llama-vid-7b-full-224-long-video / trainer_state.json
YanweiLi's picture
Upload 8 files
924a0da
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"global_step": 1911,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 3.4482758620689656e-07,
"loss": 4.9648,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 6.896551724137931e-07,
"loss": 4.6289,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 1.0344827586206898e-06,
"loss": 5.3125,
"step": 3
},
{
"epoch": 0.0,
"learning_rate": 1.3793103448275862e-06,
"loss": 5.4297,
"step": 4
},
{
"epoch": 0.0,
"learning_rate": 1.724137931034483e-06,
"loss": 5.4922,
"step": 5
},
{
"epoch": 0.0,
"learning_rate": 2.0689655172413796e-06,
"loss": 4.0117,
"step": 6
},
{
"epoch": 0.0,
"learning_rate": 2.4137931034482762e-06,
"loss": 5.2227,
"step": 7
},
{
"epoch": 0.0,
"learning_rate": 2.7586206896551725e-06,
"loss": 5.0508,
"step": 8
},
{
"epoch": 0.0,
"learning_rate": 3.103448275862069e-06,
"loss": 5.0273,
"step": 9
},
{
"epoch": 0.01,
"learning_rate": 3.448275862068966e-06,
"loss": 4.6289,
"step": 10
},
{
"epoch": 0.01,
"learning_rate": 3.793103448275862e-06,
"loss": 4.5156,
"step": 11
},
{
"epoch": 0.01,
"learning_rate": 4.137931034482759e-06,
"loss": 4.5703,
"step": 12
},
{
"epoch": 0.01,
"learning_rate": 4.482758620689656e-06,
"loss": 3.4629,
"step": 13
},
{
"epoch": 0.01,
"learning_rate": 4.8275862068965525e-06,
"loss": 4.2754,
"step": 14
},
{
"epoch": 0.01,
"learning_rate": 5.172413793103449e-06,
"loss": 4.0742,
"step": 15
},
{
"epoch": 0.01,
"learning_rate": 5.517241379310345e-06,
"loss": 3.9727,
"step": 16
},
{
"epoch": 0.01,
"learning_rate": 5.862068965517242e-06,
"loss": 3.7148,
"step": 17
},
{
"epoch": 0.01,
"learning_rate": 6.206896551724138e-06,
"loss": 3.3047,
"step": 18
},
{
"epoch": 0.01,
"learning_rate": 6.551724137931035e-06,
"loss": 3.7109,
"step": 19
},
{
"epoch": 0.01,
"learning_rate": 6.896551724137932e-06,
"loss": 3.1035,
"step": 20
},
{
"epoch": 0.01,
"learning_rate": 7.241379310344828e-06,
"loss": 4.0449,
"step": 21
},
{
"epoch": 0.01,
"learning_rate": 7.586206896551724e-06,
"loss": 3.6719,
"step": 22
},
{
"epoch": 0.01,
"learning_rate": 7.93103448275862e-06,
"loss": 3.7344,
"step": 23
},
{
"epoch": 0.01,
"learning_rate": 8.275862068965518e-06,
"loss": 3.3105,
"step": 24
},
{
"epoch": 0.01,
"learning_rate": 8.620689655172414e-06,
"loss": 3.3125,
"step": 25
},
{
"epoch": 0.01,
"learning_rate": 8.965517241379312e-06,
"loss": 3.2598,
"step": 26
},
{
"epoch": 0.01,
"learning_rate": 9.310344827586207e-06,
"loss": 2.9922,
"step": 27
},
{
"epoch": 0.01,
"learning_rate": 9.655172413793105e-06,
"loss": 3.0723,
"step": 28
},
{
"epoch": 0.02,
"learning_rate": 1e-05,
"loss": 2.5176,
"step": 29
},
{
"epoch": 0.02,
"learning_rate": 1.0344827586206898e-05,
"loss": 1.772,
"step": 30
},
{
"epoch": 0.02,
"learning_rate": 1.0689655172413792e-05,
"loss": 2.7646,
"step": 31
},
{
"epoch": 0.02,
"learning_rate": 1.103448275862069e-05,
"loss": 2.3545,
"step": 32
},
{
"epoch": 0.02,
"learning_rate": 1.1379310344827587e-05,
"loss": 2.7559,
"step": 33
},
{
"epoch": 0.02,
"learning_rate": 1.1724137931034483e-05,
"loss": 2.6309,
"step": 34
},
{
"epoch": 0.02,
"learning_rate": 1.206896551724138e-05,
"loss": 2.2471,
"step": 35
},
{
"epoch": 0.02,
"learning_rate": 1.2413793103448277e-05,
"loss": 2.4277,
"step": 36
},
{
"epoch": 0.02,
"learning_rate": 1.2758620689655174e-05,
"loss": 2.7324,
"step": 37
},
{
"epoch": 0.02,
"learning_rate": 1.310344827586207e-05,
"loss": 2.5166,
"step": 38
},
{
"epoch": 0.02,
"learning_rate": 1.3448275862068967e-05,
"loss": 2.4258,
"step": 39
},
{
"epoch": 0.02,
"learning_rate": 1.3793103448275863e-05,
"loss": 2.1025,
"step": 40
},
{
"epoch": 0.02,
"learning_rate": 1.4137931034482759e-05,
"loss": 2.6289,
"step": 41
},
{
"epoch": 0.02,
"learning_rate": 1.4482758620689657e-05,
"loss": 2.3945,
"step": 42
},
{
"epoch": 0.02,
"learning_rate": 1.4827586206896554e-05,
"loss": 2.0024,
"step": 43
},
{
"epoch": 0.02,
"learning_rate": 1.5172413793103448e-05,
"loss": 2.0791,
"step": 44
},
{
"epoch": 0.02,
"learning_rate": 1.5517241379310346e-05,
"loss": 2.1133,
"step": 45
},
{
"epoch": 0.02,
"learning_rate": 1.586206896551724e-05,
"loss": 2.0801,
"step": 46
},
{
"epoch": 0.02,
"learning_rate": 1.6206896551724137e-05,
"loss": 2.251,
"step": 47
},
{
"epoch": 0.03,
"learning_rate": 1.6551724137931037e-05,
"loss": 1.9714,
"step": 48
},
{
"epoch": 0.03,
"learning_rate": 1.6896551724137932e-05,
"loss": 2.2617,
"step": 49
},
{
"epoch": 0.03,
"learning_rate": 1.7241379310344828e-05,
"loss": 1.8398,
"step": 50
},
{
"epoch": 0.03,
"learning_rate": 1.7586206896551724e-05,
"loss": 2.0005,
"step": 51
},
{
"epoch": 0.03,
"learning_rate": 1.7931034482758623e-05,
"loss": 1.9062,
"step": 52
},
{
"epoch": 0.03,
"learning_rate": 1.827586206896552e-05,
"loss": 1.9971,
"step": 53
},
{
"epoch": 0.03,
"learning_rate": 1.8620689655172415e-05,
"loss": 1.6963,
"step": 54
},
{
"epoch": 0.03,
"learning_rate": 1.896551724137931e-05,
"loss": 1.8438,
"step": 55
},
{
"epoch": 0.03,
"learning_rate": 1.931034482758621e-05,
"loss": 1.686,
"step": 56
},
{
"epoch": 0.03,
"learning_rate": 1.9655172413793106e-05,
"loss": 2.041,
"step": 57
},
{
"epoch": 0.03,
"learning_rate": 2e-05,
"loss": 1.7822,
"step": 58
},
{
"epoch": 0.03,
"learning_rate": 1.9999985627947104e-05,
"loss": 1.9697,
"step": 59
},
{
"epoch": 0.03,
"learning_rate": 1.9999942511829714e-05,
"loss": 1.8018,
"step": 60
},
{
"epoch": 0.03,
"learning_rate": 1.999987065177177e-05,
"loss": 1.8613,
"step": 61
},
{
"epoch": 0.03,
"learning_rate": 1.9999770047979828e-05,
"loss": 1.4146,
"step": 62
},
{
"epoch": 0.03,
"learning_rate": 1.9999640700743063e-05,
"loss": 1.6846,
"step": 63
},
{
"epoch": 0.03,
"learning_rate": 1.9999482610433275e-05,
"loss": 1.7246,
"step": 64
},
{
"epoch": 0.03,
"learning_rate": 1.999929577750487e-05,
"loss": 1.8691,
"step": 65
},
{
"epoch": 0.03,
"learning_rate": 1.99990802024949e-05,
"loss": 1.7725,
"step": 66
},
{
"epoch": 0.04,
"learning_rate": 1.9998835886022997e-05,
"loss": 1.5044,
"step": 67
},
{
"epoch": 0.04,
"learning_rate": 1.999856282879144e-05,
"loss": 1.3157,
"step": 68
},
{
"epoch": 0.04,
"learning_rate": 1.9998261031585104e-05,
"loss": 1.8428,
"step": 69
},
{
"epoch": 0.04,
"learning_rate": 1.999793049527148e-05,
"loss": 2.0723,
"step": 70
},
{
"epoch": 0.04,
"learning_rate": 1.999757122080066e-05,
"loss": 1.8516,
"step": 71
},
{
"epoch": 0.04,
"learning_rate": 1.999718320920535e-05,
"loss": 1.4434,
"step": 72
},
{
"epoch": 0.04,
"learning_rate": 1.9996766461600855e-05,
"loss": 1.2244,
"step": 73
},
{
"epoch": 0.04,
"learning_rate": 1.9996320979185074e-05,
"loss": 1.6006,
"step": 74
},
{
"epoch": 0.04,
"learning_rate": 1.9995846763238514e-05,
"loss": 1.8066,
"step": 75
},
{
"epoch": 0.04,
"learning_rate": 1.999534381512426e-05,
"loss": 1.9062,
"step": 76
},
{
"epoch": 0.04,
"learning_rate": 1.9994812136287994e-05,
"loss": 2.0957,
"step": 77
},
{
"epoch": 0.04,
"learning_rate": 1.9994251728257978e-05,
"loss": 1.7969,
"step": 78
},
{
"epoch": 0.04,
"learning_rate": 1.9993662592645057e-05,
"loss": 1.958,
"step": 79
},
{
"epoch": 0.04,
"learning_rate": 1.999304473114265e-05,
"loss": 1.4932,
"step": 80
},
{
"epoch": 0.04,
"learning_rate": 1.9992398145526738e-05,
"loss": 1.4139,
"step": 81
},
{
"epoch": 0.04,
"learning_rate": 1.999172283765588e-05,
"loss": 1.6318,
"step": 82
},
{
"epoch": 0.04,
"learning_rate": 1.999101880947118e-05,
"loss": 1.9033,
"step": 83
},
{
"epoch": 0.04,
"learning_rate": 1.9990286062996316e-05,
"loss": 1.7368,
"step": 84
},
{
"epoch": 0.04,
"learning_rate": 1.9989524600337493e-05,
"loss": 1.9033,
"step": 85
},
{
"epoch": 0.05,
"learning_rate": 1.998873442368347e-05,
"loss": 1.5342,
"step": 86
},
{
"epoch": 0.05,
"learning_rate": 1.998791553530554e-05,
"loss": 1.459,
"step": 87
},
{
"epoch": 0.05,
"learning_rate": 1.9987067937557527e-05,
"loss": 1.5562,
"step": 88
},
{
"epoch": 0.05,
"learning_rate": 1.9986191632875768e-05,
"loss": 1.3936,
"step": 89
},
{
"epoch": 0.05,
"learning_rate": 1.9985286623779126e-05,
"loss": 1.494,
"step": 90
},
{
"epoch": 0.05,
"learning_rate": 1.9984352912868973e-05,
"loss": 1.312,
"step": 91
},
{
"epoch": 0.05,
"learning_rate": 1.9983390502829168e-05,
"loss": 1.8467,
"step": 92
},
{
"epoch": 0.05,
"learning_rate": 1.998239939642608e-05,
"loss": 1.4397,
"step": 93
},
{
"epoch": 0.05,
"learning_rate": 1.998137959650855e-05,
"loss": 1.8779,
"step": 94
},
{
"epoch": 0.05,
"learning_rate": 1.9980331106007912e-05,
"loss": 1.7192,
"step": 95
},
{
"epoch": 0.05,
"learning_rate": 1.9979253927937948e-05,
"loss": 1.6748,
"step": 96
},
{
"epoch": 0.05,
"learning_rate": 1.9978148065394917e-05,
"loss": 1.5498,
"step": 97
},
{
"epoch": 0.05,
"learning_rate": 1.9977013521557516e-05,
"loss": 1.2637,
"step": 98
},
{
"epoch": 0.05,
"learning_rate": 1.9975850299686894e-05,
"loss": 1.1064,
"step": 99
},
{
"epoch": 0.05,
"learning_rate": 1.9974658403126624e-05,
"loss": 1.5728,
"step": 100
},
{
"epoch": 0.05,
"learning_rate": 1.997343783530271e-05,
"loss": 1.5781,
"step": 101
},
{
"epoch": 0.05,
"learning_rate": 1.9972188599723564e-05,
"loss": 1.7412,
"step": 102
},
{
"epoch": 0.05,
"learning_rate": 1.997091069998e-05,
"loss": 1.3691,
"step": 103
},
{
"epoch": 0.05,
"learning_rate": 1.9969604139745227e-05,
"loss": 1.4824,
"step": 104
},
{
"epoch": 0.05,
"learning_rate": 1.9968268922774842e-05,
"loss": 1.5947,
"step": 105
},
{
"epoch": 0.06,
"learning_rate": 1.99669050529068e-05,
"loss": 1.5049,
"step": 106
},
{
"epoch": 0.06,
"learning_rate": 1.9965512534061426e-05,
"loss": 1.8086,
"step": 107
},
{
"epoch": 0.06,
"learning_rate": 1.9964091370241384e-05,
"loss": 1.5146,
"step": 108
},
{
"epoch": 0.06,
"learning_rate": 1.9962641565531694e-05,
"loss": 1.6699,
"step": 109
},
{
"epoch": 0.06,
"learning_rate": 1.996116312409968e-05,
"loss": 1.4238,
"step": 110
},
{
"epoch": 0.06,
"learning_rate": 1.9959656050194994e-05,
"loss": 1.3311,
"step": 111
},
{
"epoch": 0.06,
"learning_rate": 1.9958120348149585e-05,
"loss": 1.5396,
"step": 112
},
{
"epoch": 0.06,
"learning_rate": 1.9956556022377693e-05,
"loss": 1.3779,
"step": 113
},
{
"epoch": 0.06,
"learning_rate": 1.995496307737583e-05,
"loss": 1.2122,
"step": 114
},
{
"epoch": 0.06,
"learning_rate": 1.9953341517722773e-05,
"loss": 1.8545,
"step": 115
},
{
"epoch": 0.06,
"learning_rate": 1.9951691348079553e-05,
"loss": 1.0951,
"step": 116
},
{
"epoch": 0.06,
"learning_rate": 1.995001257318943e-05,
"loss": 1.6533,
"step": 117
},
{
"epoch": 0.06,
"learning_rate": 1.99483051978779e-05,
"loss": 1.6465,
"step": 118
},
{
"epoch": 0.06,
"learning_rate": 1.994656922705266e-05,
"loss": 1.6045,
"step": 119
},
{
"epoch": 0.06,
"learning_rate": 1.9944804665703593e-05,
"loss": 1.6978,
"step": 120
},
{
"epoch": 0.06,
"learning_rate": 1.994301151890278e-05,
"loss": 1.6748,
"step": 121
},
{
"epoch": 0.06,
"learning_rate": 1.9941189791804466e-05,
"loss": 1.3828,
"step": 122
},
{
"epoch": 0.06,
"learning_rate": 1.9939339489645034e-05,
"loss": 1.3296,
"step": 123
},
{
"epoch": 0.06,
"learning_rate": 1.9937460617743013e-05,
"loss": 1.6924,
"step": 124
},
{
"epoch": 0.07,
"learning_rate": 1.993555318149906e-05,
"loss": 1.2607,
"step": 125
},
{
"epoch": 0.07,
"learning_rate": 1.9933617186395917e-05,
"loss": 1.7383,
"step": 126
},
{
"epoch": 0.07,
"learning_rate": 1.9931652637998443e-05,
"loss": 1.7686,
"step": 127
},
{
"epoch": 0.07,
"learning_rate": 1.9929659541953548e-05,
"loss": 1.8975,
"step": 128
},
{
"epoch": 0.07,
"learning_rate": 1.992763790399021e-05,
"loss": 1.8799,
"step": 129
},
{
"epoch": 0.07,
"learning_rate": 1.9925587729919447e-05,
"loss": 1.5991,
"step": 130
},
{
"epoch": 0.07,
"learning_rate": 1.99235090256343e-05,
"loss": 1.999,
"step": 131
},
{
"epoch": 0.07,
"learning_rate": 1.992140179710982e-05,
"loss": 1.5149,
"step": 132
},
{
"epoch": 0.07,
"learning_rate": 1.991926605040305e-05,
"loss": 1.7344,
"step": 133
},
{
"epoch": 0.07,
"learning_rate": 1.9917101791652997e-05,
"loss": 1.1921,
"step": 134
},
{
"epoch": 0.07,
"learning_rate": 1.991490902708063e-05,
"loss": 1.6406,
"step": 135
},
{
"epoch": 0.07,
"learning_rate": 1.9912687762988855e-05,
"loss": 1.7539,
"step": 136
},
{
"epoch": 0.07,
"learning_rate": 1.9910438005762503e-05,
"loss": 1.3281,
"step": 137
},
{
"epoch": 0.07,
"learning_rate": 1.9908159761868292e-05,
"loss": 1.6001,
"step": 138
},
{
"epoch": 0.07,
"learning_rate": 1.9905853037854834e-05,
"loss": 1.3193,
"step": 139
},
{
"epoch": 0.07,
"learning_rate": 1.9903517840352605e-05,
"loss": 1.2607,
"step": 140
},
{
"epoch": 0.07,
"learning_rate": 1.9901154176073914e-05,
"loss": 2.1016,
"step": 141
},
{
"epoch": 0.07,
"learning_rate": 1.9898762051812906e-05,
"loss": 1.7861,
"step": 142
},
{
"epoch": 0.07,
"learning_rate": 1.9896341474445526e-05,
"loss": 1.2358,
"step": 143
},
{
"epoch": 0.08,
"learning_rate": 1.989389245092951e-05,
"loss": 1.8379,
"step": 144
},
{
"epoch": 0.08,
"learning_rate": 1.9891414988304357e-05,
"loss": 1.6826,
"step": 145
},
{
"epoch": 0.08,
"learning_rate": 1.988890909369131e-05,
"loss": 1.3398,
"step": 146
},
{
"epoch": 0.08,
"learning_rate": 1.988637477429334e-05,
"loss": 1.6172,
"step": 147
},
{
"epoch": 0.08,
"learning_rate": 1.9883812037395126e-05,
"loss": 1.877,
"step": 148
},
{
"epoch": 0.08,
"learning_rate": 1.988122089036302e-05,
"loss": 1.4463,
"step": 149
},
{
"epoch": 0.08,
"learning_rate": 1.987860134064504e-05,
"loss": 1.0048,
"step": 150
},
{
"epoch": 0.08,
"learning_rate": 1.987595339577085e-05,
"loss": 1.7891,
"step": 151
},
{
"epoch": 0.08,
"learning_rate": 1.9873277063351737e-05,
"loss": 1.5967,
"step": 152
},
{
"epoch": 0.08,
"learning_rate": 1.9870572351080573e-05,
"loss": 1.7192,
"step": 153
},
{
"epoch": 0.08,
"learning_rate": 1.9867839266731813e-05,
"loss": 1.8652,
"step": 154
},
{
"epoch": 0.08,
"learning_rate": 1.9865077818161464e-05,
"loss": 1.4697,
"step": 155
},
{
"epoch": 0.08,
"learning_rate": 1.9862288013307062e-05,
"loss": 1.7266,
"step": 156
},
{
"epoch": 0.08,
"learning_rate": 1.9859469860187654e-05,
"loss": 1.3608,
"step": 157
},
{
"epoch": 0.08,
"learning_rate": 1.9856623366903765e-05,
"loss": 1.1792,
"step": 158
},
{
"epoch": 0.08,
"learning_rate": 1.9853748541637387e-05,
"loss": 1.2749,
"step": 159
},
{
"epoch": 0.08,
"learning_rate": 1.985084539265195e-05,
"loss": 1.1621,
"step": 160
},
{
"epoch": 0.08,
"learning_rate": 1.9847913928292297e-05,
"loss": 1.6816,
"step": 161
},
{
"epoch": 0.08,
"learning_rate": 1.9844954156984656e-05,
"loss": 1.6172,
"step": 162
},
{
"epoch": 0.09,
"learning_rate": 1.9841966087236626e-05,
"loss": 1.4736,
"step": 163
},
{
"epoch": 0.09,
"learning_rate": 1.983894972763715e-05,
"loss": 1.1748,
"step": 164
},
{
"epoch": 0.09,
"learning_rate": 1.983590508685648e-05,
"loss": 1.3047,
"step": 165
},
{
"epoch": 0.09,
"learning_rate": 1.9832832173646164e-05,
"loss": 1.8203,
"step": 166
},
{
"epoch": 0.09,
"learning_rate": 1.982973099683902e-05,
"loss": 1.2522,
"step": 167
},
{
"epoch": 0.09,
"learning_rate": 1.98266015653491e-05,
"loss": 1.0513,
"step": 168
},
{
"epoch": 0.09,
"learning_rate": 1.9823443888171676e-05,
"loss": 0.9141,
"step": 169
},
{
"epoch": 0.09,
"learning_rate": 1.9820257974383205e-05,
"loss": 1.2417,
"step": 170
},
{
"epoch": 0.09,
"learning_rate": 1.9817043833141317e-05,
"loss": 1.8027,
"step": 171
},
{
"epoch": 0.09,
"learning_rate": 1.981380147368477e-05,
"loss": 1.6753,
"step": 172
},
{
"epoch": 0.09,
"learning_rate": 1.9810530905333437e-05,
"loss": 1.6045,
"step": 173
},
{
"epoch": 0.09,
"learning_rate": 1.980723213748828e-05,
"loss": 1.9287,
"step": 174
},
{
"epoch": 0.09,
"learning_rate": 1.9803905179631303e-05,
"loss": 1.231,
"step": 175
},
{
"epoch": 0.09,
"learning_rate": 1.9800550041325553e-05,
"loss": 1.4736,
"step": 176
},
{
"epoch": 0.09,
"learning_rate": 1.9797166732215078e-05,
"loss": 1.4077,
"step": 177
},
{
"epoch": 0.09,
"learning_rate": 1.979375526202489e-05,
"loss": 1.7012,
"step": 178
},
{
"epoch": 0.09,
"learning_rate": 1.9790315640560962e-05,
"loss": 1.4067,
"step": 179
},
{
"epoch": 0.09,
"learning_rate": 1.9786847877710177e-05,
"loss": 1.0358,
"step": 180
},
{
"epoch": 0.09,
"learning_rate": 1.9783351983440306e-05,
"loss": 1.3394,
"step": 181
},
{
"epoch": 0.1,
"learning_rate": 1.9779827967799985e-05,
"loss": 1.7329,
"step": 182
},
{
"epoch": 0.1,
"learning_rate": 1.9776275840918685e-05,
"loss": 1.2595,
"step": 183
},
{
"epoch": 0.1,
"learning_rate": 1.9772695613006676e-05,
"loss": 1.1616,
"step": 184
},
{
"epoch": 0.1,
"learning_rate": 1.9769087294355006e-05,
"loss": 1.5562,
"step": 185
},
{
"epoch": 0.1,
"learning_rate": 1.9765450895335454e-05,
"loss": 1.3149,
"step": 186
},
{
"epoch": 0.1,
"learning_rate": 1.9761786426400532e-05,
"loss": 1.5107,
"step": 187
},
{
"epoch": 0.1,
"learning_rate": 1.975809389808343e-05,
"loss": 1.7129,
"step": 188
},
{
"epoch": 0.1,
"learning_rate": 1.9754373320997987e-05,
"loss": 1.5195,
"step": 189
},
{
"epoch": 0.1,
"learning_rate": 1.9750624705838666e-05,
"loss": 1.1064,
"step": 190
},
{
"epoch": 0.1,
"learning_rate": 1.974684806338053e-05,
"loss": 1.5518,
"step": 191
},
{
"epoch": 0.1,
"learning_rate": 1.9743043404479203e-05,
"loss": 1.394,
"step": 192
},
{
"epoch": 0.1,
"learning_rate": 1.9739210740070833e-05,
"loss": 1.686,
"step": 193
},
{
"epoch": 0.1,
"learning_rate": 1.973535008117207e-05,
"loss": 1.6709,
"step": 194
},
{
"epoch": 0.1,
"learning_rate": 1.9731461438880034e-05,
"loss": 1.4941,
"step": 195
},
{
"epoch": 0.1,
"learning_rate": 1.972754482437228e-05,
"loss": 1.9531,
"step": 196
},
{
"epoch": 0.1,
"learning_rate": 1.9723600248906766e-05,
"loss": 1.6992,
"step": 197
},
{
"epoch": 0.1,
"learning_rate": 1.9719627723821818e-05,
"loss": 1.6558,
"step": 198
},
{
"epoch": 0.1,
"learning_rate": 1.971562726053611e-05,
"loss": 1.7363,
"step": 199
},
{
"epoch": 0.1,
"learning_rate": 1.971159887054862e-05,
"loss": 1.6577,
"step": 200
},
{
"epoch": 0.11,
"learning_rate": 1.970754256543858e-05,
"loss": 1.4634,
"step": 201
},
{
"epoch": 0.11,
"learning_rate": 1.9703458356865484e-05,
"loss": 1.7441,
"step": 202
},
{
"epoch": 0.11,
"learning_rate": 1.9699346256569033e-05,
"loss": 1.2324,
"step": 203
},
{
"epoch": 0.11,
"learning_rate": 1.9695206276369077e-05,
"loss": 1.4375,
"step": 204
},
{
"epoch": 0.11,
"learning_rate": 1.969103842816563e-05,
"loss": 1.8916,
"step": 205
},
{
"epoch": 0.11,
"learning_rate": 1.968684272393879e-05,
"loss": 1.2065,
"step": 206
},
{
"epoch": 0.11,
"learning_rate": 1.968261917574874e-05,
"loss": 1.4702,
"step": 207
},
{
"epoch": 0.11,
"learning_rate": 1.967836779573569e-05,
"loss": 1.4023,
"step": 208
},
{
"epoch": 0.11,
"learning_rate": 1.9674088596119853e-05,
"loss": 1.4757,
"step": 209
},
{
"epoch": 0.11,
"learning_rate": 1.96697815892014e-05,
"loss": 1.1831,
"step": 210
},
{
"epoch": 0.11,
"learning_rate": 1.9665446787360444e-05,
"loss": 1.8867,
"step": 211
},
{
"epoch": 0.11,
"learning_rate": 1.9661084203056983e-05,
"loss": 1.5752,
"step": 212
},
{
"epoch": 0.11,
"learning_rate": 1.965669384883087e-05,
"loss": 1.501,
"step": 213
},
{
"epoch": 0.11,
"learning_rate": 1.9652275737301797e-05,
"loss": 1.4746,
"step": 214
},
{
"epoch": 0.11,
"learning_rate": 1.9647829881169223e-05,
"loss": 1.5176,
"step": 215
},
{
"epoch": 0.11,
"learning_rate": 1.9643356293212363e-05,
"loss": 1.9346,
"step": 216
},
{
"epoch": 0.11,
"learning_rate": 1.963885498629015e-05,
"loss": 1.9209,
"step": 217
},
{
"epoch": 0.11,
"learning_rate": 1.9634325973341183e-05,
"loss": 1.1978,
"step": 218
},
{
"epoch": 0.11,
"learning_rate": 1.962976926738371e-05,
"loss": 1.6904,
"step": 219
},
{
"epoch": 0.12,
"learning_rate": 1.962518488151557e-05,
"loss": 1.7881,
"step": 220
},
{
"epoch": 0.12,
"learning_rate": 1.9620572828914177e-05,
"loss": 1.542,
"step": 221
},
{
"epoch": 0.12,
"learning_rate": 1.9615933122836454e-05,
"loss": 1.248,
"step": 222
},
{
"epoch": 0.12,
"learning_rate": 1.961126577661883e-05,
"loss": 1.072,
"step": 223
},
{
"epoch": 0.12,
"learning_rate": 1.960657080367717e-05,
"loss": 1.791,
"step": 224
},
{
"epoch": 0.12,
"learning_rate": 1.9601848217506755e-05,
"loss": 1.5854,
"step": 225
},
{
"epoch": 0.12,
"learning_rate": 1.9597098031682238e-05,
"loss": 1.4705,
"step": 226
},
{
"epoch": 0.12,
"learning_rate": 1.9592320259857603e-05,
"loss": 1.6816,
"step": 227
},
{
"epoch": 0.12,
"learning_rate": 1.9587514915766124e-05,
"loss": 1.9238,
"step": 228
},
{
"epoch": 0.12,
"learning_rate": 1.9582682013220337e-05,
"loss": 1.6934,
"step": 229
},
{
"epoch": 0.12,
"learning_rate": 1.957782156611199e-05,
"loss": 1.2354,
"step": 230
},
{
"epoch": 0.12,
"learning_rate": 1.9572933588411997e-05,
"loss": 1.7544,
"step": 231
},
{
"epoch": 0.12,
"learning_rate": 1.9568018094170415e-05,
"loss": 1.7842,
"step": 232
},
{
"epoch": 0.12,
"learning_rate": 1.9563075097516397e-05,
"loss": 1.0984,
"step": 233
},
{
"epoch": 0.12,
"learning_rate": 1.9558104612658137e-05,
"loss": 1.5488,
"step": 234
},
{
"epoch": 0.12,
"learning_rate": 1.955310665388286e-05,
"loss": 1.9023,
"step": 235
},
{
"epoch": 0.12,
"learning_rate": 1.954808123555674e-05,
"loss": 1.7031,
"step": 236
},
{
"epoch": 0.12,
"learning_rate": 1.9543028372124903e-05,
"loss": 1.6113,
"step": 237
},
{
"epoch": 0.12,
"learning_rate": 1.9537948078111348e-05,
"loss": 1.791,
"step": 238
},
{
"epoch": 0.13,
"learning_rate": 1.953284036811893e-05,
"loss": 1.6338,
"step": 239
},
{
"epoch": 0.13,
"learning_rate": 1.9527705256829293e-05,
"loss": 1.5161,
"step": 240
},
{
"epoch": 0.13,
"learning_rate": 1.9522542759002866e-05,
"loss": 1.5859,
"step": 241
},
{
"epoch": 0.13,
"learning_rate": 1.9517352889478787e-05,
"loss": 1.8755,
"step": 242
},
{
"epoch": 0.13,
"learning_rate": 1.951213566317487e-05,
"loss": 1.7578,
"step": 243
},
{
"epoch": 0.13,
"learning_rate": 1.9506891095087563e-05,
"loss": 1.0671,
"step": 244
},
{
"epoch": 0.13,
"learning_rate": 1.950161920029191e-05,
"loss": 1.1311,
"step": 245
},
{
"epoch": 0.13,
"learning_rate": 1.9496319993941498e-05,
"loss": 1.269,
"step": 246
},
{
"epoch": 0.13,
"learning_rate": 1.949099349126843e-05,
"loss": 1.2642,
"step": 247
},
{
"epoch": 0.13,
"learning_rate": 1.9485639707583255e-05,
"loss": 1.2371,
"step": 248
},
{
"epoch": 0.13,
"learning_rate": 1.9480258658274947e-05,
"loss": 1.6943,
"step": 249
},
{
"epoch": 0.13,
"learning_rate": 1.947485035881085e-05,
"loss": 1.7363,
"step": 250
},
{
"epoch": 0.13,
"learning_rate": 1.946941482473664e-05,
"loss": 1.9277,
"step": 251
},
{
"epoch": 0.13,
"learning_rate": 1.9463952071676268e-05,
"loss": 1.6338,
"step": 252
},
{
"epoch": 0.13,
"learning_rate": 1.9458462115331938e-05,
"loss": 1.4199,
"step": 253
},
{
"epoch": 0.13,
"learning_rate": 1.945294497148403e-05,
"loss": 1.7793,
"step": 254
},
{
"epoch": 0.13,
"learning_rate": 1.9447400655991086e-05,
"loss": 1.4854,
"step": 255
},
{
"epoch": 0.13,
"learning_rate": 1.9441829184789745e-05,
"loss": 1.2202,
"step": 256
},
{
"epoch": 0.13,
"learning_rate": 1.94362305738947e-05,
"loss": 1.3418,
"step": 257
},
{
"epoch": 0.14,
"learning_rate": 1.9430604839398657e-05,
"loss": 1.3691,
"step": 258
},
{
"epoch": 0.14,
"learning_rate": 1.942495199747229e-05,
"loss": 1.5259,
"step": 259
},
{
"epoch": 0.14,
"learning_rate": 1.9419272064364186e-05,
"loss": 1.7598,
"step": 260
},
{
"epoch": 0.14,
"learning_rate": 1.9413565056400804e-05,
"loss": 1.5083,
"step": 261
},
{
"epoch": 0.14,
"learning_rate": 1.940783098998643e-05,
"loss": 1.5337,
"step": 262
},
{
"epoch": 0.14,
"learning_rate": 1.9402069881603123e-05,
"loss": 1.29,
"step": 263
},
{
"epoch": 0.14,
"learning_rate": 1.939628174781068e-05,
"loss": 1.6855,
"step": 264
},
{
"epoch": 0.14,
"learning_rate": 1.939046660524656e-05,
"loss": 1.3389,
"step": 265
},
{
"epoch": 0.14,
"learning_rate": 1.9384624470625884e-05,
"loss": 1.4673,
"step": 266
},
{
"epoch": 0.14,
"learning_rate": 1.9378755360741342e-05,
"loss": 1.459,
"step": 267
},
{
"epoch": 0.14,
"learning_rate": 1.9372859292463163e-05,
"loss": 1.3325,
"step": 268
},
{
"epoch": 0.14,
"learning_rate": 1.9366936282739067e-05,
"loss": 1.29,
"step": 269
},
{
"epoch": 0.14,
"learning_rate": 1.9360986348594226e-05,
"loss": 1.7007,
"step": 270
},
{
"epoch": 0.14,
"learning_rate": 1.935500950713118e-05,
"loss": 1.6743,
"step": 271
},
{
"epoch": 0.14,
"learning_rate": 1.934900577552983e-05,
"loss": 1.2891,
"step": 272
},
{
"epoch": 0.14,
"learning_rate": 1.934297517104737e-05,
"loss": 1.8789,
"step": 273
},
{
"epoch": 0.14,
"learning_rate": 1.9336917711018227e-05,
"loss": 1.5244,
"step": 274
},
{
"epoch": 0.14,
"learning_rate": 1.9330833412854035e-05,
"loss": 1.4014,
"step": 275
},
{
"epoch": 0.14,
"learning_rate": 1.932472229404356e-05,
"loss": 1.1035,
"step": 276
},
{
"epoch": 0.14,
"learning_rate": 1.9318584372152668e-05,
"loss": 1.1802,
"step": 277
},
{
"epoch": 0.15,
"learning_rate": 1.9312419664824268e-05,
"loss": 1.4917,
"step": 278
},
{
"epoch": 0.15,
"learning_rate": 1.9306228189778255e-05,
"loss": 1.3477,
"step": 279
},
{
"epoch": 0.15,
"learning_rate": 1.9300009964811475e-05,
"loss": 1.7637,
"step": 280
},
{
"epoch": 0.15,
"learning_rate": 1.9293765007797662e-05,
"loss": 1.4507,
"step": 281
},
{
"epoch": 0.15,
"learning_rate": 1.928749333668738e-05,
"loss": 1.1709,
"step": 282
},
{
"epoch": 0.15,
"learning_rate": 1.928119496950799e-05,
"loss": 1.7305,
"step": 283
},
{
"epoch": 0.15,
"learning_rate": 1.9274869924363585e-05,
"loss": 1.3074,
"step": 284
},
{
"epoch": 0.15,
"learning_rate": 1.9268518219434943e-05,
"loss": 1.1116,
"step": 285
},
{
"epoch": 0.15,
"learning_rate": 1.9262139872979474e-05,
"loss": 1.8936,
"step": 286
},
{
"epoch": 0.15,
"learning_rate": 1.9255734903331157e-05,
"loss": 1.5659,
"step": 287
},
{
"epoch": 0.15,
"learning_rate": 1.9249303328900505e-05,
"loss": 1.3179,
"step": 288
},
{
"epoch": 0.15,
"learning_rate": 1.9242845168174514e-05,
"loss": 1.4568,
"step": 289
},
{
"epoch": 0.15,
"learning_rate": 1.923636043971658e-05,
"loss": 1.7562,
"step": 290
},
{
"epoch": 0.15,
"learning_rate": 1.9229849162166478e-05,
"loss": 1.6768,
"step": 291
},
{
"epoch": 0.15,
"learning_rate": 1.922331135424029e-05,
"loss": 1.2554,
"step": 292
},
{
"epoch": 0.15,
"learning_rate": 1.921674703473037e-05,
"loss": 1.8506,
"step": 293
},
{
"epoch": 0.15,
"learning_rate": 1.921015622250526e-05,
"loss": 1.6934,
"step": 294
},
{
"epoch": 0.15,
"learning_rate": 1.920353893650966e-05,
"loss": 1.7227,
"step": 295
},
{
"epoch": 0.15,
"learning_rate": 1.9196895195764363e-05,
"loss": 1.0075,
"step": 296
},
{
"epoch": 0.16,
"learning_rate": 1.919022501936622e-05,
"loss": 1.4814,
"step": 297
},
{
"epoch": 0.16,
"learning_rate": 1.918352842648804e-05,
"loss": 1.6152,
"step": 298
},
{
"epoch": 0.16,
"learning_rate": 1.9176805436378597e-05,
"loss": 2.1514,
"step": 299
},
{
"epoch": 0.16,
"learning_rate": 1.9170056068362516e-05,
"loss": 1.5288,
"step": 300
},
{
"epoch": 0.16,
"learning_rate": 1.9163280341840253e-05,
"loss": 1.666,
"step": 301
},
{
"epoch": 0.16,
"learning_rate": 1.9156478276288027e-05,
"loss": 1.6278,
"step": 302
},
{
"epoch": 0.16,
"learning_rate": 1.9149649891257772e-05,
"loss": 1.6504,
"step": 303
},
{
"epoch": 0.16,
"learning_rate": 1.9142795206377066e-05,
"loss": 1.5586,
"step": 304
},
{
"epoch": 0.16,
"learning_rate": 1.9135914241349088e-05,
"loss": 1.5718,
"step": 305
},
{
"epoch": 0.16,
"learning_rate": 1.9129007015952556e-05,
"loss": 1.4395,
"step": 306
},
{
"epoch": 0.16,
"learning_rate": 1.9122073550041675e-05,
"loss": 1.8735,
"step": 307
},
{
"epoch": 0.16,
"learning_rate": 1.911511386354607e-05,
"loss": 1.6025,
"step": 308
},
{
"epoch": 0.16,
"learning_rate": 1.9108127976470737e-05,
"loss": 1.4951,
"step": 309
},
{
"epoch": 0.16,
"learning_rate": 1.9101115908895985e-05,
"loss": 1.0557,
"step": 310
},
{
"epoch": 0.16,
"learning_rate": 1.9094077680977377e-05,
"loss": 1.8306,
"step": 311
},
{
"epoch": 0.16,
"learning_rate": 1.9087013312945668e-05,
"loss": 1.1816,
"step": 312
},
{
"epoch": 0.16,
"learning_rate": 1.907992282510675e-05,
"loss": 1.3887,
"step": 313
},
{
"epoch": 0.16,
"learning_rate": 1.90728062378416e-05,
"loss": 1.3164,
"step": 314
},
{
"epoch": 0.16,
"learning_rate": 1.906566357160621e-05,
"loss": 1.123,
"step": 315
},
{
"epoch": 0.17,
"learning_rate": 1.9058494846931538e-05,
"loss": 1.5557,
"step": 316
},
{
"epoch": 0.17,
"learning_rate": 1.9051300084423433e-05,
"loss": 1.1377,
"step": 317
},
{
"epoch": 0.17,
"learning_rate": 1.904407930476261e-05,
"loss": 1.5459,
"step": 318
},
{
"epoch": 0.17,
"learning_rate": 1.9036832528704547e-05,
"loss": 1.3032,
"step": 319
},
{
"epoch": 0.17,
"learning_rate": 1.902955977707945e-05,
"loss": 1.665,
"step": 320
},
{
"epoch": 0.17,
"learning_rate": 1.9022261070792197e-05,
"loss": 1.6816,
"step": 321
},
{
"epoch": 0.17,
"learning_rate": 1.901493643082227e-05,
"loss": 1.8926,
"step": 322
},
{
"epoch": 0.17,
"learning_rate": 1.900758587822369e-05,
"loss": 1.751,
"step": 323
},
{
"epoch": 0.17,
"learning_rate": 1.900020943412496e-05,
"loss": 1.3843,
"step": 324
},
{
"epoch": 0.17,
"learning_rate": 1.8992807119729013e-05,
"loss": 1.8936,
"step": 325
},
{
"epoch": 0.17,
"learning_rate": 1.8985378956313133e-05,
"loss": 1.709,
"step": 326
},
{
"epoch": 0.17,
"learning_rate": 1.8977924965228923e-05,
"loss": 1.4668,
"step": 327
},
{
"epoch": 0.17,
"learning_rate": 1.89704451679022e-05,
"loss": 1.4849,
"step": 328
},
{
"epoch": 0.17,
"learning_rate": 1.8962939585832985e-05,
"loss": 1.5342,
"step": 329
},
{
"epoch": 0.17,
"learning_rate": 1.8955408240595396e-05,
"loss": 1.2207,
"step": 330
},
{
"epoch": 0.17,
"learning_rate": 1.894785115383761e-05,
"loss": 1.7241,
"step": 331
},
{
"epoch": 0.17,
"learning_rate": 1.8940268347281802e-05,
"loss": 1.8047,
"step": 332
},
{
"epoch": 0.17,
"learning_rate": 1.8932659842724067e-05,
"loss": 1.4448,
"step": 333
},
{
"epoch": 0.17,
"learning_rate": 1.8925025662034376e-05,
"loss": 1.6621,
"step": 334
},
{
"epoch": 0.18,
"learning_rate": 1.8917365827156493e-05,
"loss": 1.627,
"step": 335
},
{
"epoch": 0.18,
"learning_rate": 1.890968036010793e-05,
"loss": 1.5327,
"step": 336
},
{
"epoch": 0.18,
"learning_rate": 1.8901969282979882e-05,
"loss": 1.6558,
"step": 337
},
{
"epoch": 0.18,
"learning_rate": 1.889423261793714e-05,
"loss": 1.3013,
"step": 338
},
{
"epoch": 0.18,
"learning_rate": 1.888647038721806e-05,
"loss": 1.7744,
"step": 339
},
{
"epoch": 0.18,
"learning_rate": 1.8878682613134486e-05,
"loss": 1.4575,
"step": 340
},
{
"epoch": 0.18,
"learning_rate": 1.8870869318071667e-05,
"loss": 1.1382,
"step": 341
},
{
"epoch": 0.18,
"learning_rate": 1.886303052448823e-05,
"loss": 1.3679,
"step": 342
},
{
"epoch": 0.18,
"learning_rate": 1.8855166254916082e-05,
"loss": 1.3848,
"step": 343
},
{
"epoch": 0.18,
"learning_rate": 1.8847276531960363e-05,
"loss": 1.6313,
"step": 344
},
{
"epoch": 0.18,
"learning_rate": 1.8839361378299376e-05,
"loss": 1.6836,
"step": 345
},
{
"epoch": 0.18,
"learning_rate": 1.8831420816684528e-05,
"loss": 1.7178,
"step": 346
},
{
"epoch": 0.18,
"learning_rate": 1.8823454869940243e-05,
"loss": 1.8096,
"step": 347
},
{
"epoch": 0.18,
"learning_rate": 1.8815463560963932e-05,
"loss": 1.8037,
"step": 348
},
{
"epoch": 0.18,
"learning_rate": 1.8807446912725893e-05,
"loss": 1.6787,
"step": 349
},
{
"epoch": 0.18,
"learning_rate": 1.8799404948269267e-05,
"loss": 1.4805,
"step": 350
},
{
"epoch": 0.18,
"learning_rate": 1.879133769070996e-05,
"loss": 1.5674,
"step": 351
},
{
"epoch": 0.18,
"learning_rate": 1.878324516323658e-05,
"loss": 1.9429,
"step": 352
},
{
"epoch": 0.18,
"learning_rate": 1.877512738911038e-05,
"loss": 1.6042,
"step": 353
},
{
"epoch": 0.19,
"learning_rate": 1.8766984391665173e-05,
"loss": 1.5635,
"step": 354
},
{
"epoch": 0.19,
"learning_rate": 1.8758816194307272e-05,
"loss": 1.3354,
"step": 355
},
{
"epoch": 0.19,
"learning_rate": 1.875062282051544e-05,
"loss": 1.7881,
"step": 356
},
{
"epoch": 0.19,
"learning_rate": 1.8742404293840784e-05,
"loss": 0.9346,
"step": 357
},
{
"epoch": 0.19,
"learning_rate": 1.8734160637906738e-05,
"loss": 1.7529,
"step": 358
},
{
"epoch": 0.19,
"learning_rate": 1.8725891876408946e-05,
"loss": 1.5811,
"step": 359
},
{
"epoch": 0.19,
"learning_rate": 1.8717598033115224e-05,
"loss": 1.1401,
"step": 360
},
{
"epoch": 0.19,
"learning_rate": 1.8709279131865486e-05,
"loss": 1.3574,
"step": 361
},
{
"epoch": 0.19,
"learning_rate": 1.8700935196571666e-05,
"loss": 1.2129,
"step": 362
},
{
"epoch": 0.19,
"learning_rate": 1.8692566251217665e-05,
"loss": 1.8516,
"step": 363
},
{
"epoch": 0.19,
"learning_rate": 1.8684172319859258e-05,
"loss": 1.5688,
"step": 364
},
{
"epoch": 0.19,
"learning_rate": 1.867575342662406e-05,
"loss": 1.5005,
"step": 365
},
{
"epoch": 0.19,
"learning_rate": 1.8667309595711426e-05,
"loss": 1.5867,
"step": 366
},
{
"epoch": 0.19,
"learning_rate": 1.8658840851392386e-05,
"loss": 1.1108,
"step": 367
},
{
"epoch": 0.19,
"learning_rate": 1.8650347218009596e-05,
"loss": 1.0679,
"step": 368
},
{
"epoch": 0.19,
"learning_rate": 1.8641828719977244e-05,
"loss": 0.9811,
"step": 369
},
{
"epoch": 0.19,
"learning_rate": 1.8633285381780987e-05,
"loss": 1.2744,
"step": 370
},
{
"epoch": 0.19,
"learning_rate": 1.862471722797789e-05,
"loss": 1.6221,
"step": 371
},
{
"epoch": 0.19,
"learning_rate": 1.861612428319634e-05,
"loss": 1.5977,
"step": 372
},
{
"epoch": 0.2,
"learning_rate": 1.8607506572135994e-05,
"loss": 1.332,
"step": 373
},
{
"epoch": 0.2,
"learning_rate": 1.8598864119567693e-05,
"loss": 1.2905,
"step": 374
},
{
"epoch": 0.2,
"learning_rate": 1.859019695033339e-05,
"loss": 1.1235,
"step": 375
},
{
"epoch": 0.2,
"learning_rate": 1.858150508934609e-05,
"loss": 1.1279,
"step": 376
},
{
"epoch": 0.2,
"learning_rate": 1.8572788561589766e-05,
"loss": 1.1641,
"step": 377
},
{
"epoch": 0.2,
"learning_rate": 1.8564047392119306e-05,
"loss": 1.4949,
"step": 378
},
{
"epoch": 0.2,
"learning_rate": 1.855528160606041e-05,
"loss": 1.4187,
"step": 379
},
{
"epoch": 0.2,
"learning_rate": 1.8546491228609552e-05,
"loss": 1.2559,
"step": 380
},
{
"epoch": 0.2,
"learning_rate": 1.8537676285033886e-05,
"loss": 1.6514,
"step": 381
},
{
"epoch": 0.2,
"learning_rate": 1.852883680067118e-05,
"loss": 1.7793,
"step": 382
},
{
"epoch": 0.2,
"learning_rate": 1.851997280092974e-05,
"loss": 1.459,
"step": 383
},
{
"epoch": 0.2,
"learning_rate": 1.8511084311288338e-05,
"loss": 1.6787,
"step": 384
},
{
"epoch": 0.2,
"learning_rate": 1.8502171357296144e-05,
"loss": 1.5303,
"step": 385
},
{
"epoch": 0.2,
"learning_rate": 1.8493233964572648e-05,
"loss": 1.7461,
"step": 386
},
{
"epoch": 0.2,
"learning_rate": 1.8484272158807587e-05,
"loss": 1.5869,
"step": 387
},
{
"epoch": 0.2,
"learning_rate": 1.847528596576087e-05,
"loss": 1.7012,
"step": 388
},
{
"epoch": 0.2,
"learning_rate": 1.8466275411262504e-05,
"loss": 1.6045,
"step": 389
},
{
"epoch": 0.2,
"learning_rate": 1.8457240521212525e-05,
"loss": 1.6836,
"step": 390
},
{
"epoch": 0.2,
"learning_rate": 1.8448181321580916e-05,
"loss": 1.2324,
"step": 391
},
{
"epoch": 0.21,
"learning_rate": 1.8439097838407532e-05,
"loss": 1.6055,
"step": 392
},
{
"epoch": 0.21,
"learning_rate": 1.8429990097802037e-05,
"loss": 0.9858,
"step": 393
},
{
"epoch": 0.21,
"learning_rate": 1.8420858125943818e-05,
"loss": 1.3906,
"step": 394
},
{
"epoch": 0.21,
"learning_rate": 1.8411701949081906e-05,
"loss": 1.6992,
"step": 395
},
{
"epoch": 0.21,
"learning_rate": 1.840252159353492e-05,
"loss": 1.3896,
"step": 396
},
{
"epoch": 0.21,
"learning_rate": 1.8393317085690964e-05,
"loss": 1.6509,
"step": 397
},
{
"epoch": 0.21,
"learning_rate": 1.838408845200758e-05,
"loss": 1.6553,
"step": 398
},
{
"epoch": 0.21,
"learning_rate": 1.837483571901164e-05,
"loss": 1.6118,
"step": 399
},
{
"epoch": 0.21,
"learning_rate": 1.8365558913299307e-05,
"loss": 1.6045,
"step": 400
},
{
"epoch": 0.21,
"learning_rate": 1.835625806153593e-05,
"loss": 1.168,
"step": 401
},
{
"epoch": 0.21,
"learning_rate": 1.834693319045597e-05,
"loss": 1.4414,
"step": 402
},
{
"epoch": 0.21,
"learning_rate": 1.8337584326862935e-05,
"loss": 1.0396,
"step": 403
},
{
"epoch": 0.21,
"learning_rate": 1.8328211497629304e-05,
"loss": 1.666,
"step": 404
},
{
"epoch": 0.21,
"learning_rate": 1.8318814729696427e-05,
"loss": 1.292,
"step": 405
},
{
"epoch": 0.21,
"learning_rate": 1.8309394050074482e-05,
"loss": 1.2471,
"step": 406
},
{
"epoch": 0.21,
"learning_rate": 1.8299949485842366e-05,
"loss": 1.5879,
"step": 407
},
{
"epoch": 0.21,
"learning_rate": 1.8290481064147637e-05,
"loss": 1.709,
"step": 408
},
{
"epoch": 0.21,
"learning_rate": 1.828098881220642e-05,
"loss": 1.6855,
"step": 409
},
{
"epoch": 0.21,
"learning_rate": 1.8271472757303352e-05,
"loss": 1.3252,
"step": 410
},
{
"epoch": 0.22,
"learning_rate": 1.8261932926791477e-05,
"loss": 1.6582,
"step": 411
},
{
"epoch": 0.22,
"learning_rate": 1.8252369348092184e-05,
"loss": 1.5156,
"step": 412
},
{
"epoch": 0.22,
"learning_rate": 1.824278204869513e-05,
"loss": 1.5378,
"step": 413
},
{
"epoch": 0.22,
"learning_rate": 1.823317105615815e-05,
"loss": 0.9766,
"step": 414
},
{
"epoch": 0.22,
"learning_rate": 1.8223536398107177e-05,
"loss": 1.1362,
"step": 415
},
{
"epoch": 0.22,
"learning_rate": 1.8213878102236177e-05,
"loss": 1.4385,
"step": 416
},
{
"epoch": 0.22,
"learning_rate": 1.8204196196307058e-05,
"loss": 1.6162,
"step": 417
},
{
"epoch": 0.22,
"learning_rate": 1.819449070814959e-05,
"loss": 1.4419,
"step": 418
},
{
"epoch": 0.22,
"learning_rate": 1.8184761665661335e-05,
"loss": 1.5757,
"step": 419
},
{
"epoch": 0.22,
"learning_rate": 1.8175009096807557e-05,
"loss": 1.5146,
"step": 420
},
{
"epoch": 0.22,
"learning_rate": 1.8165233029621136e-05,
"loss": 1.5449,
"step": 421
},
{
"epoch": 0.22,
"learning_rate": 1.8155433492202507e-05,
"loss": 1.6187,
"step": 422
},
{
"epoch": 0.22,
"learning_rate": 1.8145610512719565e-05,
"loss": 1.2803,
"step": 423
},
{
"epoch": 0.22,
"learning_rate": 1.8135764119407585e-05,
"loss": 1.3052,
"step": 424
},
{
"epoch": 0.22,
"learning_rate": 1.8125894340569145e-05,
"loss": 1.6846,
"step": 425
},
{
"epoch": 0.22,
"learning_rate": 1.8116001204574042e-05,
"loss": 1.8115,
"step": 426
},
{
"epoch": 0.22,
"learning_rate": 1.8106084739859206e-05,
"loss": 1.4536,
"step": 427
},
{
"epoch": 0.22,
"learning_rate": 1.8096144974928635e-05,
"loss": 1.7686,
"step": 428
},
{
"epoch": 0.22,
"learning_rate": 1.808618193835329e-05,
"loss": 1.7617,
"step": 429
},
{
"epoch": 0.23,
"learning_rate": 1.8076195658771033e-05,
"loss": 1.1416,
"step": 430
},
{
"epoch": 0.23,
"learning_rate": 1.8066186164886525e-05,
"loss": 1.0762,
"step": 431
},
{
"epoch": 0.23,
"learning_rate": 1.8056153485471167e-05,
"loss": 1.4985,
"step": 432
},
{
"epoch": 0.23,
"learning_rate": 1.8046097649362994e-05,
"loss": 1.449,
"step": 433
},
{
"epoch": 0.23,
"learning_rate": 1.8036018685466614e-05,
"loss": 1.6416,
"step": 434
},
{
"epoch": 0.23,
"learning_rate": 1.80259166227531e-05,
"loss": 1.7617,
"step": 435
},
{
"epoch": 0.23,
"learning_rate": 1.8015791490259932e-05,
"loss": 1.8594,
"step": 436
},
{
"epoch": 0.23,
"learning_rate": 1.80056433170909e-05,
"loss": 1.1938,
"step": 437
},
{
"epoch": 0.23,
"learning_rate": 1.799547213241602e-05,
"loss": 1.0913,
"step": 438
},
{
"epoch": 0.23,
"learning_rate": 1.798527796547145e-05,
"loss": 1.229,
"step": 439
},
{
"epoch": 0.23,
"learning_rate": 1.7975060845559412e-05,
"loss": 1.3936,
"step": 440
},
{
"epoch": 0.23,
"learning_rate": 1.79648208020481e-05,
"loss": 1.4009,
"step": 441
},
{
"epoch": 0.23,
"learning_rate": 1.7954557864371615e-05,
"loss": 1.7529,
"step": 442
},
{
"epoch": 0.23,
"learning_rate": 1.794427206202984e-05,
"loss": 1.6252,
"step": 443
},
{
"epoch": 0.23,
"learning_rate": 1.79339634245884e-05,
"loss": 1.7422,
"step": 444
},
{
"epoch": 0.23,
"learning_rate": 1.7923631981678552e-05,
"loss": 1.7744,
"step": 445
},
{
"epoch": 0.23,
"learning_rate": 1.7913277762997105e-05,
"loss": 1.6572,
"step": 446
},
{
"epoch": 0.23,
"learning_rate": 1.7902900798306334e-05,
"loss": 1.7285,
"step": 447
},
{
"epoch": 0.23,
"learning_rate": 1.7892501117433896e-05,
"loss": 1.4902,
"step": 448
},
{
"epoch": 0.23,
"learning_rate": 1.788207875027274e-05,
"loss": 1.8623,
"step": 449
},
{
"epoch": 0.24,
"learning_rate": 1.7871633726781042e-05,
"loss": 0.8735,
"step": 450
},
{
"epoch": 0.24,
"learning_rate": 1.786116607698207e-05,
"loss": 1.8496,
"step": 451
},
{
"epoch": 0.24,
"learning_rate": 1.785067583096416e-05,
"loss": 1.4072,
"step": 452
},
{
"epoch": 0.24,
"learning_rate": 1.784016301888058e-05,
"loss": 1.4888,
"step": 453
},
{
"epoch": 0.24,
"learning_rate": 1.7829627670949468e-05,
"loss": 1.4424,
"step": 454
},
{
"epoch": 0.24,
"learning_rate": 1.7819069817453746e-05,
"loss": 1.8037,
"step": 455
},
{
"epoch": 0.24,
"learning_rate": 1.7808489488741013e-05,
"loss": 1.4629,
"step": 456
},
{
"epoch": 0.24,
"learning_rate": 1.779788671522348e-05,
"loss": 1.5479,
"step": 457
},
{
"epoch": 0.24,
"learning_rate": 1.778726152737787e-05,
"loss": 1.1143,
"step": 458
},
{
"epoch": 0.24,
"learning_rate": 1.7776613955745337e-05,
"loss": 1.5515,
"step": 459
},
{
"epoch": 0.24,
"learning_rate": 1.776594403093138e-05,
"loss": 1.6328,
"step": 460
},
{
"epoch": 0.24,
"learning_rate": 1.7755251783605733e-05,
"loss": 1.9775,
"step": 461
},
{
"epoch": 0.24,
"learning_rate": 1.774453724450231e-05,
"loss": 1.1982,
"step": 462
},
{
"epoch": 0.24,
"learning_rate": 1.7733800444419088e-05,
"loss": 1.1538,
"step": 463
},
{
"epoch": 0.24,
"learning_rate": 1.7723041414218053e-05,
"loss": 1.3623,
"step": 464
},
{
"epoch": 0.24,
"learning_rate": 1.7712260184825063e-05,
"loss": 1.1934,
"step": 465
},
{
"epoch": 0.24,
"learning_rate": 1.7701456787229805e-05,
"loss": 1.4868,
"step": 466
},
{
"epoch": 0.24,
"learning_rate": 1.7690631252485676e-05,
"loss": 1.3418,
"step": 467
},
{
"epoch": 0.24,
"learning_rate": 1.767978361170971e-05,
"loss": 1.564,
"step": 468
},
{
"epoch": 0.25,
"learning_rate": 1.766891389608248e-05,
"loss": 1.6118,
"step": 469
},
{
"epoch": 0.25,
"learning_rate": 1.7658022136848008e-05,
"loss": 1.4795,
"step": 470
},
{
"epoch": 0.25,
"learning_rate": 1.7647108365313683e-05,
"loss": 1.5483,
"step": 471
},
{
"epoch": 0.25,
"learning_rate": 1.763617261285017e-05,
"loss": 1.8125,
"step": 472
},
{
"epoch": 0.25,
"learning_rate": 1.7625214910891307e-05,
"loss": 1.3335,
"step": 473
},
{
"epoch": 0.25,
"learning_rate": 1.761423529093403e-05,
"loss": 1.9668,
"step": 474
},
{
"epoch": 0.25,
"learning_rate": 1.7603233784538275e-05,
"loss": 1.5422,
"step": 475
},
{
"epoch": 0.25,
"learning_rate": 1.759221042332689e-05,
"loss": 1.5212,
"step": 476
},
{
"epoch": 0.25,
"learning_rate": 1.7581165238985533e-05,
"loss": 1.6543,
"step": 477
},
{
"epoch": 0.25,
"learning_rate": 1.7570098263262608e-05,
"loss": 1.6338,
"step": 478
},
{
"epoch": 0.25,
"learning_rate": 1.7559009527969145e-05,
"loss": 1.5942,
"step": 479
},
{
"epoch": 0.25,
"learning_rate": 1.754789906497872e-05,
"loss": 1.5747,
"step": 480
},
{
"epoch": 0.25,
"learning_rate": 1.7536766906227365e-05,
"loss": 1.6245,
"step": 481
},
{
"epoch": 0.25,
"learning_rate": 1.752561308371348e-05,
"loss": 1.4292,
"step": 482
},
{
"epoch": 0.25,
"learning_rate": 1.751443762949772e-05,
"loss": 1.4087,
"step": 483
},
{
"epoch": 0.25,
"learning_rate": 1.750324057570294e-05,
"loss": 1.7461,
"step": 484
},
{
"epoch": 0.25,
"learning_rate": 1.7492021954514068e-05,
"loss": 1.2061,
"step": 485
},
{
"epoch": 0.25,
"learning_rate": 1.748078179817802e-05,
"loss": 1.5674,
"step": 486
},
{
"epoch": 0.25,
"learning_rate": 1.7469520139003627e-05,
"loss": 1.4792,
"step": 487
},
{
"epoch": 0.26,
"learning_rate": 1.745823700936152e-05,
"loss": 1.6392,
"step": 488
},
{
"epoch": 0.26,
"learning_rate": 1.7446932441684044e-05,
"loss": 1.6133,
"step": 489
},
{
"epoch": 0.26,
"learning_rate": 1.743560646846517e-05,
"loss": 1.1045,
"step": 490
},
{
"epoch": 0.26,
"learning_rate": 1.7424259122260396e-05,
"loss": 1.6724,
"step": 491
},
{
"epoch": 0.26,
"learning_rate": 1.741289043568665e-05,
"loss": 1.396,
"step": 492
},
{
"epoch": 0.26,
"learning_rate": 1.740150044142221e-05,
"loss": 1.7134,
"step": 493
},
{
"epoch": 0.26,
"learning_rate": 1.7390089172206594e-05,
"loss": 1.4453,
"step": 494
},
{
"epoch": 0.26,
"learning_rate": 1.7378656660840475e-05,
"loss": 1.8496,
"step": 495
},
{
"epoch": 0.26,
"learning_rate": 1.7367202940185583e-05,
"loss": 1.2969,
"step": 496
},
{
"epoch": 0.26,
"learning_rate": 1.7355728043164613e-05,
"loss": 1.7539,
"step": 497
},
{
"epoch": 0.26,
"learning_rate": 1.7344232002761138e-05,
"loss": 1.5869,
"step": 498
},
{
"epoch": 0.26,
"learning_rate": 1.7332714852019487e-05,
"loss": 1.334,
"step": 499
},
{
"epoch": 0.26,
"learning_rate": 1.732117662404469e-05,
"loss": 1.4746,
"step": 500
},
{
"epoch": 0.26,
"learning_rate": 1.7309617352002343e-05,
"loss": 1.7705,
"step": 501
},
{
"epoch": 0.26,
"learning_rate": 1.7298037069118546e-05,
"loss": 1.3179,
"step": 502
},
{
"epoch": 0.26,
"learning_rate": 1.7286435808679787e-05,
"loss": 1.3662,
"step": 503
},
{
"epoch": 0.26,
"learning_rate": 1.7274813604032847e-05,
"loss": 1.6641,
"step": 504
},
{
"epoch": 0.26,
"learning_rate": 1.7263170488584717e-05,
"loss": 0.8788,
"step": 505
},
{
"epoch": 0.26,
"learning_rate": 1.725150649580249e-05,
"loss": 1.75,
"step": 506
},
{
"epoch": 0.27,
"learning_rate": 1.7239821659213272e-05,
"loss": 1.7539,
"step": 507
},
{
"epoch": 0.27,
"learning_rate": 1.7228116012404085e-05,
"loss": 1.8027,
"step": 508
},
{
"epoch": 0.27,
"learning_rate": 1.721638958902175e-05,
"loss": 1.585,
"step": 509
},
{
"epoch": 0.27,
"learning_rate": 1.7204642422772836e-05,
"loss": 1.4321,
"step": 510
},
{
"epoch": 0.27,
"learning_rate": 1.7192874547423514e-05,
"loss": 1.502,
"step": 511
},
{
"epoch": 0.27,
"learning_rate": 1.71810859967995e-05,
"loss": 1.4385,
"step": 512
},
{
"epoch": 0.27,
"learning_rate": 1.716927680478591e-05,
"loss": 1.2754,
"step": 513
},
{
"epoch": 0.27,
"learning_rate": 1.715744700532723e-05,
"loss": 1.2314,
"step": 514
},
{
"epoch": 0.27,
"learning_rate": 1.714559663242715e-05,
"loss": 1.2578,
"step": 515
},
{
"epoch": 0.27,
"learning_rate": 1.7133725720148513e-05,
"loss": 1.2544,
"step": 516
},
{
"epoch": 0.27,
"learning_rate": 1.712183430261319e-05,
"loss": 1.1835,
"step": 517
},
{
"epoch": 0.27,
"learning_rate": 1.7109922414001997e-05,
"loss": 1.4375,
"step": 518
},
{
"epoch": 0.27,
"learning_rate": 1.7097990088554595e-05,
"loss": 1.3574,
"step": 519
},
{
"epoch": 0.27,
"learning_rate": 1.7086037360569386e-05,
"loss": 0.7559,
"step": 520
},
{
"epoch": 0.27,
"learning_rate": 1.707406426440342e-05,
"loss": 1.8052,
"step": 521
},
{
"epoch": 0.27,
"learning_rate": 1.7062070834472287e-05,
"loss": 1.3945,
"step": 522
},
{
"epoch": 0.27,
"learning_rate": 1.7050057105250034e-05,
"loss": 0.8877,
"step": 523
},
{
"epoch": 0.27,
"learning_rate": 1.7038023111269045e-05,
"loss": 1.6357,
"step": 524
},
{
"epoch": 0.27,
"learning_rate": 1.7025968887119962e-05,
"loss": 1.5205,
"step": 525
},
{
"epoch": 0.28,
"learning_rate": 1.701389446745158e-05,
"loss": 1.627,
"step": 526
},
{
"epoch": 0.28,
"learning_rate": 1.7001799886970735e-05,
"loss": 1.8467,
"step": 527
},
{
"epoch": 0.28,
"learning_rate": 1.6989685180442212e-05,
"loss": 1.0562,
"step": 528
},
{
"epoch": 0.28,
"learning_rate": 1.697755038268866e-05,
"loss": 1.2578,
"step": 529
},
{
"epoch": 0.28,
"learning_rate": 1.696539552859046e-05,
"loss": 1.7163,
"step": 530
},
{
"epoch": 0.28,
"learning_rate": 1.6953220653085662e-05,
"loss": 1.0232,
"step": 531
},
{
"epoch": 0.28,
"learning_rate": 1.6941025791169855e-05,
"loss": 1.1108,
"step": 532
},
{
"epoch": 0.28,
"learning_rate": 1.692881097789608e-05,
"loss": 1.2886,
"step": 533
},
{
"epoch": 0.28,
"learning_rate": 1.691657624837472e-05,
"loss": 1.4741,
"step": 534
},
{
"epoch": 0.28,
"learning_rate": 1.6904321637773414e-05,
"loss": 1.4961,
"step": 535
},
{
"epoch": 0.28,
"learning_rate": 1.6892047181316952e-05,
"loss": 1.0444,
"step": 536
},
{
"epoch": 0.28,
"learning_rate": 1.687975291428715e-05,
"loss": 1.5532,
"step": 537
},
{
"epoch": 0.28,
"learning_rate": 1.6867438872022782e-05,
"loss": 1.0918,
"step": 538
},
{
"epoch": 0.28,
"learning_rate": 1.6855105089919467e-05,
"loss": 1.3818,
"step": 539
},
{
"epoch": 0.28,
"learning_rate": 1.684275160342955e-05,
"loss": 1.0767,
"step": 540
},
{
"epoch": 0.28,
"learning_rate": 1.683037844806203e-05,
"loss": 1.603,
"step": 541
},
{
"epoch": 0.28,
"learning_rate": 1.6817985659382437e-05,
"loss": 1.166,
"step": 542
},
{
"epoch": 0.28,
"learning_rate": 1.680557327301273e-05,
"loss": 1.1975,
"step": 543
},
{
"epoch": 0.28,
"learning_rate": 1.67931413246312e-05,
"loss": 1.1372,
"step": 544
},
{
"epoch": 0.29,
"learning_rate": 1.678068984997238e-05,
"loss": 1.4121,
"step": 545
},
{
"epoch": 0.29,
"learning_rate": 1.6768218884826916e-05,
"loss": 1.6565,
"step": 546
},
{
"epoch": 0.29,
"learning_rate": 1.675572846504148e-05,
"loss": 1.5146,
"step": 547
},
{
"epoch": 0.29,
"learning_rate": 1.674321862651867e-05,
"loss": 1.5928,
"step": 548
},
{
"epoch": 0.29,
"learning_rate": 1.6730689405216895e-05,
"loss": 1.7119,
"step": 549
},
{
"epoch": 0.29,
"learning_rate": 1.6718140837150286e-05,
"loss": 1.2321,
"step": 550
},
{
"epoch": 0.29,
"learning_rate": 1.6705572958388576e-05,
"loss": 1.1724,
"step": 551
},
{
"epoch": 0.29,
"learning_rate": 1.669298580505701e-05,
"loss": 1.0166,
"step": 552
},
{
"epoch": 0.29,
"learning_rate": 1.668037941333623e-05,
"loss": 1.688,
"step": 553
},
{
"epoch": 0.29,
"learning_rate": 1.6667753819462196e-05,
"loss": 1.0674,
"step": 554
},
{
"epoch": 0.29,
"learning_rate": 1.665510905972603e-05,
"loss": 1.5879,
"step": 555
},
{
"epoch": 0.29,
"learning_rate": 1.664244517047398e-05,
"loss": 0.9707,
"step": 556
},
{
"epoch": 0.29,
"learning_rate": 1.6629762188107248e-05,
"loss": 1.1733,
"step": 557
},
{
"epoch": 0.29,
"learning_rate": 1.6617060149081944e-05,
"loss": 1.3101,
"step": 558
},
{
"epoch": 0.29,
"learning_rate": 1.6604339089908935e-05,
"loss": 1.5615,
"step": 559
},
{
"epoch": 0.29,
"learning_rate": 1.6591599047153778e-05,
"loss": 1.4438,
"step": 560
},
{
"epoch": 0.29,
"learning_rate": 1.657884005743658e-05,
"loss": 1.5654,
"step": 561
},
{
"epoch": 0.29,
"learning_rate": 1.6566062157431915e-05,
"loss": 1.4775,
"step": 562
},
{
"epoch": 0.29,
"learning_rate": 1.655326538386872e-05,
"loss": 1.3428,
"step": 563
},
{
"epoch": 0.3,
"learning_rate": 1.6540449773530168e-05,
"loss": 1.5181,
"step": 564
},
{
"epoch": 0.3,
"learning_rate": 1.652761536325359e-05,
"loss": 1.0115,
"step": 565
},
{
"epoch": 0.3,
"learning_rate": 1.6514762189930353e-05,
"loss": 1.7163,
"step": 566
},
{
"epoch": 0.3,
"learning_rate": 1.650189029050575e-05,
"loss": 1.6455,
"step": 567
},
{
"epoch": 0.3,
"learning_rate": 1.6488999701978905e-05,
"loss": 1.5811,
"step": 568
},
{
"epoch": 0.3,
"learning_rate": 1.6476090461402664e-05,
"loss": 1.1797,
"step": 569
},
{
"epoch": 0.3,
"learning_rate": 1.6463162605883485e-05,
"loss": 1.5205,
"step": 570
},
{
"epoch": 0.3,
"learning_rate": 1.645021617258133e-05,
"loss": 1.1045,
"step": 571
},
{
"epoch": 0.3,
"learning_rate": 1.6437251198709567e-05,
"loss": 1.0125,
"step": 572
},
{
"epoch": 0.3,
"learning_rate": 1.6424267721534856e-05,
"loss": 1.4424,
"step": 573
},
{
"epoch": 0.3,
"learning_rate": 1.641126577837703e-05,
"loss": 1.6807,
"step": 574
},
{
"epoch": 0.3,
"learning_rate": 1.6398245406609027e-05,
"loss": 1.4336,
"step": 575
},
{
"epoch": 0.3,
"learning_rate": 1.638520664365673e-05,
"loss": 1.7842,
"step": 576
},
{
"epoch": 0.3,
"learning_rate": 1.6372149526998905e-05,
"loss": 1.6982,
"step": 577
},
{
"epoch": 0.3,
"learning_rate": 1.635907409416706e-05,
"loss": 1.0474,
"step": 578
},
{
"epoch": 0.3,
"learning_rate": 1.634598038274536e-05,
"loss": 1.8516,
"step": 579
},
{
"epoch": 0.3,
"learning_rate": 1.6332868430370508e-05,
"loss": 1.626,
"step": 580
},
{
"epoch": 0.3,
"learning_rate": 1.631973827473164e-05,
"loss": 1.0522,
"step": 581
},
{
"epoch": 0.3,
"learning_rate": 1.6306589953570207e-05,
"loss": 1.5664,
"step": 582
},
{
"epoch": 0.31,
"learning_rate": 1.6293423504679898e-05,
"loss": 1.3301,
"step": 583
},
{
"epoch": 0.31,
"learning_rate": 1.6280238965906474e-05,
"loss": 1.4199,
"step": 584
},
{
"epoch": 0.31,
"learning_rate": 1.6267036375147728e-05,
"loss": 1.6855,
"step": 585
},
{
"epoch": 0.31,
"learning_rate": 1.6253815770353316e-05,
"loss": 1.5957,
"step": 586
},
{
"epoch": 0.31,
"learning_rate": 1.6240577189524688e-05,
"loss": 1.7695,
"step": 587
},
{
"epoch": 0.31,
"learning_rate": 1.6227320670714966e-05,
"loss": 1.1191,
"step": 588
},
{
"epoch": 0.31,
"learning_rate": 1.6214046252028823e-05,
"loss": 1.6113,
"step": 589
},
{
"epoch": 0.31,
"learning_rate": 1.6200753971622387e-05,
"loss": 1.2463,
"step": 590
},
{
"epoch": 0.31,
"learning_rate": 1.618744386770313e-05,
"loss": 1.4434,
"step": 591
},
{
"epoch": 0.31,
"learning_rate": 1.617411597852976e-05,
"loss": 1.3906,
"step": 592
},
{
"epoch": 0.31,
"learning_rate": 1.6160770342412097e-05,
"loss": 1.6367,
"step": 593
},
{
"epoch": 0.31,
"learning_rate": 1.614740699771098e-05,
"loss": 1.5615,
"step": 594
},
{
"epoch": 0.31,
"learning_rate": 1.6134025982838153e-05,
"loss": 1.7871,
"step": 595
},
{
"epoch": 0.31,
"learning_rate": 1.6120627336256142e-05,
"loss": 1.1763,
"step": 596
},
{
"epoch": 0.31,
"learning_rate": 1.610721109647816e-05,
"loss": 1.5293,
"step": 597
},
{
"epoch": 0.31,
"learning_rate": 1.609377730206799e-05,
"loss": 1.4556,
"step": 598
},
{
"epoch": 0.31,
"learning_rate": 1.6080325991639866e-05,
"loss": 1.8984,
"step": 599
},
{
"epoch": 0.31,
"learning_rate": 1.6066857203858387e-05,
"loss": 1.4844,
"step": 600
},
{
"epoch": 0.31,
"learning_rate": 1.605337097743837e-05,
"loss": 1.293,
"step": 601
},
{
"epoch": 0.32,
"learning_rate": 1.6039867351144778e-05,
"loss": 1.834,
"step": 602
},
{
"epoch": 0.32,
"learning_rate": 1.6026346363792565e-05,
"loss": 1.3242,
"step": 603
},
{
"epoch": 0.32,
"learning_rate": 1.6012808054246608e-05,
"loss": 1.7012,
"step": 604
},
{
"epoch": 0.32,
"learning_rate": 1.5999252461421564e-05,
"loss": 1.6201,
"step": 605
},
{
"epoch": 0.32,
"learning_rate": 1.5985679624281772e-05,
"loss": 1.5669,
"step": 606
},
{
"epoch": 0.32,
"learning_rate": 1.597208958184114e-05,
"loss": 1.6226,
"step": 607
},
{
"epoch": 0.32,
"learning_rate": 1.595848237316303e-05,
"loss": 1.4126,
"step": 608
},
{
"epoch": 0.32,
"learning_rate": 1.5944858037360145e-05,
"loss": 1.5977,
"step": 609
},
{
"epoch": 0.32,
"learning_rate": 1.5931216613594426e-05,
"loss": 1.2842,
"step": 610
},
{
"epoch": 0.32,
"learning_rate": 1.5917558141076915e-05,
"loss": 1.4741,
"step": 611
},
{
"epoch": 0.32,
"learning_rate": 1.590388265906768e-05,
"loss": 1.7705,
"step": 612
},
{
"epoch": 0.32,
"learning_rate": 1.5890190206875663e-05,
"loss": 1.7461,
"step": 613
},
{
"epoch": 0.32,
"learning_rate": 1.58764808238586e-05,
"loss": 1.6201,
"step": 614
},
{
"epoch": 0.32,
"learning_rate": 1.5862754549422884e-05,
"loss": 1.4697,
"step": 615
},
{
"epoch": 0.32,
"learning_rate": 1.5849011423023463e-05,
"loss": 1.6387,
"step": 616
},
{
"epoch": 0.32,
"learning_rate": 1.5835251484163728e-05,
"loss": 1.6191,
"step": 617
},
{
"epoch": 0.32,
"learning_rate": 1.5821474772395385e-05,
"loss": 1.5635,
"step": 618
},
{
"epoch": 0.32,
"learning_rate": 1.5807681327318372e-05,
"loss": 1.6963,
"step": 619
},
{
"epoch": 0.32,
"learning_rate": 1.5793871188580704e-05,
"loss": 1.5742,
"step": 620
},
{
"epoch": 0.32,
"learning_rate": 1.578004439587839e-05,
"loss": 1.6318,
"step": 621
},
{
"epoch": 0.33,
"learning_rate": 1.5766200988955315e-05,
"loss": 1.469,
"step": 622
},
{
"epoch": 0.33,
"learning_rate": 1.575234100760311e-05,
"loss": 1.3774,
"step": 623
},
{
"epoch": 0.33,
"learning_rate": 1.5738464491661054e-05,
"loss": 1.2677,
"step": 624
},
{
"epoch": 0.33,
"learning_rate": 1.5724571481015953e-05,
"loss": 1.1133,
"step": 625
},
{
"epoch": 0.33,
"learning_rate": 1.5710662015602016e-05,
"loss": 1.6191,
"step": 626
},
{
"epoch": 0.33,
"learning_rate": 1.5696736135400765e-05,
"loss": 1.042,
"step": 627
},
{
"epoch": 0.33,
"learning_rate": 1.5682793880440895e-05,
"loss": 1.624,
"step": 628
},
{
"epoch": 0.33,
"learning_rate": 1.566883529079817e-05,
"loss": 1.625,
"step": 629
},
{
"epoch": 0.33,
"learning_rate": 1.5654860406595312e-05,
"loss": 1.291,
"step": 630
},
{
"epoch": 0.33,
"learning_rate": 1.5640869268001875e-05,
"loss": 1.3999,
"step": 631
},
{
"epoch": 0.33,
"learning_rate": 1.562686191523413e-05,
"loss": 1.7725,
"step": 632
},
{
"epoch": 0.33,
"learning_rate": 1.5612838388554966e-05,
"loss": 1.4971,
"step": 633
},
{
"epoch": 0.33,
"learning_rate": 1.5598798728273753e-05,
"loss": 1.4702,
"step": 634
},
{
"epoch": 0.33,
"learning_rate": 1.5584742974746242e-05,
"loss": 1.71,
"step": 635
},
{
"epoch": 0.33,
"learning_rate": 1.557067116837444e-05,
"loss": 0.9536,
"step": 636
},
{
"epoch": 0.33,
"learning_rate": 1.5556583349606492e-05,
"loss": 1.3076,
"step": 637
},
{
"epoch": 0.33,
"learning_rate": 1.5542479558936572e-05,
"loss": 1.6582,
"step": 638
},
{
"epoch": 0.33,
"learning_rate": 1.5528359836904774e-05,
"loss": 1.5371,
"step": 639
},
{
"epoch": 0.33,
"learning_rate": 1.551422422409697e-05,
"loss": 1.6641,
"step": 640
},
{
"epoch": 0.34,
"learning_rate": 1.5500072761144712e-05,
"loss": 1.5273,
"step": 641
},
{
"epoch": 0.34,
"learning_rate": 1.5485905488725126e-05,
"loss": 1.7495,
"step": 642
},
{
"epoch": 0.34,
"learning_rate": 1.5471722447560757e-05,
"loss": 1.6455,
"step": 643
},
{
"epoch": 0.34,
"learning_rate": 1.5457523678419498e-05,
"loss": 0.9026,
"step": 644
},
{
"epoch": 0.34,
"learning_rate": 1.5443309222114434e-05,
"loss": 1.4551,
"step": 645
},
{
"epoch": 0.34,
"learning_rate": 1.5429079119503752e-05,
"loss": 1.2505,
"step": 646
},
{
"epoch": 0.34,
"learning_rate": 1.5414833411490612e-05,
"loss": 1.165,
"step": 647
},
{
"epoch": 0.34,
"learning_rate": 1.5400572139023026e-05,
"loss": 1.1001,
"step": 648
},
{
"epoch": 0.34,
"learning_rate": 1.5386295343093743e-05,
"loss": 1.2588,
"step": 649
},
{
"epoch": 0.34,
"learning_rate": 1.537200306474014e-05,
"loss": 1.6746,
"step": 650
},
{
"epoch": 0.34,
"learning_rate": 1.5357695345044096e-05,
"loss": 1.3276,
"step": 651
},
{
"epoch": 0.34,
"learning_rate": 1.5343372225131863e-05,
"loss": 1.6641,
"step": 652
},
{
"epoch": 0.34,
"learning_rate": 1.5329033746173975e-05,
"loss": 1.6729,
"step": 653
},
{
"epoch": 0.34,
"learning_rate": 1.5314679949385105e-05,
"loss": 1.583,
"step": 654
},
{
"epoch": 0.34,
"learning_rate": 1.530031087602396e-05,
"loss": 1.5801,
"step": 655
},
{
"epoch": 0.34,
"learning_rate": 1.5285926567393158e-05,
"loss": 1.4902,
"step": 656
},
{
"epoch": 0.34,
"learning_rate": 1.5271527064839107e-05,
"loss": 1.8047,
"step": 657
},
{
"epoch": 0.34,
"learning_rate": 1.5257112409751886e-05,
"loss": 1.4009,
"step": 658
},
{
"epoch": 0.34,
"learning_rate": 1.5242682643565135e-05,
"loss": 1.7168,
"step": 659
},
{
"epoch": 0.35,
"learning_rate": 1.5228237807755925e-05,
"loss": 1.1797,
"step": 660
},
{
"epoch": 0.35,
"learning_rate": 1.5213777943844648e-05,
"loss": 0.9973,
"step": 661
},
{
"epoch": 0.35,
"learning_rate": 1.5199303093394883e-05,
"loss": 1.3062,
"step": 662
},
{
"epoch": 0.35,
"learning_rate": 1.5184813298013304e-05,
"loss": 1.7539,
"step": 663
},
{
"epoch": 0.35,
"learning_rate": 1.5170308599349524e-05,
"loss": 1.7422,
"step": 664
},
{
"epoch": 0.35,
"learning_rate": 1.5155789039096005e-05,
"loss": 1.5481,
"step": 665
},
{
"epoch": 0.35,
"learning_rate": 1.5141254658987922e-05,
"loss": 1.4243,
"step": 666
},
{
"epoch": 0.35,
"learning_rate": 1.5126705500803055e-05,
"loss": 1.5732,
"step": 667
},
{
"epoch": 0.35,
"learning_rate": 1.5112141606361657e-05,
"loss": 1.7812,
"step": 668
},
{
"epoch": 0.35,
"learning_rate": 1.5097563017526336e-05,
"loss": 1.6602,
"step": 669
},
{
"epoch": 0.35,
"learning_rate": 1.5082969776201948e-05,
"loss": 1.4185,
"step": 670
},
{
"epoch": 0.35,
"learning_rate": 1.5068361924335457e-05,
"loss": 1.3149,
"step": 671
},
{
"epoch": 0.35,
"learning_rate": 1.505373950391583e-05,
"loss": 1.1836,
"step": 672
},
{
"epoch": 0.35,
"learning_rate": 1.5039102556973899e-05,
"loss": 1.0845,
"step": 673
},
{
"epoch": 0.35,
"learning_rate": 1.5024451125582271e-05,
"loss": 1.6191,
"step": 674
},
{
"epoch": 0.35,
"learning_rate": 1.5009785251855164e-05,
"loss": 1.6055,
"step": 675
},
{
"epoch": 0.35,
"learning_rate": 1.4995104977948331e-05,
"loss": 1.4253,
"step": 676
},
{
"epoch": 0.35,
"learning_rate": 1.49804103460589e-05,
"loss": 1.3594,
"step": 677
},
{
"epoch": 0.35,
"learning_rate": 1.4965701398425275e-05,
"loss": 1.0459,
"step": 678
},
{
"epoch": 0.36,
"learning_rate": 1.4950978177327016e-05,
"loss": 1.2666,
"step": 679
},
{
"epoch": 0.36,
"learning_rate": 1.4936240725084701e-05,
"loss": 1.7461,
"step": 680
},
{
"epoch": 0.36,
"learning_rate": 1.4921489084059823e-05,
"loss": 1.4785,
"step": 681
},
{
"epoch": 0.36,
"learning_rate": 1.4906723296654652e-05,
"loss": 1.6719,
"step": 682
},
{
"epoch": 0.36,
"learning_rate": 1.4891943405312126e-05,
"loss": 1.144,
"step": 683
},
{
"epoch": 0.36,
"learning_rate": 1.4877149452515718e-05,
"loss": 0.9712,
"step": 684
},
{
"epoch": 0.36,
"learning_rate": 1.4862341480789322e-05,
"loss": 1.0283,
"step": 685
},
{
"epoch": 0.36,
"learning_rate": 1.484751953269713e-05,
"loss": 1.5967,
"step": 686
},
{
"epoch": 0.36,
"learning_rate": 1.483268365084351e-05,
"loss": 1.5159,
"step": 687
},
{
"epoch": 0.36,
"learning_rate": 1.481783387787287e-05,
"loss": 1.7822,
"step": 688
},
{
"epoch": 0.36,
"learning_rate": 1.4802970256469564e-05,
"loss": 1.0854,
"step": 689
},
{
"epoch": 0.36,
"learning_rate": 1.4788092829357736e-05,
"loss": 1.6226,
"step": 690
},
{
"epoch": 0.36,
"learning_rate": 1.4773201639301223e-05,
"loss": 1.7637,
"step": 691
},
{
"epoch": 0.36,
"learning_rate": 1.4758296729103413e-05,
"loss": 1.793,
"step": 692
},
{
"epoch": 0.36,
"learning_rate": 1.4743378141607146e-05,
"loss": 1.1528,
"step": 693
},
{
"epoch": 0.36,
"learning_rate": 1.4728445919694563e-05,
"loss": 0.9468,
"step": 694
},
{
"epoch": 0.36,
"learning_rate": 1.4713500106287e-05,
"loss": 1.0693,
"step": 695
},
{
"epoch": 0.36,
"learning_rate": 1.4698540744344864e-05,
"loss": 1.5283,
"step": 696
},
{
"epoch": 0.36,
"learning_rate": 1.4683567876867503e-05,
"loss": 0.8623,
"step": 697
},
{
"epoch": 0.37,
"learning_rate": 1.4668581546893086e-05,
"loss": 1.3296,
"step": 698
},
{
"epoch": 0.37,
"learning_rate": 1.4653581797498477e-05,
"loss": 1.0474,
"step": 699
},
{
"epoch": 0.37,
"learning_rate": 1.4638568671799113e-05,
"loss": 1.7026,
"step": 700
},
{
"epoch": 0.37,
"learning_rate": 1.4623542212948887e-05,
"loss": 1.5596,
"step": 701
},
{
"epoch": 0.37,
"learning_rate": 1.4608502464140002e-05,
"loss": 1.1543,
"step": 702
},
{
"epoch": 0.37,
"learning_rate": 1.4593449468602878e-05,
"loss": 1.2422,
"step": 703
},
{
"epoch": 0.37,
"learning_rate": 1.4578383269606004e-05,
"loss": 1.5244,
"step": 704
},
{
"epoch": 0.37,
"learning_rate": 1.4563303910455821e-05,
"loss": 1.6787,
"step": 705
},
{
"epoch": 0.37,
"learning_rate": 1.4548211434496598e-05,
"loss": 1.5488,
"step": 706
},
{
"epoch": 0.37,
"learning_rate": 1.4533105885110307e-05,
"loss": 1.0776,
"step": 707
},
{
"epoch": 0.37,
"learning_rate": 1.4517987305716502e-05,
"loss": 1.2285,
"step": 708
},
{
"epoch": 0.37,
"learning_rate": 1.4502855739772184e-05,
"loss": 1.0649,
"step": 709
},
{
"epoch": 0.37,
"learning_rate": 1.4487711230771686e-05,
"loss": 1.6406,
"step": 710
},
{
"epoch": 0.37,
"learning_rate": 1.4472553822246548e-05,
"loss": 1.3579,
"step": 711
},
{
"epoch": 0.37,
"learning_rate": 1.4457383557765385e-05,
"loss": 1.1411,
"step": 712
},
{
"epoch": 0.37,
"learning_rate": 1.4442200480933761e-05,
"loss": 1.1094,
"step": 713
},
{
"epoch": 0.37,
"learning_rate": 1.442700463539408e-05,
"loss": 1.0181,
"step": 714
},
{
"epoch": 0.37,
"learning_rate": 1.4411796064825436e-05,
"loss": 0.9143,
"step": 715
},
{
"epoch": 0.37,
"learning_rate": 1.439657481294351e-05,
"loss": 1.0385,
"step": 716
},
{
"epoch": 0.38,
"learning_rate": 1.4381340923500422e-05,
"loss": 1.6094,
"step": 717
},
{
"epoch": 0.38,
"learning_rate": 1.4366094440284632e-05,
"loss": 1.1348,
"step": 718
},
{
"epoch": 0.38,
"learning_rate": 1.435083540712079e-05,
"loss": 1.4775,
"step": 719
},
{
"epoch": 0.38,
"learning_rate": 1.4335563867869618e-05,
"loss": 1.3032,
"step": 720
},
{
"epoch": 0.38,
"learning_rate": 1.4320279866427798e-05,
"loss": 1.4102,
"step": 721
},
{
"epoch": 0.38,
"learning_rate": 1.430498344672782e-05,
"loss": 1.666,
"step": 722
},
{
"epoch": 0.38,
"learning_rate": 1.4289674652737877e-05,
"loss": 1.397,
"step": 723
},
{
"epoch": 0.38,
"learning_rate": 1.4274353528461728e-05,
"loss": 1.7373,
"step": 724
},
{
"epoch": 0.38,
"learning_rate": 1.4259020117938574e-05,
"loss": 1.0601,
"step": 725
},
{
"epoch": 0.38,
"learning_rate": 1.4243674465242934e-05,
"loss": 1.1255,
"step": 726
},
{
"epoch": 0.38,
"learning_rate": 1.4228316614484511e-05,
"loss": 1.4558,
"step": 727
},
{
"epoch": 0.38,
"learning_rate": 1.4212946609808079e-05,
"loss": 1.5928,
"step": 728
},
{
"epoch": 0.38,
"learning_rate": 1.4197564495393334e-05,
"loss": 1.6543,
"step": 729
},
{
"epoch": 0.38,
"learning_rate": 1.4182170315454798e-05,
"loss": 1.0896,
"step": 730
},
{
"epoch": 0.38,
"learning_rate": 1.4166764114241657e-05,
"loss": 1.3135,
"step": 731
},
{
"epoch": 0.38,
"learning_rate": 1.4151345936037664e-05,
"loss": 1.1094,
"step": 732
},
{
"epoch": 0.38,
"learning_rate": 1.413591582516099e-05,
"loss": 1.1597,
"step": 733
},
{
"epoch": 0.38,
"learning_rate": 1.4120473825964112e-05,
"loss": 0.9634,
"step": 734
},
{
"epoch": 0.38,
"learning_rate": 1.4105019982833674e-05,
"loss": 1.1577,
"step": 735
},
{
"epoch": 0.39,
"learning_rate": 1.4089554340190365e-05,
"loss": 1.7778,
"step": 736
},
{
"epoch": 0.39,
"learning_rate": 1.4074076942488794e-05,
"loss": 1.6846,
"step": 737
},
{
"epoch": 0.39,
"learning_rate": 1.4058587834217356e-05,
"loss": 1.415,
"step": 738
},
{
"epoch": 0.39,
"learning_rate": 1.404308705989811e-05,
"loss": 0.9971,
"step": 739
},
{
"epoch": 0.39,
"learning_rate": 1.4027574664086641e-05,
"loss": 1.3311,
"step": 740
},
{
"epoch": 0.39,
"learning_rate": 1.4012050691371948e-05,
"loss": 1.791,
"step": 741
},
{
"epoch": 0.39,
"learning_rate": 1.3996515186376298e-05,
"loss": 1.877,
"step": 742
},
{
"epoch": 0.39,
"learning_rate": 1.3980968193755115e-05,
"loss": 1.584,
"step": 743
},
{
"epoch": 0.39,
"learning_rate": 1.3965409758196838e-05,
"loss": 1.79,
"step": 744
},
{
"epoch": 0.39,
"learning_rate": 1.3949839924422798e-05,
"loss": 1.8086,
"step": 745
},
{
"epoch": 0.39,
"learning_rate": 1.3934258737187087e-05,
"loss": 0.8597,
"step": 746
},
{
"epoch": 0.39,
"learning_rate": 1.3918666241276442e-05,
"loss": 1.2046,
"step": 747
},
{
"epoch": 0.39,
"learning_rate": 1.390306248151009e-05,
"loss": 1.2397,
"step": 748
},
{
"epoch": 0.39,
"learning_rate": 1.3887447502739649e-05,
"loss": 1.5557,
"step": 749
},
{
"epoch": 0.39,
"learning_rate": 1.3871821349848977e-05,
"loss": 1.1875,
"step": 750
},
{
"epoch": 0.39,
"learning_rate": 1.3856184067754053e-05,
"loss": 1.9082,
"step": 751
},
{
"epoch": 0.39,
"learning_rate": 1.3840535701402847e-05,
"loss": 1.2056,
"step": 752
},
{
"epoch": 0.39,
"learning_rate": 1.3824876295775187e-05,
"loss": 1.4102,
"step": 753
},
{
"epoch": 0.39,
"learning_rate": 1.3809205895882635e-05,
"loss": 1.2913,
"step": 754
},
{
"epoch": 0.4,
"learning_rate": 1.3793524546768358e-05,
"loss": 1.4082,
"step": 755
},
{
"epoch": 0.4,
"learning_rate": 1.3777832293506985e-05,
"loss": 1.5161,
"step": 756
},
{
"epoch": 0.4,
"learning_rate": 1.3762129181204502e-05,
"loss": 1.376,
"step": 757
},
{
"epoch": 0.4,
"learning_rate": 1.3746415254998095e-05,
"loss": 1.4937,
"step": 758
},
{
"epoch": 0.4,
"learning_rate": 1.3730690560056043e-05,
"loss": 1.5908,
"step": 759
},
{
"epoch": 0.4,
"learning_rate": 1.3714955141577573e-05,
"loss": 1.4888,
"step": 760
},
{
"epoch": 0.4,
"learning_rate": 1.3699209044792743e-05,
"loss": 1.4258,
"step": 761
},
{
"epoch": 0.4,
"learning_rate": 1.3683452314962295e-05,
"loss": 1.3394,
"step": 762
},
{
"epoch": 0.4,
"learning_rate": 1.3667684997377542e-05,
"loss": 1.5742,
"step": 763
},
{
"epoch": 0.4,
"learning_rate": 1.365190713736023e-05,
"loss": 1.5703,
"step": 764
},
{
"epoch": 0.4,
"learning_rate": 1.3636118780262405e-05,
"loss": 1.4375,
"step": 765
},
{
"epoch": 0.4,
"learning_rate": 1.3620319971466288e-05,
"loss": 1.1797,
"step": 766
},
{
"epoch": 0.4,
"learning_rate": 1.3604510756384141e-05,
"loss": 1.5332,
"step": 767
},
{
"epoch": 0.4,
"learning_rate": 1.3588691180458143e-05,
"loss": 1.1279,
"step": 768
},
{
"epoch": 0.4,
"learning_rate": 1.3572861289160245e-05,
"loss": 1.0874,
"step": 769
},
{
"epoch": 0.4,
"learning_rate": 1.3557021127992055e-05,
"loss": 1.6123,
"step": 770
},
{
"epoch": 0.4,
"learning_rate": 1.3541170742484705e-05,
"loss": 1.6216,
"step": 771
},
{
"epoch": 0.4,
"learning_rate": 1.3525310178198707e-05,
"loss": 1.5615,
"step": 772
},
{
"epoch": 0.4,
"learning_rate": 1.3509439480723835e-05,
"loss": 1.0244,
"step": 773
},
{
"epoch": 0.41,
"learning_rate": 1.3493558695678992e-05,
"loss": 1.1709,
"step": 774
},
{
"epoch": 0.41,
"learning_rate": 1.347766786871207e-05,
"loss": 1.6758,
"step": 775
},
{
"epoch": 0.41,
"learning_rate": 1.3461767045499834e-05,
"loss": 1.3013,
"step": 776
},
{
"epoch": 0.41,
"learning_rate": 1.3445856271747777e-05,
"loss": 1.5396,
"step": 777
},
{
"epoch": 0.41,
"learning_rate": 1.3429935593189996e-05,
"loss": 1.7061,
"step": 778
},
{
"epoch": 0.41,
"learning_rate": 1.3414005055589057e-05,
"loss": 1.2766,
"step": 779
},
{
"epoch": 0.41,
"learning_rate": 1.3398064704735862e-05,
"loss": 1.5396,
"step": 780
},
{
"epoch": 0.41,
"learning_rate": 1.3382114586449533e-05,
"loss": 1.5503,
"step": 781
},
{
"epoch": 0.41,
"learning_rate": 1.3366154746577254e-05,
"loss": 1.377,
"step": 782
},
{
"epoch": 0.41,
"learning_rate": 1.3350185230994157e-05,
"loss": 1.064,
"step": 783
},
{
"epoch": 0.41,
"learning_rate": 1.3334206085603186e-05,
"loss": 0.9751,
"step": 784
},
{
"epoch": 0.41,
"learning_rate": 1.3318217356334967e-05,
"loss": 1.7109,
"step": 785
},
{
"epoch": 0.41,
"learning_rate": 1.3302219089147675e-05,
"loss": 1.6152,
"step": 786
},
{
"epoch": 0.41,
"learning_rate": 1.3286211330026894e-05,
"loss": 0.9763,
"step": 787
},
{
"epoch": 0.41,
"learning_rate": 1.3270194124985499e-05,
"loss": 1.1953,
"step": 788
},
{
"epoch": 0.41,
"learning_rate": 1.325416752006351e-05,
"loss": 1.7783,
"step": 789
},
{
"epoch": 0.41,
"learning_rate": 1.3238131561327976e-05,
"loss": 1.6875,
"step": 790
},
{
"epoch": 0.41,
"learning_rate": 1.3222086294872818e-05,
"loss": 0.9619,
"step": 791
},
{
"epoch": 0.41,
"learning_rate": 1.320603176681873e-05,
"loss": 1.751,
"step": 792
},
{
"epoch": 0.41,
"learning_rate": 1.318996802331301e-05,
"loss": 1.1965,
"step": 793
},
{
"epoch": 0.42,
"learning_rate": 1.3173895110529454e-05,
"loss": 1.4597,
"step": 794
},
{
"epoch": 0.42,
"learning_rate": 1.3157813074668213e-05,
"loss": 1.6191,
"step": 795
},
{
"epoch": 0.42,
"learning_rate": 1.3141721961955658e-05,
"loss": 1.3184,
"step": 796
},
{
"epoch": 0.42,
"learning_rate": 1.3125621818644258e-05,
"loss": 1.3213,
"step": 797
},
{
"epoch": 0.42,
"learning_rate": 1.3109512691012434e-05,
"loss": 1.7422,
"step": 798
},
{
"epoch": 0.42,
"learning_rate": 1.3093394625364433e-05,
"loss": 1.6865,
"step": 799
},
{
"epoch": 0.42,
"learning_rate": 1.3077267668030194e-05,
"loss": 1.2798,
"step": 800
},
{
"epoch": 0.42,
"learning_rate": 1.3061131865365211e-05,
"loss": 1.5044,
"step": 801
},
{
"epoch": 0.42,
"learning_rate": 1.3044987263750409e-05,
"loss": 1.5137,
"step": 802
},
{
"epoch": 0.42,
"learning_rate": 1.3028833909591999e-05,
"loss": 1.8379,
"step": 803
},
{
"epoch": 0.42,
"learning_rate": 1.3012671849321351e-05,
"loss": 1.4248,
"step": 804
},
{
"epoch": 0.42,
"learning_rate": 1.2996501129394872e-05,
"loss": 1.583,
"step": 805
},
{
"epoch": 0.42,
"learning_rate": 1.2980321796293838e-05,
"loss": 1.3607,
"step": 806
},
{
"epoch": 0.42,
"learning_rate": 1.2964133896524305e-05,
"loss": 1.1348,
"step": 807
},
{
"epoch": 0.42,
"learning_rate": 1.2947937476616938e-05,
"loss": 1.0129,
"step": 808
},
{
"epoch": 0.42,
"learning_rate": 1.29317325831269e-05,
"loss": 1.6416,
"step": 809
},
{
"epoch": 0.42,
"learning_rate": 1.2915519262633704e-05,
"loss": 1.5244,
"step": 810
},
{
"epoch": 0.42,
"learning_rate": 1.2899297561741097e-05,
"loss": 1.3252,
"step": 811
},
{
"epoch": 0.42,
"learning_rate": 1.2883067527076904e-05,
"loss": 1.4194,
"step": 812
},
{
"epoch": 0.43,
"learning_rate": 1.286682920529291e-05,
"loss": 1.3179,
"step": 813
},
{
"epoch": 0.43,
"learning_rate": 1.2850582643064713e-05,
"loss": 1.5791,
"step": 814
},
{
"epoch": 0.43,
"learning_rate": 1.2834327887091611e-05,
"loss": 1.1172,
"step": 815
},
{
"epoch": 0.43,
"learning_rate": 1.2818064984096443e-05,
"loss": 1.1431,
"step": 816
},
{
"epoch": 0.43,
"learning_rate": 1.280179398082547e-05,
"loss": 1.5654,
"step": 817
},
{
"epoch": 0.43,
"learning_rate": 1.2785514924048235e-05,
"loss": 1.8135,
"step": 818
},
{
"epoch": 0.43,
"learning_rate": 1.2769227860557431e-05,
"loss": 1.8877,
"step": 819
},
{
"epoch": 0.43,
"learning_rate": 1.2752932837168768e-05,
"loss": 1.5815,
"step": 820
},
{
"epoch": 0.43,
"learning_rate": 1.2736629900720832e-05,
"loss": 1.2261,
"step": 821
},
{
"epoch": 0.43,
"learning_rate": 1.2720319098074954e-05,
"loss": 1.6338,
"step": 822
},
{
"epoch": 0.43,
"learning_rate": 1.2704000476115079e-05,
"loss": 1.5332,
"step": 823
},
{
"epoch": 0.43,
"learning_rate": 1.268767408174763e-05,
"loss": 1.5574,
"step": 824
},
{
"epoch": 0.43,
"learning_rate": 1.2671339961901366e-05,
"loss": 1.1802,
"step": 825
},
{
"epoch": 0.43,
"learning_rate": 1.265499816352725e-05,
"loss": 1.2949,
"step": 826
},
{
"epoch": 0.43,
"learning_rate": 1.2638648733598325e-05,
"loss": 1.5928,
"step": 827
},
{
"epoch": 0.43,
"learning_rate": 1.2622291719109559e-05,
"loss": 1.6592,
"step": 828
},
{
"epoch": 0.43,
"learning_rate": 1.2605927167077736e-05,
"loss": 1.3447,
"step": 829
},
{
"epoch": 0.43,
"learning_rate": 1.2589555124541293e-05,
"loss": 1.5735,
"step": 830
},
{
"epoch": 0.43,
"learning_rate": 1.25731756385602e-05,
"loss": 1.6133,
"step": 831
},
{
"epoch": 0.44,
"learning_rate": 1.255678875621583e-05,
"loss": 1.4575,
"step": 832
},
{
"epoch": 0.44,
"learning_rate": 1.2540394524610808e-05,
"loss": 0.8645,
"step": 833
},
{
"epoch": 0.44,
"learning_rate": 1.252399299086889e-05,
"loss": 1.5134,
"step": 834
},
{
"epoch": 0.44,
"learning_rate": 1.2507584202134812e-05,
"loss": 1.0496,
"step": 835
},
{
"epoch": 0.44,
"learning_rate": 1.2491168205574175e-05,
"loss": 1.6445,
"step": 836
},
{
"epoch": 0.44,
"learning_rate": 1.2474745048373292e-05,
"loss": 1.0144,
"step": 837
},
{
"epoch": 0.44,
"learning_rate": 1.2458314777739056e-05,
"loss": 1.5771,
"step": 838
},
{
"epoch": 0.44,
"learning_rate": 1.2441877440898817e-05,
"loss": 1.6143,
"step": 839
},
{
"epoch": 0.44,
"learning_rate": 1.2425433085100224e-05,
"loss": 1.418,
"step": 840
},
{
"epoch": 0.44,
"learning_rate": 1.2408981757611112e-05,
"loss": 1.1021,
"step": 841
},
{
"epoch": 0.44,
"learning_rate": 1.2392523505719349e-05,
"loss": 1.4482,
"step": 842
},
{
"epoch": 0.44,
"learning_rate": 1.237605837673271e-05,
"loss": 1.5659,
"step": 843
},
{
"epoch": 0.44,
"learning_rate": 1.2359586417978733e-05,
"loss": 1.2413,
"step": 844
},
{
"epoch": 0.44,
"learning_rate": 1.2343107676804593e-05,
"loss": 0.8834,
"step": 845
},
{
"epoch": 0.44,
"learning_rate": 1.2326622200576956e-05,
"loss": 1.5615,
"step": 846
},
{
"epoch": 0.44,
"learning_rate": 1.2310130036681847e-05,
"loss": 0.9961,
"step": 847
},
{
"epoch": 0.44,
"learning_rate": 1.2293631232524524e-05,
"loss": 1.6982,
"step": 848
},
{
"epoch": 0.44,
"learning_rate": 1.2277125835529318e-05,
"loss": 1.5449,
"step": 849
},
{
"epoch": 0.44,
"learning_rate": 1.2260613893139517e-05,
"loss": 1.1851,
"step": 850
},
{
"epoch": 0.45,
"learning_rate": 1.2244095452817227e-05,
"loss": 1.6323,
"step": 851
},
{
"epoch": 0.45,
"learning_rate": 1.2227570562043223e-05,
"loss": 1.3906,
"step": 852
},
{
"epoch": 0.45,
"learning_rate": 1.221103926831683e-05,
"loss": 1.7168,
"step": 853
},
{
"epoch": 0.45,
"learning_rate": 1.219450161915577e-05,
"loss": 1.4963,
"step": 854
},
{
"epoch": 0.45,
"learning_rate": 1.2177957662096038e-05,
"loss": 1.0092,
"step": 855
},
{
"epoch": 0.45,
"learning_rate": 1.2161407444691761e-05,
"loss": 1.4785,
"step": 856
},
{
"epoch": 0.45,
"learning_rate": 1.2144851014515055e-05,
"loss": 1.2271,
"step": 857
},
{
"epoch": 0.45,
"learning_rate": 1.2128288419155903e-05,
"loss": 1.2188,
"step": 858
},
{
"epoch": 0.45,
"learning_rate": 1.2111719706222e-05,
"loss": 1.043,
"step": 859
},
{
"epoch": 0.45,
"learning_rate": 1.2095144923338633e-05,
"loss": 1.4932,
"step": 860
},
{
"epoch": 0.45,
"learning_rate": 1.2078564118148531e-05,
"loss": 1.5688,
"step": 861
},
{
"epoch": 0.45,
"learning_rate": 1.2061977338311737e-05,
"loss": 1.064,
"step": 862
},
{
"epoch": 0.45,
"learning_rate": 1.2045384631505466e-05,
"loss": 1.749,
"step": 863
},
{
"epoch": 0.45,
"learning_rate": 1.202878604542397e-05,
"loss": 1.7422,
"step": 864
},
{
"epoch": 0.45,
"learning_rate": 1.20121816277784e-05,
"loss": 1.272,
"step": 865
},
{
"epoch": 0.45,
"learning_rate": 1.1995571426296671e-05,
"loss": 1.2671,
"step": 866
},
{
"epoch": 0.45,
"learning_rate": 1.1978955488723323e-05,
"loss": 1.4087,
"step": 867
},
{
"epoch": 0.45,
"learning_rate": 1.1962333862819379e-05,
"loss": 1.5044,
"step": 868
},
{
"epoch": 0.45,
"learning_rate": 1.1945706596362216e-05,
"loss": 1.5854,
"step": 869
},
{
"epoch": 0.46,
"learning_rate": 1.1929073737145432e-05,
"loss": 1.1011,
"step": 870
},
{
"epoch": 0.46,
"learning_rate": 1.1912435332978684e-05,
"loss": 0.9824,
"step": 871
},
{
"epoch": 0.46,
"learning_rate": 1.1895791431687585e-05,
"loss": 1.6123,
"step": 872
},
{
"epoch": 0.46,
"learning_rate": 1.1879142081113535e-05,
"loss": 1.5532,
"step": 873
},
{
"epoch": 0.46,
"learning_rate": 1.1862487329113606e-05,
"loss": 1.4365,
"step": 874
},
{
"epoch": 0.46,
"learning_rate": 1.1845827223560394e-05,
"loss": 1.1333,
"step": 875
},
{
"epoch": 0.46,
"learning_rate": 1.1829161812341883e-05,
"loss": 1.7139,
"step": 876
},
{
"epoch": 0.46,
"learning_rate": 1.1812491143361305e-05,
"loss": 1.0938,
"step": 877
},
{
"epoch": 0.46,
"learning_rate": 1.1795815264537008e-05,
"loss": 1.1064,
"step": 878
},
{
"epoch": 0.46,
"learning_rate": 1.1779134223802315e-05,
"loss": 1.665,
"step": 879
},
{
"epoch": 0.46,
"learning_rate": 1.1762448069105387e-05,
"loss": 1.0991,
"step": 880
},
{
"epoch": 0.46,
"learning_rate": 1.1745756848409083e-05,
"loss": 1.6597,
"step": 881
},
{
"epoch": 0.46,
"learning_rate": 1.1729060609690824e-05,
"loss": 1.6309,
"step": 882
},
{
"epoch": 0.46,
"learning_rate": 1.171235940094245e-05,
"loss": 1.647,
"step": 883
},
{
"epoch": 0.46,
"learning_rate": 1.1695653270170102e-05,
"loss": 0.7725,
"step": 884
},
{
"epoch": 0.46,
"learning_rate": 1.1678942265394055e-05,
"loss": 1.5474,
"step": 885
},
{
"epoch": 0.46,
"learning_rate": 1.1662226434648595e-05,
"loss": 1.8389,
"step": 886
},
{
"epoch": 0.46,
"learning_rate": 1.1645505825981884e-05,
"loss": 1.2207,
"step": 887
},
{
"epoch": 0.46,
"learning_rate": 1.162878048745582e-05,
"loss": 1.5566,
"step": 888
},
{
"epoch": 0.47,
"learning_rate": 1.161205046714589e-05,
"loss": 1.6851,
"step": 889
},
{
"epoch": 0.47,
"learning_rate": 1.1595315813141041e-05,
"loss": 1.4443,
"step": 890
},
{
"epoch": 0.47,
"learning_rate": 1.1578576573543541e-05,
"loss": 1.1045,
"step": 891
},
{
"epoch": 0.47,
"learning_rate": 1.1561832796468837e-05,
"loss": 1.1906,
"step": 892
},
{
"epoch": 0.47,
"learning_rate": 1.1545084530045419e-05,
"loss": 1.4688,
"step": 893
},
{
"epoch": 0.47,
"learning_rate": 1.1528331822414681e-05,
"loss": 1.8008,
"step": 894
},
{
"epoch": 0.47,
"learning_rate": 1.1511574721730781e-05,
"loss": 1.4004,
"step": 895
},
{
"epoch": 0.47,
"learning_rate": 1.1494813276160511e-05,
"loss": 1.7275,
"step": 896
},
{
"epoch": 0.47,
"learning_rate": 1.1478047533883143e-05,
"loss": 1.1919,
"step": 897
},
{
"epoch": 0.47,
"learning_rate": 1.1461277543090308e-05,
"loss": 0.9951,
"step": 898
},
{
"epoch": 0.47,
"learning_rate": 1.144450335198584e-05,
"loss": 1.3525,
"step": 899
},
{
"epoch": 0.47,
"learning_rate": 1.1427725008785655e-05,
"loss": 1.2007,
"step": 900
},
{
"epoch": 0.47,
"learning_rate": 1.1410942561717601e-05,
"loss": 1.1406,
"step": 901
},
{
"epoch": 0.47,
"learning_rate": 1.139415605902132e-05,
"loss": 1.5728,
"step": 902
},
{
"epoch": 0.47,
"learning_rate": 1.1377365548948114e-05,
"loss": 1.5947,
"step": 903
},
{
"epoch": 0.47,
"learning_rate": 1.1360571079760798e-05,
"loss": 1.1715,
"step": 904
},
{
"epoch": 0.47,
"learning_rate": 1.1343772699733579e-05,
"loss": 1.4805,
"step": 905
},
{
"epoch": 0.47,
"learning_rate": 1.132697045715189e-05,
"loss": 1.8428,
"step": 906
},
{
"epoch": 0.47,
"learning_rate": 1.1310164400312284e-05,
"loss": 1.4932,
"step": 907
},
{
"epoch": 0.48,
"learning_rate": 1.1293354577522264e-05,
"loss": 1.2031,
"step": 908
},
{
"epoch": 0.48,
"learning_rate": 1.127654103710016e-05,
"loss": 1.1465,
"step": 909
},
{
"epoch": 0.48,
"learning_rate": 1.1259723827374996e-05,
"loss": 1.5317,
"step": 910
},
{
"epoch": 0.48,
"learning_rate": 1.1242902996686333e-05,
"loss": 1.4526,
"step": 911
},
{
"epoch": 0.48,
"learning_rate": 1.122607859338415e-05,
"loss": 1.1372,
"step": 912
},
{
"epoch": 0.48,
"learning_rate": 1.1209250665828684e-05,
"loss": 1.3848,
"step": 913
},
{
"epoch": 0.48,
"learning_rate": 1.119241926239031e-05,
"loss": 1.752,
"step": 914
},
{
"epoch": 0.48,
"learning_rate": 1.1175584431449393e-05,
"loss": 1.4453,
"step": 915
},
{
"epoch": 0.48,
"learning_rate": 1.1158746221396148e-05,
"loss": 1.4961,
"step": 916
},
{
"epoch": 0.48,
"learning_rate": 1.1141904680630504e-05,
"loss": 1.5713,
"step": 917
},
{
"epoch": 0.48,
"learning_rate": 1.1125059857561966e-05,
"loss": 1.1006,
"step": 918
},
{
"epoch": 0.48,
"learning_rate": 1.110821180060947e-05,
"loss": 1.2229,
"step": 919
},
{
"epoch": 0.48,
"learning_rate": 1.1091360558201249e-05,
"loss": 1.083,
"step": 920
},
{
"epoch": 0.48,
"learning_rate": 1.1074506178774696e-05,
"loss": 1.6172,
"step": 921
},
{
"epoch": 0.48,
"learning_rate": 1.1057648710776212e-05,
"loss": 1.7012,
"step": 922
},
{
"epoch": 0.48,
"learning_rate": 1.1040788202661084e-05,
"loss": 1.3699,
"step": 923
},
{
"epoch": 0.48,
"learning_rate": 1.1023924702893334e-05,
"loss": 1.0747,
"step": 924
},
{
"epoch": 0.48,
"learning_rate": 1.1007058259945584e-05,
"loss": 1.5249,
"step": 925
},
{
"epoch": 0.48,
"learning_rate": 1.0990188922298918e-05,
"loss": 1.1157,
"step": 926
},
{
"epoch": 0.49,
"learning_rate": 1.0973316738442738e-05,
"loss": 1.4419,
"step": 927
},
{
"epoch": 0.49,
"learning_rate": 1.0956441756874628e-05,
"loss": 1.4354,
"step": 928
},
{
"epoch": 0.49,
"learning_rate": 1.0939564026100213e-05,
"loss": 1.4668,
"step": 929
},
{
"epoch": 0.49,
"learning_rate": 1.092268359463302e-05,
"loss": 1.0879,
"step": 930
},
{
"epoch": 0.49,
"learning_rate": 1.0905800510994341e-05,
"loss": 1.0923,
"step": 931
},
{
"epoch": 0.49,
"learning_rate": 1.0888914823713092e-05,
"loss": 1.4277,
"step": 932
},
{
"epoch": 0.49,
"learning_rate": 1.0872026581325667e-05,
"loss": 1.6455,
"step": 933
},
{
"epoch": 0.49,
"learning_rate": 1.0855135832375809e-05,
"loss": 1.2954,
"step": 934
},
{
"epoch": 0.49,
"learning_rate": 1.083824262541447e-05,
"loss": 1.4199,
"step": 935
},
{
"epoch": 0.49,
"learning_rate": 1.0821347008999658e-05,
"loss": 1.1072,
"step": 936
},
{
"epoch": 0.49,
"learning_rate": 1.0804449031696315e-05,
"loss": 1.1694,
"step": 937
},
{
"epoch": 0.49,
"learning_rate": 1.0787548742076164e-05,
"loss": 1.4644,
"step": 938
},
{
"epoch": 0.49,
"learning_rate": 1.0770646188717578e-05,
"loss": 1.4917,
"step": 939
},
{
"epoch": 0.49,
"learning_rate": 1.0753741420205431e-05,
"loss": 1.3276,
"step": 940
},
{
"epoch": 0.49,
"learning_rate": 1.0736834485130972e-05,
"loss": 1.1748,
"step": 941
},
{
"epoch": 0.49,
"learning_rate": 1.0719925432091671e-05,
"loss": 1.8076,
"step": 942
},
{
"epoch": 0.49,
"learning_rate": 1.0703014309691096e-05,
"loss": 0.9493,
"step": 943
},
{
"epoch": 0.49,
"learning_rate": 1.068610116653875e-05,
"loss": 1.5107,
"step": 944
},
{
"epoch": 0.49,
"learning_rate": 1.0669186051249948e-05,
"loss": 1.4468,
"step": 945
},
{
"epoch": 0.5,
"learning_rate": 1.0652269012445684e-05,
"loss": 1.4751,
"step": 946
},
{
"epoch": 0.5,
"learning_rate": 1.0635350098752468e-05,
"loss": 1.1997,
"step": 947
},
{
"epoch": 0.5,
"learning_rate": 1.0618429358802209e-05,
"loss": 1.4517,
"step": 948
},
{
"epoch": 0.5,
"learning_rate": 1.0601506841232057e-05,
"loss": 1.4814,
"step": 949
},
{
"epoch": 0.5,
"learning_rate": 1.0584582594684274e-05,
"loss": 1.3149,
"step": 950
},
{
"epoch": 0.5,
"learning_rate": 1.0567656667806097e-05,
"loss": 1.7627,
"step": 951
},
{
"epoch": 0.5,
"learning_rate": 1.055072910924959e-05,
"loss": 1.6982,
"step": 952
},
{
"epoch": 0.5,
"learning_rate": 1.0533799967671503e-05,
"loss": 1.0278,
"step": 953
},
{
"epoch": 0.5,
"learning_rate": 1.0516869291733142e-05,
"loss": 1.5518,
"step": 954
},
{
"epoch": 0.5,
"learning_rate": 1.0499937130100222e-05,
"loss": 1.4204,
"step": 955
},
{
"epoch": 0.5,
"learning_rate": 1.0483003531442726e-05,
"loss": 1.6123,
"step": 956
},
{
"epoch": 0.5,
"learning_rate": 1.0466068544434766e-05,
"loss": 1.3213,
"step": 957
},
{
"epoch": 0.5,
"learning_rate": 1.0449132217754455e-05,
"loss": 1.5708,
"step": 958
},
{
"epoch": 0.5,
"learning_rate": 1.043219460008374e-05,
"loss": 1.4751,
"step": 959
},
{
"epoch": 0.5,
"learning_rate": 1.0415255740108299e-05,
"loss": 1.3486,
"step": 960
},
{
"epoch": 0.5,
"learning_rate": 1.0398315686517362e-05,
"loss": 1.3721,
"step": 961
},
{
"epoch": 0.5,
"learning_rate": 1.0381374488003603e-05,
"loss": 1.1133,
"step": 962
},
{
"epoch": 0.5,
"learning_rate": 1.036443219326298e-05,
"loss": 1.0,
"step": 963
},
{
"epoch": 0.5,
"learning_rate": 1.0347488850994608e-05,
"loss": 1.1328,
"step": 964
},
{
"epoch": 0.5,
"learning_rate": 1.0330544509900603e-05,
"loss": 1.6392,
"step": 965
},
{
"epoch": 0.51,
"learning_rate": 1.0313599218685961e-05,
"loss": 1.0771,
"step": 966
},
{
"epoch": 0.51,
"learning_rate": 1.029665302605841e-05,
"loss": 1.5693,
"step": 967
},
{
"epoch": 0.51,
"learning_rate": 1.0279705980728258e-05,
"loss": 1.1655,
"step": 968
},
{
"epoch": 0.51,
"learning_rate": 1.026275813140828e-05,
"loss": 0.947,
"step": 969
},
{
"epoch": 0.51,
"learning_rate": 1.0245809526813547e-05,
"loss": 1.4395,
"step": 970
},
{
"epoch": 0.51,
"learning_rate": 1.0228860215661309e-05,
"loss": 1.2583,
"step": 971
},
{
"epoch": 0.51,
"learning_rate": 1.0211910246670848e-05,
"loss": 0.8533,
"step": 972
},
{
"epoch": 0.51,
"learning_rate": 1.0194959668563332e-05,
"loss": 1.286,
"step": 973
},
{
"epoch": 0.51,
"learning_rate": 1.017800853006168e-05,
"loss": 1.1396,
"step": 974
},
{
"epoch": 0.51,
"learning_rate": 1.0161056879890427e-05,
"loss": 1.0344,
"step": 975
},
{
"epoch": 0.51,
"learning_rate": 1.0144104766775574e-05,
"loss": 1.9268,
"step": 976
},
{
"epoch": 0.51,
"learning_rate": 1.0127152239444453e-05,
"loss": 1.5415,
"step": 977
},
{
"epoch": 0.51,
"learning_rate": 1.0110199346625593e-05,
"loss": 1.1924,
"step": 978
},
{
"epoch": 0.51,
"learning_rate": 1.0093246137048563e-05,
"loss": 1.6943,
"step": 979
},
{
"epoch": 0.51,
"learning_rate": 1.0076292659443852e-05,
"loss": 1.272,
"step": 980
},
{
"epoch": 0.51,
"learning_rate": 1.0059338962542713e-05,
"loss": 1.5254,
"step": 981
},
{
"epoch": 0.51,
"learning_rate": 1.0042385095077032e-05,
"loss": 1.6367,
"step": 982
},
{
"epoch": 0.51,
"learning_rate": 1.0025431105779184e-05,
"loss": 1.6484,
"step": 983
},
{
"epoch": 0.51,
"learning_rate": 1.0008477043381896e-05,
"loss": 1.5859,
"step": 984
},
{
"epoch": 0.52,
"learning_rate": 9.991522956618107e-06,
"loss": 1.168,
"step": 985
},
{
"epoch": 0.52,
"learning_rate": 9.974568894220821e-06,
"loss": 1.6069,
"step": 986
},
{
"epoch": 0.52,
"learning_rate": 9.957614904922971e-06,
"loss": 1.5088,
"step": 987
},
{
"epoch": 0.52,
"learning_rate": 9.940661037457294e-06,
"loss": 1.1694,
"step": 988
},
{
"epoch": 0.52,
"learning_rate": 9.923707340556151e-06,
"loss": 1.5801,
"step": 989
},
{
"epoch": 0.52,
"learning_rate": 9.906753862951437e-06,
"loss": 1.7764,
"step": 990
},
{
"epoch": 0.52,
"learning_rate": 9.88980065337441e-06,
"loss": 1.625,
"step": 991
},
{
"epoch": 0.52,
"learning_rate": 9.872847760555547e-06,
"loss": 1.1943,
"step": 992
},
{
"epoch": 0.52,
"learning_rate": 9.855895233224431e-06,
"loss": 0.9604,
"step": 993
},
{
"epoch": 0.52,
"learning_rate": 9.838943120109576e-06,
"loss": 1.5815,
"step": 994
},
{
"epoch": 0.52,
"learning_rate": 9.821991469938325e-06,
"loss": 1.4517,
"step": 995
},
{
"epoch": 0.52,
"learning_rate": 9.805040331436672e-06,
"loss": 1.1016,
"step": 996
},
{
"epoch": 0.52,
"learning_rate": 9.788089753329157e-06,
"loss": 1.623,
"step": 997
},
{
"epoch": 0.52,
"learning_rate": 9.771139784338693e-06,
"loss": 1.3481,
"step": 998
},
{
"epoch": 0.52,
"learning_rate": 9.754190473186455e-06,
"loss": 1.4707,
"step": 999
},
{
"epoch": 0.52,
"learning_rate": 9.737241868591724e-06,
"loss": 1.5947,
"step": 1000
},
{
"epoch": 0.52,
"learning_rate": 9.720294019271744e-06,
"loss": 1.7764,
"step": 1001
},
{
"epoch": 0.52,
"learning_rate": 9.703346973941595e-06,
"loss": 1.7031,
"step": 1002
},
{
"epoch": 0.52,
"learning_rate": 9.686400781314042e-06,
"loss": 1.3486,
"step": 1003
},
{
"epoch": 0.53,
"learning_rate": 9.669455490099402e-06,
"loss": 1.2261,
"step": 1004
},
{
"epoch": 0.53,
"learning_rate": 9.652511149005396e-06,
"loss": 1.2632,
"step": 1005
},
{
"epoch": 0.53,
"learning_rate": 9.635567806737021e-06,
"loss": 1.7065,
"step": 1006
},
{
"epoch": 0.53,
"learning_rate": 9.618625511996398e-06,
"loss": 1.0562,
"step": 1007
},
{
"epoch": 0.53,
"learning_rate": 9.601684313482638e-06,
"loss": 1.4648,
"step": 1008
},
{
"epoch": 0.53,
"learning_rate": 9.584744259891706e-06,
"loss": 1.1753,
"step": 1009
},
{
"epoch": 0.53,
"learning_rate": 9.56780539991626e-06,
"loss": 1.4387,
"step": 1010
},
{
"epoch": 0.53,
"learning_rate": 9.55086778224555e-06,
"loss": 1.6699,
"step": 1011
},
{
"epoch": 0.53,
"learning_rate": 9.533931455565236e-06,
"loss": 1.543,
"step": 1012
},
{
"epoch": 0.53,
"learning_rate": 9.51699646855728e-06,
"loss": 1.4912,
"step": 1013
},
{
"epoch": 0.53,
"learning_rate": 9.50006286989978e-06,
"loss": 1.0615,
"step": 1014
},
{
"epoch": 0.53,
"learning_rate": 9.483130708266856e-06,
"loss": 1.0679,
"step": 1015
},
{
"epoch": 0.53,
"learning_rate": 9.466200032328499e-06,
"loss": 1.5146,
"step": 1016
},
{
"epoch": 0.53,
"learning_rate": 9.44927089075041e-06,
"loss": 1.1787,
"step": 1017
},
{
"epoch": 0.53,
"learning_rate": 9.432343332193907e-06,
"loss": 1.4238,
"step": 1018
},
{
"epoch": 0.53,
"learning_rate": 9.415417405315727e-06,
"loss": 1.5024,
"step": 1019
},
{
"epoch": 0.53,
"learning_rate": 9.39849315876795e-06,
"loss": 1.2,
"step": 1020
},
{
"epoch": 0.53,
"learning_rate": 9.381570641197795e-06,
"loss": 1.2314,
"step": 1021
},
{
"epoch": 0.53,
"learning_rate": 9.364649901247534e-06,
"loss": 1.3638,
"step": 1022
},
{
"epoch": 0.54,
"learning_rate": 9.34773098755432e-06,
"loss": 1.6426,
"step": 1023
},
{
"epoch": 0.54,
"learning_rate": 9.330813948750053e-06,
"loss": 1.5254,
"step": 1024
},
{
"epoch": 0.54,
"learning_rate": 9.313898833461255e-06,
"loss": 1.2178,
"step": 1025
},
{
"epoch": 0.54,
"learning_rate": 9.29698569030891e-06,
"loss": 1.3511,
"step": 1026
},
{
"epoch": 0.54,
"learning_rate": 9.28007456790833e-06,
"loss": 1.0898,
"step": 1027
},
{
"epoch": 0.54,
"learning_rate": 9.26316551486903e-06,
"loss": 1.4395,
"step": 1028
},
{
"epoch": 0.54,
"learning_rate": 9.246258579794574e-06,
"loss": 1.6514,
"step": 1029
},
{
"epoch": 0.54,
"learning_rate": 9.229353811282427e-06,
"loss": 1.1983,
"step": 1030
},
{
"epoch": 0.54,
"learning_rate": 9.212451257923841e-06,
"loss": 1.4531,
"step": 1031
},
{
"epoch": 0.54,
"learning_rate": 9.195550968303688e-06,
"loss": 1.5723,
"step": 1032
},
{
"epoch": 0.54,
"learning_rate": 9.178652991000342e-06,
"loss": 1.4722,
"step": 1033
},
{
"epoch": 0.54,
"learning_rate": 9.161757374585533e-06,
"loss": 1.499,
"step": 1034
},
{
"epoch": 0.54,
"learning_rate": 9.144864167624191e-06,
"loss": 1.1548,
"step": 1035
},
{
"epoch": 0.54,
"learning_rate": 9.127973418674338e-06,
"loss": 1.7676,
"step": 1036
},
{
"epoch": 0.54,
"learning_rate": 9.111085176286912e-06,
"loss": 1.6885,
"step": 1037
},
{
"epoch": 0.54,
"learning_rate": 9.094199489005664e-06,
"loss": 1.1021,
"step": 1038
},
{
"epoch": 0.54,
"learning_rate": 9.07731640536698e-06,
"loss": 1.1333,
"step": 1039
},
{
"epoch": 0.54,
"learning_rate": 9.06043597389979e-06,
"loss": 1.6533,
"step": 1040
},
{
"epoch": 0.54,
"learning_rate": 9.043558243125374e-06,
"loss": 1.2778,
"step": 1041
},
{
"epoch": 0.55,
"learning_rate": 9.026683261557263e-06,
"loss": 1.1177,
"step": 1042
},
{
"epoch": 0.55,
"learning_rate": 9.009811077701083e-06,
"loss": 1.6318,
"step": 1043
},
{
"epoch": 0.55,
"learning_rate": 8.992941740054418e-06,
"loss": 1.4219,
"step": 1044
},
{
"epoch": 0.55,
"learning_rate": 8.97607529710667e-06,
"loss": 1.6226,
"step": 1045
},
{
"epoch": 0.55,
"learning_rate": 8.959211797338919e-06,
"loss": 1.4028,
"step": 1046
},
{
"epoch": 0.55,
"learning_rate": 8.942351289223795e-06,
"loss": 0.9825,
"step": 1047
},
{
"epoch": 0.55,
"learning_rate": 8.925493821225309e-06,
"loss": 1.1401,
"step": 1048
},
{
"epoch": 0.55,
"learning_rate": 8.908639441798751e-06,
"loss": 1.3823,
"step": 1049
},
{
"epoch": 0.55,
"learning_rate": 8.891788199390533e-06,
"loss": 1.7295,
"step": 1050
},
{
"epoch": 0.55,
"learning_rate": 8.874940142438036e-06,
"loss": 1.4253,
"step": 1051
},
{
"epoch": 0.55,
"learning_rate": 8.858095319369499e-06,
"loss": 1.2427,
"step": 1052
},
{
"epoch": 0.55,
"learning_rate": 8.841253778603855e-06,
"loss": 1.5645,
"step": 1053
},
{
"epoch": 0.55,
"learning_rate": 8.824415568550612e-06,
"loss": 1.5098,
"step": 1054
},
{
"epoch": 0.55,
"learning_rate": 8.807580737609693e-06,
"loss": 1.5322,
"step": 1055
},
{
"epoch": 0.55,
"learning_rate": 8.790749334171323e-06,
"loss": 1.5254,
"step": 1056
},
{
"epoch": 0.55,
"learning_rate": 8.773921406615854e-06,
"loss": 1.0981,
"step": 1057
},
{
"epoch": 0.55,
"learning_rate": 8.757097003313665e-06,
"loss": 1.042,
"step": 1058
},
{
"epoch": 0.55,
"learning_rate": 8.740276172625007e-06,
"loss": 1.1118,
"step": 1059
},
{
"epoch": 0.55,
"learning_rate": 8.72345896289984e-06,
"loss": 1.0481,
"step": 1060
},
{
"epoch": 0.56,
"learning_rate": 8.706645422477739e-06,
"loss": 0.8436,
"step": 1061
},
{
"epoch": 0.56,
"learning_rate": 8.689835599687717e-06,
"loss": 1.4683,
"step": 1062
},
{
"epoch": 0.56,
"learning_rate": 8.67302954284811e-06,
"loss": 1.7275,
"step": 1063
},
{
"epoch": 0.56,
"learning_rate": 8.656227300266425e-06,
"loss": 1.7979,
"step": 1064
},
{
"epoch": 0.56,
"learning_rate": 8.639428920239205e-06,
"loss": 1.5439,
"step": 1065
},
{
"epoch": 0.56,
"learning_rate": 8.62263445105189e-06,
"loss": 1.6797,
"step": 1066
},
{
"epoch": 0.56,
"learning_rate": 8.60584394097868e-06,
"loss": 1.3809,
"step": 1067
},
{
"epoch": 0.56,
"learning_rate": 8.5890574382824e-06,
"loss": 1.6143,
"step": 1068
},
{
"epoch": 0.56,
"learning_rate": 8.572274991214345e-06,
"loss": 1.4644,
"step": 1069
},
{
"epoch": 0.56,
"learning_rate": 8.555496648014163e-06,
"loss": 1.3818,
"step": 1070
},
{
"epoch": 0.56,
"learning_rate": 8.538722456909694e-06,
"loss": 1.4023,
"step": 1071
},
{
"epoch": 0.56,
"learning_rate": 8.521952466116862e-06,
"loss": 1.3374,
"step": 1072
},
{
"epoch": 0.56,
"learning_rate": 8.505186723839492e-06,
"loss": 1.1001,
"step": 1073
},
{
"epoch": 0.56,
"learning_rate": 8.488425278269224e-06,
"loss": 1.0283,
"step": 1074
},
{
"epoch": 0.56,
"learning_rate": 8.471668177585324e-06,
"loss": 1.5352,
"step": 1075
},
{
"epoch": 0.56,
"learning_rate": 8.454915469954583e-06,
"loss": 1.1626,
"step": 1076
},
{
"epoch": 0.56,
"learning_rate": 8.438167203531166e-06,
"loss": 1.457,
"step": 1077
},
{
"epoch": 0.56,
"learning_rate": 8.42142342645646e-06,
"loss": 1.7764,
"step": 1078
},
{
"epoch": 0.56,
"learning_rate": 8.404684186858962e-06,
"loss": 1.0559,
"step": 1079
},
{
"epoch": 0.57,
"learning_rate": 8.387949532854113e-06,
"loss": 0.8921,
"step": 1080
},
{
"epoch": 0.57,
"learning_rate": 8.371219512544182e-06,
"loss": 1.4619,
"step": 1081
},
{
"epoch": 0.57,
"learning_rate": 8.354494174018118e-06,
"loss": 1.2764,
"step": 1082
},
{
"epoch": 0.57,
"learning_rate": 8.337773565351407e-06,
"loss": 1.6602,
"step": 1083
},
{
"epoch": 0.57,
"learning_rate": 8.321057734605949e-06,
"loss": 1.4209,
"step": 1084
},
{
"epoch": 0.57,
"learning_rate": 8.3043467298299e-06,
"loss": 0.9824,
"step": 1085
},
{
"epoch": 0.57,
"learning_rate": 8.287640599057551e-06,
"loss": 0.999,
"step": 1086
},
{
"epoch": 0.57,
"learning_rate": 8.270939390309181e-06,
"loss": 1.7012,
"step": 1087
},
{
"epoch": 0.57,
"learning_rate": 8.254243151590922e-06,
"loss": 1.4688,
"step": 1088
},
{
"epoch": 0.57,
"learning_rate": 8.237551930894616e-06,
"loss": 1.4844,
"step": 1089
},
{
"epoch": 0.57,
"learning_rate": 8.22086577619769e-06,
"loss": 1.5173,
"step": 1090
},
{
"epoch": 0.57,
"learning_rate": 8.204184735462994e-06,
"loss": 1.6831,
"step": 1091
},
{
"epoch": 0.57,
"learning_rate": 8.187508856638697e-06,
"loss": 1.0901,
"step": 1092
},
{
"epoch": 0.57,
"learning_rate": 8.17083818765812e-06,
"loss": 1.0029,
"step": 1093
},
{
"epoch": 0.57,
"learning_rate": 8.154172776439608e-06,
"loss": 1.1416,
"step": 1094
},
{
"epoch": 0.57,
"learning_rate": 8.137512670886397e-06,
"loss": 1.5396,
"step": 1095
},
{
"epoch": 0.57,
"learning_rate": 8.120857918886467e-06,
"loss": 1.2782,
"step": 1096
},
{
"epoch": 0.57,
"learning_rate": 8.10420856831242e-06,
"loss": 1.3452,
"step": 1097
},
{
"epoch": 0.57,
"learning_rate": 8.087564667021317e-06,
"loss": 1.4873,
"step": 1098
},
{
"epoch": 0.58,
"learning_rate": 8.070926262854573e-06,
"loss": 1.1948,
"step": 1099
},
{
"epoch": 0.58,
"learning_rate": 8.054293403637787e-06,
"loss": 1.7705,
"step": 1100
},
{
"epoch": 0.58,
"learning_rate": 8.037666137180624e-06,
"loss": 1.5262,
"step": 1101
},
{
"epoch": 0.58,
"learning_rate": 8.021044511276682e-06,
"loss": 1.5522,
"step": 1102
},
{
"epoch": 0.58,
"learning_rate": 8.00442857370333e-06,
"loss": 1.624,
"step": 1103
},
{
"epoch": 0.58,
"learning_rate": 7.987818372221603e-06,
"loss": 1.4229,
"step": 1104
},
{
"epoch": 0.58,
"learning_rate": 7.971213954576031e-06,
"loss": 0.9995,
"step": 1105
},
{
"epoch": 0.58,
"learning_rate": 7.954615368494538e-06,
"loss": 1.0386,
"step": 1106
},
{
"epoch": 0.58,
"learning_rate": 7.938022661688265e-06,
"loss": 1.2856,
"step": 1107
},
{
"epoch": 0.58,
"learning_rate": 7.921435881851474e-06,
"loss": 1.4165,
"step": 1108
},
{
"epoch": 0.58,
"learning_rate": 7.90485507666137e-06,
"loss": 1.238,
"step": 1109
},
{
"epoch": 0.58,
"learning_rate": 7.888280293778001e-06,
"loss": 1.2002,
"step": 1110
},
{
"epoch": 0.58,
"learning_rate": 7.871711580844102e-06,
"loss": 1.2422,
"step": 1111
},
{
"epoch": 0.58,
"learning_rate": 7.855148985484946e-06,
"loss": 1.0547,
"step": 1112
},
{
"epoch": 0.58,
"learning_rate": 7.838592555308244e-06,
"loss": 1.6445,
"step": 1113
},
{
"epoch": 0.58,
"learning_rate": 7.822042337903964e-06,
"loss": 1.5674,
"step": 1114
},
{
"epoch": 0.58,
"learning_rate": 7.805498380844236e-06,
"loss": 1.5283,
"step": 1115
},
{
"epoch": 0.58,
"learning_rate": 7.788960731683174e-06,
"loss": 1.6382,
"step": 1116
},
{
"epoch": 0.58,
"learning_rate": 7.772429437956777e-06,
"loss": 1.3384,
"step": 1117
},
{
"epoch": 0.59,
"learning_rate": 7.755904547182778e-06,
"loss": 1.3628,
"step": 1118
},
{
"epoch": 0.59,
"learning_rate": 7.739386106860485e-06,
"loss": 1.1421,
"step": 1119
},
{
"epoch": 0.59,
"learning_rate": 7.722874164470685e-06,
"loss": 1.4028,
"step": 1120
},
{
"epoch": 0.59,
"learning_rate": 7.70636876747548e-06,
"loss": 1.6289,
"step": 1121
},
{
"epoch": 0.59,
"learning_rate": 7.689869963318154e-06,
"loss": 1.2715,
"step": 1122
},
{
"epoch": 0.59,
"learning_rate": 7.673377799423047e-06,
"loss": 1.2466,
"step": 1123
},
{
"epoch": 0.59,
"learning_rate": 7.656892323195414e-06,
"loss": 1.0854,
"step": 1124
},
{
"epoch": 0.59,
"learning_rate": 7.64041358202127e-06,
"loss": 1.6826,
"step": 1125
},
{
"epoch": 0.59,
"learning_rate": 7.6239416232672904e-06,
"loss": 1.2422,
"step": 1126
},
{
"epoch": 0.59,
"learning_rate": 7.607476494280653e-06,
"loss": 1.6172,
"step": 1127
},
{
"epoch": 0.59,
"learning_rate": 7.591018242388888e-06,
"loss": 1.2598,
"step": 1128
},
{
"epoch": 0.59,
"learning_rate": 7.574566914899779e-06,
"loss": 1.415,
"step": 1129
},
{
"epoch": 0.59,
"learning_rate": 7.5581225591011864e-06,
"loss": 1.3315,
"step": 1130
},
{
"epoch": 0.59,
"learning_rate": 7.541685222260948e-06,
"loss": 1.8037,
"step": 1131
},
{
"epoch": 0.59,
"learning_rate": 7.525254951626712e-06,
"loss": 1.2563,
"step": 1132
},
{
"epoch": 0.59,
"learning_rate": 7.5088317944258305e-06,
"loss": 1.4981,
"step": 1133
},
{
"epoch": 0.59,
"learning_rate": 7.492415797865191e-06,
"loss": 1.3926,
"step": 1134
},
{
"epoch": 0.59,
"learning_rate": 7.476007009131112e-06,
"loss": 1.751,
"step": 1135
},
{
"epoch": 0.59,
"learning_rate": 7.459605475389193e-06,
"loss": 1.001,
"step": 1136
},
{
"epoch": 0.59,
"learning_rate": 7.44321124378417e-06,
"loss": 1.4009,
"step": 1137
},
{
"epoch": 0.6,
"learning_rate": 7.426824361439803e-06,
"loss": 1.354,
"step": 1138
},
{
"epoch": 0.6,
"learning_rate": 7.41044487545871e-06,
"loss": 1.2329,
"step": 1139
},
{
"epoch": 0.6,
"learning_rate": 7.394072832922268e-06,
"loss": 1.3438,
"step": 1140
},
{
"epoch": 0.6,
"learning_rate": 7.3777082808904435e-06,
"loss": 1.0591,
"step": 1141
},
{
"epoch": 0.6,
"learning_rate": 7.361351266401682e-06,
"loss": 1.4851,
"step": 1142
},
{
"epoch": 0.6,
"learning_rate": 7.345001836472754e-06,
"loss": 1.4868,
"step": 1143
},
{
"epoch": 0.6,
"learning_rate": 7.328660038098636e-06,
"loss": 1.4688,
"step": 1144
},
{
"epoch": 0.6,
"learning_rate": 7.312325918252373e-06,
"loss": 1.3389,
"step": 1145
},
{
"epoch": 0.6,
"learning_rate": 7.295999523884921e-06,
"loss": 1.5938,
"step": 1146
},
{
"epoch": 0.6,
"learning_rate": 7.279680901925051e-06,
"loss": 0.9379,
"step": 1147
},
{
"epoch": 0.6,
"learning_rate": 7.263370099279173e-06,
"loss": 1.4214,
"step": 1148
},
{
"epoch": 0.6,
"learning_rate": 7.247067162831237e-06,
"loss": 1.0825,
"step": 1149
},
{
"epoch": 0.6,
"learning_rate": 7.230772139442572e-06,
"loss": 1.124,
"step": 1150
},
{
"epoch": 0.6,
"learning_rate": 7.214485075951771e-06,
"loss": 1.666,
"step": 1151
},
{
"epoch": 0.6,
"learning_rate": 7.198206019174534e-06,
"loss": 0.8665,
"step": 1152
},
{
"epoch": 0.6,
"learning_rate": 7.181935015903558e-06,
"loss": 1.3301,
"step": 1153
},
{
"epoch": 0.6,
"learning_rate": 7.165672112908393e-06,
"loss": 1.5352,
"step": 1154
},
{
"epoch": 0.6,
"learning_rate": 7.149417356935289e-06,
"loss": 1.2217,
"step": 1155
},
{
"epoch": 0.6,
"learning_rate": 7.133170794707096e-06,
"loss": 1.0522,
"step": 1156
},
{
"epoch": 0.61,
"learning_rate": 7.116932472923098e-06,
"loss": 1.0388,
"step": 1157
},
{
"epoch": 0.61,
"learning_rate": 7.100702438258905e-06,
"loss": 1.3281,
"step": 1158
},
{
"epoch": 0.61,
"learning_rate": 7.084480737366297e-06,
"loss": 1.4253,
"step": 1159
},
{
"epoch": 0.61,
"learning_rate": 7.068267416873102e-06,
"loss": 1.5098,
"step": 1160
},
{
"epoch": 0.61,
"learning_rate": 7.052062523383065e-06,
"loss": 1.2988,
"step": 1161
},
{
"epoch": 0.61,
"learning_rate": 7.035866103475698e-06,
"loss": 1.4795,
"step": 1162
},
{
"epoch": 0.61,
"learning_rate": 7.019678203706164e-06,
"loss": 1.707,
"step": 1163
},
{
"epoch": 0.61,
"learning_rate": 7.0034988706051325e-06,
"loss": 1.6602,
"step": 1164
},
{
"epoch": 0.61,
"learning_rate": 6.987328150678652e-06,
"loss": 1.4771,
"step": 1165
},
{
"epoch": 0.61,
"learning_rate": 6.9711660904080055e-06,
"loss": 1.0928,
"step": 1166
},
{
"epoch": 0.61,
"learning_rate": 6.955012736249598e-06,
"loss": 1.4332,
"step": 1167
},
{
"epoch": 0.61,
"learning_rate": 6.938868134634793e-06,
"loss": 1.2715,
"step": 1168
},
{
"epoch": 0.61,
"learning_rate": 6.922732331969808e-06,
"loss": 1.0107,
"step": 1169
},
{
"epoch": 0.61,
"learning_rate": 6.90660537463557e-06,
"loss": 1.5151,
"step": 1170
},
{
"epoch": 0.61,
"learning_rate": 6.890487308987567e-06,
"loss": 0.9897,
"step": 1171
},
{
"epoch": 0.61,
"learning_rate": 6.8743781813557455e-06,
"loss": 1.1364,
"step": 1172
},
{
"epoch": 0.61,
"learning_rate": 6.858278038044344e-06,
"loss": 1.6895,
"step": 1173
},
{
"epoch": 0.61,
"learning_rate": 6.842186925331793e-06,
"loss": 1.6377,
"step": 1174
},
{
"epoch": 0.61,
"learning_rate": 6.8261048894705505e-06,
"loss": 1.1216,
"step": 1175
},
{
"epoch": 0.62,
"learning_rate": 6.810031976686994e-06,
"loss": 1.2217,
"step": 1176
},
{
"epoch": 0.62,
"learning_rate": 6.793968233181274e-06,
"loss": 1.6836,
"step": 1177
},
{
"epoch": 0.62,
"learning_rate": 6.777913705127181e-06,
"loss": 1.7158,
"step": 1178
},
{
"epoch": 0.62,
"learning_rate": 6.761868438672028e-06,
"loss": 1.5767,
"step": 1179
},
{
"epoch": 0.62,
"learning_rate": 6.745832479936492e-06,
"loss": 1.1069,
"step": 1180
},
{
"epoch": 0.62,
"learning_rate": 6.729805875014506e-06,
"loss": 1.3071,
"step": 1181
},
{
"epoch": 0.62,
"learning_rate": 6.713788669973108e-06,
"loss": 1.6152,
"step": 1182
},
{
"epoch": 0.62,
"learning_rate": 6.6977809108523295e-06,
"loss": 1.2695,
"step": 1183
},
{
"epoch": 0.62,
"learning_rate": 6.6817826436650335e-06,
"loss": 1.1567,
"step": 1184
},
{
"epoch": 0.62,
"learning_rate": 6.665793914396818e-06,
"loss": 1.0889,
"step": 1185
},
{
"epoch": 0.62,
"learning_rate": 6.649814769005847e-06,
"loss": 1.4072,
"step": 1186
},
{
"epoch": 0.62,
"learning_rate": 6.633845253422748e-06,
"loss": 1.4688,
"step": 1187
},
{
"epoch": 0.62,
"learning_rate": 6.617885413550471e-06,
"loss": 1.5273,
"step": 1188
},
{
"epoch": 0.62,
"learning_rate": 6.601935295264139e-06,
"loss": 1.3323,
"step": 1189
},
{
"epoch": 0.62,
"learning_rate": 6.5859949444109495e-06,
"loss": 1.4219,
"step": 1190
},
{
"epoch": 0.62,
"learning_rate": 6.570064406810008e-06,
"loss": 1.3032,
"step": 1191
},
{
"epoch": 0.62,
"learning_rate": 6.554143728252229e-06,
"loss": 1.0376,
"step": 1192
},
{
"epoch": 0.62,
"learning_rate": 6.53823295450017e-06,
"loss": 1.6396,
"step": 1193
},
{
"epoch": 0.62,
"learning_rate": 6.522332131287931e-06,
"loss": 1.4668,
"step": 1194
},
{
"epoch": 0.63,
"learning_rate": 6.506441304321012e-06,
"loss": 0.9971,
"step": 1195
},
{
"epoch": 0.63,
"learning_rate": 6.4905605192761656e-06,
"loss": 1.0859,
"step": 1196
},
{
"epoch": 0.63,
"learning_rate": 6.474689821801295e-06,
"loss": 1.5293,
"step": 1197
},
{
"epoch": 0.63,
"learning_rate": 6.4588292575152955e-06,
"loss": 1.3582,
"step": 1198
},
{
"epoch": 0.63,
"learning_rate": 6.442978872007946e-06,
"loss": 0.8848,
"step": 1199
},
{
"epoch": 0.63,
"learning_rate": 6.42713871083976e-06,
"loss": 1.6592,
"step": 1200
},
{
"epoch": 0.63,
"learning_rate": 6.411308819541863e-06,
"loss": 0.9512,
"step": 1201
},
{
"epoch": 0.63,
"learning_rate": 6.395489243615861e-06,
"loss": 1.3691,
"step": 1202
},
{
"epoch": 0.63,
"learning_rate": 6.379680028533714e-06,
"loss": 1.6475,
"step": 1203
},
{
"epoch": 0.63,
"learning_rate": 6.363881219737599e-06,
"loss": 1.1069,
"step": 1204
},
{
"epoch": 0.63,
"learning_rate": 6.348092862639772e-06,
"loss": 1.0828,
"step": 1205
},
{
"epoch": 0.63,
"learning_rate": 6.332315002622461e-06,
"loss": 1.2798,
"step": 1206
},
{
"epoch": 0.63,
"learning_rate": 6.316547685037707e-06,
"loss": 1.0975,
"step": 1207
},
{
"epoch": 0.63,
"learning_rate": 6.300790955207262e-06,
"loss": 1.3804,
"step": 1208
},
{
"epoch": 0.63,
"learning_rate": 6.285044858422429e-06,
"loss": 1.459,
"step": 1209
},
{
"epoch": 0.63,
"learning_rate": 6.269309439943963e-06,
"loss": 0.998,
"step": 1210
},
{
"epoch": 0.63,
"learning_rate": 6.253584745001908e-06,
"loss": 1.1904,
"step": 1211
},
{
"epoch": 0.63,
"learning_rate": 6.237870818795499e-06,
"loss": 1.3691,
"step": 1212
},
{
"epoch": 0.63,
"learning_rate": 6.222167706493016e-06,
"loss": 1.2441,
"step": 1213
},
{
"epoch": 0.64,
"learning_rate": 6.206475453231644e-06,
"loss": 1.6738,
"step": 1214
},
{
"epoch": 0.64,
"learning_rate": 6.190794104117368e-06,
"loss": 1.1246,
"step": 1215
},
{
"epoch": 0.64,
"learning_rate": 6.175123704224814e-06,
"loss": 1.1133,
"step": 1216
},
{
"epoch": 0.64,
"learning_rate": 6.1594642985971565e-06,
"loss": 1.4961,
"step": 1217
},
{
"epoch": 0.64,
"learning_rate": 6.143815932245951e-06,
"loss": 1.0352,
"step": 1218
},
{
"epoch": 0.64,
"learning_rate": 6.128178650151026e-06,
"loss": 1.4941,
"step": 1219
},
{
"epoch": 0.64,
"learning_rate": 6.112552497260353e-06,
"loss": 1.438,
"step": 1220
},
{
"epoch": 0.64,
"learning_rate": 6.096937518489912e-06,
"loss": 1.0801,
"step": 1221
},
{
"epoch": 0.64,
"learning_rate": 6.081333758723562e-06,
"loss": 1.0972,
"step": 1222
},
{
"epoch": 0.64,
"learning_rate": 6.065741262812914e-06,
"loss": 1.6309,
"step": 1223
},
{
"epoch": 0.64,
"learning_rate": 6.0501600755772074e-06,
"loss": 1.4404,
"step": 1224
},
{
"epoch": 0.64,
"learning_rate": 6.0345902418031645e-06,
"loss": 1.4258,
"step": 1225
},
{
"epoch": 0.64,
"learning_rate": 6.019031806244888e-06,
"loss": 1.5029,
"step": 1226
},
{
"epoch": 0.64,
"learning_rate": 6.0034848136237035e-06,
"loss": 1.5479,
"step": 1227
},
{
"epoch": 0.64,
"learning_rate": 5.987949308628054e-06,
"loss": 1.6562,
"step": 1228
},
{
"epoch": 0.64,
"learning_rate": 5.9724253359133614e-06,
"loss": 0.9883,
"step": 1229
},
{
"epoch": 0.64,
"learning_rate": 5.956912940101891e-06,
"loss": 1.1201,
"step": 1230
},
{
"epoch": 0.64,
"learning_rate": 5.941412165782645e-06,
"loss": 1.021,
"step": 1231
},
{
"epoch": 0.64,
"learning_rate": 5.925923057511207e-06,
"loss": 1.6016,
"step": 1232
},
{
"epoch": 0.65,
"learning_rate": 5.910445659809638e-06,
"loss": 1.4399,
"step": 1233
},
{
"epoch": 0.65,
"learning_rate": 5.894980017166329e-06,
"loss": 0.9529,
"step": 1234
},
{
"epoch": 0.65,
"learning_rate": 5.87952617403589e-06,
"loss": 1.542,
"step": 1235
},
{
"epoch": 0.65,
"learning_rate": 5.864084174839013e-06,
"loss": 1.6528,
"step": 1236
},
{
"epoch": 0.65,
"learning_rate": 5.848654063962338e-06,
"loss": 0.9795,
"step": 1237
},
{
"epoch": 0.65,
"learning_rate": 5.833235885758346e-06,
"loss": 1.3682,
"step": 1238
},
{
"epoch": 0.65,
"learning_rate": 5.817829684545204e-06,
"loss": 1.4277,
"step": 1239
},
{
"epoch": 0.65,
"learning_rate": 5.80243550460667e-06,
"loss": 1.1196,
"step": 1240
},
{
"epoch": 0.65,
"learning_rate": 5.787053390191925e-06,
"loss": 1.3594,
"step": 1241
},
{
"epoch": 0.65,
"learning_rate": 5.771683385515495e-06,
"loss": 0.9307,
"step": 1242
},
{
"epoch": 0.65,
"learning_rate": 5.75632553475707e-06,
"loss": 1.4297,
"step": 1243
},
{
"epoch": 0.65,
"learning_rate": 5.740979882061433e-06,
"loss": 1.3069,
"step": 1244
},
{
"epoch": 0.65,
"learning_rate": 5.7256464715382754e-06,
"loss": 1.5448,
"step": 1245
},
{
"epoch": 0.65,
"learning_rate": 5.710325347262125e-06,
"loss": 0.9534,
"step": 1246
},
{
"epoch": 0.65,
"learning_rate": 5.695016553272182e-06,
"loss": 1.583,
"step": 1247
},
{
"epoch": 0.65,
"learning_rate": 5.6797201335722064e-06,
"loss": 1.4493,
"step": 1248
},
{
"epoch": 0.65,
"learning_rate": 5.664436132130384e-06,
"loss": 1.3965,
"step": 1249
},
{
"epoch": 0.65,
"learning_rate": 5.649164592879213e-06,
"loss": 1.39,
"step": 1250
},
{
"epoch": 0.65,
"learning_rate": 5.6339055597153734e-06,
"loss": 1.8262,
"step": 1251
},
{
"epoch": 0.66,
"learning_rate": 5.618659076499578e-06,
"loss": 1.2109,
"step": 1252
},
{
"epoch": 0.66,
"learning_rate": 5.603425187056496e-06,
"loss": 1.5771,
"step": 1253
},
{
"epoch": 0.66,
"learning_rate": 5.588203935174563e-06,
"loss": 1.5557,
"step": 1254
},
{
"epoch": 0.66,
"learning_rate": 5.572995364605921e-06,
"loss": 1.125,
"step": 1255
},
{
"epoch": 0.66,
"learning_rate": 5.55779951906624e-06,
"loss": 1.2627,
"step": 1256
},
{
"epoch": 0.66,
"learning_rate": 5.542616442234618e-06,
"loss": 1.6387,
"step": 1257
},
{
"epoch": 0.66,
"learning_rate": 5.527446177753455e-06,
"loss": 1.5566,
"step": 1258
},
{
"epoch": 0.66,
"learning_rate": 5.512288769228318e-06,
"loss": 1.0342,
"step": 1259
},
{
"epoch": 0.66,
"learning_rate": 5.49714426022782e-06,
"loss": 1.5405,
"step": 1260
},
{
"epoch": 0.66,
"learning_rate": 5.482012694283503e-06,
"loss": 1.2358,
"step": 1261
},
{
"epoch": 0.66,
"learning_rate": 5.466894114889697e-06,
"loss": 1.6763,
"step": 1262
},
{
"epoch": 0.66,
"learning_rate": 5.451788565503406e-06,
"loss": 1.2627,
"step": 1263
},
{
"epoch": 0.66,
"learning_rate": 5.43669608954418e-06,
"loss": 1.4429,
"step": 1264
},
{
"epoch": 0.66,
"learning_rate": 5.421616730394e-06,
"loss": 0.9888,
"step": 1265
},
{
"epoch": 0.66,
"learning_rate": 5.406550531397123e-06,
"loss": 1.2554,
"step": 1266
},
{
"epoch": 0.66,
"learning_rate": 5.391497535859999e-06,
"loss": 1.5898,
"step": 1267
},
{
"epoch": 0.66,
"learning_rate": 5.376457787051117e-06,
"loss": 0.9497,
"step": 1268
},
{
"epoch": 0.66,
"learning_rate": 5.361431328200888e-06,
"loss": 1.3229,
"step": 1269
},
{
"epoch": 0.66,
"learning_rate": 5.346418202501526e-06,
"loss": 1.0796,
"step": 1270
},
{
"epoch": 0.67,
"learning_rate": 5.331418453106916e-06,
"loss": 1.2529,
"step": 1271
},
{
"epoch": 0.67,
"learning_rate": 5.316432123132498e-06,
"loss": 1.5146,
"step": 1272
},
{
"epoch": 0.67,
"learning_rate": 5.301459255655134e-06,
"loss": 1.4546,
"step": 1273
},
{
"epoch": 0.67,
"learning_rate": 5.286499893713003e-06,
"loss": 1.5205,
"step": 1274
},
{
"epoch": 0.67,
"learning_rate": 5.271554080305439e-06,
"loss": 1.5762,
"step": 1275
},
{
"epoch": 0.67,
"learning_rate": 5.256621858392861e-06,
"loss": 1.4277,
"step": 1276
},
{
"epoch": 0.67,
"learning_rate": 5.241703270896589e-06,
"loss": 1.6465,
"step": 1277
},
{
"epoch": 0.67,
"learning_rate": 5.226798360698786e-06,
"loss": 1.6914,
"step": 1278
},
{
"epoch": 0.67,
"learning_rate": 5.211907170642267e-06,
"loss": 1.5449,
"step": 1279
},
{
"epoch": 0.67,
"learning_rate": 5.1970297435304395e-06,
"loss": 1.3711,
"step": 1280
},
{
"epoch": 0.67,
"learning_rate": 5.18216612212713e-06,
"loss": 1.4985,
"step": 1281
},
{
"epoch": 0.67,
"learning_rate": 5.167316349156495e-06,
"loss": 0.9785,
"step": 1282
},
{
"epoch": 0.67,
"learning_rate": 5.1524804673028725e-06,
"loss": 1.5967,
"step": 1283
},
{
"epoch": 0.67,
"learning_rate": 5.1376585192106824e-06,
"loss": 1.4902,
"step": 1284
},
{
"epoch": 0.67,
"learning_rate": 5.1228505474842885e-06,
"loss": 1.6733,
"step": 1285
},
{
"epoch": 0.67,
"learning_rate": 5.10805659468788e-06,
"loss": 1.6499,
"step": 1286
},
{
"epoch": 0.67,
"learning_rate": 5.093276703345351e-06,
"loss": 1.2339,
"step": 1287
},
{
"epoch": 0.67,
"learning_rate": 5.078510915940178e-06,
"loss": 1.7031,
"step": 1288
},
{
"epoch": 0.67,
"learning_rate": 5.063759274915299e-06,
"loss": 1.4893,
"step": 1289
},
{
"epoch": 0.68,
"learning_rate": 5.0490218226729855e-06,
"loss": 1.0605,
"step": 1290
},
{
"epoch": 0.68,
"learning_rate": 5.034298601574726e-06,
"loss": 1.5371,
"step": 1291
},
{
"epoch": 0.68,
"learning_rate": 5.019589653941103e-06,
"loss": 0.8093,
"step": 1292
},
{
"epoch": 0.68,
"learning_rate": 5.0048950220516725e-06,
"loss": 1.408,
"step": 1293
},
{
"epoch": 0.68,
"learning_rate": 4.990214748144837e-06,
"loss": 1.0444,
"step": 1294
},
{
"epoch": 0.68,
"learning_rate": 4.975548874417733e-06,
"loss": 1.3027,
"step": 1295
},
{
"epoch": 0.68,
"learning_rate": 4.9608974430261025e-06,
"loss": 1.4287,
"step": 1296
},
{
"epoch": 0.68,
"learning_rate": 4.946260496084175e-06,
"loss": 1.3066,
"step": 1297
},
{
"epoch": 0.68,
"learning_rate": 4.931638075664544e-06,
"loss": 1.3317,
"step": 1298
},
{
"epoch": 0.68,
"learning_rate": 4.917030223798057e-06,
"loss": 0.9253,
"step": 1299
},
{
"epoch": 0.68,
"learning_rate": 4.9024369824736654e-06,
"loss": 1.3726,
"step": 1300
},
{
"epoch": 0.68,
"learning_rate": 4.88785839363835e-06,
"loss": 1.1211,
"step": 1301
},
{
"epoch": 0.68,
"learning_rate": 4.873294499196946e-06,
"loss": 1.4985,
"step": 1302
},
{
"epoch": 0.68,
"learning_rate": 4.85874534101208e-06,
"loss": 0.9644,
"step": 1303
},
{
"epoch": 0.68,
"learning_rate": 4.844210960903998e-06,
"loss": 1.3154,
"step": 1304
},
{
"epoch": 0.68,
"learning_rate": 4.829691400650478e-06,
"loss": 1.4136,
"step": 1305
},
{
"epoch": 0.68,
"learning_rate": 4.8151867019866985e-06,
"loss": 1.207,
"step": 1306
},
{
"epoch": 0.68,
"learning_rate": 4.800696906605117e-06,
"loss": 1.0835,
"step": 1307
},
{
"epoch": 0.68,
"learning_rate": 4.786222056155356e-06,
"loss": 1.6274,
"step": 1308
},
{
"epoch": 0.68,
"learning_rate": 4.771762192244075e-06,
"loss": 1.5713,
"step": 1309
},
{
"epoch": 0.69,
"learning_rate": 4.757317356434869e-06,
"loss": 0.8888,
"step": 1310
},
{
"epoch": 0.69,
"learning_rate": 4.742887590248114e-06,
"loss": 1.4561,
"step": 1311
},
{
"epoch": 0.69,
"learning_rate": 4.7284729351608985e-06,
"loss": 0.9648,
"step": 1312
},
{
"epoch": 0.69,
"learning_rate": 4.714073432606841e-06,
"loss": 1.4858,
"step": 1313
},
{
"epoch": 0.69,
"learning_rate": 4.699689123976039e-06,
"loss": 1.2949,
"step": 1314
},
{
"epoch": 0.69,
"learning_rate": 4.6853200506148955e-06,
"loss": 0.8962,
"step": 1315
},
{
"epoch": 0.69,
"learning_rate": 4.670966253826027e-06,
"loss": 1.6348,
"step": 1316
},
{
"epoch": 0.69,
"learning_rate": 4.65662777486814e-06,
"loss": 1.0588,
"step": 1317
},
{
"epoch": 0.69,
"learning_rate": 4.6423046549559095e-06,
"loss": 1.1338,
"step": 1318
},
{
"epoch": 0.69,
"learning_rate": 4.627996935259862e-06,
"loss": 1.3716,
"step": 1319
},
{
"epoch": 0.69,
"learning_rate": 4.6137046569062595e-06,
"loss": 1.4316,
"step": 1320
},
{
"epoch": 0.69,
"learning_rate": 4.599427860976978e-06,
"loss": 1.6514,
"step": 1321
},
{
"epoch": 0.69,
"learning_rate": 4.585166588509391e-06,
"loss": 0.8566,
"step": 1322
},
{
"epoch": 0.69,
"learning_rate": 4.570920880496247e-06,
"loss": 1.5576,
"step": 1323
},
{
"epoch": 0.69,
"learning_rate": 4.556690777885571e-06,
"loss": 1.6045,
"step": 1324
},
{
"epoch": 0.69,
"learning_rate": 4.542476321580505e-06,
"loss": 1.4712,
"step": 1325
},
{
"epoch": 0.69,
"learning_rate": 4.528277552439244e-06,
"loss": 1.3672,
"step": 1326
},
{
"epoch": 0.69,
"learning_rate": 4.514094511274877e-06,
"loss": 0.9143,
"step": 1327
},
{
"epoch": 0.69,
"learning_rate": 4.499927238855287e-06,
"loss": 1.521,
"step": 1328
},
{
"epoch": 0.7,
"learning_rate": 4.485775775903033e-06,
"loss": 1.335,
"step": 1329
},
{
"epoch": 0.7,
"learning_rate": 4.471640163095229e-06,
"loss": 0.9832,
"step": 1330
},
{
"epoch": 0.7,
"learning_rate": 4.457520441063428e-06,
"loss": 1.2444,
"step": 1331
},
{
"epoch": 0.7,
"learning_rate": 4.443416650393509e-06,
"loss": 1.1521,
"step": 1332
},
{
"epoch": 0.7,
"learning_rate": 4.429328831625565e-06,
"loss": 1.3042,
"step": 1333
},
{
"epoch": 0.7,
"learning_rate": 4.415257025253758e-06,
"loss": 0.9712,
"step": 1334
},
{
"epoch": 0.7,
"learning_rate": 4.40120127172625e-06,
"loss": 1.1553,
"step": 1335
},
{
"epoch": 0.7,
"learning_rate": 4.387161611445034e-06,
"loss": 1.4805,
"step": 1336
},
{
"epoch": 0.7,
"learning_rate": 4.373138084765875e-06,
"loss": 1.7056,
"step": 1337
},
{
"epoch": 0.7,
"learning_rate": 4.359130731998127e-06,
"loss": 1.8389,
"step": 1338
},
{
"epoch": 0.7,
"learning_rate": 4.345139593404688e-06,
"loss": 1.2964,
"step": 1339
},
{
"epoch": 0.7,
"learning_rate": 4.331164709201829e-06,
"loss": 1.0322,
"step": 1340
},
{
"epoch": 0.7,
"learning_rate": 4.317206119559108e-06,
"loss": 1.252,
"step": 1341
},
{
"epoch": 0.7,
"learning_rate": 4.303263864599239e-06,
"loss": 1.1704,
"step": 1342
},
{
"epoch": 0.7,
"learning_rate": 4.289337984397988e-06,
"loss": 1.0161,
"step": 1343
},
{
"epoch": 0.7,
"learning_rate": 4.275428518984054e-06,
"loss": 1.1133,
"step": 1344
},
{
"epoch": 0.7,
"learning_rate": 4.261535508338951e-06,
"loss": 1.377,
"step": 1345
},
{
"epoch": 0.7,
"learning_rate": 4.2476589923968946e-06,
"loss": 1.5605,
"step": 1346
},
{
"epoch": 0.7,
"learning_rate": 4.233799011044687e-06,
"loss": 1.7144,
"step": 1347
},
{
"epoch": 0.71,
"learning_rate": 4.219955604121611e-06,
"loss": 1.4209,
"step": 1348
},
{
"epoch": 0.71,
"learning_rate": 4.206128811419299e-06,
"loss": 1.4873,
"step": 1349
},
{
"epoch": 0.71,
"learning_rate": 4.192318672681631e-06,
"loss": 1.293,
"step": 1350
},
{
"epoch": 0.71,
"learning_rate": 4.178525227604615e-06,
"loss": 1.105,
"step": 1351
},
{
"epoch": 0.71,
"learning_rate": 4.164748515836276e-06,
"loss": 1.5015,
"step": 1352
},
{
"epoch": 0.71,
"learning_rate": 4.150988576976539e-06,
"loss": 1.7388,
"step": 1353
},
{
"epoch": 0.71,
"learning_rate": 4.1372454505771186e-06,
"loss": 1.6494,
"step": 1354
},
{
"epoch": 0.71,
"learning_rate": 4.1235191761414025e-06,
"loss": 1.1504,
"step": 1355
},
{
"epoch": 0.71,
"learning_rate": 4.10980979312434e-06,
"loss": 1.6826,
"step": 1356
},
{
"epoch": 0.71,
"learning_rate": 4.096117340932322e-06,
"loss": 1.0933,
"step": 1357
},
{
"epoch": 0.71,
"learning_rate": 4.082441858923088e-06,
"loss": 1.0,
"step": 1358
},
{
"epoch": 0.71,
"learning_rate": 4.068783386405577e-06,
"loss": 1.3618,
"step": 1359
},
{
"epoch": 0.71,
"learning_rate": 4.055141962639858e-06,
"loss": 1.5786,
"step": 1360
},
{
"epoch": 0.71,
"learning_rate": 4.0415176268369714e-06,
"loss": 1.4453,
"step": 1361
},
{
"epoch": 0.71,
"learning_rate": 4.027910418158861e-06,
"loss": 1.3398,
"step": 1362
},
{
"epoch": 0.71,
"learning_rate": 4.01432037571823e-06,
"loss": 1.1853,
"step": 1363
},
{
"epoch": 0.71,
"learning_rate": 4.00074753857844e-06,
"loss": 1.2422,
"step": 1364
},
{
"epoch": 0.71,
"learning_rate": 3.987191945753396e-06,
"loss": 1.0767,
"step": 1365
},
{
"epoch": 0.71,
"learning_rate": 3.973653636207437e-06,
"loss": 1.0293,
"step": 1366
},
{
"epoch": 0.72,
"learning_rate": 3.960132648855226e-06,
"loss": 1.0918,
"step": 1367
},
{
"epoch": 0.72,
"learning_rate": 3.946629022561627e-06,
"loss": 1.5762,
"step": 1368
},
{
"epoch": 0.72,
"learning_rate": 3.933142796141616e-06,
"loss": 1.3801,
"step": 1369
},
{
"epoch": 0.72,
"learning_rate": 3.919674008360133e-06,
"loss": 1.2039,
"step": 1370
},
{
"epoch": 0.72,
"learning_rate": 3.906222697932016e-06,
"loss": 1.5615,
"step": 1371
},
{
"epoch": 0.72,
"learning_rate": 3.89278890352184e-06,
"loss": 0.9858,
"step": 1372
},
{
"epoch": 0.72,
"learning_rate": 3.879372663743862e-06,
"loss": 1.0029,
"step": 1373
},
{
"epoch": 0.72,
"learning_rate": 3.8659740171618485e-06,
"loss": 1.5,
"step": 1374
},
{
"epoch": 0.72,
"learning_rate": 3.85259300228902e-06,
"loss": 1.2031,
"step": 1375
},
{
"epoch": 0.72,
"learning_rate": 3.839229657587906e-06,
"loss": 1.0615,
"step": 1376
},
{
"epoch": 0.72,
"learning_rate": 3.825884021470244e-06,
"loss": 1.5557,
"step": 1377
},
{
"epoch": 0.72,
"learning_rate": 3.812556132296873e-06,
"loss": 1.6191,
"step": 1378
},
{
"epoch": 0.72,
"learning_rate": 3.799246028377618e-06,
"loss": 1.0586,
"step": 1379
},
{
"epoch": 0.72,
"learning_rate": 3.785953747971183e-06,
"loss": 1.4219,
"step": 1380
},
{
"epoch": 0.72,
"learning_rate": 3.7726793292850382e-06,
"loss": 1.6426,
"step": 1381
},
{
"epoch": 0.72,
"learning_rate": 3.7594228104753107e-06,
"loss": 1.0276,
"step": 1382
},
{
"epoch": 0.72,
"learning_rate": 3.746184229646689e-06,
"loss": 1.3828,
"step": 1383
},
{
"epoch": 0.72,
"learning_rate": 3.732963624852275e-06,
"loss": 1.3428,
"step": 1384
},
{
"epoch": 0.72,
"learning_rate": 3.7197610340935263e-06,
"loss": 1.5903,
"step": 1385
},
{
"epoch": 0.73,
"learning_rate": 3.7065764953201066e-06,
"loss": 1.3765,
"step": 1386
},
{
"epoch": 0.73,
"learning_rate": 3.693410046429793e-06,
"loss": 1.2174,
"step": 1387
},
{
"epoch": 0.73,
"learning_rate": 3.680261725268364e-06,
"loss": 0.9263,
"step": 1388
},
{
"epoch": 0.73,
"learning_rate": 3.667131569629495e-06,
"loss": 1.2881,
"step": 1389
},
{
"epoch": 0.73,
"learning_rate": 3.654019617254644e-06,
"loss": 1.6083,
"step": 1390
},
{
"epoch": 0.73,
"learning_rate": 3.6409259058329404e-06,
"loss": 1.2178,
"step": 1391
},
{
"epoch": 0.73,
"learning_rate": 3.6278504730011e-06,
"loss": 0.9836,
"step": 1392
},
{
"epoch": 0.73,
"learning_rate": 3.614793356343269e-06,
"loss": 0.9719,
"step": 1393
},
{
"epoch": 0.73,
"learning_rate": 3.601754593390978e-06,
"loss": 1.5381,
"step": 1394
},
{
"epoch": 0.73,
"learning_rate": 3.5887342216229694e-06,
"loss": 0.958,
"step": 1395
},
{
"epoch": 0.73,
"learning_rate": 3.5757322784651506e-06,
"loss": 0.9895,
"step": 1396
},
{
"epoch": 0.73,
"learning_rate": 3.5627488012904333e-06,
"loss": 1.1216,
"step": 1397
},
{
"epoch": 0.73,
"learning_rate": 3.5497838274186737e-06,
"loss": 1.1196,
"step": 1398
},
{
"epoch": 0.73,
"learning_rate": 3.5368373941165167e-06,
"loss": 1.4614,
"step": 1399
},
{
"epoch": 0.73,
"learning_rate": 3.5239095385973377e-06,
"loss": 1.6934,
"step": 1400
},
{
"epoch": 0.73,
"learning_rate": 3.511000298021098e-06,
"loss": 1.0481,
"step": 1401
},
{
"epoch": 0.73,
"learning_rate": 3.498109709494254e-06,
"loss": 1.4438,
"step": 1402
},
{
"epoch": 0.73,
"learning_rate": 3.4852378100696505e-06,
"loss": 1.5527,
"step": 1403
},
{
"epoch": 0.73,
"learning_rate": 3.472384636746412e-06,
"loss": 1.1362,
"step": 1404
},
{
"epoch": 0.74,
"learning_rate": 3.4595502264698356e-06,
"loss": 0.8755,
"step": 1405
},
{
"epoch": 0.74,
"learning_rate": 3.446734616131282e-06,
"loss": 1.4756,
"step": 1406
},
{
"epoch": 0.74,
"learning_rate": 3.433937842568088e-06,
"loss": 1.0354,
"step": 1407
},
{
"epoch": 0.74,
"learning_rate": 3.421159942563421e-06,
"loss": 1.5942,
"step": 1408
},
{
"epoch": 0.74,
"learning_rate": 3.4084009528462227e-06,
"loss": 1.2891,
"step": 1409
},
{
"epoch": 0.74,
"learning_rate": 3.3956609100910643e-06,
"loss": 1.5771,
"step": 1410
},
{
"epoch": 0.74,
"learning_rate": 3.382939850918059e-06,
"loss": 1.1787,
"step": 1411
},
{
"epoch": 0.74,
"learning_rate": 3.3702378118927537e-06,
"loss": 1.5923,
"step": 1412
},
{
"epoch": 0.74,
"learning_rate": 3.357554829526025e-06,
"loss": 1.5088,
"step": 1413
},
{
"epoch": 0.74,
"learning_rate": 3.3448909402739716e-06,
"loss": 1.1001,
"step": 1414
},
{
"epoch": 0.74,
"learning_rate": 3.3322461805378094e-06,
"loss": 1.7988,
"step": 1415
},
{
"epoch": 0.74,
"learning_rate": 3.3196205866637667e-06,
"loss": 1.374,
"step": 1416
},
{
"epoch": 0.74,
"learning_rate": 3.3070141949429945e-06,
"loss": 1.438,
"step": 1417
},
{
"epoch": 0.74,
"learning_rate": 3.2944270416114256e-06,
"loss": 0.9634,
"step": 1418
},
{
"epoch": 0.74,
"learning_rate": 3.28185916284972e-06,
"loss": 1.7334,
"step": 1419
},
{
"epoch": 0.74,
"learning_rate": 3.269310594783106e-06,
"loss": 1.5957,
"step": 1420
},
{
"epoch": 0.74,
"learning_rate": 3.2567813734813336e-06,
"loss": 1.6094,
"step": 1421
},
{
"epoch": 0.74,
"learning_rate": 3.244271534958523e-06,
"loss": 1.4438,
"step": 1422
},
{
"epoch": 0.74,
"learning_rate": 3.231781115173088e-06,
"loss": 1.6509,
"step": 1423
},
{
"epoch": 0.75,
"learning_rate": 3.2193101500276226e-06,
"loss": 1.6836,
"step": 1424
},
{
"epoch": 0.75,
"learning_rate": 3.2068586753688014e-06,
"loss": 0.8369,
"step": 1425
},
{
"epoch": 0.75,
"learning_rate": 3.1944267269872754e-06,
"loss": 1.7139,
"step": 1426
},
{
"epoch": 0.75,
"learning_rate": 3.1820143406175642e-06,
"loss": 1.3462,
"step": 1427
},
{
"epoch": 0.75,
"learning_rate": 3.1696215519379726e-06,
"loss": 1.3364,
"step": 1428
},
{
"epoch": 0.75,
"learning_rate": 3.1572483965704503e-06,
"loss": 1.5947,
"step": 1429
},
{
"epoch": 0.75,
"learning_rate": 3.1448949100805394e-06,
"loss": 1.5063,
"step": 1430
},
{
"epoch": 0.75,
"learning_rate": 3.1325611279772194e-06,
"loss": 1.2188,
"step": 1431
},
{
"epoch": 0.75,
"learning_rate": 3.1202470857128564e-06,
"loss": 1.3438,
"step": 1432
},
{
"epoch": 0.75,
"learning_rate": 3.107952818683052e-06,
"loss": 1.3848,
"step": 1433
},
{
"epoch": 0.75,
"learning_rate": 3.095678362226585e-06,
"loss": 1.1208,
"step": 1434
},
{
"epoch": 0.75,
"learning_rate": 3.0834237516252817e-06,
"loss": 1.4932,
"step": 1435
},
{
"epoch": 0.75,
"learning_rate": 3.071189022103923e-06,
"loss": 1.4446,
"step": 1436
},
{
"epoch": 0.75,
"learning_rate": 3.058974208830147e-06,
"loss": 1.2275,
"step": 1437
},
{
"epoch": 0.75,
"learning_rate": 3.046779346914339e-06,
"loss": 1.1867,
"step": 1438
},
{
"epoch": 0.75,
"learning_rate": 3.0346044714095425e-06,
"loss": 1.5039,
"step": 1439
},
{
"epoch": 0.75,
"learning_rate": 3.022449617311346e-06,
"loss": 0.9829,
"step": 1440
},
{
"epoch": 0.75,
"learning_rate": 3.0103148195577914e-06,
"loss": 1.5352,
"step": 1441
},
{
"epoch": 0.75,
"learning_rate": 2.9982001130292703e-06,
"loss": 1.1475,
"step": 1442
},
{
"epoch": 0.76,
"learning_rate": 2.98610553254842e-06,
"loss": 0.9854,
"step": 1443
},
{
"epoch": 0.76,
"learning_rate": 2.974031112880037e-06,
"loss": 1.377,
"step": 1444
},
{
"epoch": 0.76,
"learning_rate": 2.961976888730956e-06,
"loss": 1.5234,
"step": 1445
},
{
"epoch": 0.76,
"learning_rate": 2.9499428947499696e-06,
"loss": 1.2074,
"step": 1446
},
{
"epoch": 0.76,
"learning_rate": 2.9379291655277143e-06,
"loss": 1.6924,
"step": 1447
},
{
"epoch": 0.76,
"learning_rate": 2.925935735596582e-06,
"loss": 1.5874,
"step": 1448
},
{
"epoch": 0.76,
"learning_rate": 2.913962639430615e-06,
"loss": 1.5659,
"step": 1449
},
{
"epoch": 0.76,
"learning_rate": 2.9020099114454046e-06,
"loss": 1.4717,
"step": 1450
},
{
"epoch": 0.76,
"learning_rate": 2.890077585998007e-06,
"loss": 1.6348,
"step": 1451
},
{
"epoch": 0.76,
"learning_rate": 2.878165697386812e-06,
"loss": 1.6162,
"step": 1452
},
{
"epoch": 0.76,
"learning_rate": 2.866274279851491e-06,
"loss": 1.5415,
"step": 1453
},
{
"epoch": 0.76,
"learning_rate": 2.854403367572849e-06,
"loss": 1.3303,
"step": 1454
},
{
"epoch": 0.76,
"learning_rate": 2.8425529946727725e-06,
"loss": 1.0278,
"step": 1455
},
{
"epoch": 0.76,
"learning_rate": 2.830723195214088e-06,
"loss": 0.9565,
"step": 1456
},
{
"epoch": 0.76,
"learning_rate": 2.818914003200508e-06,
"loss": 1.2854,
"step": 1457
},
{
"epoch": 0.76,
"learning_rate": 2.807125452576487e-06,
"loss": 0.9307,
"step": 1458
},
{
"epoch": 0.76,
"learning_rate": 2.7953575772271667e-06,
"loss": 1.5864,
"step": 1459
},
{
"epoch": 0.76,
"learning_rate": 2.7836104109782525e-06,
"loss": 1.7065,
"step": 1460
},
{
"epoch": 0.76,
"learning_rate": 2.7718839875959213e-06,
"loss": 0.9014,
"step": 1461
},
{
"epoch": 0.77,
"learning_rate": 2.760178340786729e-06,
"loss": 1.4844,
"step": 1462
},
{
"epoch": 0.77,
"learning_rate": 2.7484935041975113e-06,
"loss": 1.7832,
"step": 1463
},
{
"epoch": 0.77,
"learning_rate": 2.7368295114152852e-06,
"loss": 1.1343,
"step": 1464
},
{
"epoch": 0.77,
"learning_rate": 2.7251863959671533e-06,
"loss": 1.1357,
"step": 1465
},
{
"epoch": 0.77,
"learning_rate": 2.713564191320217e-06,
"loss": 1.6123,
"step": 1466
},
{
"epoch": 0.77,
"learning_rate": 2.7019629308814544e-06,
"loss": 1.6113,
"step": 1467
},
{
"epoch": 0.77,
"learning_rate": 2.6903826479976582e-06,
"loss": 1.0098,
"step": 1468
},
{
"epoch": 0.77,
"learning_rate": 2.678823375955314e-06,
"loss": 1.3789,
"step": 1469
},
{
"epoch": 0.77,
"learning_rate": 2.667285147980515e-06,
"loss": 0.9153,
"step": 1470
},
{
"epoch": 0.77,
"learning_rate": 2.6557679972388674e-06,
"loss": 1.0869,
"step": 1471
},
{
"epoch": 0.77,
"learning_rate": 2.6442719568353894e-06,
"loss": 0.9136,
"step": 1472
},
{
"epoch": 0.77,
"learning_rate": 2.6327970598144214e-06,
"loss": 0.9133,
"step": 1473
},
{
"epoch": 0.77,
"learning_rate": 2.62134333915953e-06,
"loss": 1.0286,
"step": 1474
},
{
"epoch": 0.77,
"learning_rate": 2.6099108277934105e-06,
"loss": 1.4067,
"step": 1475
},
{
"epoch": 0.77,
"learning_rate": 2.598499558577794e-06,
"loss": 1.5781,
"step": 1476
},
{
"epoch": 0.77,
"learning_rate": 2.5871095643133502e-06,
"loss": 1.7129,
"step": 1477
},
{
"epoch": 0.77,
"learning_rate": 2.57574087773961e-06,
"loss": 1.4795,
"step": 1478
},
{
"epoch": 0.77,
"learning_rate": 2.564393531534832e-06,
"loss": 1.4468,
"step": 1479
},
{
"epoch": 0.77,
"learning_rate": 2.553067558315958e-06,
"loss": 1.0122,
"step": 1480
},
{
"epoch": 0.77,
"learning_rate": 2.5417629906384834e-06,
"loss": 1.3931,
"step": 1481
},
{
"epoch": 0.78,
"learning_rate": 2.5304798609963754e-06,
"loss": 2.0332,
"step": 1482
},
{
"epoch": 0.78,
"learning_rate": 2.5192182018219826e-06,
"loss": 1.2471,
"step": 1483
},
{
"epoch": 0.78,
"learning_rate": 2.507978045485937e-06,
"loss": 1.3945,
"step": 1484
},
{
"epoch": 0.78,
"learning_rate": 2.496759424297062e-06,
"loss": 0.9614,
"step": 1485
},
{
"epoch": 0.78,
"learning_rate": 2.485562370502279e-06,
"loss": 1.5908,
"step": 1486
},
{
"epoch": 0.78,
"learning_rate": 2.474386916286526e-06,
"loss": 1.22,
"step": 1487
},
{
"epoch": 0.78,
"learning_rate": 2.4632330937726356e-06,
"loss": 0.9996,
"step": 1488
},
{
"epoch": 0.78,
"learning_rate": 2.4521009350212854e-06,
"loss": 1.4648,
"step": 1489
},
{
"epoch": 0.78,
"learning_rate": 2.4409904720308576e-06,
"loss": 1.4282,
"step": 1490
},
{
"epoch": 0.78,
"learning_rate": 2.4299017367373966e-06,
"loss": 1.5293,
"step": 1491
},
{
"epoch": 0.78,
"learning_rate": 2.4188347610144693e-06,
"loss": 1.2124,
"step": 1492
},
{
"epoch": 0.78,
"learning_rate": 2.407789576673115e-06,
"loss": 1.5078,
"step": 1493
},
{
"epoch": 0.78,
"learning_rate": 2.396766215461728e-06,
"loss": 1.4512,
"step": 1494
},
{
"epoch": 0.78,
"learning_rate": 2.385764709065973e-06,
"loss": 1.2773,
"step": 1495
},
{
"epoch": 0.78,
"learning_rate": 2.3747850891086956e-06,
"loss": 0.9736,
"step": 1496
},
{
"epoch": 0.78,
"learning_rate": 2.363827387149833e-06,
"loss": 1.3001,
"step": 1497
},
{
"epoch": 0.78,
"learning_rate": 2.3528916346863185e-06,
"loss": 1.499,
"step": 1498
},
{
"epoch": 0.78,
"learning_rate": 2.3419778631519963e-06,
"loss": 1.5103,
"step": 1499
},
{
"epoch": 0.78,
"learning_rate": 2.331086103917525e-06,
"loss": 1.6318,
"step": 1500
},
{
"epoch": 0.79,
"learning_rate": 2.320216388290293e-06,
"loss": 1.0039,
"step": 1501
},
{
"epoch": 0.79,
"learning_rate": 2.3093687475143236e-06,
"loss": 1.0791,
"step": 1502
},
{
"epoch": 0.79,
"learning_rate": 2.2985432127701945e-06,
"loss": 1.1825,
"step": 1503
},
{
"epoch": 0.79,
"learning_rate": 2.2877398151749375e-06,
"loss": 1.0322,
"step": 1504
},
{
"epoch": 0.79,
"learning_rate": 2.27695858578195e-06,
"loss": 1.0483,
"step": 1505
},
{
"epoch": 0.79,
"learning_rate": 2.2661995555809123e-06,
"loss": 1.0518,
"step": 1506
},
{
"epoch": 0.79,
"learning_rate": 2.255462755497696e-06,
"loss": 1.3364,
"step": 1507
},
{
"epoch": 0.79,
"learning_rate": 2.2447482163942725e-06,
"loss": 1.0654,
"step": 1508
},
{
"epoch": 0.79,
"learning_rate": 2.2340559690686247e-06,
"loss": 0.9849,
"step": 1509
},
{
"epoch": 0.79,
"learning_rate": 2.223386044254663e-06,
"loss": 1.1753,
"step": 1510
},
{
"epoch": 0.79,
"learning_rate": 2.2127384726221303e-06,
"loss": 0.9249,
"step": 1511
},
{
"epoch": 0.79,
"learning_rate": 2.2021132847765246e-06,
"loss": 1.4648,
"step": 1512
},
{
"epoch": 0.79,
"learning_rate": 2.1915105112589894e-06,
"loss": 1.0049,
"step": 1513
},
{
"epoch": 0.79,
"learning_rate": 2.1809301825462593e-06,
"loss": 1.0864,
"step": 1514
},
{
"epoch": 0.79,
"learning_rate": 2.170372329050533e-06,
"loss": 1.5977,
"step": 1515
},
{
"epoch": 0.79,
"learning_rate": 2.159836981119425e-06,
"loss": 1.4272,
"step": 1516
},
{
"epoch": 0.79,
"learning_rate": 2.1493241690358437e-06,
"loss": 1.3457,
"step": 1517
},
{
"epoch": 0.79,
"learning_rate": 2.138833923017931e-06,
"loss": 0.9218,
"step": 1518
},
{
"epoch": 0.79,
"learning_rate": 2.1283662732189624e-06,
"loss": 1.0132,
"step": 1519
},
{
"epoch": 0.8,
"learning_rate": 2.1179212497272582e-06,
"loss": 1.6523,
"step": 1520
},
{
"epoch": 0.8,
"learning_rate": 2.107498882566107e-06,
"loss": 1.6416,
"step": 1521
},
{
"epoch": 0.8,
"learning_rate": 2.0970992016936685e-06,
"loss": 1.5435,
"step": 1522
},
{
"epoch": 0.8,
"learning_rate": 2.0867222370028985e-06,
"loss": 1.3032,
"step": 1523
},
{
"epoch": 0.8,
"learning_rate": 2.0763680183214485e-06,
"loss": 1.479,
"step": 1524
},
{
"epoch": 0.8,
"learning_rate": 2.0660365754116042e-06,
"loss": 1.5186,
"step": 1525
},
{
"epoch": 0.8,
"learning_rate": 2.0557279379701632e-06,
"loss": 1.1572,
"step": 1526
},
{
"epoch": 0.8,
"learning_rate": 2.045442135628389e-06,
"loss": 1.0177,
"step": 1527
},
{
"epoch": 0.8,
"learning_rate": 2.0351791979518997e-06,
"loss": 1.0464,
"step": 1528
},
{
"epoch": 0.8,
"learning_rate": 2.0249391544405916e-06,
"loss": 1.2803,
"step": 1529
},
{
"epoch": 0.8,
"learning_rate": 2.0147220345285524e-06,
"loss": 1.2764,
"step": 1530
},
{
"epoch": 0.8,
"learning_rate": 2.004527867583982e-06,
"loss": 1.1138,
"step": 1531
},
{
"epoch": 0.8,
"learning_rate": 1.9943566829091e-06,
"loss": 1.4082,
"step": 1532
},
{
"epoch": 0.8,
"learning_rate": 1.9842085097400678e-06,
"loss": 1.5366,
"step": 1533
},
{
"epoch": 0.8,
"learning_rate": 1.9740833772469025e-06,
"loss": 1.7998,
"step": 1534
},
{
"epoch": 0.8,
"learning_rate": 1.963981314533391e-06,
"loss": 1.5234,
"step": 1535
},
{
"epoch": 0.8,
"learning_rate": 1.953902350637006e-06,
"loss": 1.1631,
"step": 1536
},
{
"epoch": 0.8,
"learning_rate": 1.9438465145288377e-06,
"loss": 0.9619,
"step": 1537
},
{
"epoch": 0.8,
"learning_rate": 1.9338138351134773e-06,
"loss": 1.5249,
"step": 1538
},
{
"epoch": 0.81,
"learning_rate": 1.9238043412289696e-06,
"loss": 1.7227,
"step": 1539
},
{
"epoch": 0.81,
"learning_rate": 1.9138180616467105e-06,
"loss": 1.6191,
"step": 1540
},
{
"epoch": 0.81,
"learning_rate": 1.903855025071366e-06,
"loss": 0.9546,
"step": 1541
},
{
"epoch": 0.81,
"learning_rate": 1.8939152601407952e-06,
"loss": 1.4814,
"step": 1542
},
{
"epoch": 0.81,
"learning_rate": 1.8839987954259609e-06,
"loss": 1.5742,
"step": 1543
},
{
"epoch": 0.81,
"learning_rate": 1.8741056594308571e-06,
"loss": 1.2129,
"step": 1544
},
{
"epoch": 0.81,
"learning_rate": 1.8642358805924144e-06,
"loss": 1.6606,
"step": 1545
},
{
"epoch": 0.81,
"learning_rate": 1.8543894872804368e-06,
"loss": 0.9497,
"step": 1546
},
{
"epoch": 0.81,
"learning_rate": 1.8445665077974928e-06,
"loss": 1.5234,
"step": 1547
},
{
"epoch": 0.81,
"learning_rate": 1.8347669703788683e-06,
"loss": 0.8857,
"step": 1548
},
{
"epoch": 0.81,
"learning_rate": 1.824990903192445e-06,
"loss": 1.0371,
"step": 1549
},
{
"epoch": 0.81,
"learning_rate": 1.815238334338667e-06,
"loss": 1.0686,
"step": 1550
},
{
"epoch": 0.81,
"learning_rate": 1.8055092918504103e-06,
"loss": 1.6206,
"step": 1551
},
{
"epoch": 0.81,
"learning_rate": 1.795803803692947e-06,
"loss": 0.9238,
"step": 1552
},
{
"epoch": 0.81,
"learning_rate": 1.7861218977638252e-06,
"loss": 1.6865,
"step": 1553
},
{
"epoch": 0.81,
"learning_rate": 1.7764636018928249e-06,
"loss": 1.0547,
"step": 1554
},
{
"epoch": 0.81,
"learning_rate": 1.766828943841853e-06,
"loss": 1.4316,
"step": 1555
},
{
"epoch": 0.81,
"learning_rate": 1.7572179513048703e-06,
"loss": 1.103,
"step": 1556
},
{
"epoch": 0.81,
"learning_rate": 1.7476306519078168e-06,
"loss": 1.2836,
"step": 1557
},
{
"epoch": 0.82,
"learning_rate": 1.738067073208527e-06,
"loss": 1.4331,
"step": 1558
},
{
"epoch": 0.82,
"learning_rate": 1.7285272426966526e-06,
"loss": 1.6335,
"step": 1559
},
{
"epoch": 0.82,
"learning_rate": 1.7190111877935833e-06,
"loss": 1.0571,
"step": 1560
},
{
"epoch": 0.82,
"learning_rate": 1.7095189358523668e-06,
"loss": 1.5205,
"step": 1561
},
{
"epoch": 0.82,
"learning_rate": 1.700050514157635e-06,
"loss": 0.8805,
"step": 1562
},
{
"epoch": 0.82,
"learning_rate": 1.690605949925519e-06,
"loss": 1.6807,
"step": 1563
},
{
"epoch": 0.82,
"learning_rate": 1.6811852703035737e-06,
"loss": 1.4707,
"step": 1564
},
{
"epoch": 0.82,
"learning_rate": 1.6717885023707004e-06,
"loss": 1.5596,
"step": 1565
},
{
"epoch": 0.82,
"learning_rate": 1.6624156731370666e-06,
"loss": 1.1389,
"step": 1566
},
{
"epoch": 0.82,
"learning_rate": 1.6530668095440339e-06,
"loss": 1.1719,
"step": 1567
},
{
"epoch": 0.82,
"learning_rate": 1.643741938464073e-06,
"loss": 1.5112,
"step": 1568
},
{
"epoch": 0.82,
"learning_rate": 1.6344410867006944e-06,
"loss": 1.0,
"step": 1569
},
{
"epoch": 0.82,
"learning_rate": 1.6251642809883595e-06,
"loss": 1.1484,
"step": 1570
},
{
"epoch": 0.82,
"learning_rate": 1.6159115479924259e-06,
"loss": 0.8926,
"step": 1571
},
{
"epoch": 0.82,
"learning_rate": 1.606682914309038e-06,
"loss": 1.2764,
"step": 1572
},
{
"epoch": 0.82,
"learning_rate": 1.597478406465085e-06,
"loss": 1.0288,
"step": 1573
},
{
"epoch": 0.82,
"learning_rate": 1.5882980509180956e-06,
"loss": 0.9993,
"step": 1574
},
{
"epoch": 0.82,
"learning_rate": 1.579141874056187e-06,
"loss": 1.4414,
"step": 1575
},
{
"epoch": 0.82,
"learning_rate": 1.5700099021979642e-06,
"loss": 1.7783,
"step": 1576
},
{
"epoch": 0.83,
"learning_rate": 1.5609021615924702e-06,
"loss": 1.0283,
"step": 1577
},
{
"epoch": 0.83,
"learning_rate": 1.551818678419087e-06,
"loss": 1.4639,
"step": 1578
},
{
"epoch": 0.83,
"learning_rate": 1.5427594787874766e-06,
"loss": 1.2378,
"step": 1579
},
{
"epoch": 0.83,
"learning_rate": 1.5337245887374962e-06,
"loss": 1.3423,
"step": 1580
},
{
"epoch": 0.83,
"learning_rate": 1.524714034239132e-06,
"loss": 1.0757,
"step": 1581
},
{
"epoch": 0.83,
"learning_rate": 1.5157278411924148e-06,
"loss": 0.9536,
"step": 1582
},
{
"epoch": 0.83,
"learning_rate": 1.5067660354273527e-06,
"loss": 1.0134,
"step": 1583
},
{
"epoch": 0.83,
"learning_rate": 1.4978286427038602e-06,
"loss": 1.0161,
"step": 1584
},
{
"epoch": 0.83,
"learning_rate": 1.4889156887116652e-06,
"loss": 1.249,
"step": 1585
},
{
"epoch": 0.83,
"learning_rate": 1.4800271990702664e-06,
"loss": 1.5391,
"step": 1586
},
{
"epoch": 0.83,
"learning_rate": 1.4711631993288222e-06,
"loss": 1.3828,
"step": 1587
},
{
"epoch": 0.83,
"learning_rate": 1.462323714966114e-06,
"loss": 1.4619,
"step": 1588
},
{
"epoch": 0.83,
"learning_rate": 1.4535087713904484e-06,
"loss": 1.3036,
"step": 1589
},
{
"epoch": 0.83,
"learning_rate": 1.4447183939395925e-06,
"loss": 1.3345,
"step": 1590
},
{
"epoch": 0.83,
"learning_rate": 1.4359526078806985e-06,
"loss": 1.8623,
"step": 1591
},
{
"epoch": 0.83,
"learning_rate": 1.4272114384102353e-06,
"loss": 1.4834,
"step": 1592
},
{
"epoch": 0.83,
"learning_rate": 1.4184949106539137e-06,
"loss": 1.5137,
"step": 1593
},
{
"epoch": 0.83,
"learning_rate": 1.4098030496666126e-06,
"loss": 1.166,
"step": 1594
},
{
"epoch": 0.83,
"learning_rate": 1.4011358804323095e-06,
"loss": 1.626,
"step": 1595
},
{
"epoch": 0.84,
"learning_rate": 1.3924934278640068e-06,
"loss": 1.1748,
"step": 1596
},
{
"epoch": 0.84,
"learning_rate": 1.3838757168036598e-06,
"loss": 1.0117,
"step": 1597
},
{
"epoch": 0.84,
"learning_rate": 1.3752827720221128e-06,
"loss": 0.7277,
"step": 1598
},
{
"epoch": 0.84,
"learning_rate": 1.366714618219015e-06,
"loss": 1.4199,
"step": 1599
},
{
"epoch": 0.84,
"learning_rate": 1.3581712800227576e-06,
"loss": 1.6641,
"step": 1600
},
{
"epoch": 0.84,
"learning_rate": 1.3496527819904038e-06,
"loss": 1.3232,
"step": 1601
},
{
"epoch": 0.84,
"learning_rate": 1.3411591486076136e-06,
"loss": 1.3037,
"step": 1602
},
{
"epoch": 0.84,
"learning_rate": 1.3326904042885757e-06,
"loss": 0.8843,
"step": 1603
},
{
"epoch": 0.84,
"learning_rate": 1.324246573375938e-06,
"loss": 1.3393,
"step": 1604
},
{
"epoch": 0.84,
"learning_rate": 1.3158276801407432e-06,
"loss": 1.4644,
"step": 1605
},
{
"epoch": 0.84,
"learning_rate": 1.3074337487823374e-06,
"loss": 0.9814,
"step": 1606
},
{
"epoch": 0.84,
"learning_rate": 1.299064803428336e-06,
"loss": 1.0357,
"step": 1607
},
{
"epoch": 0.84,
"learning_rate": 1.290720868134515e-06,
"loss": 1.3647,
"step": 1608
},
{
"epoch": 0.84,
"learning_rate": 1.2824019668847775e-06,
"loss": 1.3374,
"step": 1609
},
{
"epoch": 0.84,
"learning_rate": 1.2741081235910547e-06,
"loss": 1.1929,
"step": 1610
},
{
"epoch": 0.84,
"learning_rate": 1.2658393620932652e-06,
"loss": 0.8875,
"step": 1611
},
{
"epoch": 0.84,
"learning_rate": 1.2575957061592159e-06,
"loss": 1.1807,
"step": 1612
},
{
"epoch": 0.84,
"learning_rate": 1.2493771794845644e-06,
"loss": 1.293,
"step": 1613
},
{
"epoch": 0.84,
"learning_rate": 1.2411838056927295e-06,
"loss": 1.2812,
"step": 1614
},
{
"epoch": 0.85,
"learning_rate": 1.233015608334831e-06,
"loss": 1.0635,
"step": 1615
},
{
"epoch": 0.85,
"learning_rate": 1.2248726108896225e-06,
"loss": 1.2139,
"step": 1616
},
{
"epoch": 0.85,
"learning_rate": 1.216754836763422e-06,
"loss": 1.4155,
"step": 1617
},
{
"epoch": 0.85,
"learning_rate": 1.2086623092900441e-06,
"loss": 0.9958,
"step": 1618
},
{
"epoch": 0.85,
"learning_rate": 1.2005950517307374e-06,
"loss": 1.5215,
"step": 1619
},
{
"epoch": 0.85,
"learning_rate": 1.1925530872741098e-06,
"loss": 1.1436,
"step": 1620
},
{
"epoch": 0.85,
"learning_rate": 1.1845364390360692e-06,
"loss": 1.3711,
"step": 1621
},
{
"epoch": 0.85,
"learning_rate": 1.1765451300597574e-06,
"loss": 1.4653,
"step": 1622
},
{
"epoch": 0.85,
"learning_rate": 1.1685791833154748e-06,
"loss": 1.3442,
"step": 1623
},
{
"epoch": 0.85,
"learning_rate": 1.160638621700624e-06,
"loss": 1.0979,
"step": 1624
},
{
"epoch": 0.85,
"learning_rate": 1.1527234680396392e-06,
"loss": 1.2544,
"step": 1625
},
{
"epoch": 0.85,
"learning_rate": 1.1448337450839208e-06,
"loss": 1.2729,
"step": 1626
},
{
"epoch": 0.85,
"learning_rate": 1.1369694755117733e-06,
"loss": 1.4575,
"step": 1627
},
{
"epoch": 0.85,
"learning_rate": 1.1291306819283353e-06,
"loss": 1.2195,
"step": 1628
},
{
"epoch": 0.85,
"learning_rate": 1.121317386865518e-06,
"loss": 1.6055,
"step": 1629
},
{
"epoch": 0.85,
"learning_rate": 1.1135296127819407e-06,
"loss": 1.0593,
"step": 1630
},
{
"epoch": 0.85,
"learning_rate": 1.1057673820628612e-06,
"loss": 1.626,
"step": 1631
},
{
"epoch": 0.85,
"learning_rate": 1.0980307170201222e-06,
"loss": 1.2798,
"step": 1632
},
{
"epoch": 0.85,
"learning_rate": 1.090319639892069e-06,
"loss": 1.543,
"step": 1633
},
{
"epoch": 0.86,
"learning_rate": 1.0826341728435096e-06,
"loss": 1.0747,
"step": 1634
},
{
"epoch": 0.86,
"learning_rate": 1.0749743379656263e-06,
"loss": 1.4746,
"step": 1635
},
{
"epoch": 0.86,
"learning_rate": 1.0673401572759335e-06,
"loss": 1.6484,
"step": 1636
},
{
"epoch": 0.86,
"learning_rate": 1.0597316527182e-06,
"loss": 1.0044,
"step": 1637
},
{
"epoch": 0.86,
"learning_rate": 1.0521488461623919e-06,
"loss": 1.3647,
"step": 1638
},
{
"epoch": 0.86,
"learning_rate": 1.0445917594046073e-06,
"loss": 1.4268,
"step": 1639
},
{
"epoch": 0.86,
"learning_rate": 1.037060414167017e-06,
"loss": 1.4287,
"step": 1640
},
{
"epoch": 0.86,
"learning_rate": 1.029554832097801e-06,
"loss": 1.103,
"step": 1641
},
{
"epoch": 0.86,
"learning_rate": 1.0220750347710795e-06,
"loss": 1.0405,
"step": 1642
},
{
"epoch": 0.86,
"learning_rate": 1.0146210436868675e-06,
"loss": 1.007,
"step": 1643
},
{
"epoch": 0.86,
"learning_rate": 1.0071928802709886e-06,
"loss": 1.6016,
"step": 1644
},
{
"epoch": 0.86,
"learning_rate": 9.997905658750428e-07,
"loss": 1.0405,
"step": 1645
},
{
"epoch": 0.86,
"learning_rate": 9.924141217763117e-07,
"loss": 1.5381,
"step": 1646
},
{
"epoch": 0.86,
"learning_rate": 9.8506356917773e-07,
"loss": 1.2305,
"step": 1647
},
{
"epoch": 0.86,
"learning_rate": 9.777389292078033e-07,
"loss": 1.1289,
"step": 1648
},
{
"epoch": 0.86,
"learning_rate": 9.704402229205534e-07,
"loss": 1.1543,
"step": 1649
},
{
"epoch": 0.86,
"learning_rate": 9.631674712954586e-07,
"loss": 1.7832,
"step": 1650
},
{
"epoch": 0.86,
"learning_rate": 9.559206952373933e-07,
"loss": 1.645,
"step": 1651
},
{
"epoch": 0.86,
"learning_rate": 9.486999155765675e-07,
"loss": 1.5034,
"step": 1652
},
{
"epoch": 0.86,
"learning_rate": 9.415051530684671e-07,
"loss": 1.6152,
"step": 1653
},
{
"epoch": 0.87,
"learning_rate": 9.343364283937928e-07,
"loss": 0.8618,
"step": 1654
},
{
"epoch": 0.87,
"learning_rate": 9.271937621584026e-07,
"loss": 1.6191,
"step": 1655
},
{
"epoch": 0.87,
"learning_rate": 9.200771748932513e-07,
"loss": 0.9402,
"step": 1656
},
{
"epoch": 0.87,
"learning_rate": 9.129866870543347e-07,
"loss": 0.8081,
"step": 1657
},
{
"epoch": 0.87,
"learning_rate": 9.059223190226251e-07,
"loss": 1.459,
"step": 1658
},
{
"epoch": 0.87,
"learning_rate": 8.988840911040165e-07,
"loss": 0.906,
"step": 1659
},
{
"epoch": 0.87,
"learning_rate": 8.918720235292655e-07,
"loss": 1.3604,
"step": 1660
},
{
"epoch": 0.87,
"learning_rate": 8.84886136453933e-07,
"loss": 1.0078,
"step": 1661
},
{
"epoch": 0.87,
"learning_rate": 8.779264499583284e-07,
"loss": 0.9829,
"step": 1662
},
{
"epoch": 0.87,
"learning_rate": 8.709929840474463e-07,
"loss": 1.5176,
"step": 1663
},
{
"epoch": 0.87,
"learning_rate": 8.640857586509155e-07,
"loss": 1.3147,
"step": 1664
},
{
"epoch": 0.87,
"learning_rate": 8.572047936229355e-07,
"loss": 1.2969,
"step": 1665
},
{
"epoch": 0.87,
"learning_rate": 8.503501087422305e-07,
"loss": 1.6924,
"step": 1666
},
{
"epoch": 0.87,
"learning_rate": 8.435217237119719e-07,
"loss": 1.5505,
"step": 1667
},
{
"epoch": 0.87,
"learning_rate": 8.367196581597503e-07,
"loss": 1.4619,
"step": 1668
},
{
"epoch": 0.87,
"learning_rate": 8.299439316374858e-07,
"loss": 0.9233,
"step": 1669
},
{
"epoch": 0.87,
"learning_rate": 8.231945636214068e-07,
"loss": 1.6211,
"step": 1670
},
{
"epoch": 0.87,
"learning_rate": 8.164715735119599e-07,
"loss": 1.1113,
"step": 1671
},
{
"epoch": 0.87,
"learning_rate": 8.097749806337851e-07,
"loss": 1.2598,
"step": 1672
},
{
"epoch": 0.88,
"learning_rate": 8.031048042356393e-07,
"loss": 1.3989,
"step": 1673
},
{
"epoch": 0.88,
"learning_rate": 7.964610634903458e-07,
"loss": 1.0234,
"step": 1674
},
{
"epoch": 0.88,
"learning_rate": 7.898437774947442e-07,
"loss": 1.1946,
"step": 1675
},
{
"epoch": 0.88,
"learning_rate": 7.832529652696319e-07,
"loss": 0.7934,
"step": 1676
},
{
"epoch": 0.88,
"learning_rate": 7.766886457597089e-07,
"loss": 1.1946,
"step": 1677
},
{
"epoch": 0.88,
"learning_rate": 7.701508378335243e-07,
"loss": 1.4277,
"step": 1678
},
{
"epoch": 0.88,
"learning_rate": 7.636395602834223e-07,
"loss": 1.7637,
"step": 1679
},
{
"epoch": 0.88,
"learning_rate": 7.571548318254874e-07,
"loss": 1.1538,
"step": 1680
},
{
"epoch": 0.88,
"learning_rate": 7.506966710994945e-07,
"loss": 0.9434,
"step": 1681
},
{
"epoch": 0.88,
"learning_rate": 7.442650966688469e-07,
"loss": 1.541,
"step": 1682
},
{
"epoch": 0.88,
"learning_rate": 7.378601270205299e-07,
"loss": 1.6367,
"step": 1683
},
{
"epoch": 0.88,
"learning_rate": 7.314817805650576e-07,
"loss": 1.5957,
"step": 1684
},
{
"epoch": 0.88,
"learning_rate": 7.251300756364154e-07,
"loss": 1.6123,
"step": 1685
},
{
"epoch": 0.88,
"learning_rate": 7.188050304920113e-07,
"loss": 1.4922,
"step": 1686
},
{
"epoch": 0.88,
"learning_rate": 7.125066633126221e-07,
"loss": 1.4092,
"step": 1687
},
{
"epoch": 0.88,
"learning_rate": 7.062349922023414e-07,
"loss": 1.3467,
"step": 1688
},
{
"epoch": 0.88,
"learning_rate": 6.999900351885258e-07,
"loss": 1.0742,
"step": 1689
},
{
"epoch": 0.88,
"learning_rate": 6.937718102217461e-07,
"loss": 1.4021,
"step": 1690
},
{
"epoch": 0.88,
"learning_rate": 6.875803351757371e-07,
"loss": 1.5396,
"step": 1691
},
{
"epoch": 0.89,
"learning_rate": 6.814156278473338e-07,
"loss": 0.842,
"step": 1692
},
{
"epoch": 0.89,
"learning_rate": 6.752777059564431e-07,
"loss": 1.5283,
"step": 1693
},
{
"epoch": 0.89,
"learning_rate": 6.691665871459674e-07,
"loss": 1.6631,
"step": 1694
},
{
"epoch": 0.89,
"learning_rate": 6.630822889817734e-07,
"loss": 1.2393,
"step": 1695
},
{
"epoch": 0.89,
"learning_rate": 6.570248289526326e-07,
"loss": 1.6309,
"step": 1696
},
{
"epoch": 0.89,
"learning_rate": 6.509942244701717e-07,
"loss": 1.5322,
"step": 1697
},
{
"epoch": 0.89,
"learning_rate": 6.449904928688244e-07,
"loss": 1.083,
"step": 1698
},
{
"epoch": 0.89,
"learning_rate": 6.390136514057788e-07,
"loss": 1.1235,
"step": 1699
},
{
"epoch": 0.89,
"learning_rate": 6.330637172609333e-07,
"loss": 1.5137,
"step": 1700
},
{
"epoch": 0.89,
"learning_rate": 6.271407075368385e-07,
"loss": 1.5005,
"step": 1701
},
{
"epoch": 0.89,
"learning_rate": 6.212446392586613e-07,
"loss": 1.5918,
"step": 1702
},
{
"epoch": 0.89,
"learning_rate": 6.15375529374116e-07,
"loss": 1.415,
"step": 1703
},
{
"epoch": 0.89,
"learning_rate": 6.095333947534421e-07,
"loss": 1.1992,
"step": 1704
},
{
"epoch": 0.89,
"learning_rate": 6.037182521893247e-07,
"loss": 1.3105,
"step": 1705
},
{
"epoch": 0.89,
"learning_rate": 5.979301183968789e-07,
"loss": 1.3635,
"step": 1706
},
{
"epoch": 0.89,
"learning_rate": 5.921690100135713e-07,
"loss": 1.5068,
"step": 1707
},
{
"epoch": 0.89,
"learning_rate": 5.864349435991978e-07,
"loss": 0.9624,
"step": 1708
},
{
"epoch": 0.89,
"learning_rate": 5.807279356358164e-07,
"loss": 1.7383,
"step": 1709
},
{
"epoch": 0.89,
"learning_rate": 5.750480025277116e-07,
"loss": 1.3501,
"step": 1710
},
{
"epoch": 0.9,
"learning_rate": 5.693951606013448e-07,
"loss": 1.6592,
"step": 1711
},
{
"epoch": 0.9,
"learning_rate": 5.637694261053029e-07,
"loss": 0.8959,
"step": 1712
},
{
"epoch": 0.9,
"learning_rate": 5.581708152102583e-07,
"loss": 1.397,
"step": 1713
},
{
"epoch": 0.9,
"learning_rate": 5.525993440089161e-07,
"loss": 1.3096,
"step": 1714
},
{
"epoch": 0.9,
"learning_rate": 5.470550285159704e-07,
"loss": 1.4268,
"step": 1715
},
{
"epoch": 0.9,
"learning_rate": 5.415378846680641e-07,
"loss": 1.1704,
"step": 1716
},
{
"epoch": 0.9,
"learning_rate": 5.360479283237319e-07,
"loss": 0.9492,
"step": 1717
},
{
"epoch": 0.9,
"learning_rate": 5.305851752633628e-07,
"loss": 1.2798,
"step": 1718
},
{
"epoch": 0.9,
"learning_rate": 5.251496411891521e-07,
"loss": 1.4629,
"step": 1719
},
{
"epoch": 0.9,
"learning_rate": 5.197413417250563e-07,
"loss": 0.9551,
"step": 1720
},
{
"epoch": 0.9,
"learning_rate": 5.143602924167479e-07,
"loss": 1.2041,
"step": 1721
},
{
"epoch": 0.9,
"learning_rate": 5.090065087315721e-07,
"loss": 0.9448,
"step": 1722
},
{
"epoch": 0.9,
"learning_rate": 5.036800060585023e-07,
"loss": 1.4932,
"step": 1723
},
{
"epoch": 0.9,
"learning_rate": 4.983807997080925e-07,
"loss": 1.4126,
"step": 1724
},
{
"epoch": 0.9,
"learning_rate": 4.931089049124405e-07,
"loss": 1.0737,
"step": 1725
},
{
"epoch": 0.9,
"learning_rate": 4.878643368251323e-07,
"loss": 1.5322,
"step": 1726
},
{
"epoch": 0.9,
"learning_rate": 4.826471105212149e-07,
"loss": 1.4697,
"step": 1727
},
{
"epoch": 0.9,
"learning_rate": 4.774572409971334e-07,
"loss": 0.9136,
"step": 1728
},
{
"epoch": 0.9,
"learning_rate": 4.722947431707092e-07,
"loss": 0.953,
"step": 1729
},
{
"epoch": 0.91,
"learning_rate": 4.671596318810745e-07,
"loss": 1.0669,
"step": 1730
},
{
"epoch": 0.91,
"learning_rate": 4.620519218886543e-07,
"loss": 1.4785,
"step": 1731
},
{
"epoch": 0.91,
"learning_rate": 4.5697162787509686e-07,
"loss": 1.3447,
"step": 1732
},
{
"epoch": 0.91,
"learning_rate": 4.5191876444325986e-07,
"loss": 1.0176,
"step": 1733
},
{
"epoch": 0.91,
"learning_rate": 4.468933461171432e-07,
"loss": 1.314,
"step": 1734
},
{
"epoch": 0.91,
"learning_rate": 4.418953873418641e-07,
"loss": 1.5366,
"step": 1735
},
{
"epoch": 0.91,
"learning_rate": 4.3692490248360755e-07,
"loss": 0.8837,
"step": 1736
},
{
"epoch": 0.91,
"learning_rate": 4.319819058295893e-07,
"loss": 1.4487,
"step": 1737
},
{
"epoch": 0.91,
"learning_rate": 4.2706641158800855e-07,
"loss": 1.0796,
"step": 1738
},
{
"epoch": 0.91,
"learning_rate": 4.221784338880153e-07,
"loss": 1.0431,
"step": 1739
},
{
"epoch": 0.91,
"learning_rate": 4.173179867796662e-07,
"loss": 1.373,
"step": 1740
},
{
"epoch": 0.91,
"learning_rate": 4.124850842338779e-07,
"loss": 1.2847,
"step": 1741
},
{
"epoch": 0.91,
"learning_rate": 4.076797401423993e-07,
"loss": 1.2471,
"step": 1742
},
{
"epoch": 0.91,
"learning_rate": 4.0290196831776264e-07,
"loss": 1.0312,
"step": 1743
},
{
"epoch": 0.91,
"learning_rate": 3.981517824932446e-07,
"loss": 1.214,
"step": 1744
},
{
"epoch": 0.91,
"learning_rate": 3.9342919632283093e-07,
"loss": 1.4346,
"step": 1745
},
{
"epoch": 0.91,
"learning_rate": 3.8873422338117084e-07,
"loss": 1.4993,
"step": 1746
},
{
"epoch": 0.91,
"learning_rate": 3.8406687716354695e-07,
"loss": 0.8267,
"step": 1747
},
{
"epoch": 0.91,
"learning_rate": 3.7942717108582663e-07,
"loss": 1.0364,
"step": 1748
},
{
"epoch": 0.92,
"learning_rate": 3.748151184844295e-07,
"loss": 0.7578,
"step": 1749
},
{
"epoch": 0.92,
"learning_rate": 3.702307326162924e-07,
"loss": 1.5908,
"step": 1750
},
{
"epoch": 0.92,
"learning_rate": 3.656740266588177e-07,
"loss": 1.2075,
"step": 1751
},
{
"epoch": 0.92,
"learning_rate": 3.611450137098538e-07,
"loss": 1.0254,
"step": 1752
},
{
"epoch": 0.92,
"learning_rate": 3.5664370678763825e-07,
"loss": 1.2026,
"step": 1753
},
{
"epoch": 0.92,
"learning_rate": 3.5217011883077914e-07,
"loss": 1.3481,
"step": 1754
},
{
"epoch": 0.92,
"learning_rate": 3.477242626982047e-07,
"loss": 0.9426,
"step": 1755
},
{
"epoch": 0.92,
"learning_rate": 3.4330615116912933e-07,
"loss": 1.4053,
"step": 1756
},
{
"epoch": 0.92,
"learning_rate": 3.389157969430201e-07,
"loss": 1.4771,
"step": 1757
},
{
"epoch": 0.92,
"learning_rate": 3.345532126395579e-07,
"loss": 1.6562,
"step": 1758
},
{
"epoch": 0.92,
"learning_rate": 3.302184107986017e-07,
"loss": 0.9888,
"step": 1759
},
{
"epoch": 0.92,
"learning_rate": 3.259114038801503e-07,
"loss": 1.2437,
"step": 1760
},
{
"epoch": 0.92,
"learning_rate": 3.216322042643127e-07,
"loss": 1.5208,
"step": 1761
},
{
"epoch": 0.92,
"learning_rate": 3.17380824251261e-07,
"loss": 0.9585,
"step": 1762
},
{
"epoch": 0.92,
"learning_rate": 3.1315727606121226e-07,
"loss": 1.3574,
"step": 1763
},
{
"epoch": 0.92,
"learning_rate": 3.0896157183437305e-07,
"loss": 1.4399,
"step": 1764
},
{
"epoch": 0.92,
"learning_rate": 3.0479372363092506e-07,
"loss": 1.7236,
"step": 1765
},
{
"epoch": 0.92,
"learning_rate": 3.0065374343096974e-07,
"loss": 1.5107,
"step": 1766
},
{
"epoch": 0.92,
"learning_rate": 2.965416431345147e-07,
"loss": 1.3484,
"step": 1767
},
{
"epoch": 0.93,
"learning_rate": 2.9245743456142283e-07,
"loss": 1.6045,
"step": 1768
},
{
"epoch": 0.93,
"learning_rate": 2.8840112945138553e-07,
"loss": 1.335,
"step": 1769
},
{
"epoch": 0.93,
"learning_rate": 2.8437273946388954e-07,
"loss": 1.4961,
"step": 1770
},
{
"epoch": 0.93,
"learning_rate": 2.803722761781824e-07,
"loss": 1.6963,
"step": 1771
},
{
"epoch": 0.93,
"learning_rate": 2.763997510932381e-07,
"loss": 1.1687,
"step": 1772
},
{
"epoch": 0.93,
"learning_rate": 2.7245517562772385e-07,
"loss": 1.5439,
"step": 1773
},
{
"epoch": 0.93,
"learning_rate": 2.685385611199698e-07,
"loss": 1.437,
"step": 1774
},
{
"epoch": 0.93,
"learning_rate": 2.646499188279328e-07,
"loss": 1.5298,
"step": 1775
},
{
"epoch": 0.93,
"learning_rate": 2.607892599291695e-07,
"loss": 1.5469,
"step": 1776
},
{
"epoch": 0.93,
"learning_rate": 2.5695659552079866e-07,
"loss": 1.3394,
"step": 1777
},
{
"epoch": 0.93,
"learning_rate": 2.5315193661947013e-07,
"loss": 0.9432,
"step": 1778
},
{
"epoch": 0.93,
"learning_rate": 2.493752941613348e-07,
"loss": 1.3271,
"step": 1779
},
{
"epoch": 0.93,
"learning_rate": 2.4562667900201677e-07,
"loss": 1.3784,
"step": 1780
},
{
"epoch": 0.93,
"learning_rate": 2.4190610191657246e-07,
"loss": 1.3979,
"step": 1781
},
{
"epoch": 0.93,
"learning_rate": 2.3821357359946838e-07,
"loss": 1.0366,
"step": 1782
},
{
"epoch": 0.93,
"learning_rate": 2.345491046645465e-07,
"loss": 1.5308,
"step": 1783
},
{
"epoch": 0.93,
"learning_rate": 2.309127056449989e-07,
"loss": 1.4268,
"step": 1784
},
{
"epoch": 0.93,
"learning_rate": 2.2730438699332446e-07,
"loss": 1.52,
"step": 1785
},
{
"epoch": 0.93,
"learning_rate": 2.237241590813155e-07,
"loss": 1.3594,
"step": 1786
},
{
"epoch": 0.94,
"learning_rate": 2.2017203220001559e-07,
"loss": 1.0269,
"step": 1787
},
{
"epoch": 0.94,
"learning_rate": 2.1664801655969735e-07,
"loss": 1.5781,
"step": 1788
},
{
"epoch": 0.94,
"learning_rate": 2.1315212228982584e-07,
"loss": 0.9819,
"step": 1789
},
{
"epoch": 0.94,
"learning_rate": 2.0968435943904075e-07,
"loss": 1.3218,
"step": 1790
},
{
"epoch": 0.94,
"learning_rate": 2.062447379751109e-07,
"loss": 1.4053,
"step": 1791
},
{
"epoch": 0.94,
"learning_rate": 2.028332677849254e-07,
"loss": 1.5898,
"step": 1792
},
{
"epoch": 0.94,
"learning_rate": 1.9944995867444806e-07,
"loss": 1.5996,
"step": 1793
},
{
"epoch": 0.94,
"learning_rate": 1.9609482036869853e-07,
"loss": 1.5527,
"step": 1794
},
{
"epoch": 0.94,
"learning_rate": 1.927678625117224e-07,
"loss": 0.9492,
"step": 1795
},
{
"epoch": 0.94,
"learning_rate": 1.8946909466656115e-07,
"loss": 1.1089,
"step": 1796
},
{
"epoch": 0.94,
"learning_rate": 1.8619852631523105e-07,
"loss": 1.7178,
"step": 1797
},
{
"epoch": 0.94,
"learning_rate": 1.8295616685868434e-07,
"loss": 1.3276,
"step": 1798
},
{
"epoch": 0.94,
"learning_rate": 1.797420256167981e-07,
"loss": 1.0581,
"step": 1799
},
{
"epoch": 0.94,
"learning_rate": 1.7655611182832767e-07,
"loss": 1.4346,
"step": 1800
},
{
"epoch": 0.94,
"learning_rate": 1.7339843465090322e-07,
"loss": 0.9688,
"step": 1801
},
{
"epoch": 0.94,
"learning_rate": 1.7026900316098217e-07,
"loss": 0.8027,
"step": 1802
},
{
"epoch": 0.94,
"learning_rate": 1.6716782635383677e-07,
"loss": 0.9595,
"step": 1803
},
{
"epoch": 0.94,
"learning_rate": 1.640949131435221e-07,
"loss": 1.5947,
"step": 1804
},
{
"epoch": 0.94,
"learning_rate": 1.6105027236285265e-07,
"loss": 1.2339,
"step": 1805
},
{
"epoch": 0.95,
"learning_rate": 1.5803391276337565e-07,
"loss": 1.3301,
"step": 1806
},
{
"epoch": 0.95,
"learning_rate": 1.5504584301534677e-07,
"loss": 1.6084,
"step": 1807
},
{
"epoch": 0.95,
"learning_rate": 1.520860717077055e-07,
"loss": 1.417,
"step": 1808
},
{
"epoch": 0.95,
"learning_rate": 1.49154607348051e-07,
"loss": 1.5889,
"step": 1809
},
{
"epoch": 0.95,
"learning_rate": 1.462514583626129e-07,
"loss": 1.3867,
"step": 1810
},
{
"epoch": 0.95,
"learning_rate": 1.4337663309623718e-07,
"loss": 1.2871,
"step": 1811
},
{
"epoch": 0.95,
"learning_rate": 1.405301398123482e-07,
"loss": 1.2622,
"step": 1812
},
{
"epoch": 0.95,
"learning_rate": 1.3771198669293885e-07,
"loss": 1.7227,
"step": 1813
},
{
"epoch": 0.95,
"learning_rate": 1.3492218183853712e-07,
"loss": 1.6777,
"step": 1814
},
{
"epoch": 0.95,
"learning_rate": 1.3216073326818736e-07,
"loss": 1.5366,
"step": 1815
},
{
"epoch": 0.95,
"learning_rate": 1.2942764891942793e-07,
"loss": 1.5947,
"step": 1816
},
{
"epoch": 0.95,
"learning_rate": 1.2672293664826364e-07,
"loss": 1.5918,
"step": 1817
},
{
"epoch": 0.95,
"learning_rate": 1.2404660422914993e-07,
"loss": 1.4131,
"step": 1818
},
{
"epoch": 0.95,
"learning_rate": 1.213986593549621e-07,
"loss": 1.6553,
"step": 1819
},
{
"epoch": 0.95,
"learning_rate": 1.1877910963698502e-07,
"loss": 1.0903,
"step": 1820
},
{
"epoch": 0.95,
"learning_rate": 1.1618796260487675e-07,
"loss": 1.4561,
"step": 1821
},
{
"epoch": 0.95,
"learning_rate": 1.1362522570665945e-07,
"loss": 1.2747,
"step": 1822
},
{
"epoch": 0.95,
"learning_rate": 1.1109090630868958e-07,
"loss": 0.9189,
"step": 1823
},
{
"epoch": 0.95,
"learning_rate": 1.0858501169564439e-07,
"loss": 1.0294,
"step": 1824
},
{
"epoch": 0.95,
"learning_rate": 1.0610754907049104e-07,
"loss": 1.2437,
"step": 1825
},
{
"epoch": 0.96,
"learning_rate": 1.0365852555447642e-07,
"loss": 1.4097,
"step": 1826
},
{
"epoch": 0.96,
"learning_rate": 1.0123794818709732e-07,
"loss": 1.5571,
"step": 1827
},
{
"epoch": 0.96,
"learning_rate": 9.884582392608921e-08,
"loss": 1.0684,
"step": 1828
},
{
"epoch": 0.96,
"learning_rate": 9.64821596473986e-08,
"loss": 1.3369,
"step": 1829
},
{
"epoch": 0.96,
"learning_rate": 9.414696214516628e-08,
"loss": 1.0073,
"step": 1830
},
{
"epoch": 0.96,
"learning_rate": 9.184023813170961e-08,
"loss": 0.988,
"step": 1831
},
{
"epoch": 0.96,
"learning_rate": 8.956199423750033e-08,
"loss": 0.8818,
"step": 1832
},
{
"epoch": 0.96,
"learning_rate": 8.731223701114566e-08,
"loss": 1.6055,
"step": 1833
},
{
"epoch": 0.96,
"learning_rate": 8.509097291937273e-08,
"loss": 1.4214,
"step": 1834
},
{
"epoch": 0.96,
"learning_rate": 8.289820834700757e-08,
"loss": 1.5654,
"step": 1835
},
{
"epoch": 0.96,
"learning_rate": 8.073394959695391e-08,
"loss": 1.4844,
"step": 1836
},
{
"epoch": 0.96,
"learning_rate": 7.859820289018105e-08,
"loss": 1.1074,
"step": 1837
},
{
"epoch": 0.96,
"learning_rate": 7.649097436570164e-08,
"loss": 1.3979,
"step": 1838
},
{
"epoch": 0.96,
"learning_rate": 7.441227008055607e-08,
"loss": 1.1084,
"step": 1839
},
{
"epoch": 0.96,
"learning_rate": 7.23620960097926e-08,
"loss": 0.7485,
"step": 1840
},
{
"epoch": 0.96,
"learning_rate": 7.034045804645506e-08,
"loss": 1.0649,
"step": 1841
},
{
"epoch": 0.96,
"learning_rate": 6.834736200155956e-08,
"loss": 1.2095,
"step": 1842
},
{
"epoch": 0.96,
"learning_rate": 6.638281360408339e-08,
"loss": 1.3506,
"step": 1843
},
{
"epoch": 0.96,
"learning_rate": 6.444681850094392e-08,
"loss": 1.3369,
"step": 1844
},
{
"epoch": 0.97,
"learning_rate": 6.253938225698863e-08,
"loss": 0.9146,
"step": 1845
},
{
"epoch": 0.97,
"learning_rate": 6.066051035496956e-08,
"loss": 1.0986,
"step": 1846
},
{
"epoch": 0.97,
"learning_rate": 5.881020819553773e-08,
"loss": 1.4478,
"step": 1847
},
{
"epoch": 0.97,
"learning_rate": 5.6988481097219886e-08,
"loss": 1.2568,
"step": 1848
},
{
"epoch": 0.97,
"learning_rate": 5.519533429640955e-08,
"loss": 1.1743,
"step": 1849
},
{
"epoch": 0.97,
"learning_rate": 5.343077294734378e-08,
"loss": 1.5654,
"step": 1850
},
{
"epoch": 0.97,
"learning_rate": 5.1694802122099763e-08,
"loss": 1.4131,
"step": 1851
},
{
"epoch": 0.97,
"learning_rate": 4.9987426810569337e-08,
"loss": 1.4297,
"step": 1852
},
{
"epoch": 0.97,
"learning_rate": 4.8308651920448976e-08,
"loss": 1.0591,
"step": 1853
},
{
"epoch": 0.97,
"learning_rate": 4.665848227722869e-08,
"loss": 1.117,
"step": 1854
},
{
"epoch": 0.97,
"learning_rate": 4.503692262417203e-08,
"loss": 1.6279,
"step": 1855
},
{
"epoch": 0.97,
"learning_rate": 4.344397762230834e-08,
"loss": 1.5068,
"step": 1856
},
{
"epoch": 0.97,
"learning_rate": 4.1879651850414984e-08,
"loss": 1.4214,
"step": 1857
},
{
"epoch": 0.97,
"learning_rate": 4.0343949805006224e-08,
"loss": 1.5132,
"step": 1858
},
{
"epoch": 0.97,
"learning_rate": 3.8836875900321036e-08,
"loss": 1.4668,
"step": 1859
},
{
"epoch": 0.97,
"learning_rate": 3.735843446830867e-08,
"loss": 1.3789,
"step": 1860
},
{
"epoch": 0.97,
"learning_rate": 3.590862975861642e-08,
"loss": 1.0439,
"step": 1861
},
{
"epoch": 0.97,
"learning_rate": 3.448746593857744e-08,
"loss": 1.5684,
"step": 1862
},
{
"epoch": 0.97,
"learning_rate": 3.309494709320182e-08,
"loss": 1.4878,
"step": 1863
},
{
"epoch": 0.98,
"learning_rate": 3.173107722515889e-08,
"loss": 1.0449,
"step": 1864
},
{
"epoch": 0.98,
"learning_rate": 3.0395860254772705e-08,
"loss": 1.0469,
"step": 1865
},
{
"epoch": 0.98,
"learning_rate": 2.9089300020002097e-08,
"loss": 0.9731,
"step": 1866
},
{
"epoch": 0.98,
"learning_rate": 2.781140027643958e-08,
"loss": 1.3989,
"step": 1867
},
{
"epoch": 0.98,
"learning_rate": 2.656216469729356e-08,
"loss": 1.2993,
"step": 1868
},
{
"epoch": 0.98,
"learning_rate": 2.5341596873379447e-08,
"loss": 1.0331,
"step": 1869
},
{
"epoch": 0.98,
"learning_rate": 2.4149700313110813e-08,
"loss": 1.6216,
"step": 1870
},
{
"epoch": 0.98,
"learning_rate": 2.2986478442486026e-08,
"loss": 1.3672,
"step": 1871
},
{
"epoch": 0.98,
"learning_rate": 2.1851934605086057e-08,
"loss": 1.3789,
"step": 1872
},
{
"epoch": 0.98,
"learning_rate": 2.074607206205226e-08,
"loss": 1.5688,
"step": 1873
},
{
"epoch": 0.98,
"learning_rate": 1.96688939920886e-08,
"loss": 1.0527,
"step": 1874
},
{
"epoch": 0.98,
"learning_rate": 1.8620403491448334e-08,
"loss": 1.6191,
"step": 1875
},
{
"epoch": 0.98,
"learning_rate": 1.7600603573921794e-08,
"loss": 1.4985,
"step": 1876
},
{
"epoch": 0.98,
"learning_rate": 1.6609497170834154e-08,
"loss": 1.3945,
"step": 1877
},
{
"epoch": 0.98,
"learning_rate": 1.5647087131031023e-08,
"loss": 1.4707,
"step": 1878
},
{
"epoch": 0.98,
"learning_rate": 1.4713376220875097e-08,
"loss": 1.0735,
"step": 1879
},
{
"epoch": 0.98,
"learning_rate": 1.3808367124233945e-08,
"loss": 1.2207,
"step": 1880
},
{
"epoch": 0.98,
"learning_rate": 1.2932062442475579e-08,
"loss": 1.7051,
"step": 1881
},
{
"epoch": 0.98,
"learning_rate": 1.2084464694459563e-08,
"loss": 1.5303,
"step": 1882
},
{
"epoch": 0.99,
"learning_rate": 1.1265576316530358e-08,
"loss": 1.0596,
"step": 1883
},
{
"epoch": 0.99,
"learning_rate": 1.0475399662508434e-08,
"loss": 1.0874,
"step": 1884
},
{
"epoch": 0.99,
"learning_rate": 9.713937003685836e-09,
"loss": 1.3599,
"step": 1885
},
{
"epoch": 0.99,
"learning_rate": 8.981190528819517e-09,
"loss": 1.5312,
"step": 1886
},
{
"epoch": 0.99,
"learning_rate": 8.277162344123568e-09,
"loss": 1.5804,
"step": 1887
},
{
"epoch": 0.99,
"learning_rate": 7.60185447326478e-09,
"loss": 1.4702,
"step": 1888
},
{
"epoch": 0.99,
"learning_rate": 6.955268857353759e-09,
"loss": 1.4707,
"step": 1889
},
{
"epoch": 0.99,
"learning_rate": 6.3374073549438144e-09,
"loss": 1.3647,
"step": 1890
},
{
"epoch": 0.99,
"learning_rate": 5.748271742023192e-09,
"loss": 1.561,
"step": 1891
},
{
"epoch": 0.99,
"learning_rate": 5.187863712007302e-09,
"loss": 1.5264,
"step": 1892
},
{
"epoch": 0.99,
"learning_rate": 4.6561848757420425e-09,
"loss": 1.5068,
"step": 1893
},
{
"epoch": 0.99,
"learning_rate": 4.153236761488266e-09,
"loss": 1.5322,
"step": 1894
},
{
"epoch": 0.99,
"learning_rate": 3.679020814926215e-09,
"loss": 1.394,
"step": 1895
},
{
"epoch": 0.99,
"learning_rate": 3.2335383991477507e-09,
"loss": 1.0673,
"step": 1896
},
{
"epoch": 0.99,
"learning_rate": 2.8167907946519134e-09,
"loss": 1.1443,
"step": 1897
},
{
"epoch": 0.99,
"learning_rate": 2.428779199341591e-09,
"loss": 1.7559,
"step": 1898
},
{
"epoch": 0.99,
"learning_rate": 2.069504728522409e-09,
"loss": 1.1357,
"step": 1899
},
{
"epoch": 0.99,
"learning_rate": 1.7389684148960694e-09,
"loss": 1.2192,
"step": 1900
},
{
"epoch": 0.99,
"learning_rate": 1.4371712085603506e-09,
"loss": 1.2759,
"step": 1901
},
{
"epoch": 1.0,
"learning_rate": 1.164113977003556e-09,
"loss": 1.2883,
"step": 1902
},
{
"epoch": 1.0,
"learning_rate": 9.197975051045138e-10,
"loss": 1.4995,
"step": 1903
},
{
"epoch": 1.0,
"learning_rate": 7.042224951292475e-10,
"loss": 1.0126,
"step": 1904
},
{
"epoch": 1.0,
"learning_rate": 5.173895667287543e-10,
"loss": 0.9268,
"step": 1905
},
{
"epoch": 1.0,
"learning_rate": 3.5929925693789504e-10,
"loss": 1.0811,
"step": 1906
},
{
"epoch": 1.0,
"learning_rate": 2.2995202017317418e-10,
"loss": 1.2202,
"step": 1907
},
{
"epoch": 1.0,
"learning_rate": 1.2934822823051917e-10,
"loss": 1.3159,
"step": 1908
},
{
"epoch": 1.0,
"learning_rate": 5.748817028750075e-11,
"loss": 1.5366,
"step": 1909
},
{
"epoch": 1.0,
"learning_rate": 1.4372052900002432e-11,
"loss": 1.2114,
"step": 1910
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 1.5439,
"step": 1911
},
{
"epoch": 1.0,
"step": 1911,
"total_flos": 1.3552964313840353e+19,
"train_loss": 1.450651798617584,
"train_runtime": 41416.7741,
"train_samples_per_second": 0.369,
"train_steps_per_second": 0.046
}
],
"max_steps": 1911,
"num_train_epochs": 1,
"total_flos": 1.3552964313840353e+19,
"trial_name": null,
"trial_params": null
}