alpaca-native / checkpoint-400 /trainer_state.json
chavinlo's picture
Upload checkpoint-400/trainer_state.json with huggingface_hub
53ff912
raw
history blame
48.4 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.9843125192248539,
"global_step": 400,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 5.405405405405406e-07,
"loss": 1.5039,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 1.0810810810810812e-06,
"loss": 1.5785,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.6216216216216219e-06,
"loss": 1.2981,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 2.1621621621621623e-06,
"loss": 1.1631,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 2.702702702702703e-06,
"loss": 1.2493,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 3.2432432432432437e-06,
"loss": 1.2413,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 3.7837837837837844e-06,
"loss": 1.2368,
"step": 7
},
{
"epoch": 0.02,
"learning_rate": 4.324324324324325e-06,
"loss": 1.2152,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 4.864864864864866e-06,
"loss": 1.1487,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 5.405405405405406e-06,
"loss": 1.116,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 5.945945945945947e-06,
"loss": 1.179,
"step": 11
},
{
"epoch": 0.03,
"learning_rate": 6.486486486486487e-06,
"loss": 1.2134,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 7.027027027027028e-06,
"loss": 1.1641,
"step": 13
},
{
"epoch": 0.03,
"learning_rate": 7.567567567567569e-06,
"loss": 1.0623,
"step": 14
},
{
"epoch": 0.04,
"learning_rate": 8.108108108108109e-06,
"loss": 1.0922,
"step": 15
},
{
"epoch": 0.04,
"learning_rate": 8.64864864864865e-06,
"loss": 1.0979,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 9.189189189189191e-06,
"loss": 1.1572,
"step": 17
},
{
"epoch": 0.04,
"learning_rate": 9.729729729729732e-06,
"loss": 1.0788,
"step": 18
},
{
"epoch": 0.05,
"learning_rate": 1.027027027027027e-05,
"loss": 1.1467,
"step": 19
},
{
"epoch": 0.05,
"learning_rate": 1.0810810810810812e-05,
"loss": 1.0852,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 1.1351351351351352e-05,
"loss": 1.1176,
"step": 21
},
{
"epoch": 0.05,
"learning_rate": 1.1891891891891894e-05,
"loss": 1.1049,
"step": 22
},
{
"epoch": 0.06,
"learning_rate": 1.2432432432432433e-05,
"loss": 1.0972,
"step": 23
},
{
"epoch": 0.06,
"learning_rate": 1.2972972972972975e-05,
"loss": 1.1404,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 1.3513513513513515e-05,
"loss": 1.1416,
"step": 25
},
{
"epoch": 0.06,
"learning_rate": 1.4054054054054055e-05,
"loss": 1.1458,
"step": 26
},
{
"epoch": 0.07,
"learning_rate": 1.4594594594594596e-05,
"loss": 1.1497,
"step": 27
},
{
"epoch": 0.07,
"learning_rate": 1.5135135135135138e-05,
"loss": 1.14,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 1.5675675675675676e-05,
"loss": 1.0808,
"step": 29
},
{
"epoch": 0.07,
"learning_rate": 1.6216216216216218e-05,
"loss": 1.0996,
"step": 30
},
{
"epoch": 0.08,
"learning_rate": 1.6756756756756757e-05,
"loss": 1.1064,
"step": 31
},
{
"epoch": 0.08,
"learning_rate": 1.72972972972973e-05,
"loss": 1.1823,
"step": 32
},
{
"epoch": 0.08,
"learning_rate": 1.783783783783784e-05,
"loss": 1.0483,
"step": 33
},
{
"epoch": 0.08,
"learning_rate": 1.8378378378378383e-05,
"loss": 1.1019,
"step": 34
},
{
"epoch": 0.09,
"learning_rate": 1.891891891891892e-05,
"loss": 1.0325,
"step": 35
},
{
"epoch": 0.09,
"learning_rate": 1.9459459459459463e-05,
"loss": 1.1006,
"step": 36
},
{
"epoch": 0.09,
"learning_rate": 2e-05,
"loss": 1.0715,
"step": 37
},
{
"epoch": 0.09,
"learning_rate": 1.999996461903301e-05,
"loss": 1.1542,
"step": 38
},
{
"epoch": 0.1,
"learning_rate": 1.9999858476382388e-05,
"loss": 1.1402,
"step": 39
},
{
"epoch": 0.1,
"learning_rate": 1.9999681572799226e-05,
"loss": 1.0163,
"step": 40
},
{
"epoch": 0.1,
"learning_rate": 1.9999433909535333e-05,
"loss": 1.0851,
"step": 41
},
{
"epoch": 0.1,
"learning_rate": 1.9999115488343213e-05,
"loss": 1.1119,
"step": 42
},
{
"epoch": 0.11,
"learning_rate": 1.999872631147608e-05,
"loss": 1.0726,
"step": 43
},
{
"epoch": 0.11,
"learning_rate": 1.999826638168783e-05,
"loss": 1.0927,
"step": 44
},
{
"epoch": 0.11,
"learning_rate": 1.9997735702233006e-05,
"loss": 1.0697,
"step": 45
},
{
"epoch": 0.11,
"learning_rate": 1.99971342768668e-05,
"loss": 1.0363,
"step": 46
},
{
"epoch": 0.12,
"learning_rate": 1.999646210984502e-05,
"loss": 1.1251,
"step": 47
},
{
"epoch": 0.12,
"learning_rate": 1.999571920592405e-05,
"loss": 1.127,
"step": 48
},
{
"epoch": 0.12,
"learning_rate": 1.9994905570360817e-05,
"loss": 1.0994,
"step": 49
},
{
"epoch": 0.12,
"learning_rate": 1.999402120891276e-05,
"loss": 1.0688,
"step": 50
},
{
"epoch": 0.13,
"learning_rate": 1.99930661278378e-05,
"loss": 1.0544,
"step": 51
},
{
"epoch": 0.13,
"learning_rate": 1.9992040333894273e-05,
"loss": 1.0908,
"step": 52
},
{
"epoch": 0.13,
"learning_rate": 1.9990943834340893e-05,
"loss": 1.1394,
"step": 53
},
{
"epoch": 0.13,
"learning_rate": 1.9989776636936705e-05,
"loss": 1.1065,
"step": 54
},
{
"epoch": 0.14,
"learning_rate": 1.9988538749941024e-05,
"loss": 1.0752,
"step": 55
},
{
"epoch": 0.14,
"learning_rate": 1.9987230182113374e-05,
"loss": 1.076,
"step": 56
},
{
"epoch": 0.14,
"learning_rate": 1.998585094271344e-05,
"loss": 1.0504,
"step": 57
},
{
"epoch": 0.14,
"learning_rate": 1.998440104150098e-05,
"loss": 1.0934,
"step": 58
},
{
"epoch": 0.15,
"learning_rate": 1.998288048873578e-05,
"loss": 1.1355,
"step": 59
},
{
"epoch": 0.15,
"learning_rate": 1.9981289295177566e-05,
"loss": 1.1693,
"step": 60
},
{
"epoch": 0.15,
"learning_rate": 1.9979627472085927e-05,
"loss": 1.0514,
"step": 61
},
{
"epoch": 0.15,
"learning_rate": 1.997789503122025e-05,
"loss": 1.119,
"step": 62
},
{
"epoch": 0.16,
"learning_rate": 1.9976091984839616e-05,
"loss": 1.0635,
"step": 63
},
{
"epoch": 0.16,
"learning_rate": 1.9974218345702733e-05,
"loss": 1.1111,
"step": 64
},
{
"epoch": 0.16,
"learning_rate": 1.9972274127067838e-05,
"loss": 1.1732,
"step": 65
},
{
"epoch": 0.16,
"learning_rate": 1.997025934269259e-05,
"loss": 1.149,
"step": 66
},
{
"epoch": 0.16,
"learning_rate": 1.9968174006833996e-05,
"loss": 1.1011,
"step": 67
},
{
"epoch": 0.17,
"learning_rate": 1.9966018134248296e-05,
"loss": 1.1367,
"step": 68
},
{
"epoch": 0.17,
"learning_rate": 1.9963791740190863e-05,
"loss": 1.1354,
"step": 69
},
{
"epoch": 0.17,
"learning_rate": 1.996149484041609e-05,
"loss": 1.0343,
"step": 70
},
{
"epoch": 0.17,
"learning_rate": 1.9959127451177287e-05,
"loss": 1.1997,
"step": 71
},
{
"epoch": 0.18,
"learning_rate": 1.9956689589226555e-05,
"loss": 1.1131,
"step": 72
},
{
"epoch": 0.18,
"learning_rate": 1.9954181271814673e-05,
"loss": 1.0335,
"step": 73
},
{
"epoch": 0.18,
"learning_rate": 1.9951602516690988e-05,
"loss": 1.0786,
"step": 74
},
{
"epoch": 0.18,
"learning_rate": 1.9948953342103268e-05,
"loss": 1.0926,
"step": 75
},
{
"epoch": 0.19,
"learning_rate": 1.994623376679758e-05,
"loss": 1.1353,
"step": 76
},
{
"epoch": 0.19,
"learning_rate": 1.9943443810018174e-05,
"loss": 1.0605,
"step": 77
},
{
"epoch": 0.19,
"learning_rate": 1.9940583491507314e-05,
"loss": 1.1255,
"step": 78
},
{
"epoch": 0.19,
"learning_rate": 1.993765283150517e-05,
"loss": 1.0541,
"step": 79
},
{
"epoch": 0.2,
"learning_rate": 1.9934651850749663e-05,
"loss": 1.1197,
"step": 80
},
{
"epoch": 0.2,
"learning_rate": 1.9931580570476306e-05,
"loss": 1.1297,
"step": 81
},
{
"epoch": 0.2,
"learning_rate": 1.9928439012418076e-05,
"loss": 1.1324,
"step": 82
},
{
"epoch": 0.2,
"learning_rate": 1.9925227198805247e-05,
"loss": 1.0846,
"step": 83
},
{
"epoch": 0.21,
"learning_rate": 1.9921945152365235e-05,
"loss": 1.1831,
"step": 84
},
{
"epoch": 0.21,
"learning_rate": 1.9918592896322432e-05,
"loss": 1.0921,
"step": 85
},
{
"epoch": 0.21,
"learning_rate": 1.9915170454398045e-05,
"loss": 1.1037,
"step": 86
},
{
"epoch": 0.21,
"learning_rate": 1.9911677850809943e-05,
"loss": 1.125,
"step": 87
},
{
"epoch": 0.22,
"learning_rate": 1.9908115110272463e-05,
"loss": 1.1279,
"step": 88
},
{
"epoch": 0.22,
"learning_rate": 1.9904482257996244e-05,
"loss": 1.1427,
"step": 89
},
{
"epoch": 0.22,
"learning_rate": 1.990077931968805e-05,
"loss": 1.1475,
"step": 90
},
{
"epoch": 0.22,
"learning_rate": 1.9897006321550592e-05,
"loss": 1.136,
"step": 91
},
{
"epoch": 0.23,
"learning_rate": 1.9893163290282335e-05,
"loss": 1.0469,
"step": 92
},
{
"epoch": 0.23,
"learning_rate": 1.9889250253077306e-05,
"loss": 1.1028,
"step": 93
},
{
"epoch": 0.23,
"learning_rate": 1.9885267237624923e-05,
"loss": 1.1775,
"step": 94
},
{
"epoch": 0.23,
"learning_rate": 1.988121427210976e-05,
"loss": 1.1744,
"step": 95
},
{
"epoch": 0.24,
"learning_rate": 1.98770913852114e-05,
"loss": 1.1049,
"step": 96
},
{
"epoch": 0.24,
"learning_rate": 1.9872898606104175e-05,
"loss": 1.1085,
"step": 97
},
{
"epoch": 0.24,
"learning_rate": 1.9868635964457007e-05,
"loss": 1.1153,
"step": 98
},
{
"epoch": 0.24,
"learning_rate": 1.986430349043317e-05,
"loss": 1.1637,
"step": 99
},
{
"epoch": 0.25,
"learning_rate": 1.9859901214690094e-05,
"loss": 1.0971,
"step": 100
},
{
"epoch": 0.25,
"learning_rate": 1.9855429168379127e-05,
"loss": 1.1537,
"step": 101
},
{
"epoch": 0.25,
"learning_rate": 1.985088738314534e-05,
"loss": 1.0882,
"step": 102
},
{
"epoch": 0.25,
"learning_rate": 1.9846275891127275e-05,
"loss": 1.128,
"step": 103
},
{
"epoch": 0.26,
"learning_rate": 1.9841594724956746e-05,
"loss": 1.14,
"step": 104
},
{
"epoch": 0.26,
"learning_rate": 1.9836843917758593e-05,
"loss": 1.0273,
"step": 105
},
{
"epoch": 0.26,
"learning_rate": 1.983202350315044e-05,
"loss": 1.0891,
"step": 106
},
{
"epoch": 0.26,
"learning_rate": 1.982713351524248e-05,
"loss": 1.106,
"step": 107
},
{
"epoch": 0.27,
"learning_rate": 1.982217398863721e-05,
"loss": 1.0599,
"step": 108
},
{
"epoch": 0.27,
"learning_rate": 1.98171449584292e-05,
"loss": 1.0396,
"step": 109
},
{
"epoch": 0.27,
"learning_rate": 1.9812046460204837e-05,
"loss": 1.0767,
"step": 110
},
{
"epoch": 0.27,
"learning_rate": 1.9806878530042083e-05,
"loss": 1.1568,
"step": 111
},
{
"epoch": 0.28,
"learning_rate": 1.9801641204510216e-05,
"loss": 1.1534,
"step": 112
},
{
"epoch": 0.28,
"learning_rate": 1.9796334520669555e-05,
"loss": 1.0332,
"step": 113
},
{
"epoch": 0.28,
"learning_rate": 1.9790958516071228e-05,
"loss": 1.1018,
"step": 114
},
{
"epoch": 0.28,
"learning_rate": 1.978551322875688e-05,
"loss": 1.0634,
"step": 115
},
{
"epoch": 0.29,
"learning_rate": 1.977999869725842e-05,
"loss": 1.0766,
"step": 116
},
{
"epoch": 0.29,
"learning_rate": 1.977441496059774e-05,
"loss": 1.0456,
"step": 117
},
{
"epoch": 0.29,
"learning_rate": 1.9768762058286433e-05,
"loss": 1.0047,
"step": 118
},
{
"epoch": 0.29,
"learning_rate": 1.976304003032554e-05,
"loss": 1.0798,
"step": 119
},
{
"epoch": 0.3,
"learning_rate": 1.9757248917205228e-05,
"loss": 1.1055,
"step": 120
},
{
"epoch": 0.3,
"learning_rate": 1.975138875990454e-05,
"loss": 1.1921,
"step": 121
},
{
"epoch": 0.3,
"learning_rate": 1.974545959989108e-05,
"loss": 1.0465,
"step": 122
},
{
"epoch": 0.3,
"learning_rate": 1.9739461479120727e-05,
"loss": 1.129,
"step": 123
},
{
"epoch": 0.31,
"learning_rate": 1.973339444003735e-05,
"loss": 1.0773,
"step": 124
},
{
"epoch": 0.31,
"learning_rate": 1.9727258525572487e-05,
"loss": 1.0955,
"step": 125
},
{
"epoch": 0.31,
"learning_rate": 1.9721053779145057e-05,
"loss": 1.0941,
"step": 126
},
{
"epoch": 0.31,
"learning_rate": 1.9714780244661044e-05,
"loss": 1.1672,
"step": 127
},
{
"epoch": 0.31,
"learning_rate": 1.9708437966513196e-05,
"loss": 1.1014,
"step": 128
},
{
"epoch": 0.32,
"learning_rate": 1.9702026989580694e-05,
"loss": 1.1272,
"step": 129
},
{
"epoch": 0.32,
"learning_rate": 1.969554735922885e-05,
"loss": 1.0398,
"step": 130
},
{
"epoch": 0.32,
"learning_rate": 1.968899912130879e-05,
"loss": 1.106,
"step": 131
},
{
"epoch": 0.32,
"learning_rate": 1.9682382322157103e-05,
"loss": 1.1033,
"step": 132
},
{
"epoch": 0.33,
"learning_rate": 1.9675697008595545e-05,
"loss": 1.0154,
"step": 133
},
{
"epoch": 0.33,
"learning_rate": 1.9668943227930686e-05,
"loss": 1.1219,
"step": 134
},
{
"epoch": 0.33,
"learning_rate": 1.966212102795358e-05,
"loss": 1.0824,
"step": 135
},
{
"epoch": 0.33,
"learning_rate": 1.965523045693944e-05,
"loss": 1.0581,
"step": 136
},
{
"epoch": 0.34,
"learning_rate": 1.964827156364728e-05,
"loss": 1.1327,
"step": 137
},
{
"epoch": 0.34,
"learning_rate": 1.964124439731957e-05,
"loss": 1.1144,
"step": 138
},
{
"epoch": 0.34,
"learning_rate": 1.9634149007681894e-05,
"loss": 1.1436,
"step": 139
},
{
"epoch": 0.34,
"learning_rate": 1.962698544494261e-05,
"loss": 1.0457,
"step": 140
},
{
"epoch": 0.35,
"learning_rate": 1.9619753759792466e-05,
"loss": 1.1053,
"step": 141
},
{
"epoch": 0.35,
"learning_rate": 1.961245400340427e-05,
"loss": 1.0881,
"step": 142
},
{
"epoch": 0.35,
"learning_rate": 1.9605086227432512e-05,
"loss": 1.1058,
"step": 143
},
{
"epoch": 0.35,
"learning_rate": 1.9597650484012997e-05,
"loss": 1.1418,
"step": 144
},
{
"epoch": 0.36,
"learning_rate": 1.9590146825762476e-05,
"loss": 1.0623,
"step": 145
},
{
"epoch": 0.36,
"learning_rate": 1.9582575305778297e-05,
"loss": 1.0987,
"step": 146
},
{
"epoch": 0.36,
"learning_rate": 1.9574935977637994e-05,
"loss": 1.0764,
"step": 147
},
{
"epoch": 0.36,
"learning_rate": 1.9567228895398936e-05,
"loss": 1.153,
"step": 148
},
{
"epoch": 0.37,
"learning_rate": 1.955945411359792e-05,
"loss": 1.1233,
"step": 149
},
{
"epoch": 0.37,
"learning_rate": 1.9551611687250808e-05,
"loss": 1.0827,
"step": 150
},
{
"epoch": 0.37,
"learning_rate": 1.9543701671852127e-05,
"loss": 1.0487,
"step": 151
},
{
"epoch": 0.37,
"learning_rate": 1.9535724123374674e-05,
"loss": 1.0795,
"step": 152
},
{
"epoch": 0.38,
"learning_rate": 1.952767909826913e-05,
"loss": 1.15,
"step": 153
},
{
"epoch": 0.38,
"learning_rate": 1.951956665346364e-05,
"loss": 1.0645,
"step": 154
},
{
"epoch": 0.38,
"learning_rate": 1.951138684636344e-05,
"loss": 1.0764,
"step": 155
},
{
"epoch": 0.38,
"learning_rate": 1.9503139734850426e-05,
"loss": 1.1185,
"step": 156
},
{
"epoch": 0.39,
"learning_rate": 1.9494825377282746e-05,
"loss": 1.0879,
"step": 157
},
{
"epoch": 0.39,
"learning_rate": 1.9486443832494414e-05,
"loss": 1.0558,
"step": 158
},
{
"epoch": 0.39,
"learning_rate": 1.9477995159794854e-05,
"loss": 1.1042,
"step": 159
},
{
"epoch": 0.39,
"learning_rate": 1.9469479418968506e-05,
"loss": 1.1029,
"step": 160
},
{
"epoch": 0.4,
"learning_rate": 1.9460896670274408e-05,
"loss": 0.9836,
"step": 161
},
{
"epoch": 0.4,
"learning_rate": 1.9452246974445743e-05,
"loss": 1.1325,
"step": 162
},
{
"epoch": 0.4,
"learning_rate": 1.9443530392689434e-05,
"loss": 1.1018,
"step": 163
},
{
"epoch": 0.4,
"learning_rate": 1.94347469866857e-05,
"loss": 1.1525,
"step": 164
},
{
"epoch": 0.41,
"learning_rate": 1.9425896818587615e-05,
"loss": 1.0719,
"step": 165
},
{
"epoch": 0.41,
"learning_rate": 1.941697995102069e-05,
"loss": 1.1595,
"step": 166
},
{
"epoch": 0.41,
"learning_rate": 1.9407996447082394e-05,
"loss": 1.1698,
"step": 167
},
{
"epoch": 0.41,
"learning_rate": 1.939894637034174e-05,
"loss": 1.131,
"step": 168
},
{
"epoch": 0.42,
"learning_rate": 1.9389829784838833e-05,
"loss": 1.0604,
"step": 169
},
{
"epoch": 0.42,
"learning_rate": 1.938064675508438e-05,
"loss": 1.0422,
"step": 170
},
{
"epoch": 0.42,
"learning_rate": 1.9371397346059286e-05,
"loss": 1.1098,
"step": 171
},
{
"epoch": 0.42,
"learning_rate": 1.936208162321415e-05,
"loss": 1.0682,
"step": 172
},
{
"epoch": 0.43,
"learning_rate": 1.9352699652468835e-05,
"loss": 1.0763,
"step": 173
},
{
"epoch": 0.43,
"learning_rate": 1.9343251500211977e-05,
"loss": 1.1038,
"step": 174
},
{
"epoch": 0.43,
"learning_rate": 1.933373723330053e-05,
"loss": 1.0613,
"step": 175
},
{
"epoch": 0.43,
"learning_rate": 1.9324156919059286e-05,
"loss": 1.1157,
"step": 176
},
{
"epoch": 0.44,
"learning_rate": 1.93145106252804e-05,
"loss": 1.1264,
"step": 177
},
{
"epoch": 0.44,
"learning_rate": 1.9304798420222918e-05,
"loss": 1.0787,
"step": 178
},
{
"epoch": 0.44,
"learning_rate": 1.9295020372612276e-05,
"loss": 1.1298,
"step": 179
},
{
"epoch": 0.44,
"learning_rate": 1.9285176551639826e-05,
"loss": 1.1175,
"step": 180
},
{
"epoch": 0.45,
"learning_rate": 1.9275267026962358e-05,
"loss": 1.1686,
"step": 181
},
{
"epoch": 0.45,
"learning_rate": 1.9265291868701584e-05,
"loss": 1.0939,
"step": 182
},
{
"epoch": 0.45,
"learning_rate": 1.9255251147443646e-05,
"loss": 1.1674,
"step": 183
},
{
"epoch": 0.45,
"learning_rate": 1.924514493423864e-05,
"loss": 1.0747,
"step": 184
},
{
"epoch": 0.46,
"learning_rate": 1.9234973300600074e-05,
"loss": 1.1085,
"step": 185
},
{
"epoch": 0.46,
"learning_rate": 1.92247363185044e-05,
"loss": 1.1157,
"step": 186
},
{
"epoch": 0.46,
"learning_rate": 1.9214434060390484e-05,
"loss": 1.1827,
"step": 187
},
{
"epoch": 0.46,
"learning_rate": 1.9204066599159094e-05,
"loss": 1.1126,
"step": 188
},
{
"epoch": 0.47,
"learning_rate": 1.9193634008172396e-05,
"loss": 1.1158,
"step": 189
},
{
"epoch": 0.47,
"learning_rate": 1.9183136361253417e-05,
"loss": 1.0609,
"step": 190
},
{
"epoch": 0.47,
"learning_rate": 1.917257373268554e-05,
"loss": 1.0617,
"step": 191
},
{
"epoch": 0.47,
"learning_rate": 1.916194619721196e-05,
"loss": 1.0724,
"step": 192
},
{
"epoch": 0.47,
"learning_rate": 1.915125383003518e-05,
"loss": 1.0784,
"step": 193
},
{
"epoch": 0.48,
"learning_rate": 1.914049670681646e-05,
"loss": 1.1141,
"step": 194
},
{
"epoch": 0.48,
"learning_rate": 1.912967490367528e-05,
"loss": 1.0305,
"step": 195
},
{
"epoch": 0.48,
"learning_rate": 1.9118788497188815e-05,
"loss": 1.026,
"step": 196
},
{
"epoch": 0.48,
"learning_rate": 1.9107837564391376e-05,
"loss": 1.0663,
"step": 197
},
{
"epoch": 0.49,
"learning_rate": 1.9096822182773887e-05,
"loss": 1.0368,
"step": 198
},
{
"epoch": 0.49,
"learning_rate": 1.9085742430283322e-05,
"loss": 1.1052,
"step": 199
},
{
"epoch": 0.49,
"learning_rate": 1.907459838532215e-05,
"loss": 1.0834,
"step": 200
},
{
"epoch": 0.49,
"learning_rate": 1.9063390126747778e-05,
"loss": 1.1579,
"step": 201
},
{
"epoch": 0.5,
"learning_rate": 1.9052117733872025e-05,
"loss": 1.1034,
"step": 202
},
{
"epoch": 0.5,
"learning_rate": 1.904078128646052e-05,
"loss": 1.1173,
"step": 203
},
{
"epoch": 0.5,
"learning_rate": 1.902938086473215e-05,
"loss": 1.1546,
"step": 204
},
{
"epoch": 0.5,
"learning_rate": 1.901791654935852e-05,
"loss": 1.1274,
"step": 205
},
{
"epoch": 0.51,
"learning_rate": 1.9006388421463322e-05,
"loss": 1.0896,
"step": 206
},
{
"epoch": 0.51,
"learning_rate": 1.899479656262183e-05,
"loss": 1.1937,
"step": 207
},
{
"epoch": 0.51,
"learning_rate": 1.898314105486028e-05,
"loss": 1.0613,
"step": 208
},
{
"epoch": 0.51,
"learning_rate": 1.8971421980655295e-05,
"loss": 1.0896,
"step": 209
},
{
"epoch": 0.52,
"learning_rate": 1.8959639422933316e-05,
"loss": 1.0578,
"step": 210
},
{
"epoch": 0.52,
"learning_rate": 1.894779346506999e-05,
"loss": 1.0935,
"step": 211
},
{
"epoch": 0.52,
"learning_rate": 1.893588419088962e-05,
"loss": 1.0655,
"step": 212
},
{
"epoch": 0.52,
"learning_rate": 1.892391168466452e-05,
"loss": 1.1562,
"step": 213
},
{
"epoch": 0.53,
"learning_rate": 1.891187603111447e-05,
"loss": 1.1434,
"step": 214
},
{
"epoch": 0.53,
"learning_rate": 1.8899777315406073e-05,
"loss": 1.0462,
"step": 215
},
{
"epoch": 0.53,
"learning_rate": 1.8887615623152188e-05,
"loss": 1.1151,
"step": 216
},
{
"epoch": 0.53,
"learning_rate": 1.88753910404113e-05,
"loss": 1.1457,
"step": 217
},
{
"epoch": 0.54,
"learning_rate": 1.8863103653686917e-05,
"loss": 1.0401,
"step": 218
},
{
"epoch": 0.54,
"learning_rate": 1.8850753549926967e-05,
"loss": 1.074,
"step": 219
},
{
"epoch": 0.54,
"learning_rate": 1.8838340816523175e-05,
"loss": 1.14,
"step": 220
},
{
"epoch": 0.54,
"learning_rate": 1.8825865541310438e-05,
"loss": 1.1483,
"step": 221
},
{
"epoch": 0.55,
"learning_rate": 1.8813327812566217e-05,
"loss": 1.156,
"step": 222
},
{
"epoch": 0.55,
"learning_rate": 1.880072771900991e-05,
"loss": 1.0334,
"step": 223
},
{
"epoch": 0.55,
"learning_rate": 1.878806534980221e-05,
"loss": 1.1854,
"step": 224
},
{
"epoch": 0.55,
"learning_rate": 1.8775340794544497e-05,
"loss": 1.0593,
"step": 225
},
{
"epoch": 0.56,
"learning_rate": 1.876255414327818e-05,
"loss": 1.0863,
"step": 226
},
{
"epoch": 0.56,
"learning_rate": 1.8749705486484074e-05,
"loss": 1.2038,
"step": 227
},
{
"epoch": 0.56,
"learning_rate": 1.8736794915081765e-05,
"loss": 1.0991,
"step": 228
},
{
"epoch": 0.56,
"learning_rate": 1.8723822520428954e-05,
"loss": 1.0572,
"step": 229
},
{
"epoch": 0.57,
"learning_rate": 1.8710788394320807e-05,
"loss": 1.0798,
"step": 230
},
{
"epoch": 0.57,
"learning_rate": 1.8697692628989327e-05,
"loss": 1.082,
"step": 231
},
{
"epoch": 0.57,
"learning_rate": 1.868453531710268e-05,
"loss": 1.1121,
"step": 232
},
{
"epoch": 0.57,
"learning_rate": 1.8671316551764552e-05,
"loss": 1.089,
"step": 233
},
{
"epoch": 0.58,
"learning_rate": 1.865803642651348e-05,
"loss": 1.1132,
"step": 234
},
{
"epoch": 0.58,
"learning_rate": 1.8644695035322203e-05,
"loss": 1.1323,
"step": 235
},
{
"epoch": 0.58,
"learning_rate": 1.8631292472596978e-05,
"loss": 1.1352,
"step": 236
},
{
"epoch": 0.58,
"learning_rate": 1.8617828833176935e-05,
"loss": 1.0801,
"step": 237
},
{
"epoch": 0.59,
"learning_rate": 1.860430421233339e-05,
"loss": 1.1049,
"step": 238
},
{
"epoch": 0.59,
"learning_rate": 1.859071870576918e-05,
"loss": 1.0518,
"step": 239
},
{
"epoch": 0.59,
"learning_rate": 1.857707240961797e-05,
"loss": 1.0691,
"step": 240
},
{
"epoch": 0.59,
"learning_rate": 1.8563365420443594e-05,
"loss": 1.0941,
"step": 241
},
{
"epoch": 0.6,
"learning_rate": 1.854959783523936e-05,
"loss": 1.1016,
"step": 242
},
{
"epoch": 0.6,
"learning_rate": 1.853576975142736e-05,
"loss": 1.0896,
"step": 243
},
{
"epoch": 0.6,
"learning_rate": 1.852188126685779e-05,
"loss": 1.0822,
"step": 244
},
{
"epoch": 0.6,
"learning_rate": 1.8507932479808254e-05,
"loss": 0.9657,
"step": 245
},
{
"epoch": 0.61,
"learning_rate": 1.8493923488983066e-05,
"loss": 1.0548,
"step": 246
},
{
"epoch": 0.61,
"learning_rate": 1.847985439351256e-05,
"loss": 1.0908,
"step": 247
},
{
"epoch": 0.61,
"learning_rate": 1.846572529295237e-05,
"loss": 1.1709,
"step": 248
},
{
"epoch": 0.61,
"learning_rate": 1.845153628728274e-05,
"loss": 1.1686,
"step": 249
},
{
"epoch": 0.62,
"learning_rate": 1.8437287476907828e-05,
"loss": 1.1704,
"step": 250
},
{
"epoch": 0.62,
"learning_rate": 1.842297896265497e-05,
"loss": 1.054,
"step": 251
},
{
"epoch": 0.62,
"learning_rate": 1.8408610845773974e-05,
"loss": 1.1077,
"step": 252
},
{
"epoch": 0.62,
"learning_rate": 1.8394183227936418e-05,
"loss": 1.0797,
"step": 253
},
{
"epoch": 0.63,
"learning_rate": 1.8379696211234918e-05,
"loss": 1.1079,
"step": 254
},
{
"epoch": 0.63,
"learning_rate": 1.8365149898182403e-05,
"loss": 1.11,
"step": 255
},
{
"epoch": 0.63,
"learning_rate": 1.8350544391711396e-05,
"loss": 1.0777,
"step": 256
},
{
"epoch": 0.63,
"learning_rate": 1.833587979517329e-05,
"loss": 1.0761,
"step": 257
},
{
"epoch": 0.63,
"learning_rate": 1.8321156212337604e-05,
"loss": 1.0941,
"step": 258
},
{
"epoch": 0.64,
"learning_rate": 1.830637374739126e-05,
"loss": 1.128,
"step": 259
},
{
"epoch": 0.64,
"learning_rate": 1.829153250493783e-05,
"loss": 1.0112,
"step": 260
},
{
"epoch": 0.64,
"learning_rate": 1.827663258999683e-05,
"loss": 1.0975,
"step": 261
},
{
"epoch": 0.64,
"learning_rate": 1.8261674108002925e-05,
"loss": 1.0442,
"step": 262
},
{
"epoch": 0.65,
"learning_rate": 1.824665716480524e-05,
"loss": 1.0653,
"step": 263
},
{
"epoch": 0.65,
"learning_rate": 1.823158186666656e-05,
"loss": 1.0561,
"step": 264
},
{
"epoch": 0.65,
"learning_rate": 1.821644832026261e-05,
"loss": 1.1615,
"step": 265
},
{
"epoch": 0.65,
"learning_rate": 1.82012566326813e-05,
"loss": 1.1045,
"step": 266
},
{
"epoch": 0.66,
"learning_rate": 1.8186006911421937e-05,
"loss": 1.1486,
"step": 267
},
{
"epoch": 0.66,
"learning_rate": 1.817069926439451e-05,
"loss": 1.1622,
"step": 268
},
{
"epoch": 0.66,
"learning_rate": 1.8155333799918883e-05,
"loss": 1.0986,
"step": 269
},
{
"epoch": 0.66,
"learning_rate": 1.8139910626724058e-05,
"loss": 1.0416,
"step": 270
},
{
"epoch": 0.67,
"learning_rate": 1.8124429853947387e-05,
"loss": 1.1328,
"step": 271
},
{
"epoch": 0.67,
"learning_rate": 1.8108891591133812e-05,
"loss": 1.0921,
"step": 272
},
{
"epoch": 0.67,
"learning_rate": 1.809329594823509e-05,
"loss": 1.0795,
"step": 273
},
{
"epoch": 0.67,
"learning_rate": 1.8077643035609006e-05,
"loss": 1.1597,
"step": 274
},
{
"epoch": 0.68,
"learning_rate": 1.806193296401859e-05,
"loss": 1.0555,
"step": 275
},
{
"epoch": 0.68,
"learning_rate": 1.804616584463136e-05,
"loss": 1.1308,
"step": 276
},
{
"epoch": 0.68,
"learning_rate": 1.803034178901849e-05,
"loss": 1.0728,
"step": 277
},
{
"epoch": 0.68,
"learning_rate": 1.8014460909154058e-05,
"loss": 1.0919,
"step": 278
},
{
"epoch": 0.69,
"learning_rate": 1.799852331741425e-05,
"loss": 1.1573,
"step": 279
},
{
"epoch": 0.69,
"learning_rate": 1.7982529126576543e-05,
"loss": 1.0903,
"step": 280
},
{
"epoch": 0.69,
"learning_rate": 1.7966478449818925e-05,
"loss": 1.1511,
"step": 281
},
{
"epoch": 0.69,
"learning_rate": 1.7950371400719087e-05,
"loss": 1.1447,
"step": 282
},
{
"epoch": 0.7,
"learning_rate": 1.7934208093253625e-05,
"loss": 1.1149,
"step": 283
},
{
"epoch": 0.7,
"learning_rate": 1.7917988641797227e-05,
"loss": 1.1297,
"step": 284
},
{
"epoch": 0.7,
"learning_rate": 1.7901713161121873e-05,
"loss": 1.0903,
"step": 285
},
{
"epoch": 0.7,
"learning_rate": 1.7885381766396008e-05,
"loss": 1.0677,
"step": 286
},
{
"epoch": 0.71,
"learning_rate": 1.786899457318374e-05,
"loss": 1.0687,
"step": 287
},
{
"epoch": 0.71,
"learning_rate": 1.7852551697444017e-05,
"loss": 1.0536,
"step": 288
},
{
"epoch": 0.71,
"learning_rate": 1.783605325552981e-05,
"loss": 1.0364,
"step": 289
},
{
"epoch": 0.71,
"learning_rate": 1.7819499364187282e-05,
"loss": 1.1249,
"step": 290
},
{
"epoch": 0.72,
"learning_rate": 1.780289014055497e-05,
"loss": 1.0534,
"step": 291
},
{
"epoch": 0.72,
"learning_rate": 1.7786225702162955e-05,
"loss": 1.0512,
"step": 292
},
{
"epoch": 0.72,
"learning_rate": 1.7769506166932026e-05,
"loss": 1.1828,
"step": 293
},
{
"epoch": 0.72,
"learning_rate": 1.7752731653172847e-05,
"loss": 1.064,
"step": 294
},
{
"epoch": 0.73,
"learning_rate": 1.7735902279585118e-05,
"loss": 1.0973,
"step": 295
},
{
"epoch": 0.73,
"learning_rate": 1.7719018165256745e-05,
"loss": 1.0576,
"step": 296
},
{
"epoch": 0.73,
"learning_rate": 1.7702079429662986e-05,
"loss": 1.1355,
"step": 297
},
{
"epoch": 0.73,
"learning_rate": 1.7685086192665605e-05,
"loss": 0.9462,
"step": 298
},
{
"epoch": 0.74,
"learning_rate": 1.7668038574512045e-05,
"loss": 1.0778,
"step": 299
},
{
"epoch": 0.74,
"learning_rate": 1.7650936695834536e-05,
"loss": 1.1013,
"step": 300
},
{
"epoch": 0.74,
"learning_rate": 1.763378067764929e-05,
"loss": 1.0889,
"step": 301
},
{
"epoch": 0.74,
"learning_rate": 1.7616570641355602e-05,
"loss": 1.013,
"step": 302
},
{
"epoch": 0.75,
"learning_rate": 1.759930670873502e-05,
"loss": 1.0847,
"step": 303
},
{
"epoch": 0.75,
"learning_rate": 1.758198900195047e-05,
"loss": 1.1345,
"step": 304
},
{
"epoch": 0.75,
"learning_rate": 1.7564617643545395e-05,
"loss": 1.0481,
"step": 305
},
{
"epoch": 0.75,
"learning_rate": 1.7547192756442887e-05,
"loss": 1.0952,
"step": 306
},
{
"epoch": 0.76,
"learning_rate": 1.7529714463944815e-05,
"loss": 1.0869,
"step": 307
},
{
"epoch": 0.76,
"learning_rate": 1.751218288973096e-05,
"loss": 1.1197,
"step": 308
},
{
"epoch": 0.76,
"learning_rate": 1.7494598157858127e-05,
"loss": 1.1011,
"step": 309
},
{
"epoch": 0.76,
"learning_rate": 1.7476960392759284e-05,
"loss": 1.079,
"step": 310
},
{
"epoch": 0.77,
"learning_rate": 1.7459269719242665e-05,
"loss": 1.0569,
"step": 311
},
{
"epoch": 0.77,
"learning_rate": 1.74415262624909e-05,
"loss": 1.1322,
"step": 312
},
{
"epoch": 0.77,
"learning_rate": 1.742373014806012e-05,
"loss": 1.121,
"step": 313
},
{
"epoch": 0.77,
"learning_rate": 1.740588150187907e-05,
"loss": 1.1544,
"step": 314
},
{
"epoch": 0.78,
"learning_rate": 1.7387980450248222e-05,
"loss": 1.1591,
"step": 315
},
{
"epoch": 0.78,
"learning_rate": 1.7370027119838884e-05,
"loss": 1.0956,
"step": 316
},
{
"epoch": 0.78,
"learning_rate": 1.735202163769229e-05,
"loss": 1.1096,
"step": 317
},
{
"epoch": 0.78,
"learning_rate": 1.7333964131218714e-05,
"loss": 1.0962,
"step": 318
},
{
"epoch": 0.78,
"learning_rate": 1.7315854728196568e-05,
"loss": 1.1088,
"step": 319
},
{
"epoch": 0.79,
"learning_rate": 1.729769355677149e-05,
"loss": 1.0693,
"step": 320
},
{
"epoch": 0.79,
"learning_rate": 1.7279480745455433e-05,
"loss": 1.0909,
"step": 321
},
{
"epoch": 0.79,
"learning_rate": 1.7261216423125782e-05,
"loss": 1.0636,
"step": 322
},
{
"epoch": 0.79,
"learning_rate": 1.724290071902441e-05,
"loss": 1.131,
"step": 323
},
{
"epoch": 0.8,
"learning_rate": 1.7224533762756775e-05,
"loss": 1.1824,
"step": 324
},
{
"epoch": 0.8,
"learning_rate": 1.720611568429103e-05,
"loss": 1.002,
"step": 325
},
{
"epoch": 0.8,
"learning_rate": 1.718764661395704e-05,
"loss": 1.0908,
"step": 326
},
{
"epoch": 0.8,
"learning_rate": 1.716912668244553e-05,
"loss": 1.1213,
"step": 327
},
{
"epoch": 0.81,
"learning_rate": 1.715055602080711e-05,
"loss": 1.1056,
"step": 328
},
{
"epoch": 0.81,
"learning_rate": 1.7131934760451385e-05,
"loss": 1.073,
"step": 329
},
{
"epoch": 0.81,
"learning_rate": 1.7113263033145985e-05,
"loss": 1.1817,
"step": 330
},
{
"epoch": 0.81,
"learning_rate": 1.7094540971015663e-05,
"loss": 1.0157,
"step": 331
},
{
"epoch": 0.82,
"learning_rate": 1.7075768706541355e-05,
"loss": 1.0835,
"step": 332
},
{
"epoch": 0.82,
"learning_rate": 1.7056946372559234e-05,
"loss": 0.9943,
"step": 333
},
{
"epoch": 0.82,
"learning_rate": 1.7038074102259775e-05,
"loss": 1.1713,
"step": 334
},
{
"epoch": 0.82,
"learning_rate": 1.7019152029186817e-05,
"loss": 1.0421,
"step": 335
},
{
"epoch": 0.83,
"learning_rate": 1.70001802872366e-05,
"loss": 1.1,
"step": 336
},
{
"epoch": 0.83,
"learning_rate": 1.6981159010656847e-05,
"loss": 1.0711,
"step": 337
},
{
"epoch": 0.83,
"learning_rate": 1.6962088334045785e-05,
"loss": 1.1572,
"step": 338
},
{
"epoch": 0.83,
"learning_rate": 1.694296839235121e-05,
"loss": 1.0448,
"step": 339
},
{
"epoch": 0.84,
"learning_rate": 1.692379932086953e-05,
"loss": 1.0735,
"step": 340
},
{
"epoch": 0.84,
"learning_rate": 1.6904581255244802e-05,
"loss": 1.0977,
"step": 341
},
{
"epoch": 0.84,
"learning_rate": 1.688531433146777e-05,
"loss": 1.0777,
"step": 342
},
{
"epoch": 0.84,
"learning_rate": 1.6865998685874923e-05,
"loss": 1.077,
"step": 343
},
{
"epoch": 0.85,
"learning_rate": 1.6846634455147498e-05,
"loss": 1.0797,
"step": 344
},
{
"epoch": 0.85,
"learning_rate": 1.6827221776310532e-05,
"loss": 1.0048,
"step": 345
},
{
"epoch": 0.85,
"learning_rate": 1.6807760786731905e-05,
"loss": 1.0392,
"step": 346
},
{
"epoch": 0.85,
"learning_rate": 1.6788251624121335e-05,
"loss": 1.1693,
"step": 347
},
{
"epoch": 0.86,
"learning_rate": 1.6768694426529432e-05,
"loss": 1.1036,
"step": 348
},
{
"epoch": 0.86,
"learning_rate": 1.6749089332346714e-05,
"loss": 1.0874,
"step": 349
},
{
"epoch": 0.86,
"learning_rate": 1.672943648030261e-05,
"loss": 1.1224,
"step": 350
},
{
"epoch": 0.86,
"learning_rate": 1.6709736009464504e-05,
"loss": 1.0791,
"step": 351
},
{
"epoch": 0.87,
"learning_rate": 1.668998805923675e-05,
"loss": 1.0822,
"step": 352
},
{
"epoch": 0.87,
"learning_rate": 1.6670192769359643e-05,
"loss": 1.0818,
"step": 353
},
{
"epoch": 0.87,
"learning_rate": 1.6650350279908497e-05,
"loss": 1.1335,
"step": 354
},
{
"epoch": 0.87,
"learning_rate": 1.6630460731292597e-05,
"loss": 1.0699,
"step": 355
},
{
"epoch": 0.88,
"learning_rate": 1.661052426425424e-05,
"loss": 1.1318,
"step": 356
},
{
"epoch": 0.88,
"learning_rate": 1.6590541019867722e-05,
"loss": 1.1362,
"step": 357
},
{
"epoch": 0.88,
"learning_rate": 1.6570511139538348e-05,
"loss": 1.0223,
"step": 358
},
{
"epoch": 0.88,
"learning_rate": 1.655043476500142e-05,
"loss": 1.0471,
"step": 359
},
{
"epoch": 0.89,
"learning_rate": 1.6530312038321247e-05,
"loss": 1.1665,
"step": 360
},
{
"epoch": 0.89,
"learning_rate": 1.6510143101890136e-05,
"loss": 1.0409,
"step": 361
},
{
"epoch": 0.89,
"learning_rate": 1.6489928098427383e-05,
"loss": 1.1231,
"step": 362
},
{
"epoch": 0.89,
"learning_rate": 1.6469667170978258e-05,
"loss": 1.0062,
"step": 363
},
{
"epoch": 0.9,
"learning_rate": 1.6449360462913005e-05,
"loss": 1.0764,
"step": 364
},
{
"epoch": 0.9,
"learning_rate": 1.642900811792582e-05,
"loss": 1.1421,
"step": 365
},
{
"epoch": 0.9,
"learning_rate": 1.640861028003383e-05,
"loss": 1.0809,
"step": 366
},
{
"epoch": 0.9,
"learning_rate": 1.6388167093576083e-05,
"loss": 1.1295,
"step": 367
},
{
"epoch": 0.91,
"learning_rate": 1.6367678703212515e-05,
"loss": 1.0813,
"step": 368
},
{
"epoch": 0.91,
"learning_rate": 1.6347145253922942e-05,
"loss": 1.1542,
"step": 369
},
{
"epoch": 0.91,
"learning_rate": 1.632656689100602e-05,
"loss": 1.0917,
"step": 370
},
{
"epoch": 0.91,
"learning_rate": 1.6305943760078226e-05,
"loss": 1.1101,
"step": 371
},
{
"epoch": 0.92,
"learning_rate": 1.628527600707283e-05,
"loss": 1.0929,
"step": 372
},
{
"epoch": 0.92,
"learning_rate": 1.6264563778238834e-05,
"loss": 1.074,
"step": 373
},
{
"epoch": 0.92,
"learning_rate": 1.6243807220139988e-05,
"loss": 1.0869,
"step": 374
},
{
"epoch": 0.92,
"learning_rate": 1.6223006479653708e-05,
"loss": 1.0546,
"step": 375
},
{
"epoch": 0.93,
"learning_rate": 1.6202161703970057e-05,
"loss": 1.0218,
"step": 376
},
{
"epoch": 0.93,
"learning_rate": 1.6181273040590696e-05,
"loss": 1.054,
"step": 377
},
{
"epoch": 0.93,
"learning_rate": 1.616034063732785e-05,
"loss": 0.9993,
"step": 378
},
{
"epoch": 0.93,
"learning_rate": 1.613936464230325e-05,
"loss": 1.0481,
"step": 379
},
{
"epoch": 0.94,
"learning_rate": 1.6118345203947093e-05,
"loss": 1.1722,
"step": 380
},
{
"epoch": 0.94,
"learning_rate": 1.6097282470996997e-05,
"loss": 1.1137,
"step": 381
},
{
"epoch": 0.94,
"learning_rate": 1.6076176592496926e-05,
"loss": 1.0729,
"step": 382
},
{
"epoch": 0.94,
"learning_rate": 1.605502771779616e-05,
"loss": 1.0825,
"step": 383
},
{
"epoch": 0.94,
"learning_rate": 1.603383599654823e-05,
"loss": 1.0767,
"step": 384
},
{
"epoch": 0.95,
"learning_rate": 1.601260157870985e-05,
"loss": 1.1263,
"step": 385
},
{
"epoch": 0.95,
"learning_rate": 1.599132461453987e-05,
"loss": 1.0745,
"step": 386
},
{
"epoch": 0.95,
"learning_rate": 1.5970005254598204e-05,
"loss": 1.1024,
"step": 387
},
{
"epoch": 0.95,
"learning_rate": 1.594864364974476e-05,
"loss": 1.1268,
"step": 388
},
{
"epoch": 0.96,
"learning_rate": 1.592723995113839e-05,
"loss": 1.0998,
"step": 389
},
{
"epoch": 0.96,
"learning_rate": 1.5905794310235808e-05,
"loss": 1.0569,
"step": 390
},
{
"epoch": 0.96,
"learning_rate": 1.5884306878790512e-05,
"loss": 1.0528,
"step": 391
},
{
"epoch": 0.96,
"learning_rate": 1.586277780885172e-05,
"loss": 1.0959,
"step": 392
},
{
"epoch": 0.97,
"learning_rate": 1.58412072527633e-05,
"loss": 1.167,
"step": 393
},
{
"epoch": 0.97,
"learning_rate": 1.5819595363162682e-05,
"loss": 1.11,
"step": 394
},
{
"epoch": 0.97,
"learning_rate": 1.5797942292979767e-05,
"loss": 1.1125,
"step": 395
},
{
"epoch": 0.97,
"learning_rate": 1.577624819543587e-05,
"loss": 1.1226,
"step": 396
},
{
"epoch": 0.98,
"learning_rate": 1.5754513224042625e-05,
"loss": 1.134,
"step": 397
},
{
"epoch": 0.98,
"learning_rate": 1.573273753260089e-05,
"loss": 1.1455,
"step": 398
},
{
"epoch": 0.98,
"learning_rate": 1.571092127519967e-05,
"loss": 1.0762,
"step": 399
},
{
"epoch": 0.98,
"learning_rate": 1.568906460621502e-05,
"loss": 1.0632,
"step": 400
}
],
"max_steps": 1218,
"num_train_epochs": 3,
"total_flos": 9.670402784414925e+16,
"trial_name": null,
"trial_params": null
}