,
"learning_rate": 7.858267716535434e-05,
"loss": 0.2984,
"step": 142
},
{
"epoch": 4.47,
"learning_rate": 7.84251968503937e-05,
"loss": 0.2319,
"step": 143
},
{
"epoch": 4.5,
"learning_rate": 7.826771653543308e-05,
"loss": 0.2731,
"step": 144
},
{
"epoch": 4.53,
"learning_rate": 7.811023622047245e-05,
"loss": 0.2904,
"step": 145
},
{
"epoch": 4.56,
"learning_rate": 7.795275590551181e-05,
"loss": 0.2438,
"step": 146
},
{
"epoch": 4.59,
"learning_rate": 7.779527559055119e-05,
"loss": 0.3245,
"step": 147
},
{
"epoch": 4.62,
"learning_rate": 7.763779527559056e-05,
"loss": 0.2738,
"step": 148
},
{
"epoch": 4.66,
"learning_rate": 7.748031496062993e-05,
"loss": 0.2536,
"step": 149
},
{
"epoch": 4.69,
"learning_rate": 7.73228346456693e-05,
"loss": 0.3255,
"step": 150
},
{
"epoch": 4.72,
"learning_rate": 7.716535433070867e-05,
"loss": 0.2891,
"step": 151
},
{
"epoch": 4.75,
"learning_rate": 7.700787401574804e-05,
"loss": 0.2648,
"step": 152
},
{
"epoch": 4.78,
"learning_rate": 7.685039370078741e-05,
"loss": 0.2663,
"step": 153
},
{
"epoch": 4.81,
"learning_rate": 7.669291338582677e-05,
"loss": 0.318,
"step": 154
},
{
"epoch": 4.84,
"learning_rate": 7.653543307086615e-05,
"loss": 0.3181,
"step": 155
},
{
"epoch": 4.88,
"learning_rate": 7.637795275590552e-05,
"loss": 0.2486,
"step": 156
},
{
"epoch": 4.91,
"learning_rate": 7.622047244094488e-05,
"loss": 0.2911,
"step": 157
},
{
"epoch": 4.94,
"learning_rate": 7.606299212598425e-05,
"loss": 0.3189,
"step": 158
},
{
"epoch": 4.97,
"learning_rate": 7.590551181102362e-05,
"loss": 0.3104,
"step": 159
},
{
"epoch": 5.0,
"learning_rate": 7.5748031496063e-05,
"loss": 0.2286,
"step": 160
},
{
"epoch": 5.03,
"learning_rate": 7.559055118110236e-05,
"loss": 0.1934,
"step": 161
},
{
"epoch": 5.06,
"learning_rate": 7.543307086614173e-05,
"loss": 0.1535,
"step": 162
},
{
"epoch": 5.09,
"learning_rate": 7.52755905511811e-05,
"loss": 0.1791,
"step": 163
},
{
"epoch": 5.12,
"learning_rate": 7.511811023622047e-05,
"loss": 0.144,
"step": 164
},
{
"epoch": 5.16,
"learning_rate": 7.496062992125985e-05,
"loss": 0.1143,
"step": 165
},
{
"epoch": 5.19,
"learning_rate": 7.480314960629921e-05,
"loss": 0.1244,
"step": 166
},
{
"epoch": 5.22,
"learning_rate": 7.464566929133858e-05,
"loss": 0.0934,
"step": 167
},
{
"epoch": 5.25,
"learning_rate": 7.448818897637796e-05,
"loss": 0.1276,
"step": 168
},
{
"epoch": 5.28,
"learning_rate": 7.433070866141732e-05,
"loss": 0.1416,
"step": 169
},
{
"epoch": 5.31,
"learning_rate": 7.417322834645669e-05,
"loss": 0.1335,
"step": 170
},
{
"epoch": 5.34,
"learning_rate": 7.401574803149607e-05,
"loss": 0.1368,
"step": 171
},
{
"epoch": 5.38,
"learning_rate": 7.385826771653543e-05,
"loss": 0.1422,
"step": 172
},
{
"epoch": 5.41,
"learning_rate": 7.370078740157481e-05,
"loss": 0.1306,
"step": 173
},
{
"epoch": 5.44,
"learning_rate": 7.354330708661418e-05,
"loss": 0.1287,
"step": 174
},
{
"epoch": 5.47,
"learning_rate": 7.338582677165354e-05,
"loss": 0.1026,
"step": 175
},
{
"epoch": 5.5,
"learning_rate": 7.322834645669292e-05,
"loss": 0.1304,
"step": 176
},
{
"epoch": 5.53,
"learning_rate": 7.307086614173229e-05,
"loss": 0.1497,
"step": 177
},
{
"epoch": 5.56,
"learning_rate": 7.291338582677165e-05,
"loss": 0.1985,
"step": 178
},
{
"epoch": 5.59,
"learning_rate": 7.275590551181103e-05,
"loss": 0.1248,
"step": 179
},
{
"epoch": 5.62,
"learning_rate": 7.25984251968504e-05,
"loss": 0.1705,
"step": 180
},
{
"epoch": 5.66,
"learning_rate": 7.244094488188978e-05,
"loss": 0.1435,
"step": 181
},
{
"epoch": 5.69,
"learning_rate": 7.228346456692914e-05,
"loss": 0.1864,
"step": 182
},
{
"epoch": 5.72,
"learning_rate": 7.212598425196851e-05,
"loss": 0.1399,
"step": 183
},
{
"epoch": 5.75,
"learning_rate": 7.196850393700789e-05,
"loss": 0.1604,
"step": 184
},
{
"epoch": 5.78,
"learning_rate": 7.181102362204725e-05,
"loss": 0.1391,
"step": 185
},
{
"epoch": 5.81,
"learning_rate": 7.165354330708662e-05,
"loss": 0.1578,
"step": 186
},
{
"epoch": 5.84,
"learning_rate": 7.1496062992126e-05,
"loss": 0.1467,
"step": 187
},
{
"epoch": 5.88,
"learning_rate": 7.133858267716536e-05,
"loss": 0.1577,
"step": 188
},
{
"epoch": 5.91,
"learning_rate": 7.118110236220473e-05,
"loss": 0.159,
"step": 189
},
{
"epoch": 5.94,
"learning_rate": 7.102362204724409e-05,
"loss": 0.1589,
"step": 190
},
{
"epoch": 5.97,
"learning_rate": 7.086614173228347e-05,
"loss": 0.1333,
"step": 191
},
{
"epoch": 6.0,
"learning_rate": 7.070866141732284e-05,
"loss": 0.2101,
"step": 192
},
{
"epoch": 6.03,
"learning_rate": 7.05511811023622e-05,
"loss": 0.0931,
"step": 193
},
{
"epoch": 6.06,
"learning_rate": 7.039370078740157e-05,
"loss": 0.0615,
"step": 194
},
{
"epoch": 6.09,
"learning_rate": 7.023622047244095e-05,
"loss": 0.0841,
"step": 195
},
{
"epoch": 6.12,
"learning_rate": 7.007874015748031e-05,
"loss": 0.0833,
"step": 196
},
{
"epoch": 6.16,
"learning_rate": 6.992125984251969e-05,
"loss": 0.1018,
"step": 197
},
{
"epoch": 6.19,
"learning_rate": 6.976377952755906e-05,
"loss": 0.0827,
"step": 198
},
{
"epoch": 6.22,
"learning_rate": 6.960629921259842e-05,
"loss": 0.0786,
"step": 199
},
{
"epoch": 6.25,
"learning_rate": 6.94488188976378e-05,
"loss": 0.0741,
"step": 200
},
{
"epoch": 6.28,
"learning_rate": 6.929133858267717e-05,
"loss": 0.0911,
"step": 201
},
{
"epoch": 6.31,
"learning_rate": 6.913385826771653e-05,
"loss": 0.1057,
"step": 202
},
{
"epoch": 6.34,
"learning_rate": 6.897637795275591e-05,
"loss": 0.0598,
"step": 203
},
{
"epoch": 6.38,
"learning_rate": 6.881889763779528e-05,
"loss": 0.1423,
"step": 204
},
{
"epoch": 6.41,
"learning_rate": 6.866141732283465e-05,
"loss": 0.0843,
"step": 205
},
{
"epoch": 6.44,
"learning_rate": 6.850393700787402e-05,
"loss": 0.0664,
"step": 206
},
{
"epoch": 6.47,
"learning_rate": 6.834645669291338e-05,
"loss": 0.0685,
"step": 207
},
{
"epoch": 6.5,
"learning_rate": 6.818897637795276e-05,
"loss": 0.0974,
"step": 208
},
{
"epoch": 6.53,
"learning_rate": 6.803149606299213e-05,
"loss": 0.1523,
"step": 209
},
{
"epoch": 6.56,
"learning_rate": 6.78740157480315e-05,
"loss": 0.0851,
"step": 210
},
{
"epoch": 6.59,
"learning_rate": 6.771653543307087e-05,
"loss": 0.1009,
"step": 211
},
{
"epoch": 6.62,
"learning_rate": 6.755905511811024e-05,
"loss": 0.1057,
"step": 212
},
{
"epoch": 6.66,
"learning_rate": 6.740157480314962e-05,
"loss": 0.0816,
"step": 213
},
{
"epoch": 6.69,
"learning_rate": 6.724409448818898e-05,
"loss": 0.1062,
"step": 214
},
{
"epoch": 6.72,
"learning_rate": 6.708661417322835e-05,
"loss": 0.1022,
"step": 215
},
{
"epoch": 6.75,
"learning_rate": 6.692913385826773e-05,
"loss": 0.1159,
"step": 216
},
{
"epoch": 6.78,
"learning_rate": 6.677165354330709e-05,
"loss": 0.0964,
"step": 217
},
{
"epoch": 6.81,
"learning_rate": 6.661417322834646e-05,
"loss": 0.1399,
"step": 218
},
{
"epoch": 6.84,
"learning_rate": 6.645669291338584e-05,
"loss": 0.1076,
"step": 219
},
{
"epoch": 6.88,
"learning_rate": 6.62992125984252e-05,
"loss": 0.1024,
"step": 220
},
{
"epoch": 6.91,
"learning_rate": 6.614173228346457e-05,
"loss": 0.1096,
"step": 221
},
{
"epoch": 6.94,
"learning_rate": 6.598425196850395e-05,
"loss": 0.1007,
"step": 222
},
{
"epoch": 6.97,
"learning_rate": 6.582677165354331e-05,
"loss": 0.1329,
"step": 223
},
{
"epoch": 7.0,
"learning_rate": 6.566929133858268e-05,
"loss": 0.0809,
"step": 224
},
{
"epoch": 7.03,
"learning_rate": 6.551181102362204e-05,
"loss": 0.0602,
"step": 225
},
{
"epoch": 7.06,
"learning_rate": 6.535433070866141e-05,
"loss": 0.0753,
"step": 226
},
{
"epoch": 7.09,
"learning_rate": 6.519685039370079e-05,
"loss": 0.0541,
"step": 227
},
{
"epoch": 7.12,
"learning_rate": 6.503937007874015e-05,
"loss": 0.068,
"step": 228
},
{
"epoch": 7.16,
"learning_rate": 6.488188976377953e-05,
"loss": 0.0741,
"step": 229
},
{
"epoch": 7.19,
"learning_rate": 6.47244094488189e-05,
"loss": 0.0823,
"step": 230
},
{
"epoch": 7.22,
"learning_rate": 6.456692913385826e-05,
"loss": 0.068,
"step": 231
},
{
"epoch": 7.25,
"learning_rate": 6.440944881889764e-05,
"loss": 0.0605,
"step": 232
},
{
"epoch": 7.28,
"learning_rate": 6.425196850393701e-05,
"loss": 0.0493,
"step": 233
},
{
"epoch": 7.31,
"learning_rate": 6.409448818897637e-05,
"loss": 0.0728,
"step": 234
},
{
"epoch": 7.34,
"learning_rate": 6.393700787401575e-05,
"loss": 0.0506,
"step": 235
},
{
"epoch": 7.38,
"learning_rate": 6.377952755905512e-05,
"loss": 0.0552,
"step": 236
},
{
"epoch": 7.41,
"learning_rate": 6.36220472440945e-05,
"loss": 0.058,
"step": 237
},
{
"epoch": 7.44,
"learning_rate": 6.346456692913386e-05,
"loss": 0.0592,
"step": 238
},
{
"epoch": 7.47,
"learning_rate": 6.330708661417323e-05,
"loss": 0.0803,
"step": 239
},
{
"epoch": 7.5,
"learning_rate": 6.31496062992126e-05,
"loss": 0.0531,
"step": 240
},
{
"epoch": 7.53,
"learning_rate": 6.299212598425197e-05,
"loss": 0.0693,
"step": 241
},
{
"epoch": 7.56,
"learning_rate": 6.283464566929134e-05,
"loss": 0.091,
"step": 242
},
{
"epoch": 7.59,
"learning_rate": 6.267716535433072e-05,
"loss": 0.0845,
"step": 243
},
{
"epoch": 7.62,
"learning_rate": 6.251968503937008e-05,
"loss": 0.0642,
"step": 244
},
{
"epoch": 7.66,
"learning_rate": 6.236220472440946e-05,
"loss": 0.0887,
"step": 245
},
{
"epoch": 7.69,
"learning_rate": 6.220472440944882e-05,
"loss": 0.0602,
"step": 246
},
{
"epoch": 7.72,
"learning_rate": 6.204724409448819e-05,
"loss": 0.0803,
"step": 247
},
{
"epoch": 7.75,
"learning_rate": 6.188976377952757e-05,
"loss": 0.091,
"step": 248
},
{
"epoch": 7.78,
"learning_rate": 6.173228346456693e-05,
"loss": 0.0935,
"step": 249
},
{
"epoch": 7.81,
"learning_rate": 6.15748031496063e-05,
"loss": 0.1126,
"step": 250
},
{
"epoch": 7.84,
"learning_rate": 6.141732283464568e-05,
"loss": 0.0717,
"step": 251
},
{
"epoch": 7.88,
"learning_rate": 6.125984251968504e-05,
"loss": 0.0888,
"step": 252
},
{
"epoch": 7.91,
"learning_rate": 6.110236220472442e-05,
"loss": 0.0775,
"step": 253
},
{
"epoch": 7.94,
"learning_rate": 6.094488188976378e-05,
"loss": 0.0713,
"step": 254
},
{
"epoch": 7.97,
"learning_rate": 6.078740157480315e-05,
"loss": 0.0697,
"step": 255
},
{
"epoch": 8.0,
"learning_rate": 6.0629921259842526e-05,
"loss": 0.0904,
"step": 256
},
{
"epoch": 8.03,
"learning_rate": 6.047244094488189e-05,
"loss": 0.0527,
"step": 257
},
{
"epoch": 8.06,
"learning_rate": 6.031496062992126e-05,
"loss": 0.0398,
"step": 258
},
{
"epoch": 8.09,
"learning_rate": 6.0157480314960636e-05,
"loss": 0.0475,
"step": 259
},
{
"epoch": 8.12,
"learning_rate": 6e-05,
"loss": 0.0493,
"step": 260
},
{
"epoch": 8.16,
"learning_rate": 5.984251968503938e-05,
"loss": 0.057,
"step": 261
},
{
"epoch": 8.19,
"learning_rate": 5.9685039370078746e-05,
"loss": 0.0535,
"step": 262
},
{
"epoch": 8.22,
"learning_rate": 5.952755905511811e-05,
"loss": 0.0554,
"step": 263
},
{
"epoch": 8.25,
"learning_rate": 5.9370078740157483e-05,
"loss": 0.0593,
"step": 264
},
{
"epoch": 8.28,
"learning_rate": 5.9212598425196856e-05,
"loss": 0.0591,
"step": 265
},
{
"epoch": 8.31,
"learning_rate": 5.905511811023622e-05,
"loss": 0.0589,
"step": 266
},
{
"epoch": 8.34,
"learning_rate": 5.889763779527559e-05,
"loss": 0.0449,
"step": 267
},
{
"epoch": 8.38,
"learning_rate": 5.874015748031496e-05,
"loss": 0.0541,
"step": 268
},
{
"epoch": 8.41,
"learning_rate": 5.858267716535434e-05,
"loss": 0.0507,
"step": 269
},
{
"epoch": 8.44,
"learning_rate": 5.84251968503937e-05,
"loss": 0.0411,
"step": 270
},
{
"epoch": 8.47,
"learning_rate": 5.826771653543307e-05,
"loss": 0.0617,
"step": 271
},
{
"epoch": 8.5,
"learning_rate": 5.811023622047245e-05,
"loss": 0.0565,
"step": 272
},
{
"epoch": 8.53,
"learning_rate": 5.795275590551181e-05,
"loss": 0.083,
"step": 273
},
{
"epoch": 8.56,
"learning_rate": 5.779527559055118e-05,
"loss": 0.0588,
"step": 274
},
{
"epoch": 8.59,
"learning_rate": 5.763779527559056e-05,
"loss": 0.0686,
"step": 275
},
{
"epoch": 8.62,
"learning_rate": 5.748031496062992e-05,
"loss": 0.0712,
"step": 276
},
{
"epoch": 8.66,
"learning_rate": 5.73228346456693e-05,
"loss": 0.0708,
"step": 277
},
{
"epoch": 8.69,
"learning_rate": 5.716535433070867e-05,
"loss": 0.0573,
"step": 278
},
{
"epoch": 8.72,
"learning_rate": 5.700787401574803e-05,
"loss": 0.0563,
"step": 279
},
{
"epoch": 8.75,
"learning_rate": 5.6850393700787404e-05,
"loss": 0.0567,
"step": 280
},
{
"epoch": 8.78,
"learning_rate": 5.6692913385826777e-05,
"loss": 0.0517,
"step": 281
},
{
"epoch": 8.81,
"learning_rate": 5.653543307086614e-05,
"loss": 0.0571,
"step": 282
},
{
"epoch": 8.84,
"learning_rate": 5.6377952755905514e-05,
"loss": 0.0605,
"step": 283
},
{
"epoch": 8.88,
"learning_rate": 5.622047244094488e-05,
"loss": 0.0469,
"step": 284
},
{
"epoch": 8.91,
"learning_rate": 5.606299212598426e-05,
"loss": 0.0653,
"step": 285
},
{
"epoch": 8.94,
"learning_rate": 5.5905511811023624e-05,
"loss": 0.0578,
"step": 286
},
{
"epoch": 8.97,
"learning_rate": 5.574803149606299e-05,
"loss": 0.0547,
"step": 287
},
{
"epoch": 9.0,
"learning_rate": 5.559055118110237e-05,
"loss": 0.0602,
"step": 288
},
{
"epoch": 9.03,
"learning_rate": 5.5433070866141734e-05,
"loss": 0.0668,
"step": 289
},
{
"epoch": 9.06,
"learning_rate": 5.52755905511811e-05,
"loss": 0.0458,
"step": 290
},
{
"epoch": 9.09,
"learning_rate": 5.511811023622048e-05,
"loss": 0.0438,
"step": 291
},
{
"epoch": 9.12,
"learning_rate": 5.496062992125984e-05,
"loss": 0.0657,
"step": 292
},
{
"epoch": 9.16,
"learning_rate": 5.480314960629922e-05,
"loss": 0.052,
"step": 293
},
{
"epoch": 9.19,
"learning_rate": 5.464566929133859e-05,
"loss": 0.0569,
"step": 294
},
{
"epoch": 9.22,
"learning_rate": 5.448818897637795e-05,
"loss": 0.0357,
"step": 295
},
{
"epoch": 9.25,
"learning_rate": 5.433070866141733e-05,
"loss": 0.0358,
"step": 296
},
{
"epoch": 9.28,
"learning_rate": 5.41732283464567e-05,
"loss": 0.0414,
"step": 297
},
{
"epoch": 9.31,
"learning_rate": 5.401574803149606e-05,
"loss": 0.0384,
"step": 298
},
{
"epoch": 9.34,
"learning_rate": 5.3858267716535435e-05,
"loss": 0.0662,
"step": 299
},
{
"epoch": 9.38,
"learning_rate": 5.37007874015748e-05,
"loss": 0.0354,
"step": 300
},
{
"epoch": 9.41,
"learning_rate": 5.354330708661418e-05,
"loss": 0.0441,
"step": 301
},
{
"epoch": 9.44,
"learning_rate": 5.3385826771653545e-05,
"loss": 0.0448,
"step": 302
},
{
"epoch": 9.47,
"learning_rate": 5.322834645669291e-05,
"loss": 0.0478,
"step": 303
},
{
"epoch": 9.5,
"learning_rate": 5.307086614173229e-05,
"loss": 0.0414,
"step": 304
},
{
"epoch": 9.53,
"learning_rate": 5.2913385826771654e-05,
"loss": 0.0401,
"step": 305
},
{
"epoch": 9.56,
"learning_rate": 5.275590551181102e-05,
"loss": 0.0404,
"step": 306
},
{
"epoch": 9.59,
"learning_rate": 5.25984251968504e-05,
"loss": 0.0544,
"step": 307
},
{
"epoch": 9.62,
"learning_rate": 5.2440944881889764e-05,
"loss": 0.037,
"step": 308
},
{
"epoch": 9.66,
"learning_rate": 5.228346456692914e-05,
"loss": 0.0776,
"step": 309
},
{
"epoch": 9.69,
"learning_rate": 5.212598425196851e-05,
"loss": 0.0722,
"step": 310
},
{
"epoch": 9.72,
"learning_rate": 5.1968503937007874e-05,
"loss": 0.043,
"step": 311
},
{
"epoch": 9.75,
"learning_rate": 5.181102362204725e-05,
"loss": 0.0515,
"step": 312
},
{
"epoch": 9.78,
"learning_rate": 5.165354330708662e-05,
"loss": 0.0584,
"step": 313
},
{
"epoch": 9.81,
"learning_rate": 5.1496062992125984e-05,
"loss": 0.0542,
"step": 314
},
{
"epoch": 9.84,
"learning_rate": 5.1338582677165356e-05,
"loss": 0.042,
"step": 315
},
{
"epoch": 9.88,
"learning_rate": 5.118110236220473e-05,
"loss": 0.0497,
"step": 316
},
{
"epoch": 9.91,
"learning_rate": 5.10236220472441e-05,
"loss": 0.0558,
"step": 317
},
{
"epoch": 9.94,
"learning_rate": 5.0866141732283466e-05,
"loss": 0.0696,
"step": 318
},
{
"epoch": 9.97,
"learning_rate": 5.070866141732283e-05,
"loss": 0.0555,
"step": 319
},
{
"epoch": 10.0,
"learning_rate": 5.055118110236221e-05,
"loss": 0.0611,
"step": 320
},
{
"epoch": 10.03,
"learning_rate": 5.0393700787401575e-05,
"loss": 0.0418,
"step": 321
},
{
"epoch": 10.06,
"learning_rate": 5.023622047244094e-05,
"loss": 0.049,
"step": 322
},
{
"epoch": 10.09,
"learning_rate": 5.007874015748032e-05,
"loss": 0.0444,
"step": 323
},
{
"epoch": 10.12,
"learning_rate": 4.9921259842519685e-05,
"loss": 0.0426,
"step": 324
},
{
"epoch": 10.16,
"learning_rate": 4.976377952755906e-05,
"loss": 0.0338,
"step": 325
},
{
"epoch": 10.19,
"learning_rate": 4.960629921259843e-05,
"loss": 0.0354,
"step": 326
},
{
"epoch": 10.22,
"learning_rate": 4.94488188976378e-05,
"loss": 0.0344,
"step": 327
},
{
"epoch": 10.25,
"learning_rate": 4.929133858267717e-05,
"loss": 0.0312,
"step": 328
},
{
"epoch": 10.28,
"learning_rate": 4.913385826771654e-05,
"loss": 0.0514,
"step": 329
},
{
"epoch": 10.31,
"learning_rate": 4.897637795275591e-05,
"loss": 0.0377,
"step": 330
},
{
"epoch": 10.34,
"learning_rate": 4.881889763779528e-05,
"loss": 0.0398,
"step": 331
},
{
"epoch": 10.38,
"learning_rate": 4.866141732283465e-05,
"loss": 0.0618,
"step": 332
},
{
"epoch": 10.41,
"learning_rate": 4.8503937007874014e-05,
"loss": 0.0415,
"step": 333
},
{
"epoch": 10.44,
"learning_rate": 4.8346456692913387e-05,
"loss": 0.0518,
"step": 334
},
{
"epoch": 10.47,
"learning_rate": 4.818897637795276e-05,
"loss": 0.0606,
"step": 335
},
{
"epoch": 10.5,
"learning_rate": 4.8031496062992124e-05,
"loss": 0.0317,
"step": 336
},
{
"epoch": 10.53,
"learning_rate": 4.7874015748031496e-05,
"loss": 0.0508,
"step": 337
},
{
"epoch": 10.56,
"learning_rate": 4.771653543307087e-05,
"loss": 0.0311,
"step": 338
},
{
"epoch": 10.59,
"learning_rate": 4.755905511811024e-05,
"loss": 0.0422,
"step": 339
},
{
"epoch": 10.62,
"learning_rate": 4.7401574803149606e-05,
"loss": 0.0364,
"step": 340
},
{
"epoch": 10.66,
"learning_rate": 4.724409448818898e-05,
"loss": 0.0521,
"step": 341
},
{
"epoch": 10.69,
"learning_rate": 4.708661417322835e-05,
"loss": 0.0359,
"step": 342
},
{
"epoch": 10.72,
"learning_rate": 4.692913385826772e-05,
"loss": 0.0495,
"step": 343
},
{
"epoch": 10.75,
"learning_rate": 4.677165354330709e-05,
"loss": 0.0397,
"step": 344
},
{
"epoch": 10.78,
"learning_rate": 4.661417322834646e-05,
"loss": 0.0475,
"step": 345
},
{
"epoch": 10.81,
"learning_rate": 4.645669291338583e-05,
"loss": 0.0406,
"step": 346
},
{
"epoch": 10.84,
"learning_rate": 4.6299212598425204e-05,
"loss": 0.0624,
"step": 347
},
{
"epoch": 10.88,
"learning_rate": 4.614173228346457e-05,
"loss": 0.0423,
"step": 348
},
{
"epoch": 10.91,
"learning_rate": 4.5984251968503935e-05,
"loss": 0.053,
"step": 349
},
{
"epoch": 10.94,
"learning_rate": 4.582677165354331e-05,
"loss": 0.0732,
"step": 350
},
{
"epoch": 10.97,
"learning_rate": 4.566929133858268e-05,
"loss": 0.0517,
"step": 351
},
{
"epoch": 11.0,
"learning_rate": 4.5511811023622045e-05,
"loss": 0.0678,
"step": 352
},
{
"epoch": 11.03,
"learning_rate": 4.535433070866142e-05,
"loss": 0.0337,
"step": 353
},
{
"epoch": 11.06,
"learning_rate": 4.519685039370079e-05,
"loss": 0.0304,
"step": 354
},
{
"epoch": 11.09,
"learning_rate": 4.503937007874016e-05,
"loss": 0.0385,
"step": 355
},
{
"epoch": 11.12,
"learning_rate": 4.488188976377953e-05,
"loss": 0.0327,
"step": 356
},
{
"epoch": 11.16,
"learning_rate": 4.47244094488189e-05,
"loss": 0.0481,
"step": 357
},
{
"epoch": 11.19,
"learning_rate": 4.456692913385827e-05,
"loss": 0.0273,
"step": 358
},
{
"epoch": 11.22,
"learning_rate": 4.4409448818897643e-05,
"loss": 0.0422,
"step": 359
},
{
"epoch": 11.25,
"learning_rate": 4.425196850393701e-05,
"loss": 0.0368,
"step": 360
},
{
"epoch": 11.28,
"learning_rate": 4.409448818897638e-05,
"loss": 0.0349,
"step": 361
},
{
"epoch": 11.31,
"learning_rate": 4.393700787401575e-05,
"loss": 0.0316,
"step": 362
},
{
"epoch": 11.34,
"learning_rate": 4.3779527559055125e-05,
"loss": 0.0305,
"step": 363
},
{
"epoch": 11.38,
"learning_rate": 4.362204724409449e-05,
"loss": 0.0434,
"step": 364
},
{
"epoch": 11.41,
"learning_rate": 4.346456692913386e-05,
"loss": 0.041,
"step": 365
},
{
"epoch": 11.44,
"learning_rate": 4.330708661417323e-05,
"loss": 0.0446,
"step": 366
},
{
"epoch": 11.47,
"learning_rate": 4.31496062992126e-05,
"loss": 0.0382,
"step": 367
},
{
"epoch": 11.5,
"learning_rate": 4.2992125984251966e-05,
"loss": 0.0379,
"step": 368
},
{
"epoch": 11.53,
"learning_rate": 4.283464566929134e-05,
"loss": 0.0363,
"step": 369
},
{
"epoch": 11.56,
"learning_rate": 4.267716535433071e-05,
"loss": 0.0386,
"step": 370
},
{
"epoch": 11.59,
"learning_rate": 4.251968503937008e-05,
"loss": 0.0251,
"step": 371
},
{
"epoch": 11.62,
"learning_rate": 4.236220472440945e-05,
"loss": 0.0336,
"step": 372
},
{
"epoch": 11.66,
"learning_rate": 4.220472440944882e-05,
"loss": 0.0438,
"step": 373
},
{
"epoch": 11.69,
"learning_rate": 4.204724409448819e-05,
"loss": 0.0442,
"step": 374
},
{
"epoch": 11.72,
"learning_rate": 4.1889763779527564e-05,
"loss": 0.0285,
"step": 375
},
{
"epoch": 11.75,
"learning_rate": 4.173228346456693e-05,
"loss": 0.0388,
"step": 376
},
{
"epoch": 11.78,
"learning_rate": 4.15748031496063e-05,
"loss": 0.0442,
"step": 377
},
{
"epoch": 11.81,
"learning_rate": 4.1417322834645674e-05,
"loss": 0.0371,
"step": 378
},
{
"epoch": 11.84,
"learning_rate": 4.1259842519685046e-05,
"loss": 0.0488,
"step": 379
},
{
"epoch": 11.88,
"learning_rate": 4.110236220472441e-05,
"loss": 0.0508,
"step": 380
},
{
"epoch": 11.91,
"learning_rate": 4.0944881889763784e-05,
"loss": 0.0455,
"step": 381
},
{
"epoch": 11.94,
"learning_rate": 4.078740157480315e-05,
"loss": 0.0432,
"step": 382
},
{
"epoch": 11.97,
"learning_rate": 4.062992125984252e-05,
"loss": 0.033,
"step": 383
},
{
"epoch": 12.0,
"learning_rate": 4.047244094488189e-05,
"loss": 0.0446,
"step": 384
},
{
"epoch": 12.03,
"learning_rate": 4.031496062992126e-05,
"loss": 0.0355,
"step": 385
},
{
"epoch": 12.06,
"learning_rate": 4.015748031496063e-05,
"loss": 0.0468,
"step": 386
},
{
"epoch": 12.09,
"learning_rate": 4e-05,
"loss": 0.0273,
"step": 387
},
{
"epoch": 12.12,
"learning_rate": 3.984251968503937e-05,
"loss": 0.0344,
"step": 388
},
{
"epoch": 12.16,
"learning_rate": 3.968503937007874e-05,
"loss": 0.0331,
"step": 389
},
{
"epoch": 12.19,
"learning_rate": 3.952755905511811e-05,
"loss": 0.0369,
"step": 390
},
{
"epoch": 12.22,
"learning_rate": 3.9370078740157485e-05,
"loss": 0.0324,
"step": 391
},
{
"epoch": 12.25,
"learning_rate": 3.921259842519685e-05,
"loss": 0.0294,
"step": 392
},
{
"epoch": 12.28,
"learning_rate": 3.905511811023622e-05,
"loss": 0.0441,
"step": 393
},
{
"epoch": 12.31,
"learning_rate": 3.8897637795275595e-05,
"loss": 0.0326,
"step": 394
},
{
"epoch": 12.34,
"learning_rate": 3.874015748031497e-05,
"loss": 0.0288,
"step": 395
},
{
"epoch": 12.38,
"learning_rate": 3.858267716535433e-05,
"loss": 0.0305,
"step": 396
},
{
"epoch": 12.41,
"learning_rate": 3.8425196850393705e-05,
"loss": 0.0286,
"step": 397
},
{
"epoch": 12.44,
"learning_rate": 3.826771653543308e-05,
"loss": 0.0327,
"step": 398
},
{
"epoch": 12.47,
"learning_rate": 3.811023622047244e-05,
"loss": 0.0814,
"step": 399
},
{
"epoch": 12.5,
"learning_rate": 3.795275590551181e-05,
"loss": 0.0269,
"step": 400
},
{
"epoch": 12.53,
"learning_rate": 3.779527559055118e-05,
"loss": 0.0352,
"step": 401
},
{
"epoch": 12.56,
"learning_rate": 3.763779527559055e-05,
"loss": 0.0331,
"step": 402
},
{
"epoch": 12.59,
"learning_rate": 3.7480314960629924e-05,
"loss": 0.0327,
"step": 403
},
{
"epoch": 12.62,
"learning_rate": 3.732283464566929e-05,
"loss": 0.0544,
"step": 404
},
{
"epoch": 12.66,
"learning_rate": 3.716535433070866e-05,
"loss": 0.0395,
"step": 405
},
{
"epoch": 12.69,
"learning_rate": 3.7007874015748034e-05,
"loss": 0.0323,
"step": 406
},
{
"epoch": 12.72,
"learning_rate": 3.6850393700787406e-05,
"loss": 0.037,
"step": 407
},
{
"epoch": 12.75,
"learning_rate": 3.669291338582677e-05,
"loss": 0.0344,
"step": 408
},
{
"epoch": 12.78,
"learning_rate": 3.6535433070866144e-05,
"loss": 0.0358,
"step": 409
},
{
"epoch": 12.81,
"learning_rate": 3.6377952755905516e-05,
"loss": 0.0311,
"step": 410
},
{
"epoch": 12.84,
"learning_rate": 3.622047244094489e-05,
"loss": 0.0467,
"step": 411
},
{
"epoch": 12.88,
"learning_rate": 3.6062992125984253e-05,
"loss": 0.0316,
"step": 412
},
{
"epoch": 12.91,
"learning_rate": 3.5905511811023626e-05,
"loss": 0.0433,
"step": 413
},
{
"epoch": 12.94,
"learning_rate": 3.5748031496063e-05,
"loss": 0.0489,
"step": 414
},
{
"epoch": 12.97,
"learning_rate": 3.559055118110236e-05,
"loss": 0.0383,
"step": 415
},
{
"epoch": 13.0,
"learning_rate": 3.5433070866141735e-05,
"loss": 0.0399,
"step": 416
},
{
"epoch": 13.03,
"learning_rate": 3.52755905511811e-05,
"loss": 0.0297,
"step": 417
},
{
"epoch": 13.06,
"learning_rate": 3.511811023622047e-05,
"loss": 0.0382,
"step": 418
},
{
"epoch": 13.09,
"learning_rate": 3.4960629921259845e-05,
"loss": 0.0259,
"step": 419
},
{
"epoch": 13.12,
"learning_rate": 3.480314960629921e-05,
"loss": 0.0287,
"step": 420
},
{
"epoch": 13.16,
"learning_rate": 3.464566929133858e-05,
"loss": 0.0335,
"step": 421
},
{
"epoch": 13.19,
"learning_rate": 3.4488188976377955e-05,
"loss": 0.0313,
"step": 422
},
{
"epoch": 13.22,
"learning_rate": 3.433070866141733e-05,
"loss": 0.0278,
"step": 423
},
{
"epoch": 13.25,
"learning_rate": 3.417322834645669e-05,
"loss": 0.0327,
"step": 424
},
{
"epoch": 13.28,
"learning_rate": 3.4015748031496065e-05,
"loss": 0.0327,
"step": 425
},
{
"epoch": 13.31,
"learning_rate": 3.385826771653544e-05,
"loss": 0.0246,
"step": 426
},
{
"epoch": 13.34,
"learning_rate": 3.370078740157481e-05,
"loss": 0.0327,
"step": 427
},
{
"epoch": 13.38,
"learning_rate": 3.3543307086614174e-05,
"loss": 0.0261,
"step": 428
},
{
"epoch": 13.41,
"learning_rate": 3.3385826771653546e-05,
"loss": 0.0359,
"step": 429
},
{
"epoch": 13.44,
"learning_rate": 3.322834645669292e-05,
"loss": 0.0321,
"step": 430
},
{
"epoch": 13.47,
"learning_rate": 3.3070866141732284e-05,
"loss": 0.0329,
"step": 431
},
{
"epoch": 13.5,
"learning_rate": 3.2913385826771656e-05,
"loss": 0.0303,
"step": 432
},
{
"epoch": 13.53,
"learning_rate": 3.275590551181102e-05,
"loss": 0.0372,
"step": 433
},
{
"epoch": 13.56,
"learning_rate": 3.2598425196850394e-05,
"loss": 0.0255,
"step": 434
},
{
"epoch": 13.59,
"learning_rate": 3.2440944881889766e-05,
"loss": 0.0357,
"step": 435
},
{
"epoch": 13.62,
"learning_rate": 3.228346456692913e-05,
"loss": 0.0536,
"step": 436
},
{
"epoch": 13.66,
"learning_rate": 3.2125984251968504e-05,
"loss": 0.0323,
"step": 437
},
{
"epoch": 13.69,
"learning_rate": 3.1968503937007876e-05,
"loss": 0.0273,
"step": 438
},
{
"epoch": 13.72,
"learning_rate": 3.181102362204725e-05,
"loss": 0.0366,
"step": 439
},
{
"epoch": 13.75,
"learning_rate": 3.165354330708661e-05,
"loss": 0.0339,
"step": 440
},
{
"epoch": 13.78,
"learning_rate": 3.1496062992125985e-05,
"loss": 0.0312,
"step": 441
},
{
"epoch": 13.81,
"learning_rate": 3.133858267716536e-05,
"loss": 0.0355,
"step": 442
},
{
"epoch": 13.84,
"learning_rate": 3.118110236220473e-05,
"loss": 0.029,
"step": 443
},
{
"epoch": 13.88,
"learning_rate": 3.1023622047244095e-05,
"loss": 0.0336,
"step": 444
},
{
"epoch": 13.91,
"learning_rate": 3.086614173228347e-05,
"loss": 0.0374,
"step": 445
},
{
"epoch": 13.94,
"learning_rate": 3.070866141732284e-05,
"loss": 0.0276,
"step": 446
},
{
"epoch": 13.97,
"learning_rate": 3.055118110236221e-05,
"loss": 0.0282,
"step": 447
},
{
"epoch": 14.0,
"learning_rate": 3.0393700787401574e-05,
"loss": 0.0377,
"step": 448
},
{
"epoch": 14.03,
"learning_rate": 3.0236220472440946e-05,
"loss": 0.0327,
"step": 449
},
{
"epoch": 14.06,
"learning_rate": 3.0078740157480318e-05,
"loss": 0.0286,
"step": 450
},
{
"epoch": 14.09,
"learning_rate": 2.992125984251969e-05,
"loss": 0.0307,
"step": 451
},
{
"epoch": 14.12,
"learning_rate": 2.9763779527559056e-05,
"loss": 0.0247,
"step": 452
},
{
"epoch": 14.16,
"learning_rate": 2.9606299212598428e-05,
"loss": 0.0259,
"step": 453
},
{
"epoch": 14.19,
"learning_rate": 2.9448818897637797e-05,
"loss": 0.0227,
"step": 454
},
{
"epoch": 14.22,
"learning_rate": 2.929133858267717e-05,
"loss": 0.0264,
"step": 455
},
{
"epoch": 14.25,
"learning_rate": 2.9133858267716534e-05,
"loss": 0.0292,
"step": 456
},
{
"epoch": 14.28,
"learning_rate": 2.8976377952755906e-05,
"loss": 0.025,
"step": 457
},
{
"epoch": 14.31,
"learning_rate": 2.881889763779528e-05,
"loss": 0.0321,
"step": 458
},
{
"epoch": 14.34,
"learning_rate": 2.866141732283465e-05,
"loss": 0.0296,
"step": 459
},
{
"epoch": 14.38,
"learning_rate": 2.8503937007874016e-05,
"loss": 0.0326,
"step": 460
},
{
"epoch": 14.41,
"learning_rate": 2.8346456692913388e-05,
"loss": 0.0284,
"step": 461
},
{
"epoch": 14.44,
"learning_rate": 2.8188976377952757e-05,
"loss": 0.0307,
"step": 462
},
{
"epoch": 14.47,
"learning_rate": 2.803149606299213e-05,
"loss": 0.0285,
"step": 463
},
{
"epoch": 14.5,
"learning_rate": 2.7874015748031495e-05,
"loss": 0.0252,
"step": 464
},
{
"epoch": 14.53,
"learning_rate": 2.7716535433070867e-05,
"loss": 0.0424,
"step": 465
},
{
"epoch": 14.56,
"learning_rate": 2.755905511811024e-05,
"loss": 0.0381,
"step": 466
},
{
"epoch": 14.59,
"learning_rate": 2.740157480314961e-05,
"loss": 0.032,
"step": 467
},
{
"epoch": 14.62,
"learning_rate": 2.7244094488188977e-05,
"loss": 0.027,
"step": 468
},
{
"epoch": 14.66,
"learning_rate": 2.708661417322835e-05,
"loss": 0.0295,
"step": 469
},
{
"epoch": 14.69,
"learning_rate": 2.6929133858267717e-05,
"loss": 0.0283,
"step": 470
},
{
"epoch": 14.72,
"learning_rate": 2.677165354330709e-05,
"loss": 0.0292,
"step": 471
},
{
"epoch": 14.75,
"learning_rate": 2.6614173228346455e-05,
"loss": 0.0368,
"step": 472
},
{
"epoch": 14.78,
"learning_rate": 2.6456692913385827e-05,
"loss": 0.04,
"step": 473
},
{
"epoch": 14.81,
"learning_rate": 2.62992125984252e-05,
"loss": 0.0361,
"step": 474
},
{
"epoch": 14.84,
"learning_rate": 2.614173228346457e-05,
"loss": 0.031,
"step": 475
},
{
"epoch": 14.88,
"learning_rate": 2.5984251968503937e-05,
"loss": 0.0318,
"step": 476
},
{
"epoch": 14.91,
"learning_rate": 2.582677165354331e-05,
"loss": 0.0329,
"step": 477
},
{
"epoch": 14.94,
"learning_rate": 2.5669291338582678e-05,
"loss": 0.0322,
"step": 478
},
{
"epoch": 14.97,
"learning_rate": 2.551181102362205e-05,
"loss": 0.0262,
"step": 479
},
{
"epoch": 15.0,
"learning_rate": 2.5354330708661416e-05,
"loss": 0.0279,
"step": 480
},
{
"epoch": 15.03,
"learning_rate": 2.5196850393700788e-05,
"loss": 0.0291,
"step": 481
},
{
"epoch": 15.06,
"learning_rate": 2.503937007874016e-05,
"loss": 0.0281,
"step": 482
},
{
"epoch": 15.09,
"learning_rate": 2.488188976377953e-05,
"loss": 0.0301,
"step": 483
},
{
"epoch": 15.12,
"learning_rate": 2.47244094488189e-05,
"loss": 0.0269,
"step": 484
},
{
"epoch": 15.16,
"learning_rate": 2.456692913385827e-05,
"loss": 0.026,
"step": 485
},
{
"epoch": 15.19,
"learning_rate": 2.440944881889764e-05,
"loss": 0.0273,
"step": 486
},
{
"epoch": 15.22,
"learning_rate": 2.4251968503937007e-05,
"loss": 0.0281,
"step": 487
},
{
"epoch": 15.25,
"learning_rate": 2.409448818897638e-05,
"loss": 0.0257,
"step": 488
},
{
"epoch": 15.28,
"learning_rate": 2.3937007874015748e-05,
"loss": 0.0281,
"step": 489
},
{
"epoch": 15.31,
"learning_rate": 2.377952755905512e-05,
"loss": 0.0313,
"step": 490
},
{
"epoch": 15.34,
"learning_rate": 2.362204724409449e-05,
"loss": 0.0376,
"step": 491
},
{
"epoch": 15.38,
"learning_rate": 2.346456692913386e-05,
"loss": 0.0304,
"step": 492
},
{
"epoch": 15.41,
"learning_rate": 2.330708661417323e-05,
"loss": 0.0221,
"step": 493
},
{
"epoch": 15.44,
"learning_rate": 2.3149606299212602e-05,
"loss": 0.0289,
"step": 494
},
{
"epoch": 15.47,
"learning_rate": 2.2992125984251968e-05,
"loss": 0.0281,
"step": 495
},
{
"epoch": 15.5,
"learning_rate": 2.283464566929134e-05,
"loss": 0.0272,
"step": 496
},
{
"epoch": 15.53,
"learning_rate": 2.267716535433071e-05,
"loss": 0.022,
"step": 497
},
{
"epoch": 15.56,
"learning_rate": 2.251968503937008e-05,
"loss": 0.0327,
"step": 498
},
{
"epoch": 15.59,
"learning_rate": 2.236220472440945e-05,
"loss": 0.0328,
"step": 499
},
{
"epoch": 15.62,
"learning_rate": 2.2204724409448822e-05,
"loss": 0.0329,
"step": 500
},
{
"epoch": 15.66,
"learning_rate": 2.204724409448819e-05,
"loss": 0.0322,
"step": 501
},
{
"epoch": 15.69,
"learning_rate": 2.1889763779527563e-05,
"loss": 0.0317,
"step": 502
},
{
"epoch": 15.72,
"learning_rate": 2.173228346456693e-05,
"loss": 0.0261,
"step": 503
},
{
"epoch": 15.75,
"learning_rate": 2.15748031496063e-05,
"loss": 0.0342,
"step": 504
},
{
"epoch": 15.78,
"learning_rate": 2.141732283464567e-05,
"loss": 0.0308,
"step": 505
},
{
"epoch": 15.81,
"learning_rate": 2.125984251968504e-05,
"loss": 0.0291,
"step": 506
},
{
"epoch": 15.84,
"learning_rate": 2.110236220472441e-05,
"loss": 0.0301,
"step": 507
},
{
"epoch": 15.88,
"learning_rate": 2.0944881889763782e-05,
"loss": 0.0301,
"step": 508
},
{
"epoch": 15.91,
"learning_rate": 2.078740157480315e-05,
"loss": 0.0271,
"step": 509
},
{
"epoch": 15.94,
"learning_rate": 2.0629921259842523e-05,
"loss": 0.0314,
"step": 510
},
{
"epoch": 15.97,
"learning_rate": 2.0472440944881892e-05,
"loss": 0.0294,
"step": 511
},
{
"epoch": 16.0,
"learning_rate": 2.031496062992126e-05,
"loss": 0.0293,
"step": 512
},
{
"epoch": 16.03,
"learning_rate": 2.015748031496063e-05,
"loss": 0.0267,
"step": 513
},
{
"epoch": 16.06,
"learning_rate": 2e-05,
"loss": 0.0228,
"step": 514
},
{
"epoch": 16.09,
"learning_rate": 1.984251968503937e-05,
"loss": 0.0275,
"step": 515
},
{
"epoch": 16.12,
"learning_rate": 1.9685039370078743e-05,
"loss": 0.0269,
"step": 516
},
{
"epoch": 16.16,
"learning_rate": 1.952755905511811e-05,
"loss": 0.0225,
"step": 517
},
{
"epoch": 16.19,
"learning_rate": 1.9370078740157484e-05,
"loss": 0.0293,
"step": 518
},
{
"epoch": 16.22,
"learning_rate": 1.9212598425196852e-05,
"loss": 0.0313,
"step": 519
},
{
"epoch": 16.25,
"learning_rate": 1.905511811023622e-05,
"loss": 0.0259,
"step": 520
},
{
"epoch": 16.28,
"learning_rate": 1.889763779527559e-05,
"loss": 0.0325,
"step": 521
},
{
"epoch": 16.31,
"learning_rate": 1.8740157480314962e-05,
"loss": 0.0275,
"step": 522
},
{
"epoch": 16.34,
"learning_rate": 1.858267716535433e-05,
"loss": 0.0307,
"step": 523
},
{
"epoch": 16.38,
"learning_rate": 1.8425196850393703e-05,
"loss": 0.0292,
"step": 524
},
{
"epoch": 16.41,
"learning_rate": 1.8267716535433072e-05,
"loss": 0.0298,
"step": 525
},
{
"epoch": 16.44,
"learning_rate": 1.8110236220472444e-05,
"loss": 0.0247,
"step": 526
},
{
"epoch": 16.47,
"learning_rate": 1.7952755905511813e-05,
"loss": 0.0267,
"step": 527
},
{
"epoch": 16.5,
"learning_rate": 1.779527559055118e-05,
"loss": 0.0272,
"step": 528
},
{
"epoch": 16.53,
"learning_rate": 1.763779527559055e-05,
"loss": 0.0334,
"step": 529
},
{
"epoch": 16.56,
"learning_rate": 1.7480314960629923e-05,
"loss": 0.0277,
"step": 530
},
{
"epoch": 16.59,
"learning_rate": 1.732283464566929e-05,
"loss": 0.0313,
"step": 531
},
{
"epoch": 16.62,
"learning_rate": 1.7165354330708663e-05,
"loss": 0.0261,
"step": 532
},
{
"epoch": 16.66,
"learning_rate": 1.7007874015748032e-05,
"loss": 0.0299,
"step": 533
},
{
"epoch": 16.69,
"learning_rate": 1.6850393700787404e-05,
"loss": 0.0238,
"step": 534
},
{
"epoch": 16.72,
"learning_rate": 1.6692913385826773e-05,
"loss": 0.0264,
"step": 535
},
{
"epoch": 16.75,
"learning_rate": 1.6535433070866142e-05,
"loss": 0.0292,
"step": 536
},
{
"epoch": 16.78,
"learning_rate": 1.637795275590551e-05,
"loss": 0.0255,
"step": 537
},
{
"epoch": 16.81,
"learning_rate": 1.6220472440944883e-05,
"loss": 0.0328,
"step": 538
},
{
"epoch": 16.84,
"learning_rate": 1.6062992125984252e-05,
"loss": 0.0309,
"step": 539
},
{
"epoch": 16.88,
"learning_rate": 1.5905511811023624e-05,
"loss": 0.0287,
"step": 540
},
{
"epoch": 16.91,
"learning_rate": 1.5748031496062993e-05,
"loss": 0.0261,
"step": 541
},
{
"epoch": 16.94,
"learning_rate": 1.5590551181102365e-05,
"loss": 0.0323,
"step": 542
},
{
"epoch": 16.97,
"learning_rate": 1.5433070866141734e-05,
"loss": 0.0248,
"step": 543
},
{
"epoch": 17.0,
"learning_rate": 1.5275590551181106e-05,
"loss": 0.0305,
"step": 544
},
{
"epoch": 17.03,
"learning_rate": 1.5118110236220473e-05,
"loss": 0.0231,
"step": 545
},
{
"epoch": 17.06,
"learning_rate": 1.4960629921259845e-05,
"loss": 0.0244,
"step": 546
},
{
"epoch": 17.09,
"learning_rate": 1.4803149606299214e-05,
"loss": 0.03,
"step": 547
},
{
"epoch": 17.12,
"learning_rate": 1.4645669291338584e-05,
"loss": 0.026,
"step": 548
},
{
"epoch": 17.16,
"learning_rate": 1.4488188976377953e-05,
"loss": 0.0297,
"step": 549
},
{
"epoch": 17.19,
"learning_rate": 1.4330708661417325e-05,
"loss": 0.0274,
"step": 550
},
{
"epoch": 17.22,
"learning_rate": 1.4173228346456694e-05,
"loss": 0.0258,
"step": 551
},
{
"epoch": 17.25,
"learning_rate": 1.4015748031496065e-05,
"loss": 0.0304,
"step": 552
},
{
"epoch": 17.28,
"learning_rate": 1.3858267716535433e-05,
"loss": 0.0312,
"step": 553
},
{
"epoch": 17.31,
"learning_rate": 1.3700787401574806e-05,
"loss": 0.0287,
"step": 554
},
{
"epoch": 17.34,
"learning_rate": 1.3543307086614174e-05,
"loss": 0.0236,
"step": 555
},
{
"epoch": 17.38,
"learning_rate": 1.3385826771653545e-05,
"loss": 0.0264,
"step": 556
},
{
"epoch": 17.41,
"learning_rate": 1.3228346456692914e-05,
"loss": 0.0299,
"step": 557
},
{
"epoch": 17.44,
"learning_rate": 1.3070866141732286e-05,
"loss": 0.0246,
"step": 558
},
{
"epoch": 17.47,
"learning_rate": 1.2913385826771655e-05,
"loss": 0.029,
"step": 559
},
{
"epoch": 17.5,
"learning_rate": 1.2755905511811025e-05,
"loss": 0.0267,
"step": 560
},
{
"epoch": 17.53,
"learning_rate": 1.2598425196850394e-05,
"loss": 0.0239,
"step": 561
},
{
"epoch": 17.56,
"learning_rate": 1.2440944881889764e-05,
"loss": 0.0241,
"step": 562
},
{
"epoch": 17.59,
"learning_rate": 1.2283464566929135e-05,
"loss": 0.0274,
"step": 563
},
{
"epoch": 17.62,
"learning_rate": 1.2125984251968504e-05,
"loss": 0.0309,
"step": 564
},
{
"epoch": 17.66,
"learning_rate": 1.1968503937007874e-05,
"loss": 0.0283,
"step": 565
},
{
"epoch": 17.69,
"learning_rate": 1.1811023622047245e-05,
"loss": 0.0238,
"step": 566
},
{
"epoch": 17.72,
"learning_rate": 1.1653543307086615e-05,
"loss": 0.0267,
"step": 567
},
{
"epoch": 17.75,
"learning_rate": 1.1496062992125984e-05,
"loss": 0.024,
"step": 568
},
{
"epoch": 17.78,
"learning_rate": 1.1338582677165354e-05,
"loss": 0.0289,
"step": 569
},
{
"epoch": 17.81,
"learning_rate": 1.1181102362204725e-05,
"loss": 0.03,
"step": 570
},
{
"epoch": 17.84,
"learning_rate": 1.1023622047244095e-05,
"loss": 0.0342,
"step": 571
},
{
"epoch": 17.88,
"learning_rate": 1.0866141732283466e-05,
"loss": 0.0307,
"step": 572
},
{
"epoch": 17.91,
"learning_rate": 1.0708661417322835e-05,
"loss": 0.0294,
"step": 573
},
{
"epoch": 17.94,
"learning_rate": 1.0551181102362205e-05,
"loss": 0.0284,
"step": 574
},
{
"epoch": 17.97,
"learning_rate": 1.0393700787401575e-05,
"loss": 0.031,
"step": 575
},
{
"epoch": 18.0,
"learning_rate": 1.0236220472440946e-05,
"loss": 0.0263,
"step": 576
},
{
"epoch": 18.03,
"learning_rate": 1.0078740157480315e-05,
"loss": 0.0224,
"step": 577
},
{
"epoch": 18.06,
"learning_rate": 9.921259842519685e-06,
"loss": 0.024,
"step": 578
},
{
"epoch": 18.09,
"learning_rate": 9.763779527559056e-06,
"loss": 0.0274,
"step": 579
},
{
"epoch": 18.12,
"learning_rate": 9.606299212598426e-06,
"loss": 0.025,
"step": 580
},
{
"epoch": 18.16,
"learning_rate": 9.448818897637795e-06,
"loss": 0.0244,
"step": 581
},
{
"epoch": 18.19,
"learning_rate": 9.291338582677165e-06,
"loss": 0.0253,
"step": 582
},
{
"epoch": 18.22,
"learning_rate": 9.133858267716536e-06,
"loss": 0.0238,
"step": 583
},
{
"epoch": 18.25,
"learning_rate": 8.976377952755906e-06,
"loss": 0.0265,
"step": 584
},
{
"epoch": 18.28,
"learning_rate": 8.818897637795275e-06,
"loss": 0.0278,
"step": 585
},
{
"epoch": 18.31,
"learning_rate": 8.661417322834646e-06,
"loss": 0.0242,
"step": 586
},
{
"epoch": 18.34,
"learning_rate": 8.503937007874016e-06,
"loss": 0.026,
"step": 587
},
{
"epoch": 18.38,
"learning_rate": 8.346456692913387e-06,
"loss": 0.0196,
"step": 588
},
{
"epoch": 18.41,
"learning_rate": 8.188976377952755e-06,
"loss": 0.0245,
"step": 589
},
{
"epoch": 18.44,
"learning_rate": 8.031496062992126e-06,
"loss": 0.0259,
"step": 590
},
{
"epoch": 18.47,
"learning_rate": 7.874015748031496e-06,
"loss": 0.033,
"step": 591
},
{
"epoch": 18.5,
"learning_rate": 7.716535433070867e-06,
"loss": 0.0291,
"step": 592
},
{
"epoch": 18.53,
"learning_rate": 7.5590551181102365e-06,
"loss": 0.0272,
"step": 593
},
{
"epoch": 18.56,
"learning_rate": 7.401574803149607e-06,
"loss": 0.0266,
"step": 594
},
{
"epoch": 18.59,
"learning_rate": 7.244094488188977e-06,
"loss": 0.0274,
"step": 595
},
{
"epoch": 18.62,
"learning_rate": 7.086614173228347e-06,
"loss": 0.0261,
"step": 596
},
{
"epoch": 18.66,
"learning_rate": 6.929133858267717e-06,
"loss": 0.0259,
"step": 597
},
{
"epoch": 18.69,
"learning_rate": 6.771653543307087e-06,
"loss": 0.0328,
"step": 598
},
{
"epoch": 18.72,
"learning_rate": 6.614173228346457e-06,
"loss": 0.0321,
"step": 599
},
{
"epoch": 18.75,
"learning_rate": 6.456692913385827e-06,
"loss": 0.0283,
"step": 600
},
{
"epoch": 18.78,
"learning_rate": 6.299212598425197e-06,
"loss": 0.0253,
"step": 601
},
{
"epoch": 18.81,
"learning_rate": 6.141732283464567e-06,
"loss": 0.0346,
"step": 602
},
{
"epoch": 18.84,
"learning_rate": 5.984251968503937e-06,
"loss": 0.0264,
"step": 603
},
{
"epoch": 18.88,
"learning_rate": 5.8267716535433075e-06,
"loss": 0.026,
"step": 604
},
{
"epoch": 18.91,
"learning_rate": 5.669291338582677e-06,
"loss": 0.0303,
"step": 605
},
{
"epoch": 18.94,
"learning_rate": 5.511811023622048e-06,
"loss": 0.0326,
"step": 606
},
{
"epoch": 18.97,
"learning_rate": 5.354330708661417e-06,
"loss": 0.0302,
"step": 607
},
{
"epoch": 19.0,
"learning_rate": 5.196850393700788e-06,
"loss": 0.0288,
"step": 608
},
{
"epoch": 19.03,
"learning_rate": 5.039370078740157e-06,
"loss": 0.0279,
"step": 609
},
{
"epoch": 19.06,
"learning_rate": 4.881889763779528e-06,
"loss": 0.0207,
"step": 610
},
{
"epoch": 19.09,
"learning_rate": 4.7244094488188975e-06,
"loss": 0.0282,
"step": 611
},
{
"epoch": 19.12,
"learning_rate": 4.566929133858268e-06,
"loss": 0.0234,
"step": 612
},
{
"epoch": 19.16,
"learning_rate": 4.409448818897638e-06,
"loss": 0.0305,
"step": 613
},
{
"epoch": 19.19,
"learning_rate": 4.251968503937008e-06,
"loss": 0.0269,
"step": 614
},
{
"epoch": 19.22,
"learning_rate": 4.094488188976378e-06,
"loss": 0.0254,
"step": 615
},
{
"epoch": 19.25,
"learning_rate": 3.937007874015748e-06,
"loss": 0.029,
"step": 616
},
{
"epoch": 19.28,
"learning_rate": 3.7795275590551182e-06,
"loss": 0.0275,
"step": 617
},
{
"epoch": 19.31,
"learning_rate": 3.6220472440944883e-06,
"loss": 0.024,
"step": 618
},
{
"epoch": 19.34,
"learning_rate": 3.4645669291338583e-06,
"loss": 0.0265,
"step": 619
},
{
"epoch": 19.38,
"learning_rate": 3.3070866141732284e-06,
"loss": 0.0235,
"step": 620
},
{
"epoch": 19.41,
"learning_rate": 3.1496062992125985e-06,
"loss": 0.0288,
"step": 621
},
{
"epoch": 19.44,
"learning_rate": 2.9921259842519685e-06,
"loss": 0.0296,
"step": 622
},
{
"epoch": 19.47,
"learning_rate": 2.8346456692913386e-06,
"loss": 0.0247,
"step": 623
},
{
"epoch": 19.5,
"learning_rate": 2.6771653543307086e-06,
"loss": 0.0282,
"step": 624
},
{
"epoch": 19.53,
"learning_rate": 2.5196850393700787e-06,
"loss": 0.0309,
"step": 625
},
{
"epoch": 19.56,
"learning_rate": 2.3622047244094487e-06,
"loss": 0.0215,
"step": 626
},
{
"epoch": 19.59,
"learning_rate": 2.204724409448819e-06,
"loss": 0.0312,
"step": 627
},
{
"epoch": 19.62,
"learning_rate": 2.047244094488189e-06,
"loss": 0.0225,
"step": 628
},
{
"epoch": 19.66,
"learning_rate": 1.8897637795275591e-06,
"loss": 0.0255,
"step": 629
},
{
"epoch": 19.69,
"learning_rate": 1.7322834645669292e-06,
"loss": 0.0265,
"step": 630
},
{
"epoch": 19.72,
"learning_rate": 1.5748031496062992e-06,
"loss": 0.0319,
"step": 631
},
{
"epoch": 19.75,
"learning_rate": 1.4173228346456693e-06,
"loss": 0.0267,
"step": 632
},
{
"epoch": 19.78,
"learning_rate": 1.2598425196850393e-06,
"loss": 0.0254,
"step": 633
},
{
"epoch": 19.81,
"learning_rate": 1.1023622047244094e-06,
"loss": 0.0237,
"step": 634
},
{
"epoch": 19.84,
"learning_rate": 9.448818897637796e-07,
"loss": 0.0253,
"step": 635
},
{
"epoch": 19.88,
"learning_rate": 7.874015748031496e-07,
"loss": 0.0236,
"step": 636
},
{
"epoch": 19.91,
"learning_rate": 6.299212598425197e-07,
"loss": 0.0303,
"step": 637
},
{
"epoch": 19.94,
"learning_rate": 4.724409448818898e-07,
"loss": 0.0232,
"step": 638
},
{
"epoch": 19.97,
"learning_rate": 3.1496062992125984e-07,
"loss": 0.0314,
"step": 639
},
{
"epoch": 20.0,
"learning_rate": 1.5748031496062992e-07,
"loss": 0.028,
"step": 640
},
{
"epoch": 20.0,
"step": 640,
"total_flos": 4.839674807844864e+16,
"train_loss": 0.2799676021706546,
"train_runtime": 1494.9511,
"train_samples_per_second": 1.712,
"train_steps_per_second": 0.428
}
],
"logging_steps": 1,
"max_steps": 640,
"num_train_epochs": 20,
"save_steps": 500,
"total_flos": 4.839674807844864e+16,
"trial_name": null,
"trial_params": null
}