{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999360204734485,
  "global_step": 781,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 8.333333333333333e-07,
      "loss": 3.0762,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 3.1973,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.5e-06,
      "loss": 3.2324,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 3.21,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.166666666666667e-06,
      "loss": 3.0703,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-06,
      "loss": 2.9287,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.833333333333334e-06,
      "loss": 2.9326,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.666666666666667e-06,
      "loss": 2.7061,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.500000000000001e-06,
      "loss": 2.6777,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.333333333333334e-06,
      "loss": 2.6816,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 9.166666666666666e-06,
      "loss": 2.3823,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 1e-05,
      "loss": 2.4517,
      "step": 12
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 2.5,
      "step": 13
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 2.4087,
      "step": 14
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.25e-05,
      "loss": 2.4521,
      "step": 15
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 2.4209,
      "step": 16
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.416666666666667e-05,
      "loss": 2.3804,
      "step": 17
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 2.3135,
      "step": 18
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 2.3203,
      "step": 19
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 2.3721,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 2.3086,
      "step": 21
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 2.3511,
      "step": 22
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.916666666666667e-05,
      "loss": 2.3682,
      "step": 23
    },
    {
      "epoch": 0.03,
      "learning_rate": 2e-05,
      "loss": 2.1953,
      "step": 24
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9999913885285243e-05,
      "loss": 2.209,
      "step": 25
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.9999655542624107e-05,
      "loss": 2.2524,
      "step": 26
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.999922497646602e-05,
      "loss": 2.2725,
      "step": 27
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9998622194226605e-05,
      "loss": 2.1963,
      "step": 28
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9997847206287534e-05,
      "loss": 2.2461,
      "step": 29
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9996900025996385e-05,
      "loss": 2.3765,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999578066966639e-05,
      "loss": 2.2871,
      "step": 31
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9994489156576158e-05,
      "loss": 2.333,
      "step": 32
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999302550896935e-05,
      "loss": 2.2876,
      "step": 33
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.999138975205428e-05,
      "loss": 2.1973,
      "step": 34
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.9989581914003496e-05,
      "loss": 2.2651,
      "step": 35
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9987602025953293e-05,
      "loss": 2.3105,
      "step": 36
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.998545012200317e-05,
      "loss": 2.2871,
      "step": 37
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9983126239215243e-05,
      "loss": 2.2607,
      "step": 38
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9980630417613615e-05,
      "loss": 2.2407,
      "step": 39
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.997796270018368e-05,
      "loss": 2.2241,
      "step": 40
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9975123132871376e-05,
      "loss": 2.2578,
      "step": 41
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9972111764582416e-05,
      "loss": 2.1167,
      "step": 42
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9968928647181423e-05,
      "loss": 2.2305,
      "step": 43
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9965573835491046e-05,
      "loss": 2.2188,
      "step": 44
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9962047387291017e-05,
      "loss": 2.2451,
      "step": 45
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.995834936331715e-05,
      "loss": 2.2041,
      "step": 46
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9954479827260297e-05,
      "loss": 2.1416,
      "step": 47
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9950438845765263e-05,
      "loss": 2.165,
      "step": 48
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.994622648842964e-05,
      "loss": 2.2251,
      "step": 49
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.994184282780262e-05,
      "loss": 2.1714,
      "step": 50
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9937287939383737e-05,
      "loss": 2.186,
      "step": 51
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9932561901621574e-05,
      "loss": 2.2227,
      "step": 52
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9927664795912412e-05,
      "loss": 2.144,
      "step": 53
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.992259670659882e-05,
      "loss": 2.1797,
      "step": 54
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9917357720968213e-05,
      "loss": 2.1357,
      "step": 55
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9911947929251342e-05,
      "loss": 2.229,
      "step": 56
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9906367424620747e-05,
      "loss": 2.1748,
      "step": 57
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9900616303189126e-05,
      "loss": 2.1392,
      "step": 58
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.989469466400773e-05,
      "loss": 2.105,
      "step": 59
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9888602609064604e-05,
      "loss": 2.1284,
      "step": 60
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.988234024328287e-05,
      "loss": 2.103,
      "step": 61
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.987590767451889e-05,
      "loss": 2.1504,
      "step": 62
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9869305013560428e-05,
      "loss": 2.1836,
      "step": 63
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9862532374124745e-05,
      "loss": 2.1982,
      "step": 64
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9855589872856617e-05,
      "loss": 2.1484,
      "step": 65
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.984847762932635e-05,
      "loss": 2.189,
      "step": 66
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.984119576602771e-05,
      "loss": 2.2363,
      "step": 67
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9833744408375804e-05,
      "loss": 2.1582,
      "step": 68
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.982612368470495e-05,
      "loss": 2.1382,
      "step": 69
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9818333726266435e-05,
      "loss": 2.1602,
      "step": 70
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.981037466722627e-05,
      "loss": 2.1313,
      "step": 71
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.980224664466287e-05,
      "loss": 2.1372,
      "step": 72
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9793949798564703e-05,
      "loss": 2.1133,
      "step": 73
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9785484271827884e-05,
      "loss": 2.1406,
      "step": 74
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.977685021025369e-05,
      "loss": 2.0957,
      "step": 75
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9768047762546076e-05,
      "loss": 2.1333,
      "step": 76
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.975907708030909e-05,
      "loss": 2.1313,
      "step": 77
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9749938318044286e-05,
      "loss": 2.123,
      "step": 78
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9740631633148045e-05,
      "loss": 2.1191,
      "step": 79
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9731157185908866e-05,
      "loss": 2.165,
      "step": 80
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9721515139504617e-05,
      "loss": 2.1245,
      "step": 81
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9711705659999714e-05,
      "loss": 2.0884,
      "step": 82
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.970172891634226e-05,
      "loss": 2.1914,
      "step": 83
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9691585080361138e-05,
      "loss": 2.1206,
      "step": 84
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9681274326763068e-05,
      "loss": 2.1738,
      "step": 85
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9670796833129563e-05,
      "loss": 2.1753,
      "step": 86
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9660152779913898e-05,
      "loss": 2.1592,
      "step": 87
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9649342350437998e-05,
      "loss": 2.1284,
      "step": 88
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9638365730889266e-05,
      "loss": 2.0859,
      "step": 89
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9627223110317402e-05,
      "loss": 2.1558,
      "step": 90
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.961591468063112e-05,
      "loss": 2.2393,
      "step": 91
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9604440636594863e-05,
      "loss": 2.1064,
      "step": 92
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9592801175825435e-05,
      "loss": 2.0718,
      "step": 93
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9580996498788602e-05,
      "loss": 2.0894,
      "step": 94
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9569026808795647e-05,
      "loss": 2.1514,
      "step": 95
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.955689231199986e-05,
      "loss": 2.1646,
      "step": 96
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.954459321739298e-05,
      "loss": 2.124,
      "step": 97
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9532129736801616e-05,
      "loss": 2.0737,
      "step": 98
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9519502084883585e-05,
      "loss": 2.0039,
      "step": 99
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9506710479124212e-05,
      "loss": 2.1216,
      "step": 100
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9493755139832593e-05,
      "loss": 2.1719,
      "step": 101
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9480636290137803e-05,
      "loss": 2.1074,
      "step": 102
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.946735415598503e-05,
      "loss": 2.1665,
      "step": 103
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9453908966131728e-05,
      "loss": 2.0825,
      "step": 104
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9440300952143628e-05,
      "loss": 2.1929,
      "step": 105
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9426530348390777e-05,
      "loss": 2.1001,
      "step": 106
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.94125973920435e-05,
      "loss": 2.1572,
      "step": 107
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9398502323068306e-05,
      "loss": 2.1309,
      "step": 108
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9384245384223767e-05,
      "loss": 2.124,
      "step": 109
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9369826821056327e-05,
      "loss": 2.1812,
      "step": 110
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9355246881896078e-05,
      "loss": 2.1392,
      "step": 111
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.934050581785248e-05,
      "loss": 2.1357,
      "step": 112
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.932560388281003e-05,
      "loss": 2.1147,
      "step": 113
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.931054133342392e-05,
      "loss": 2.1904,
      "step": 114
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.929531842911557e-05,
      "loss": 2.1313,
      "step": 115
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.927993543206819e-05,
      "loss": 2.1089,
      "step": 116
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9264392607222263e-05,
      "loss": 2.022,
      "step": 117
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9248690222270983e-05,
      "loss": 2.0171,
      "step": 118
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9232828547655616e-05,
      "loss": 2.1177,
      "step": 119
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.921680785656089e-05,
      "loss": 2.0728,
      "step": 120
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.920062842491025e-05,
      "loss": 2.0376,
      "step": 121
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.918429053136112e-05,
      "loss": 2.0513,
      "step": 122
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.916779445730011e-05,
      "loss": 2.0405,
      "step": 123
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9151140486838173e-05,
      "loss": 2.0898,
      "step": 124
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.913432890680568e-05,
      "loss": 2.1895,
      "step": 125
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9117360006747524e-05,
      "loss": 2.0532,
      "step": 126
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.910023407891809e-05,
      "loss": 2.1719,
      "step": 127
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.908295141827627e-05,
      "loss": 2.1235,
      "step": 128
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9065512322480332e-05,
      "loss": 2.0361,
      "step": 129
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9047917091882835e-05,
      "loss": 2.1084,
      "step": 130
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9030166029525427e-05,
      "loss": 2.0547,
      "step": 131
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9012259441133646e-05,
      "loss": 2.1025,
      "step": 132
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8994197635111637e-05,
      "loss": 2.1279,
      "step": 133
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8975980922536862e-05,
      "loss": 2.1646,
      "step": 134
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.8957609617154718e-05,
      "loss": 2.1016,
      "step": 135
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.893908403537315e-05,
      "loss": 2.0938,
      "step": 136
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8920404496257197e-05,
      "loss": 2.0996,
      "step": 137
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8901571321523494e-05,
      "loss": 2.0811,
      "step": 138
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.888258483553474e-05,
      "loss": 2.0811,
      "step": 139
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8863445365294088e-05,
      "loss": 2.0747,
      "step": 140
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8844153240439555e-05,
      "loss": 2.1177,
      "step": 141
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8824708793238303e-05,
      "loss": 2.0962,
      "step": 142
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8805112358580936e-05,
      "loss": 1.9995,
      "step": 143
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.878536427397573e-05,
      "loss": 2.0405,
      "step": 144
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8765464879542817e-05,
      "loss": 2.0479,
      "step": 145
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8745414518008335e-05,
      "loss": 2.0933,
      "step": 146
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.872521353469852e-05,
      "loss": 2.021,
      "step": 147
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.870486227753375e-05,
      "loss": 2.0752,
      "step": 148
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8684361097022566e-05,
      "loss": 2.1606,
      "step": 149
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8663710346255633e-05,
      "loss": 2.0,
      "step": 150
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8642910380899656e-05,
      "loss": 2.0757,
      "step": 151
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8621961559191245e-05,
      "loss": 2.0078,
      "step": 152
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.860086424193077e-05,
      "loss": 2.0132,
      "step": 153
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8579618792476107e-05,
      "loss": 2.0513,
      "step": 154
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8558225576736436e-05,
      "loss": 2.0806,
      "step": 155
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.853668496316589e-05,
      "loss": 2.0444,
      "step": 156
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8514997322757215e-05,
      "loss": 2.0415,
      "step": 157
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.849316302903542e-05,
      "loss": 1.9946,
      "step": 158
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8471182458051285e-05,
      "loss": 2.1104,
      "step": 159
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8449055988374943e-05,
      "loss": 2.0034,
      "step": 160
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8426784001089312e-05,
      "loss": 2.0327,
      "step": 161
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.840436687978356e-05,
      "loss": 2.0664,
      "step": 162
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8381805010546493e-05,
      "loss": 2.1455,
      "step": 163
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.835909878195989e-05,
      "loss": 2.0742,
      "step": 164
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8336248585091834e-05,
      "loss": 2.1191,
      "step": 165
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.831325481348996e-05,
      "loss": 2.0776,
      "step": 166
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.829011786317469e-05,
      "loss": 2.0229,
      "step": 167
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8266838132632396e-05,
      "loss": 2.083,
      "step": 168
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.824341602280855e-05,
      "loss": 2.0801,
      "step": 169
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.821985193710081e-05,
      "loss": 1.9814,
      "step": 170
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8196146281352086e-05,
      "loss": 2.0107,
      "step": 171
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8172299463843525e-05,
      "loss": 2.0088,
      "step": 172
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8148311895287513e-05,
      "loss": 2.0024,
      "step": 173
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8124183988820572e-05,
      "loss": 1.9897,
      "step": 174
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8099916159996263e-05,
      "loss": 2.103,
      "step": 175
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8075508826778013e-05,
      "loss": 2.0537,
      "step": 176
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8050962409531926e-05,
      "loss": 2.0156,
      "step": 177
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8026277331019552e-05,
      "loss": 2.0649,
      "step": 178
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8001454016390587e-05,
      "loss": 2.064,
      "step": 179
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7976492893175563e-05,
      "loss": 2.0864,
      "step": 180
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7951394391278484e-05,
      "loss": 2.0498,
      "step": 181
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7926158942969413e-05,
      "loss": 2.1182,
      "step": 182
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.7900786982877037e-05,
      "loss": 2.0962,
      "step": 183
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.787527894798118e-05,
      "loss": 2.1284,
      "step": 184
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.784963527760527e-05,
      "loss": 2.0625,
      "step": 185
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7823856413408775e-05,
      "loss": 1.8911,
      "step": 186
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7797942799379608e-05,
      "loss": 2.043,
      "step": 187
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7771894881826464e-05,
      "loss": 2.0127,
      "step": 188
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7745713109371138e-05,
      "loss": 2.123,
      "step": 189
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7719397932940808e-05,
      "loss": 2.0132,
      "step": 190
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.7692949805760256e-05,
      "loss": 1.877,
      "step": 191
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7666369183344065e-05,
      "loss": 1.9692,
      "step": 192
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7639656523488777e-05,
      "loss": 2.04,
      "step": 193
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7612812286265015e-05,
      "loss": 2.1187,
      "step": 194
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7585836934009535e-05,
      "loss": 2.0469,
      "step": 195
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.75587309313173e-05,
      "loss": 2.0776,
      "step": 196
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.7531494745033446e-05,
      "loss": 2.0845,
      "step": 197
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.750412884424525e-05,
      "loss": 2.0469,
      "step": 198
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.747663370027407e-05,
      "loss": 2.0996,
      "step": 199
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7449009786667192e-05,
      "loss": 1.9468,
      "step": 200
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.742125757918971e-05,
      "loss": 1.9131,
      "step": 201
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7393377555816307e-05,
      "loss": 2.1069,
      "step": 202
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7365370196723035e-05,
      "loss": 2.0781,
      "step": 203
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7337235984279047e-05,
      "loss": 2.0801,
      "step": 204
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7308975403038278e-05,
      "loss": 2.0181,
      "step": 205
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7280588939731097e-05,
      "loss": 1.9966,
      "step": 206
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7252077083255953e-05,
      "loss": 2.0308,
      "step": 207
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7223440324670918e-05,
      "loss": 2.064,
      "step": 208
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.719467915718526e-05,
      "loss": 2.0063,
      "step": 209
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.716579407615091e-05,
      "loss": 2.1196,
      "step": 210
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7136785579053983e-05,
      "loss": 1.998,
      "step": 211
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.710765416550617e-05,
      "loss": 2.166,
      "step": 212
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7078400337236135e-05,
      "loss": 2.1338,
      "step": 213
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7049024598080902e-05,
      "loss": 2.1411,
      "step": 214
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7019527453977146e-05,
      "loss": 1.9976,
      "step": 215
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.69899094129525e-05,
      "loss": 2.0835,
      "step": 216
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6960170985116796e-05,
      "loss": 2.0518,
      "step": 217
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6930312682653277e-05,
      "loss": 2.1416,
      "step": 218
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6900335019809784e-05,
      "loss": 2.0566,
      "step": 219
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6870238512889895e-05,
      "loss": 1.8994,
      "step": 220
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6840023680244033e-05,
      "loss": 2.2329,
      "step": 221
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.6809691042260535e-05,
      "loss": 1.9653,
      "step": 222
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.677924112135669e-05,
      "loss": 2.0039,
      "step": 223
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6748674441969757e-05,
      "loss": 1.9346,
      "step": 224
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6717991530547906e-05,
      "loss": 2.0918,
      "step": 225
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6687192915541172e-05,
      "loss": 2.063,
      "step": 226
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6656279127392343e-05,
      "loss": 2.0088,
      "step": 227
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.662525069852783e-05,
      "loss": 1.9795,
      "step": 228
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6594108163348494e-05,
      "loss": 1.9824,
      "step": 229
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6562852058220436e-05,
      "loss": 2.0737,
      "step": 230
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.653148292146578e-05,
      "loss": 2.0659,
      "step": 231
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6500001293353375e-05,
      "loss": 2.0552,
      "step": 232
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6468407716089503e-05,
      "loss": 1.9307,
      "step": 233
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6436702733808546e-05,
      "loss": 2.0146,
      "step": 234
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.640488689256361e-05,
      "loss": 2.0645,
      "step": 235
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6372960740317104e-05,
      "loss": 1.9868,
      "step": 236
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.634092482693133e-05,
      "loss": 1.9507,
      "step": 237
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6308779704159002e-05,
      "loss": 1.9897,
      "step": 238
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.627652592563373e-05,
      "loss": 2.0942,
      "step": 239
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6244164046860503e-05,
      "loss": 2.0078,
      "step": 240
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6211694625206114e-05,
      "loss": 1.9844,
      "step": 241
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6179118219889555e-05,
      "loss": 2.0039,
      "step": 242
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6146435391972405e-05,
      "loss": 1.9844,
      "step": 243
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.611364670434914e-05,
      "loss": 1.9854,
      "step": 244
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6080752721737454e-05,
      "loss": 1.9741,
      "step": 245
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6047754010668536e-05,
      "loss": 2.0376,
      "step": 246
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6014651139477307e-05,
      "loss": 1.915,
      "step": 247
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.598144467829262e-05,
      "loss": 1.9546,
      "step": 248
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5948135199027474e-05,
      "loss": 1.9824,
      "step": 249
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.591472327536912e-05,
      "loss": 2.0122,
      "step": 250
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5881209482769214e-05,
      "loss": 2.0542,
      "step": 251
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5847594398433904e-05,
      "loss": 2.0176,
      "step": 252
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.5813878601313857e-05,
      "loss": 2.0449,
      "step": 253
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5780062672094332e-05,
      "loss": 2.0352,
      "step": 254
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5746147193185142e-05,
      "loss": 1.9858,
      "step": 255
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5712132748710653e-05,
      "loss": 1.957,
      "step": 256
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5678019924499697e-05,
      "loss": 1.9795,
      "step": 257
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.56438093080755e-05,
      "loss": 2.0571,
      "step": 258
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5609501488645555e-05,
      "loss": 1.9717,
      "step": 259
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5575097057091485e-05,
      "loss": 2.0132,
      "step": 260
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5540596605958843e-05,
      "loss": 2.0527,
      "step": 261
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5506000729446943e-05,
      "loss": 1.9658,
      "step": 262
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.547131002339858e-05,
      "loss": 2.0225,
      "step": 263
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5436525085289813e-05,
      "loss": 2.0601,
      "step": 264
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5401646514219642e-05,
      "loss": 2.1587,
      "step": 265
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.536667491089971e-05,
      "loss": 1.9219,
      "step": 266
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5331610877643947e-05,
      "loss": 2.0254,
      "step": 267
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.529645501835819e-05,
      "loss": 1.9697,
      "step": 268
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.526120793852981e-05,
      "loss": 2.0352,
      "step": 269
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5225870245217245e-05,
      "loss": 1.999,
      "step": 270
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5190442547039567e-05,
      "loss": 2.0054,
      "step": 271
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5154925454166005e-05,
      "loss": 2.1294,
      "step": 272
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5119319578305427e-05,
      "loss": 1.9604,
      "step": 273
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5083625532695796e-05,
      "loss": 2.0996,
      "step": 274
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5047843932093627e-05,
      "loss": 2.0474,
      "step": 275
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5011975392763383e-05,
      "loss": 1.9824,
      "step": 276
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.4976020532466872e-05,
      "loss": 2.0278,
      "step": 277
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.49399799704526e-05,
      "loss": 2.0366,
      "step": 278
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4903854327445115e-05,
      "loss": 2.0146,
      "step": 279
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4867644225634302e-05,
      "loss": 1.9131,
      "step": 280
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.483135028866468e-05,
      "loss": 2.0068,
      "step": 281
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4794973141624655e-05,
      "loss": 2.0952,
      "step": 282
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4758513411035753e-05,
      "loss": 2.0737,
      "step": 283
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4721971724841837e-05,
      "loss": 1.9829,
      "step": 284
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.4685348712398281e-05,
      "loss": 1.9634,
      "step": 285
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4648645004461142e-05,
      "loss": 1.9995,
      "step": 286
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4611861233176288e-05,
      "loss": 2.1074,
      "step": 287
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.457499803206851e-05,
      "loss": 1.9814,
      "step": 288
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4538056036030622e-05,
      "loss": 2.0332,
      "step": 289
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4501035881312506e-05,
      "loss": 1.9326,
      "step": 290
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4463938205510186e-05,
      "loss": 1.9854,
      "step": 291
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4426763647554809e-05,
      "loss": 1.9429,
      "step": 292
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4389512847701666e-05,
      "loss": 2.0698,
      "step": 293
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4352186447519163e-05,
      "loss": 1.9658,
      "step": 294
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4314785089877754e-05,
      "loss": 1.9731,
      "step": 295
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4277309418938892e-05,
      "loss": 1.9824,
      "step": 296
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.423976008014392e-05,
      "loss": 1.9312,
      "step": 297
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4202137720202957e-05,
      "loss": 2.0278,
      "step": 298
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4164442987083763e-05,
      "loss": 1.814,
      "step": 299
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4126676530000571e-05,
      "loss": 1.9238,
      "step": 300
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4088838999402924e-05,
      "loss": 2.083,
      "step": 301
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.405093104696445e-05,
      "loss": 1.9365,
      "step": 302
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.401295332557165e-05,
      "loss": 1.9995,
      "step": 303
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3974906489312657e-05,
      "loss": 2.083,
      "step": 304
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3936791193465954e-05,
      "loss": 2.0029,
      "step": 305
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3898608094489115e-05,
      "loss": 2.063,
      "step": 306
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.386035785000747e-05,
      "loss": 2.0127,
      "step": 307
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.3822041118802798e-05,
      "loss": 2.0518,
      "step": 308
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.378365856080198e-05,
      "loss": 1.8789,
      "step": 309
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3745210837065617e-05,
      "loss": 1.9512,
      "step": 310
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3706698609776662e-05,
      "loss": 2.0664,
      "step": 311
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3668122542229014e-05,
      "loss": 1.9902,
      "step": 312
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3629483298816077e-05,
      "loss": 2.0483,
      "step": 313
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.359078154501934e-05,
      "loss": 2.0166,
      "step": 314
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3552017947396898e-05,
      "loss": 1.9863,
      "step": 315
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3513193173571983e-05,
      "loss": 2.0244,
      "step": 316
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3474307892221464e-05,
      "loss": 1.9409,
      "step": 317
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3435362773064316e-05,
      "loss": 2.0181,
      "step": 318
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3396358486850106e-05,
      "loss": 1.9707,
      "step": 319
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3357295705347433e-05,
      "loss": 1.9966,
      "step": 320
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3318175101332353e-05,
      "loss": 1.8218,
      "step": 321
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.32789973485768e-05,
      "loss": 1.9785,
      "step": 322
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.323976312183697e-05,
      "loss": 1.998,
      "step": 323
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3200473096841715e-05,
      "loss": 1.9785,
      "step": 324
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3161127950280894e-05,
      "loss": 2.0083,
      "step": 325
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3121728359793719e-05,
      "loss": 1.8789,
      "step": 326
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3082275003957094e-05,
      "loss": 1.9678,
      "step": 327
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.304276856227391e-05,
      "loss": 2.0073,
      "step": 328
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3003209715161362e-05,
      "loss": 2.1016,
      "step": 329
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.296359914393922e-05,
      "loss": 2.0684,
      "step": 330
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2923937530818082e-05,
      "loss": 1.8833,
      "step": 331
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.2884225558887664e-05,
      "loss": 2.0171,
      "step": 332
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2844463912104979e-05,
      "loss": 1.9766,
      "step": 333
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2804653275282605e-05,
      "loss": 1.9922,
      "step": 334
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2764794334076876e-05,
      "loss": 1.936,
      "step": 335
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2724887774976051e-05,
      "loss": 1.9175,
      "step": 336
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2684934285288527e-05,
      "loss": 1.8794,
      "step": 337
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2644934553130981e-05,
      "loss": 1.9541,
      "step": 338
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.2604889267416511e-05,
      "loss": 1.8853,
      "step": 339
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2564799117842796e-05,
      "loss": 1.9663,
      "step": 340
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2524664794880189e-05,
      "loss": 1.9648,
      "step": 341
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2484486989759846e-05,
      "loss": 2.0068,
      "step": 342
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2444266394461808e-05,
      "loss": 2.0991,
      "step": 343
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2404003701703102e-05,
      "loss": 1.8794,
      "step": 344
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.236369960492579e-05,
      "loss": 1.9502,
      "step": 345
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.232335479828502e-05,
      "loss": 2.0928,
      "step": 346
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2282969976637107e-05,
      "loss": 1.8931,
      "step": 347
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2242545835527519e-05,
      "loss": 2.0435,
      "step": 348
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2202083071178938e-05,
      "loss": 1.9668,
      "step": 349
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2161582380479251e-05,
      "loss": 2.04,
      "step": 350
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2121044460969538e-05,
      "loss": 1.9336,
      "step": 351
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2080470010832074e-05,
      "loss": 2.0298,
      "step": 352
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2039859728878298e-05,
      "loss": 1.9668,
      "step": 353
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1999214314536783e-05,
      "loss": 1.9595,
      "step": 354
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.1958534467841178e-05,
      "loss": 2.0029,
      "step": 355
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1917820889418165e-05,
      "loss": 2.0039,
      "step": 356
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1877074280475383e-05,
      "loss": 2.0479,
      "step": 357
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1836295342789348e-05,
      "loss": 1.9512,
      "step": 358
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1795484778693382e-05,
      "loss": 2.0303,
      "step": 359
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1754643291065503e-05,
      "loss": 1.8911,
      "step": 360
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1713771583316315e-05,
      "loss": 1.9609,
      "step": 361
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1672870359376922e-05,
      "loss": 1.938,
      "step": 362
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.1631940323686758e-05,
      "loss": 1.9536,
      "step": 363
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.15909821811815e-05,
      "loss": 1.8066,
      "step": 364
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.15499966372809e-05,
      "loss": 2.0933,
      "step": 365
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1508984397876636e-05,
      "loss": 1.9395,
      "step": 366
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1467946169320175e-05,
      "loss": 1.9658,
      "step": 367
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1426882658410578e-05,
      "loss": 1.9121,
      "step": 368
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1385794572382359e-05,
      "loss": 2.0469,
      "step": 369
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1344682618893279e-05,
      "loss": 1.9639,
      "step": 370
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1303547506012163e-05,
      "loss": 1.9038,
      "step": 371
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1262389942206713e-05,
      "loss": 2.0273,
      "step": 372
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1221210636331307e-05,
      "loss": 1.9492,
      "step": 373
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1180010297614779e-05,
      "loss": 2.0322,
      "step": 374
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1138789635648212e-05,
      "loss": 1.9922,
      "step": 375
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1097549360372718e-05,
      "loss": 1.8462,
      "step": 376
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.10562901820672e-05,
      "loss": 2.0283,
      "step": 377
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.101501281133614e-05,
      "loss": 1.9414,
      "step": 378
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.0973717959097327e-05,
      "loss": 1.9673,
      "step": 379
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0932406336569658e-05,
      "loss": 1.8564,
      "step": 380
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0891078655260845e-05,
      "loss": 1.9932,
      "step": 381
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0849735626955186e-05,
      "loss": 2.0469,
      "step": 382
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0808377963701302e-05,
      "loss": 1.9185,
      "step": 383
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0767006377799864e-05,
      "loss": 1.8628,
      "step": 384
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0725621581791336e-05,
      "loss": 1.979,
      "step": 385
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.0684224288443701e-05,
      "loss": 1.9805,
      "step": 386
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0642815210740181e-05,
      "loss": 1.979,
      "step": 387
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0601395061866957e-05,
      "loss": 1.8887,
      "step": 388
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.055996455520089e-05,
      "loss": 1.8184,
      "step": 389
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0518524404297232e-05,
      "loss": 1.9312,
      "step": 390
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0477075322877342e-05,
      "loss": 1.9297,
      "step": 391
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0435618024816381e-05,
      "loss": 1.9692,
      "step": 392
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0394153224131031e-05,
      "loss": 1.9863,
      "step": 393
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0352681634967185e-05,
      "loss": 1.9209,
      "step": 394
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0311203971587662e-05,
      "loss": 1.9927,
      "step": 395
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0269720948359891e-05,
      "loss": 1.9634,
      "step": 396
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0228233279743608e-05,
      "loss": 1.9648,
      "step": 397
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0186741680278573e-05,
      "loss": 1.8765,
      "step": 398
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0145246864572231e-05,
      "loss": 1.8594,
      "step": 399
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0103749547287425e-05,
      "loss": 1.8989,
      "step": 400
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0062250443130087e-05,
      "loss": 1.9629,
      "step": 401
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0020750266836919e-05,
      "loss": 2.0103,
      "step": 402
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.979249733163088e-06,
      "loss": 1.8828,
      "step": 403
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.937749556869914e-06,
      "loss": 1.9453,
      "step": 404
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.896250452712576e-06,
      "loss": 1.9536,
      "step": 405
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.854753135427772e-06,
      "loss": 1.979,
      "step": 406
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.813258319721425e-06,
      "loss": 1.9497,
      "step": 407
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.771766720256393e-06,
      "loss": 1.9351,
      "step": 408
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.73027905164011e-06,
      "loss": 1.9131,
      "step": 409
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.68879602841234e-06,
      "loss": 1.9775,
      "step": 410
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.647318365032817e-06,
      "loss": 1.8779,
      "step": 411
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.605846775868969e-06,
      "loss": 1.9995,
      "step": 412
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.564381975183622e-06,
      "loss": 2.083,
      "step": 413
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.522924677122658e-06,
      "loss": 2.0142,
      "step": 414
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.48147559570277e-06,
      "loss": 1.8374,
      "step": 415
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.440035444799113e-06,
      "loss": 1.9844,
      "step": 416
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.398604938133048e-06,
      "loss": 1.8804,
      "step": 417
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.357184789259824e-06,
      "loss": 2.0132,
      "step": 418
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.315775711556299e-06,
      "loss": 2.0073,
      "step": 419
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.274378418208666e-06,
      "loss": 1.9912,
      "step": 420
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.232993622200139e-06,
      "loss": 1.9824,
      "step": 421
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.191622036298701e-06,
      "loss": 1.9194,
      "step": 422
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.150264373044815e-06,
      "loss": 2.1201,
      "step": 423
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.108921344739156e-06,
      "loss": 1.9688,
      "step": 424
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.067593663430343e-06,
      "loss": 1.96,
      "step": 425
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.026282040902675e-06,
      "loss": 1.9531,
      "step": 426
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.984987188663867e-06,
      "loss": 1.9429,
      "step": 427
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.943709817932803e-06,
      "loss": 1.9868,
      "step": 428
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.902450639627287e-06,
      "loss": 1.9922,
      "step": 429
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.86121036435179e-06,
      "loss": 1.8721,
      "step": 430
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.819989702385222e-06,
      "loss": 1.9849,
      "step": 431
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.778789363668696e-06,
      "loss": 1.8711,
      "step": 432
    },
    {
      "epoch": 0.55,
      "learning_rate": 8.73761005779329e-06,
      "loss": 1.9146,
      "step": 433
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.696452493987844e-06,
      "loss": 1.8477,
      "step": 434
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.655317381106725e-06,
      "loss": 1.8242,
      "step": 435
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.614205427617641e-06,
      "loss": 1.9556,
      "step": 436
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.573117341589423e-06,
      "loss": 2.0742,
      "step": 437
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.53205383067983e-06,
      "loss": 2.0029,
      "step": 438
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.491015602123369e-06,
      "loss": 2.0884,
      "step": 439
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.450003362719104e-06,
      "loss": 1.9546,
      "step": 440
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.409017818818505e-06,
      "loss": 1.877,
      "step": 441
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.368059676313243e-06,
      "loss": 1.978,
      "step": 442
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.327129640623081e-06,
      "loss": 1.9209,
      "step": 443
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.286228416683686e-06,
      "loss": 1.9639,
      "step": 444
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.2453567089345e-06,
      "loss": 1.9077,
      "step": 445
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.204515221306623e-06,
      "loss": 1.8823,
      "step": 446
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.163704657210653e-06,
      "loss": 1.8364,
      "step": 447
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.12292571952462e-06,
      "loss": 1.9404,
      "step": 448
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.082179110581838e-06,
      "loss": 2.0288,
      "step": 449
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.041465532158822e-06,
      "loss": 1.8667,
      "step": 450
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.000785685463222e-06,
      "loss": 1.8896,
      "step": 451
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.960140271121704e-06,
      "loss": 1.9355,
      "step": 452
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.91952998916793e-06,
      "loss": 2.0508,
      "step": 453
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.878955539030467e-06,
      "loss": 1.877,
      "step": 454
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.838417619520749e-06,
      "loss": 1.9707,
      "step": 455
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.797916928821064e-06,
      "loss": 1.9692,
      "step": 456
    },
    {
      "epoch": 0.58,
      "learning_rate": 7.757454164472483e-06,
      "loss": 1.8623,
      "step": 457
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.7170300233629e-06,
      "loss": 1.9805,
      "step": 458
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.676645201714982e-06,
      "loss": 1.8755,
      "step": 459
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.636300395074212e-06,
      "loss": 1.9907,
      "step": 460
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.595996298296899e-06,
      "loss": 1.8579,
      "step": 461
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.555733605538191e-06,
      "loss": 1.9761,
      "step": 462
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.51551301024016e-06,
      "loss": 1.9346,
      "step": 463
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.475335205119816e-06,
      "loss": 1.7603,
      "step": 464
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.435200882157209e-06,
      "loss": 1.9668,
      "step": 465
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.395110732583491e-06,
      "loss": 1.9907,
      "step": 466
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.35506544686902e-06,
      "loss": 1.894,
      "step": 467
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.315065714711473e-06,
      "loss": 1.8774,
      "step": 468
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.2751122250239505e-06,
      "loss": 1.8784,
      "step": 469
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.235205665923131e-06,
      "loss": 2.002,
      "step": 470
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.195346724717397e-06,
      "loss": 2.019,
      "step": 471
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.155536087895023e-06,
      "loss": 1.8442,
      "step": 472
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.11577444111234e-06,
      "loss": 1.9863,
      "step": 473
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.0760624691819164e-06,
      "loss": 1.8501,
      "step": 474
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.0364008560607854e-06,
      "loss": 1.9819,
      "step": 475
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.996790284838639e-06,
      "loss": 1.9263,
      "step": 476
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.957231437726094e-06,
      "loss": 1.9229,
      "step": 477
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.917724996042909e-06,
      "loss": 1.9482,
      "step": 478
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.8782716402062825e-06,
      "loss": 1.8799,
      "step": 479
    },
    {
      "epoch": 0.61,
      "learning_rate": 6.838872049719109e-06,
      "loss": 2.0112,
      "step": 480
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.799526903158288e-06,
      "loss": 1.8721,
      "step": 481
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.760236878163035e-06,
      "loss": 1.8999,
      "step": 482
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.721002651423203e-06,
      "loss": 1.9355,
      "step": 483
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.681824898667648e-06,
      "loss": 1.9556,
      "step": 484
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.64270429465257e-06,
      "loss": 1.8853,
      "step": 485
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.603641513149897e-06,
      "loss": 2.0312,
      "step": 486
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.564637226935691e-06,
      "loss": 1.8989,
      "step": 487
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.52569210777854e-06,
      "loss": 1.9609,
      "step": 488
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.486806826428017e-06,
      "loss": 2.0303,
      "step": 489
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.447982052603105e-06,
      "loss": 1.9087,
      "step": 490
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.409218454980663e-06,
      "loss": 1.9595,
      "step": 491
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.370516701183926e-06,
      "loss": 1.9961,
      "step": 492
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.331877457770987e-06,
      "loss": 1.9521,
      "step": 493
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.293301390223339e-06,
      "loss": 1.9722,
      "step": 494
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.254789162934386e-06,
      "loss": 1.9175,
      "step": 495
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.216341439198023e-06,
      "loss": 1.9355,
      "step": 496
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.177958881197205e-06,
      "loss": 1.9277,
      "step": 497
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.139642149992532e-06,
      "loss": 1.8589,
      "step": 498
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.101391905510889e-06,
      "loss": 2.0146,
      "step": 499
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.063208806534049e-06,
      "loss": 1.79,
      "step": 500
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.0250935106873474e-06,
      "loss": 1.9204,
      "step": 501
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.987046674428354e-06,
      "loss": 1.9365,
      "step": 502
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.949068953035551e-06,
      "loss": 1.896,
      "step": 503
    },
    {
      "epoch": 0.64,
      "learning_rate": 5.911161000597079e-06,
      "loss": 1.9868,
      "step": 504
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.87332346999943e-06,
      "loss": 1.8936,
      "step": 505
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.835557012916242e-06,
      "loss": 1.9629,
      "step": 506
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.797862279797046e-06,
      "loss": 1.8838,
      "step": 507
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.7602399198560835e-06,
      "loss": 2.0098,
      "step": 508
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.72269058106111e-06,
      "loss": 1.8672,
      "step": 509
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.6852149101222475e-06,
      "loss": 1.834,
      "step": 510
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.647813552480843e-06,
      "loss": 1.918,
      "step": 511
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.610487152298334e-06,
      "loss": 1.9829,
      "step": 512
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.573236352445193e-06,
      "loss": 1.8784,
      "step": 513
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.536061794489818e-06,
      "loss": 2.0059,
      "step": 514
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.498964118687497e-06,
      "loss": 2.0068,
      "step": 515
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.461943963969384e-06,
      "loss": 1.9331,
      "step": 516
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.425001967931494e-06,
      "loss": 1.873,
      "step": 517
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.388138766823714e-06,
      "loss": 1.8662,
      "step": 518
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.351354995538859e-06,
      "loss": 1.9282,
      "step": 519
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.314651287601719e-06,
      "loss": 1.9414,
      "step": 520
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.278028275158167e-06,
      "loss": 1.855,
      "step": 521
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.241486588964246e-06,
      "loss": 1.9526,
      "step": 522
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.205026858375349e-06,
      "loss": 1.9907,
      "step": 523
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.168649711335323e-06,
      "loss": 1.9683,
      "step": 524
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.132355774365701e-06,
      "loss": 1.8896,
      "step": 525
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.096145672554886e-06,
      "loss": 1.9541,
      "step": 526
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.060020029547401e-06,
      "loss": 1.9482,
      "step": 527
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.02397946753313e-06,
      "loss": 1.9028,
      "step": 528
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.988024607236619e-06,
      "loss": 1.9028,
      "step": 529
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.952156067906378e-06,
      "loss": 1.9551,
      "step": 530
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.916374467304208e-06,
      "loss": 1.9941,
      "step": 531
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.880680421694573e-06,
      "loss": 1.9458,
      "step": 532
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.845074545833996e-06,
      "loss": 1.9141,
      "step": 533
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.809557452960436e-06,
      "loss": 2.0146,
      "step": 534
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.774129754782759e-06,
      "loss": 1.9121,
      "step": 535
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.738792061470191e-06,
      "loss": 1.9287,
      "step": 536
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.703544981641807e-06,
      "loss": 1.8999,
      "step": 537
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.668389122356054e-06,
      "loss": 1.9165,
      "step": 538
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.633325089100289e-06,
      "loss": 1.9824,
      "step": 539
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.598353485780362e-06,
      "loss": 1.8184,
      "step": 540
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.563474914710191e-06,
      "loss": 1.8813,
      "step": 541
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.528689976601423e-06,
      "loss": 1.8662,
      "step": 542
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.4939992705530615e-06,
      "loss": 1.8066,
      "step": 543
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.459403394041156e-06,
      "loss": 2.0332,
      "step": 544
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.424902942908518e-06,
      "loss": 1.8779,
      "step": 545
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.390498511354445e-06,
      "loss": 1.8657,
      "step": 546
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.356190691924505e-06,
      "loss": 1.9136,
      "step": 547
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.321980075500305e-06,
      "loss": 2.0371,
      "step": 548
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.287867251289348e-06,
      "loss": 1.834,
      "step": 549
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.25385280681486e-06,
      "loss": 2.0015,
      "step": 550
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.219937327905672e-06,
      "loss": 1.9365,
      "step": 551
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.1861213986861445e-06,
      "loss": 1.8994,
      "step": 552
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.1524056015661e-06,
      "loss": 1.8975,
      "step": 553
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.118790517230788e-06,
      "loss": 1.8511,
      "step": 554
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.085276724630882e-06,
      "loss": 1.8867,
      "step": 555
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.051864800972528e-06,
      "loss": 1.9102,
      "step": 556
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.018555321707382e-06,
      "loss": 1.9824,
      "step": 557
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.985348860522698e-06,
      "loss": 1.9912,
      "step": 558
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.952245989331466e-06,
      "loss": 1.9609,
      "step": 559
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.919247278262548e-06,
      "loss": 1.9033,
      "step": 560
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.886353295650863e-06,
      "loss": 1.8657,
      "step": 561
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.8535646080275965e-06,
      "loss": 1.979,
      "step": 562
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.820881780110445e-06,
      "loss": 1.9585,
      "step": 563
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7883053747938927e-06,
      "loss": 1.9097,
      "step": 564
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7558359531394983e-06,
      "loss": 1.916,
      "step": 565
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.7234740743662746e-06,
      "loss": 1.9028,
      "step": 566
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6912202958410024e-06,
      "loss": 1.9531,
      "step": 567
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6590751730686726e-06,
      "loss": 1.9165,
      "step": 568
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6270392596828995e-06,
      "loss": 1.9932,
      "step": 569
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5951131074363942e-06,
      "loss": 1.9619,
      "step": 570
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5632972661914543e-06,
      "loss": 1.9092,
      "step": 571
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.5315922839104976e-06,
      "loss": 1.9297,
      "step": 572
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4999987066466268e-06,
      "loss": 2.0015,
      "step": 573
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.4685170785342235e-06,
      "loss": 1.915,
      "step": 574
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.437147941779564e-06,
      "loss": 1.9634,
      "step": 575
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.4058918366515114e-06,
      "loss": 1.9263,
      "step": 576
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.374749301472173e-06,
      "loss": 1.9575,
      "step": 577
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.34372087260766e-06,
      "loss": 2.022,
      "step": 578
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.312807084458831e-06,
      "loss": 1.9121,
      "step": 579
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2820084694520973e-06,
      "loss": 1.9155,
      "step": 580
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.251325558030245e-06,
      "loss": 1.9087,
      "step": 581
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.2207588786433108e-06,
      "loss": 1.8374,
      "step": 582
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1903089577394707e-06,
      "loss": 1.8921,
      "step": 583
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1599763197559705e-06,
      "loss": 1.9136,
      "step": 584
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.1297614871101035e-06,
      "loss": 1.8564,
      "step": 585
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0996649801902167e-06,
      "loss": 1.9043,
      "step": 586
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0696873173467245e-06,
      "loss": 1.9072,
      "step": 587
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0398290148832056e-06,
      "loss": 1.8618,
      "step": 588
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.0100905870475004e-06,
      "loss": 1.9229,
      "step": 589
    },
    {
      "epoch": 0.75,
      "learning_rate": 2.980472546022859e-06,
      "loss": 1.8867,
      "step": 590
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9509754019191017e-06,
      "loss": 1.9492,
      "step": 591
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.921599662763868e-06,
      "loss": 1.8779,
      "step": 592
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8923458344938383e-06,
      "loss": 1.793,
      "step": 593
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.863214420946021e-06,
      "loss": 1.8242,
      "step": 594
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8342059238490937e-06,
      "loss": 1.8652,
      "step": 595
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.8053208428147462e-06,
      "loss": 1.9077,
      "step": 596
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.7765596753290826e-06,
      "loss": 1.8579,
      "step": 597
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.747922916744048e-06,
      "loss": 1.9146,
      "step": 598
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.719411060268903e-06,
      "loss": 1.8804,
      "step": 599
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.691024596961729e-06,
      "loss": 1.8955,
      "step": 600
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6627640157209523e-06,
      "loss": 1.8413,
      "step": 601
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6346298032769657e-06,
      "loss": 1.9165,
      "step": 602
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.6066224441836963e-06,
      "loss": 1.9927,
      "step": 603
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.578742420810294e-06,
      "loss": 1.9009,
      "step": 604
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.5509902133328113e-06,
      "loss": 1.9141,
      "step": 605
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.5233662997259336e-06,
      "loss": 1.9541,
      "step": 606
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4958711557547523e-06,
      "loss": 1.9219,
      "step": 607
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4685052549665556e-06,
      "loss": 1.9409,
      "step": 608
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.4412690686827e-06,
      "loss": 1.9028,
      "step": 609
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.414163065990468e-06,
      "loss": 1.8433,
      "step": 610
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3871877137349887e-06,
      "loss": 1.8901,
      "step": 611
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3603434765112266e-06,
      "loss": 1.915,
      "step": 612
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.3336308166559385e-06,
      "loss": 1.8887,
      "step": 613
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.307050194239746e-06,
      "loss": 1.9546,
      "step": 614
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.280602067059192e-06,
      "loss": 2.0059,
      "step": 615
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2542868906288617e-06,
      "loss": 1.9775,
      "step": 616
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.22810511817354e-06,
      "loss": 1.9321,
      "step": 617
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.2020572006203934e-06,
      "loss": 1.8691,
      "step": 618
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1761435865912296e-06,
      "loss": 1.9443,
      "step": 619
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.1503647223947355e-06,
      "loss": 1.8301,
      "step": 620
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.124721052018821e-06,
      "loss": 1.939,
      "step": 621
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0992130171229653e-06,
      "loss": 1.8184,
      "step": 622
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.07384105703059e-06,
      "loss": 1.8618,
      "step": 623
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0486056087215177e-06,
      "loss": 1.9854,
      "step": 624
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.0235071068244383e-06,
      "loss": 1.8643,
      "step": 625
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.998545983609417e-06,
      "loss": 1.9487,
      "step": 626
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9737226689804524e-06,
      "loss": 1.9106,
      "step": 627
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9490375904680747e-06,
      "loss": 1.8037,
      "step": 628
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.9244911732219917e-06,
      "loss": 1.8472,
      "step": 629
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.900083840003739e-06,
      "loss": 1.8271,
      "step": 630
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8758160111794266e-06,
      "loss": 1.96,
      "step": 631
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8516881047124869e-06,
      "loss": 1.9526,
      "step": 632
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.827700536156476e-06,
      "loss": 1.8633,
      "step": 633
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.8038537186479178e-06,
      "loss": 1.8286,
      "step": 634
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7801480628991907e-06,
      "loss": 1.9351,
      "step": 635
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7565839771914551e-06,
      "loss": 1.8481,
      "step": 636
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7331618673676064e-06,
      "loss": 1.7974,
      "step": 637
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7098821368253094e-06,
      "loss": 1.9756,
      "step": 638
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.6867451865100414e-06,
      "loss": 1.9395,
      "step": 639
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.663751414908169e-06,
      "loss": 1.8945,
      "step": 640
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.640901218040113e-06,
      "loss": 1.8223,
      "step": 641
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.618194989453511e-06,
      "loss": 1.8765,
      "step": 642
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5956331202164422e-06,
      "loss": 1.9194,
      "step": 643
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.5732159989106888e-06,
      "loss": 1.9219,
      "step": 644
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5509440116250586e-06,
      "loss": 1.9521,
      "step": 645
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5288175419487172e-06,
      "loss": 1.8164,
      "step": 646
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5068369709645859e-06,
      "loss": 1.9409,
      "step": 647
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4850026772427862e-06,
      "loss": 1.9082,
      "step": 648
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4633150368341154e-06,
      "loss": 1.9541,
      "step": 649
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4417744232635644e-06,
      "loss": 1.8682,
      "step": 650
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.4203812075238932e-06,
      "loss": 1.9004,
      "step": 651
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3991357580692356e-06,
      "loss": 1.8926,
      "step": 652
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3780384408087577e-06,
      "loss": 1.9297,
      "step": 653
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3570896191003457e-06,
      "loss": 1.9326,
      "step": 654
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3362896537443682e-06,
      "loss": 1.8862,
      "step": 655
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3156389029774352e-06,
      "loss": 1.8867,
      "step": 656
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.295137722466253e-06,
      "loss": 1.8989,
      "step": 657
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.274786465301483e-06,
      "loss": 1.9619,
      "step": 658
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2545854819916646e-06,
      "loss": 1.9019,
      "step": 659
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.2345351204571866e-06,
      "loss": 1.9165,
      "step": 660
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.2146357260242726e-06,
      "loss": 1.9258,
      "step": 661
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1948876414190657e-06,
      "loss": 1.8911,
      "step": 662
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1752912067616995e-06,
      "loss": 1.9722,
      "step": 663
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.155846759560444e-06,
      "loss": 1.9258,
      "step": 664
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.136554634705913e-06,
      "loss": 1.9492,
      "step": 665
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.1174151644652653e-06,
      "loss": 1.8486,
      "step": 666
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0984286784765075e-06,
      "loss": 1.8589,
      "step": 667
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.0795955037428052e-06,
      "loss": 1.9229,
      "step": 668
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0609159646268507e-06,
      "loss": 1.8252,
      "step": 669
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.042390382845284e-06,
      "loss": 1.8135,
      "step": 670
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0240190774631376e-06,
      "loss": 1.9185,
      "step": 671
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0058023648883642e-06,
      "loss": 1.7939,
      "step": 672
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.87740558866357e-07,
      "loss": 1.8838,
      "step": 673
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.698339704745729e-07,
      "loss": 2.0044,
      "step": 674
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.520829081171678e-07,
      "loss": 1.9146,
      "step": 675
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.344876775196688e-07,
      "loss": 1.9492,
      "step": 676
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.170485817237329e-07,
      "loss": 1.9033,
      "step": 677
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.997659210819099e-07,
      "loss": 1.8765,
      "step": 678
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.82639993252482e-07,
      "loss": 1.7529,
      "step": 679
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.656710931943213e-07,
      "loss": 1.8853,
      "step": 680
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.488595131618272e-07,
      "loss": 1.8647,
      "step": 681
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.322055426998888e-07,
      "loss": 1.7764,
      "step": 682
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.157094686388833e-07,
      "loss": 1.8828,
      "step": 683
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.993715750897557e-07,
      "loss": 1.8081,
      "step": 684
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.831921434391132e-07,
      "loss": 1.8779,
      "step": 685
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.671714523443852e-07,
      "loss": 1.9888,
      "step": 686
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.513097777290201e-07,
      "loss": 2.0195,
      "step": 687
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.356073927777352e-07,
      "loss": 1.856,
      "step": 688
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.20064567931813e-07,
      "loss": 1.8218,
      "step": 689
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.046815708844346e-07,
      "loss": 1.8584,
      "step": 690
    },
    {
      "epoch": 0.88,
      "learning_rate": 6.894586665760827e-07,
      "loss": 1.98,
      "step": 691
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.743961171899693e-07,
      "loss": 1.8188,
      "step": 692
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.594941821475243e-07,
      "loss": 1.9395,
      "step": 693
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.447531181039246e-07,
      "loss": 1.874,
      "step": 694
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.301731789436738e-07,
      "loss": 1.937,
      "step": 695
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.157546157762361e-07,
      "loss": 1.832,
      "step": 696
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.01497676931696e-07,
      "loss": 1.9634,
      "step": 697
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.87402607956502e-07,
      "loss": 1.8179,
      "step": 698
    },
    {
      "epoch": 0.89,
      "learning_rate": 5.734696516092253e-07,
      "loss": 1.9551,
      "step": 699
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.596990478563735e-07,
      "loss": 1.8711,
      "step": 700
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.460910338682723e-07,
      "loss": 1.9565,
      "step": 701
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.326458440149684e-07,
      "loss": 1.9307,
      "step": 702
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.193637098622007e-07,
      "loss": 1.9829,
      "step": 703
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.062448601674086e-07,
      "loss": 1.8389,
      "step": 704
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.932895208757905e-07,
      "loss": 1.7654,
      "step": 705
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.804979151164191e-07,
      "loss": 1.8574,
      "step": 706
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.6787026319838445e-07,
      "loss": 1.9209,
      "step": 707
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.55406782607023e-07,
      "loss": 1.7476,
      "step": 708
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.431076880001439e-07,
      "loss": 1.9224,
      "step": 709
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.3097319120435356e-07,
      "loss": 1.9302,
      "step": 710
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.190035012113991e-07,
      "loss": 1.9175,
      "step": 711
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.0719882417456726e-07,
      "loss": 1.9766,
      "step": 712
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.9555936340514025e-07,
      "loss": 1.8901,
      "step": 713
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.8408531936888095e-07,
      "loss": 1.9224,
      "step": 714
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.7277688968260117e-07,
      "loss": 1.9058,
      "step": 715
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.616342691107366e-07,
      "loss": 1.9355,
      "step": 716
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.506576495620062e-07,
      "loss": 1.8789,
      "step": 717
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.398472200861047e-07,
      "loss": 1.8716,
      "step": 718
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.2920316687043986e-07,
      "loss": 1.9707,
      "step": 719
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.187256732369337e-07,
      "loss": 1.8418,
      "step": 720
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.0841491963886083e-07,
      "loss": 1.7754,
      "step": 721
    },
    {
      "epoch": 0.92,
      "learning_rate": 2.9827108365774294e-07,
      "loss": 1.8809,
      "step": 722
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.882943400002891e-07,
      "loss": 1.8335,
      "step": 723
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.784848604953827e-07,
      "loss": 1.8765,
      "step": 724
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.688428140911359e-07,
      "loss": 1.98,
      "step": 725
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.593683668519575e-07,
      "loss": 1.8096,
      "step": 726
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.5006168195571555e-07,
      "loss": 1.8789,
      "step": 727
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.409229196909113e-07,
      "loss": 1.9224,
      "step": 728
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.3195223745392736e-07,
      "loss": 1.9131,
      "step": 729
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.2314978974631084e-07,
      "loss": 1.9897,
      "step": 730
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.1451572817211774e-07,
      "loss": 1.8564,
      "step": 731
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.060502014352972e-07,
      "loss": 1.8457,
      "step": 732
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9775335533713357e-07,
      "loss": 1.8062,
      "step": 733
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8962533277373185e-07,
      "loss": 1.9756,
      "step": 734
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.8166627373356504e-07,
      "loss": 1.9424,
      "step": 735
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.7387631529504955e-07,
      "loss": 1.8335,
      "step": 736
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6625559162419703e-07,
      "loss": 1.918,
      "step": 737
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.5880423397229506e-07,
      "loss": 1.8936,
      "step": 738
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.515223706736524e-07,
      "loss": 1.8784,
      "step": 739
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.444101271433851e-07,
      "loss": 1.957,
      "step": 740
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3746762587525718e-07,
      "loss": 1.9399,
      "step": 741
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3069498643957235e-07,
      "loss": 1.8203,
      "step": 742
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.2409232548111327e-07,
      "loss": 1.8911,
      "step": 743
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1765975671713336e-07,
      "loss": 1.8179,
      "step": 744
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1139739093539714e-07,
      "loss": 1.9487,
      "step": 745
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.0530533599227177e-07,
      "loss": 1.9131,
      "step": 746
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.938369681087412e-08,
      "loss": 1.9028,
      "step": 747
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.363257537925774e-08,
      "loss": 1.9678,
      "step": 748
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.805207074865874e-08,
      "loss": 1.854,
      "step": 749
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.264227903178823e-08,
      "loss": 1.9575,
      "step": 750
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.740329340118258e-08,
      "loss": 1.8394,
      "step": 751
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.233520408759242e-08,
      "loss": 1.8657,
      "step": 752
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.74380983784284e-08,
      "loss": 1.8408,
      "step": 753
    },
    {
      "epoch": 0.96,
      "learning_rate": 6.271206061626567e-08,
      "loss": 1.8823,
      "step": 754
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.81571721973817e-08,
      "loss": 1.8818,
      "step": 755
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.37735115703597e-08,
      "loss": 1.8403,
      "step": 756
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.95611542347374e-08,
      "loss": 1.7656,
      "step": 757
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.5520172739703664e-08,
      "loss": 1.8252,
      "step": 758
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.1650636682853964e-08,
      "loss": 1.8237,
      "step": 759
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.795261270898465e-08,
      "loss": 1.9946,
      "step": 760
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.442616450895386e-08,
      "loss": 2.0083,
      "step": 761
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.107135281857687e-08,
      "loss": 1.9736,
      "step": 762
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.7888235417584676e-08,
      "loss": 1.8921,
      "step": 763
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.4876867128625916e-08,
      "loss": 1.9229,
      "step": 764
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.2037299816324297e-08,
      "loss": 1.8872,
      "step": 765
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.9369582386387085e-08,
      "loss": 1.8955,
      "step": 766
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.6873760784758e-08,
      "loss": 1.9434,
      "step": 767
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.4549877996831162e-08,
      "loss": 1.8501,
      "step": 768
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.2397974046707284e-08,
      "loss": 1.8843,
      "step": 769
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.041808599650529e-08,
      "loss": 1.769,
      "step": 770
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.610247945722849e-09,
      "loss": 1.8657,
      "step": 771
    },
    {
      "epoch": 0.99,
      "learning_rate": 6.974491030653507e-09,
      "loss": 1.8413,
      "step": 772
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.510843423842671e-09,
      "loss": 1.8325,
      "step": 773
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.219330333612437e-09,
      "loss": 1.9419,
      "step": 774
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.099974003618611e-09,
      "loss": 1.8892,
      "step": 775
    },
    {
      "epoch": 0.99,
      "learning_rate": 2.1527937124687927e-09,
      "loss": 2.0103,
      "step": 776
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3778057733981886e-09,
      "loss": 1.9126,
      "step": 777
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.750235339787359e-10,
      "loss": 1.917,
      "step": 778
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.444573758937253e-10,
      "loss": 1.9062,
      "step": 779
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.611471476016598e-11,
      "loss": 1.8867,
      "step": 780
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 1.9629,
      "step": 781
    },
    {
      "epoch": 1.0,
      "step": 781,
      "total_flos": 1.5635750350529495e+18,
      "train_loss": 2.002131619618278,
      "train_runtime": 19394.4142,
      "train_samples_per_second": 10.312,
      "train_steps_per_second": 0.04
    }
  ],
  "max_steps": 781,
  "num_train_epochs": 1,
  "total_flos": 1.5635750350529495e+18,
  "trial_name": null,
  "trial_params": null
}