6936 lines
136 KiB
JSON
6936 lines
136 KiB
JSON
{
|
|
"best_metric": null,
|
|
"best_model_checkpoint": null,
|
|
"epoch": 3.0,
|
|
"global_step": 1149,
|
|
"is_hyper_param_search": false,
|
|
"is_local_process_zero": true,
|
|
"is_world_process_zero": true,
|
|
"log_history": [
|
|
{
|
|
"epoch": 0.0,
|
|
"learning_rate": 4.347826086956522e-07,
|
|
"loss": 1.4604,
|
|
"step": 1
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 8.695652173913044e-07,
|
|
"loss": 1.4463,
|
|
"step": 2
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.3043478260869566e-06,
|
|
"loss": 1.4575,
|
|
"step": 3
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 1.7391304347826088e-06,
|
|
"loss": 1.4507,
|
|
"step": 4
|
|
},
|
|
{
|
|
"epoch": 0.01,
|
|
"learning_rate": 2.173913043478261e-06,
|
|
"loss": 1.4331,
|
|
"step": 5
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 2.6086956521739132e-06,
|
|
"loss": 1.3506,
|
|
"step": 6
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.043478260869566e-06,
|
|
"loss": 1.3384,
|
|
"step": 7
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.4782608695652175e-06,
|
|
"loss": 1.1938,
|
|
"step": 8
|
|
},
|
|
{
|
|
"epoch": 0.02,
|
|
"learning_rate": 3.91304347826087e-06,
|
|
"loss": 1.1484,
|
|
"step": 9
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 4.347826086956522e-06,
|
|
"loss": 1.1099,
|
|
"step": 10
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 4.782608695652174e-06,
|
|
"loss": 1.1001,
|
|
"step": 11
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 5.2173913043478265e-06,
|
|
"loss": 1.0183,
|
|
"step": 12
|
|
},
|
|
{
|
|
"epoch": 0.03,
|
|
"learning_rate": 5.652173913043479e-06,
|
|
"loss": 0.9934,
|
|
"step": 13
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 6.086956521739132e-06,
|
|
"loss": 0.9536,
|
|
"step": 14
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 6.521739130434783e-06,
|
|
"loss": 0.9614,
|
|
"step": 15
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 6.956521739130435e-06,
|
|
"loss": 0.9238,
|
|
"step": 16
|
|
},
|
|
{
|
|
"epoch": 0.04,
|
|
"learning_rate": 7.391304347826087e-06,
|
|
"loss": 0.9182,
|
|
"step": 17
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 7.82608695652174e-06,
|
|
"loss": 0.8984,
|
|
"step": 18
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 8.260869565217392e-06,
|
|
"loss": 0.8828,
|
|
"step": 19
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 8.695652173913044e-06,
|
|
"loss": 0.8708,
|
|
"step": 20
|
|
},
|
|
{
|
|
"epoch": 0.05,
|
|
"learning_rate": 9.130434782608697e-06,
|
|
"loss": 0.8652,
|
|
"step": 21
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 9.565217391304349e-06,
|
|
"loss": 0.8638,
|
|
"step": 22
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 1e-05,
|
|
"loss": 0.8467,
|
|
"step": 23
|
|
},
|
|
{
|
|
"epoch": 0.06,
|
|
"learning_rate": 1.0434782608695653e-05,
|
|
"loss": 0.856,
|
|
"step": 24
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 1.0869565217391305e-05,
|
|
"loss": 0.8391,
|
|
"step": 25
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 1.1304347826086957e-05,
|
|
"loss": 0.8276,
|
|
"step": 26
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 1.1739130434782611e-05,
|
|
"loss": 0.8257,
|
|
"step": 27
|
|
},
|
|
{
|
|
"epoch": 0.07,
|
|
"learning_rate": 1.2173913043478263e-05,
|
|
"loss": 0.8196,
|
|
"step": 28
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 1.2608695652173915e-05,
|
|
"loss": 0.8081,
|
|
"step": 29
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 1.3043478260869566e-05,
|
|
"loss": 0.8042,
|
|
"step": 30
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 1.3478260869565218e-05,
|
|
"loss": 0.802,
|
|
"step": 31
|
|
},
|
|
{
|
|
"epoch": 0.08,
|
|
"learning_rate": 1.391304347826087e-05,
|
|
"loss": 0.7998,
|
|
"step": 32
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 1.4347826086956522e-05,
|
|
"loss": 0.7898,
|
|
"step": 33
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 1.4782608695652174e-05,
|
|
"loss": 0.7788,
|
|
"step": 34
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 1.5217391304347828e-05,
|
|
"loss": 0.781,
|
|
"step": 35
|
|
},
|
|
{
|
|
"epoch": 0.09,
|
|
"learning_rate": 1.565217391304348e-05,
|
|
"loss": 0.77,
|
|
"step": 36
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 1.6086956521739132e-05,
|
|
"loss": 0.7664,
|
|
"step": 37
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 1.6521739130434785e-05,
|
|
"loss": 0.7664,
|
|
"step": 38
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 1.6956521739130437e-05,
|
|
"loss": 0.7668,
|
|
"step": 39
|
|
},
|
|
{
|
|
"epoch": 0.1,
|
|
"learning_rate": 1.739130434782609e-05,
|
|
"loss": 0.7615,
|
|
"step": 40
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 1.782608695652174e-05,
|
|
"loss": 0.752,
|
|
"step": 41
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 1.8260869565217393e-05,
|
|
"loss": 0.7583,
|
|
"step": 42
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 1.8695652173913045e-05,
|
|
"loss": 0.7517,
|
|
"step": 43
|
|
},
|
|
{
|
|
"epoch": 0.11,
|
|
"learning_rate": 1.9130434782608697e-05,
|
|
"loss": 0.7561,
|
|
"step": 44
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 1.956521739130435e-05,
|
|
"loss": 0.75,
|
|
"step": 45
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 2e-05,
|
|
"loss": 0.7412,
|
|
"step": 46
|
|
},
|
|
{
|
|
"epoch": 0.12,
|
|
"learning_rate": 1.9999959438086812e-05,
|
|
"loss": 0.7393,
|
|
"step": 47
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 1.9999837752676295e-05,
|
|
"loss": 0.7349,
|
|
"step": 48
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 1.999963494475561e-05,
|
|
"loss": 0.7329,
|
|
"step": 49
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 1.999935101597001e-05,
|
|
"loss": 0.7336,
|
|
"step": 50
|
|
},
|
|
{
|
|
"epoch": 0.13,
|
|
"learning_rate": 1.9998985968622838e-05,
|
|
"loss": 0.7424,
|
|
"step": 51
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 1.9998539805675495e-05,
|
|
"loss": 0.7375,
|
|
"step": 52
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 1.999801253074743e-05,
|
|
"loss": 0.7383,
|
|
"step": 53
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 1.9997404148116088e-05,
|
|
"loss": 0.7334,
|
|
"step": 54
|
|
},
|
|
{
|
|
"epoch": 0.14,
|
|
"learning_rate": 1.9996714662716914e-05,
|
|
"loss": 0.7322,
|
|
"step": 55
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 1.999594408014327e-05,
|
|
"loss": 0.7236,
|
|
"step": 56
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 1.9995092406646422e-05,
|
|
"loss": 0.728,
|
|
"step": 57
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 1.999415964913547e-05,
|
|
"loss": 0.7271,
|
|
"step": 58
|
|
},
|
|
{
|
|
"epoch": 0.15,
|
|
"learning_rate": 1.999314581517729e-05,
|
|
"loss": 0.7234,
|
|
"step": 59
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 1.9992050912996507e-05,
|
|
"loss": 0.7244,
|
|
"step": 60
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 1.9990874951475377e-05,
|
|
"loss": 0.718,
|
|
"step": 61
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 1.9989617940153753e-05,
|
|
"loss": 0.717,
|
|
"step": 62
|
|
},
|
|
{
|
|
"epoch": 0.16,
|
|
"learning_rate": 1.9988279889228987e-05,
|
|
"loss": 0.7253,
|
|
"step": 63
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 1.9986860809555867e-05,
|
|
"loss": 0.7217,
|
|
"step": 64
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 1.9985360712646502e-05,
|
|
"loss": 0.7153,
|
|
"step": 65
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 1.998377961067026e-05,
|
|
"loss": 0.7212,
|
|
"step": 66
|
|
},
|
|
{
|
|
"epoch": 0.17,
|
|
"learning_rate": 1.998211751645364e-05,
|
|
"loss": 0.7319,
|
|
"step": 67
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 1.998037444348019e-05,
|
|
"loss": 0.7188,
|
|
"step": 68
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 1.9978550405890384e-05,
|
|
"loss": 0.7144,
|
|
"step": 69
|
|
},
|
|
{
|
|
"epoch": 0.18,
|
|
"learning_rate": 1.997664541848151e-05,
|
|
"loss": 0.7078,
|
|
"step": 70
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 1.997465949670756e-05,
|
|
"loss": 0.7183,
|
|
"step": 71
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 1.9972592656679085e-05,
|
|
"loss": 0.7041,
|
|
"step": 72
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 1.997044491516309e-05,
|
|
"loss": 0.7119,
|
|
"step": 73
|
|
},
|
|
{
|
|
"epoch": 0.19,
|
|
"learning_rate": 1.9968216289582865e-05,
|
|
"loss": 0.7109,
|
|
"step": 74
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 1.9965906798017883e-05,
|
|
"loss": 0.7134,
|
|
"step": 75
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 1.996351645920362e-05,
|
|
"loss": 0.707,
|
|
"step": 76
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 1.996104529253142e-05,
|
|
"loss": 0.709,
|
|
"step": 77
|
|
},
|
|
{
|
|
"epoch": 0.2,
|
|
"learning_rate": 1.9958493318048332e-05,
|
|
"loss": 0.7124,
|
|
"step": 78
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 1.9955860556456947e-05,
|
|
"loss": 0.7119,
|
|
"step": 79
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 1.995314702911524e-05,
|
|
"loss": 0.7002,
|
|
"step": 80
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 1.9950352758036378e-05,
|
|
"loss": 0.7104,
|
|
"step": 81
|
|
},
|
|
{
|
|
"epoch": 0.21,
|
|
"learning_rate": 1.994747776588856e-05,
|
|
"loss": 0.7153,
|
|
"step": 82
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 1.9944522075994822e-05,
|
|
"loss": 0.71,
|
|
"step": 83
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 1.994148571233285e-05,
|
|
"loss": 0.7061,
|
|
"step": 84
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 1.9938368699534788e-05,
|
|
"loss": 0.7139,
|
|
"step": 85
|
|
},
|
|
{
|
|
"epoch": 0.22,
|
|
"learning_rate": 1.9935171062887035e-05,
|
|
"loss": 0.6963,
|
|
"step": 86
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 1.993189282833005e-05,
|
|
"loss": 0.7107,
|
|
"step": 87
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 1.9928534022458117e-05,
|
|
"loss": 0.7048,
|
|
"step": 88
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 1.992509467251916e-05,
|
|
"loss": 0.707,
|
|
"step": 89
|
|
},
|
|
{
|
|
"epoch": 0.23,
|
|
"learning_rate": 1.99215748064145e-05,
|
|
"loss": 0.7048,
|
|
"step": 90
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 1.991797445269864e-05,
|
|
"loss": 0.71,
|
|
"step": 91
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 1.9914293640579025e-05,
|
|
"loss": 0.7102,
|
|
"step": 92
|
|
},
|
|
{
|
|
"epoch": 0.24,
|
|
"learning_rate": 1.991053239991581e-05,
|
|
"loss": 0.7009,
|
|
"step": 93
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 1.990669076122162e-05,
|
|
"loss": 0.7024,
|
|
"step": 94
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 1.99027687556613e-05,
|
|
"loss": 0.7046,
|
|
"step": 95
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 1.9898766415051654e-05,
|
|
"loss": 0.6946,
|
|
"step": 96
|
|
},
|
|
{
|
|
"epoch": 0.25,
|
|
"learning_rate": 1.9894683771861207e-05,
|
|
"loss": 0.7024,
|
|
"step": 97
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 1.989052085920992e-05,
|
|
"loss": 0.7102,
|
|
"step": 98
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 1.988627771086893e-05,
|
|
"loss": 0.699,
|
|
"step": 99
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 1.988195436126029e-05,
|
|
"loss": 0.7061,
|
|
"step": 100
|
|
},
|
|
{
|
|
"epoch": 0.26,
|
|
"learning_rate": 1.9877550845456654e-05,
|
|
"loss": 0.6987,
|
|
"step": 101
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 1.9873067199181032e-05,
|
|
"loss": 0.7014,
|
|
"step": 102
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 1.986850345880648e-05,
|
|
"loss": 0.6982,
|
|
"step": 103
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 1.986385966135581e-05,
|
|
"loss": 0.7112,
|
|
"step": 104
|
|
},
|
|
{
|
|
"epoch": 0.27,
|
|
"learning_rate": 1.9859135844501274e-05,
|
|
"loss": 0.7021,
|
|
"step": 105
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 1.985433204656429e-05,
|
|
"loss": 0.6912,
|
|
"step": 106
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 1.9849448306515096e-05,
|
|
"loss": 0.7024,
|
|
"step": 107
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 1.984448466397247e-05,
|
|
"loss": 0.7031,
|
|
"step": 108
|
|
},
|
|
{
|
|
"epoch": 0.28,
|
|
"learning_rate": 1.9839441159203374e-05,
|
|
"loss": 0.6963,
|
|
"step": 109
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 1.983431783312265e-05,
|
|
"loss": 0.696,
|
|
"step": 110
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 1.982911472729268e-05,
|
|
"loss": 0.7097,
|
|
"step": 111
|
|
},
|
|
{
|
|
"epoch": 0.29,
|
|
"learning_rate": 1.9823831883923044e-05,
|
|
"loss": 0.6951,
|
|
"step": 112
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 1.98184693458702e-05,
|
|
"loss": 0.696,
|
|
"step": 113
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 1.9813027156637102e-05,
|
|
"loss": 0.6985,
|
|
"step": 114
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 1.9807505360372868e-05,
|
|
"loss": 0.6956,
|
|
"step": 115
|
|
},
|
|
{
|
|
"epoch": 0.3,
|
|
"learning_rate": 1.9801904001872426e-05,
|
|
"loss": 0.7058,
|
|
"step": 116
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 1.979622312657614e-05,
|
|
"loss": 0.7043,
|
|
"step": 117
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 1.9790462780569442e-05,
|
|
"loss": 0.7024,
|
|
"step": 118
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 1.978462301058247e-05,
|
|
"loss": 0.6992,
|
|
"step": 119
|
|
},
|
|
{
|
|
"epoch": 0.31,
|
|
"learning_rate": 1.977870386398966e-05,
|
|
"loss": 0.6948,
|
|
"step": 120
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 1.9772705388809402e-05,
|
|
"loss": 0.6987,
|
|
"step": 121
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 1.9766627633703622e-05,
|
|
"loss": 0.6902,
|
|
"step": 122
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 1.9760470647977395e-05,
|
|
"loss": 0.7026,
|
|
"step": 123
|
|
},
|
|
{
|
|
"epoch": 0.32,
|
|
"learning_rate": 1.9754234481578544e-05,
|
|
"loss": 0.6919,
|
|
"step": 124
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 1.9747919185097238e-05,
|
|
"loss": 0.6897,
|
|
"step": 125
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 1.9741524809765574e-05,
|
|
"loss": 0.6926,
|
|
"step": 126
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 1.973505140745718e-05,
|
|
"loss": 0.7046,
|
|
"step": 127
|
|
},
|
|
{
|
|
"epoch": 0.33,
|
|
"learning_rate": 1.972849903068676e-05,
|
|
"loss": 0.6963,
|
|
"step": 128
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 1.9721867732609714e-05,
|
|
"loss": 0.6968,
|
|
"step": 129
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 1.9715157567021663e-05,
|
|
"loss": 0.6833,
|
|
"step": 130
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 1.970836858835804e-05,
|
|
"loss": 0.7017,
|
|
"step": 131
|
|
},
|
|
{
|
|
"epoch": 0.34,
|
|
"learning_rate": 1.9701500851693635e-05,
|
|
"loss": 0.6946,
|
|
"step": 132
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 1.969455441274216e-05,
|
|
"loss": 0.6938,
|
|
"step": 133
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 1.968752932785578e-05,
|
|
"loss": 0.7012,
|
|
"step": 134
|
|
},
|
|
{
|
|
"epoch": 0.35,
|
|
"learning_rate": 1.968042565402468e-05,
|
|
"loss": 0.6956,
|
|
"step": 135
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 1.967324344887657e-05,
|
|
"loss": 0.7097,
|
|
"step": 136
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 1.9665982770676256e-05,
|
|
"loss": 0.6978,
|
|
"step": 137
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 1.965864367832513e-05,
|
|
"loss": 0.6921,
|
|
"step": 138
|
|
},
|
|
{
|
|
"epoch": 0.36,
|
|
"learning_rate": 1.9651226231360724e-05,
|
|
"loss": 0.7034,
|
|
"step": 139
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 1.9643730489956198e-05,
|
|
"loss": 0.6921,
|
|
"step": 140
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 1.9636156514919883e-05,
|
|
"loss": 0.6951,
|
|
"step": 141
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 1.9628504367694754e-05,
|
|
"loss": 0.6924,
|
|
"step": 142
|
|
},
|
|
{
|
|
"epoch": 0.37,
|
|
"learning_rate": 1.9620774110357967e-05,
|
|
"loss": 0.687,
|
|
"step": 143
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 1.9612965805620318e-05,
|
|
"loss": 0.698,
|
|
"step": 144
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 1.960507951682577e-05,
|
|
"loss": 0.6882,
|
|
"step": 145
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 1.9597115307950914e-05,
|
|
"loss": 0.6982,
|
|
"step": 146
|
|
},
|
|
{
|
|
"epoch": 0.38,
|
|
"learning_rate": 1.9589073243604454e-05,
|
|
"loss": 0.6873,
|
|
"step": 147
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 1.95809533890267e-05,
|
|
"loss": 0.7053,
|
|
"step": 148
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 1.9572755810089016e-05,
|
|
"loss": 0.6951,
|
|
"step": 149
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 1.9564480573293302e-05,
|
|
"loss": 0.6777,
|
|
"step": 150
|
|
},
|
|
{
|
|
"epoch": 0.39,
|
|
"learning_rate": 1.9556127745771437e-05,
|
|
"loss": 0.687,
|
|
"step": 151
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 1.9547697395284765e-05,
|
|
"loss": 0.6831,
|
|
"step": 152
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 1.9539189590223512e-05,
|
|
"loss": 0.6919,
|
|
"step": 153
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 1.953060439960624e-05,
|
|
"loss": 0.6953,
|
|
"step": 154
|
|
},
|
|
{
|
|
"epoch": 0.4,
|
|
"learning_rate": 1.9521941893079314e-05,
|
|
"loss": 0.6946,
|
|
"step": 155
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 1.9513202140916284e-05,
|
|
"loss": 0.6887,
|
|
"step": 156
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 1.950438521401738e-05,
|
|
"loss": 0.6858,
|
|
"step": 157
|
|
},
|
|
{
|
|
"epoch": 0.41,
|
|
"learning_rate": 1.949549118390888e-05,
|
|
"loss": 0.6941,
|
|
"step": 158
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 1.9486520122742557e-05,
|
|
"loss": 0.6855,
|
|
"step": 159
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 1.9477472103295097e-05,
|
|
"loss": 0.6912,
|
|
"step": 160
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 1.9468347198967494e-05,
|
|
"loss": 0.6892,
|
|
"step": 161
|
|
},
|
|
{
|
|
"epoch": 0.42,
|
|
"learning_rate": 1.945914548378446e-05,
|
|
"loss": 0.6917,
|
|
"step": 162
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 1.9449867032393834e-05,
|
|
"loss": 0.6904,
|
|
"step": 163
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 1.9440511920065962e-05,
|
|
"loss": 0.6921,
|
|
"step": 164
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 1.9431080222693096e-05,
|
|
"loss": 0.696,
|
|
"step": 165
|
|
},
|
|
{
|
|
"epoch": 0.43,
|
|
"learning_rate": 1.9421572016788773e-05,
|
|
"loss": 0.6816,
|
|
"step": 166
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 1.9411987379487197e-05,
|
|
"loss": 0.6907,
|
|
"step": 167
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 1.9402326388542615e-05,
|
|
"loss": 0.6804,
|
|
"step": 168
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 1.9392589122328682e-05,
|
|
"loss": 0.6826,
|
|
"step": 169
|
|
},
|
|
{
|
|
"epoch": 0.44,
|
|
"learning_rate": 1.9382775659837824e-05,
|
|
"loss": 0.6895,
|
|
"step": 170
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 1.9372886080680607e-05,
|
|
"loss": 0.6868,
|
|
"step": 171
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 1.9362920465085078e-05,
|
|
"loss": 0.6943,
|
|
"step": 172
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 1.935287889389613e-05,
|
|
"loss": 0.6836,
|
|
"step": 173
|
|
},
|
|
{
|
|
"epoch": 0.45,
|
|
"learning_rate": 1.9342761448574826e-05,
|
|
"loss": 0.6973,
|
|
"step": 174
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 1.9332568211197756e-05,
|
|
"loss": 0.6914,
|
|
"step": 175
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 1.9322299264456358e-05,
|
|
"loss": 0.6909,
|
|
"step": 176
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 1.9311954691656264e-05,
|
|
"loss": 0.6907,
|
|
"step": 177
|
|
},
|
|
{
|
|
"epoch": 0.46,
|
|
"learning_rate": 1.93015345767166e-05,
|
|
"loss": 0.6836,
|
|
"step": 178
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 1.9291039004169332e-05,
|
|
"loss": 0.6997,
|
|
"step": 179
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 1.928046805915855e-05,
|
|
"loss": 0.6938,
|
|
"step": 180
|
|
},
|
|
{
|
|
"epoch": 0.47,
|
|
"learning_rate": 1.9269821827439822e-05,
|
|
"loss": 0.6899,
|
|
"step": 181
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 1.9259100395379435e-05,
|
|
"loss": 0.6965,
|
|
"step": 182
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 1.924830384995376e-05,
|
|
"loss": 0.6875,
|
|
"step": 183
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 1.9237432278748498e-05,
|
|
"loss": 0.6982,
|
|
"step": 184
|
|
},
|
|
{
|
|
"epoch": 0.48,
|
|
"learning_rate": 1.9226485769958e-05,
|
|
"loss": 0.689,
|
|
"step": 185
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 1.921546441238453e-05,
|
|
"loss": 0.6965,
|
|
"step": 186
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 1.920436829543756e-05,
|
|
"loss": 0.6858,
|
|
"step": 187
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 1.9193197509133035e-05,
|
|
"loss": 0.6985,
|
|
"step": 188
|
|
},
|
|
{
|
|
"epoch": 0.49,
|
|
"learning_rate": 1.918195214409265e-05,
|
|
"loss": 0.6853,
|
|
"step": 189
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 1.9170632291543106e-05,
|
|
"loss": 0.6895,
|
|
"step": 190
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 1.9159238043315383e-05,
|
|
"loss": 0.6934,
|
|
"step": 191
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 1.9147769491843977e-05,
|
|
"loss": 0.6938,
|
|
"step": 192
|
|
},
|
|
{
|
|
"epoch": 0.5,
|
|
"learning_rate": 1.9136226730166168e-05,
|
|
"loss": 0.6804,
|
|
"step": 193
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 1.9124609851921257e-05,
|
|
"loss": 0.6902,
|
|
"step": 194
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 1.9112918951349805e-05,
|
|
"loss": 0.6853,
|
|
"step": 195
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 1.9101154123292866e-05,
|
|
"loss": 0.679,
|
|
"step": 196
|
|
},
|
|
{
|
|
"epoch": 0.51,
|
|
"learning_rate": 1.9089315463191234e-05,
|
|
"loss": 0.677,
|
|
"step": 197
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 1.9077403067084646e-05,
|
|
"loss": 0.6924,
|
|
"step": 198
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 1.9065417031611014e-05,
|
|
"loss": 0.6858,
|
|
"step": 199
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 1.9053357454005654e-05,
|
|
"loss": 0.6863,
|
|
"step": 200
|
|
},
|
|
{
|
|
"epoch": 0.52,
|
|
"learning_rate": 1.9041224432100464e-05,
|
|
"loss": 0.6853,
|
|
"step": 201
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 1.9029018064323167e-05,
|
|
"loss": 0.6848,
|
|
"step": 202
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 1.9016738449696484e-05,
|
|
"loss": 0.6848,
|
|
"step": 203
|
|
},
|
|
{
|
|
"epoch": 0.53,
|
|
"learning_rate": 1.9004385687837346e-05,
|
|
"loss": 0.6836,
|
|
"step": 204
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 1.8991959878956096e-05,
|
|
"loss": 0.6855,
|
|
"step": 205
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 1.8979461123855634e-05,
|
|
"loss": 0.6738,
|
|
"step": 206
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 1.8966889523930657e-05,
|
|
"loss": 0.6804,
|
|
"step": 207
|
|
},
|
|
{
|
|
"epoch": 0.54,
|
|
"learning_rate": 1.8954245181166787e-05,
|
|
"loss": 0.6804,
|
|
"step": 208
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 1.894152819813977e-05,
|
|
"loss": 0.6785,
|
|
"step": 209
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 1.8928738678014646e-05,
|
|
"loss": 0.6829,
|
|
"step": 210
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 1.8915876724544886e-05,
|
|
"loss": 0.6814,
|
|
"step": 211
|
|
},
|
|
{
|
|
"epoch": 0.55,
|
|
"learning_rate": 1.890294244207158e-05,
|
|
"loss": 0.6838,
|
|
"step": 212
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 1.8889935935522587e-05,
|
|
"loss": 0.6885,
|
|
"step": 213
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 1.887685731041165e-05,
|
|
"loss": 0.6873,
|
|
"step": 214
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 1.8863706672837593e-05,
|
|
"loss": 0.6843,
|
|
"step": 215
|
|
},
|
|
{
|
|
"epoch": 0.56,
|
|
"learning_rate": 1.885048412948341e-05,
|
|
"loss": 0.6826,
|
|
"step": 216
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 1.883718978761544e-05,
|
|
"loss": 0.6814,
|
|
"step": 217
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 1.8823823755082463e-05,
|
|
"loss": 0.6746,
|
|
"step": 218
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 1.8810386140314857e-05,
|
|
"loss": 0.6819,
|
|
"step": 219
|
|
},
|
|
{
|
|
"epoch": 0.57,
|
|
"learning_rate": 1.879687705232369e-05,
|
|
"loss": 0.6843,
|
|
"step": 220
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 1.8783296600699858e-05,
|
|
"loss": 0.6843,
|
|
"step": 221
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 1.8769644895613175e-05,
|
|
"loss": 0.677,
|
|
"step": 222
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 1.87559220478115e-05,
|
|
"loss": 0.686,
|
|
"step": 223
|
|
},
|
|
{
|
|
"epoch": 0.58,
|
|
"learning_rate": 1.8742128168619822e-05,
|
|
"loss": 0.6775,
|
|
"step": 224
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 1.8728263369939373e-05,
|
|
"loss": 0.6814,
|
|
"step": 225
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 1.8714327764246697e-05,
|
|
"loss": 0.6758,
|
|
"step": 226
|
|
},
|
|
{
|
|
"epoch": 0.59,
|
|
"learning_rate": 1.8700321464592766e-05,
|
|
"loss": 0.6836,
|
|
"step": 227
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 1.868624458460204e-05,
|
|
"loss": 0.677,
|
|
"step": 228
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 1.8672097238471554e-05,
|
|
"loss": 0.679,
|
|
"step": 229
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 1.8657879540969995e-05,
|
|
"loss": 0.6836,
|
|
"step": 230
|
|
},
|
|
{
|
|
"epoch": 0.6,
|
|
"learning_rate": 1.8643591607436767e-05,
|
|
"loss": 0.6929,
|
|
"step": 231
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 1.8629233553781052e-05,
|
|
"loss": 0.6809,
|
|
"step": 232
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 1.8614805496480875e-05,
|
|
"loss": 0.6831,
|
|
"step": 233
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 1.8600307552582158e-05,
|
|
"loss": 0.6792,
|
|
"step": 234
|
|
},
|
|
{
|
|
"epoch": 0.61,
|
|
"learning_rate": 1.8585739839697768e-05,
|
|
"loss": 0.686,
|
|
"step": 235
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 1.857110247600657e-05,
|
|
"loss": 0.6875,
|
|
"step": 236
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 1.8556395580252457e-05,
|
|
"loss": 0.6777,
|
|
"step": 237
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 1.8541619271743393e-05,
|
|
"loss": 0.6821,
|
|
"step": 238
|
|
},
|
|
{
|
|
"epoch": 0.62,
|
|
"learning_rate": 1.8526773670350446e-05,
|
|
"loss": 0.6863,
|
|
"step": 239
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 1.851185889650682e-05,
|
|
"loss": 0.675,
|
|
"step": 240
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 1.8496875071206862e-05,
|
|
"loss": 0.6724,
|
|
"step": 241
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 1.84818223160051e-05,
|
|
"loss": 0.6777,
|
|
"step": 242
|
|
},
|
|
{
|
|
"epoch": 0.63,
|
|
"learning_rate": 1.8466700753015242e-05,
|
|
"loss": 0.6855,
|
|
"step": 243
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 1.8451510504909193e-05,
|
|
"loss": 0.6853,
|
|
"step": 244
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 1.843625169491606e-05,
|
|
"loss": 0.677,
|
|
"step": 245
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 1.8420924446821146e-05,
|
|
"loss": 0.6775,
|
|
"step": 246
|
|
},
|
|
{
|
|
"epoch": 0.64,
|
|
"learning_rate": 1.840552888496495e-05,
|
|
"loss": 0.6672,
|
|
"step": 247
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 1.839006513424217e-05,
|
|
"loss": 0.6821,
|
|
"step": 248
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 1.8374533320100658e-05,
|
|
"loss": 0.6843,
|
|
"step": 249
|
|
},
|
|
{
|
|
"epoch": 0.65,
|
|
"learning_rate": 1.835893356854044e-05,
|
|
"loss": 0.6755,
|
|
"step": 250
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 1.8343266006112665e-05,
|
|
"loss": 0.6799,
|
|
"step": 251
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 1.83275307599186e-05,
|
|
"loss": 0.6785,
|
|
"step": 252
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 1.831172795760858e-05,
|
|
"loss": 0.6824,
|
|
"step": 253
|
|
},
|
|
{
|
|
"epoch": 0.66,
|
|
"learning_rate": 1.8295857727380983e-05,
|
|
"loss": 0.6753,
|
|
"step": 254
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 1.8279920197981194e-05,
|
|
"loss": 0.6858,
|
|
"step": 255
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 1.8263915498700542e-05,
|
|
"loss": 0.6885,
|
|
"step": 256
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 1.824784375937528e-05,
|
|
"loss": 0.6851,
|
|
"step": 257
|
|
},
|
|
{
|
|
"epoch": 0.67,
|
|
"learning_rate": 1.8231705110385497e-05,
|
|
"loss": 0.677,
|
|
"step": 258
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 1.8215499682654097e-05,
|
|
"loss": 0.6775,
|
|
"step": 259
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 1.8199227607645713e-05,
|
|
"loss": 0.6758,
|
|
"step": 260
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 1.8182889017365637e-05,
|
|
"loss": 0.6775,
|
|
"step": 261
|
|
},
|
|
{
|
|
"epoch": 0.68,
|
|
"learning_rate": 1.8166484044358762e-05,
|
|
"loss": 0.6819,
|
|
"step": 262
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 1.815001282170852e-05,
|
|
"loss": 0.677,
|
|
"step": 263
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 1.8133475483035758e-05,
|
|
"loss": 0.6829,
|
|
"step": 264
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 1.81168721624977e-05,
|
|
"loss": 0.6851,
|
|
"step": 265
|
|
},
|
|
{
|
|
"epoch": 0.69,
|
|
"learning_rate": 1.8100202994786836e-05,
|
|
"loss": 0.6848,
|
|
"step": 266
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 1.8083468115129833e-05,
|
|
"loss": 0.6782,
|
|
"step": 267
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 1.8066667659286435e-05,
|
|
"loss": 0.6775,
|
|
"step": 268
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 1.8049801763548374e-05,
|
|
"loss": 0.6909,
|
|
"step": 269
|
|
},
|
|
{
|
|
"epoch": 0.7,
|
|
"learning_rate": 1.8032870564738244e-05,
|
|
"loss": 0.6831,
|
|
"step": 270
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 1.8015874200208414e-05,
|
|
"loss": 0.6824,
|
|
"step": 271
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 1.7998812807839893e-05,
|
|
"loss": 0.6697,
|
|
"step": 272
|
|
},
|
|
{
|
|
"epoch": 0.71,
|
|
"learning_rate": 1.7981686526041226e-05,
|
|
"loss": 0.6738,
|
|
"step": 273
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 1.7964495493747357e-05,
|
|
"loss": 0.6719,
|
|
"step": 274
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 1.7947239850418524e-05,
|
|
"loss": 0.6885,
|
|
"step": 275
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 1.7929919736039113e-05,
|
|
"loss": 0.6716,
|
|
"step": 276
|
|
},
|
|
{
|
|
"epoch": 0.72,
|
|
"learning_rate": 1.7912535291116507e-05,
|
|
"loss": 0.676,
|
|
"step": 277
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 1.789508665667999e-05,
|
|
"loss": 0.6736,
|
|
"step": 278
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 1.787757397427955e-05,
|
|
"loss": 0.6812,
|
|
"step": 279
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 1.785999738598477e-05,
|
|
"loss": 0.6816,
|
|
"step": 280
|
|
},
|
|
{
|
|
"epoch": 0.73,
|
|
"learning_rate": 1.7842357034383662e-05,
|
|
"loss": 0.6702,
|
|
"step": 281
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 1.78246530625815e-05,
|
|
"loss": 0.6755,
|
|
"step": 282
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 1.7806885614199693e-05,
|
|
"loss": 0.6848,
|
|
"step": 283
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 1.778905483337457e-05,
|
|
"loss": 0.6772,
|
|
"step": 284
|
|
},
|
|
{
|
|
"epoch": 0.74,
|
|
"learning_rate": 1.777116086475625e-05,
|
|
"loss": 0.6765,
|
|
"step": 285
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 1.775320385350745e-05,
|
|
"loss": 0.6787,
|
|
"step": 286
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 1.7735183945302324e-05,
|
|
"loss": 0.6816,
|
|
"step": 287
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 1.7717101286325254e-05,
|
|
"loss": 0.6777,
|
|
"step": 288
|
|
},
|
|
{
|
|
"epoch": 0.75,
|
|
"learning_rate": 1.7698956023269693e-05,
|
|
"loss": 0.6855,
|
|
"step": 289
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 1.768074830333696e-05,
|
|
"loss": 0.6797,
|
|
"step": 290
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 1.766247827423504e-05,
|
|
"loss": 0.6819,
|
|
"step": 291
|
|
},
|
|
{
|
|
"epoch": 0.76,
|
|
"learning_rate": 1.7644146084177408e-05,
|
|
"loss": 0.6843,
|
|
"step": 292
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 1.7625751881881796e-05,
|
|
"loss": 0.6765,
|
|
"step": 293
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 1.7607295816569017e-05,
|
|
"loss": 0.6702,
|
|
"step": 294
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 1.7588778037961733e-05,
|
|
"loss": 0.6631,
|
|
"step": 295
|
|
},
|
|
{
|
|
"epoch": 0.77,
|
|
"learning_rate": 1.757019869628325e-05,
|
|
"loss": 0.6709,
|
|
"step": 296
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 1.7551557942256295e-05,
|
|
"loss": 0.6777,
|
|
"step": 297
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 1.7532855927101796e-05,
|
|
"loss": 0.6853,
|
|
"step": 298
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 1.7514092802537663e-05,
|
|
"loss": 0.6724,
|
|
"step": 299
|
|
},
|
|
{
|
|
"epoch": 0.78,
|
|
"learning_rate": 1.749526872077754e-05,
|
|
"loss": 0.6765,
|
|
"step": 300
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 1.7476383834529577e-05,
|
|
"loss": 0.6641,
|
|
"step": 301
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 1.7457438296995197e-05,
|
|
"loss": 0.6809,
|
|
"step": 302
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 1.7438432261867855e-05,
|
|
"loss": 0.6689,
|
|
"step": 303
|
|
},
|
|
{
|
|
"epoch": 0.79,
|
|
"learning_rate": 1.7419365883331774e-05,
|
|
"loss": 0.6824,
|
|
"step": 304
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 1.740023931606072e-05,
|
|
"loss": 0.6682,
|
|
"step": 305
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 1.7381052715216715e-05,
|
|
"loss": 0.6697,
|
|
"step": 306
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 1.7361806236448817e-05,
|
|
"loss": 0.6763,
|
|
"step": 307
|
|
},
|
|
{
|
|
"epoch": 0.8,
|
|
"learning_rate": 1.734250003589182e-05,
|
|
"loss": 0.6777,
|
|
"step": 308
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1.732313427016501e-05,
|
|
"loss": 0.6755,
|
|
"step": 309
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1.7303709096370895e-05,
|
|
"loss": 0.6804,
|
|
"step": 310
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1.728422467209391e-05,
|
|
"loss": 0.6814,
|
|
"step": 311
|
|
},
|
|
{
|
|
"epoch": 0.81,
|
|
"learning_rate": 1.7264681155399167e-05,
|
|
"loss": 0.666,
|
|
"step": 312
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 1.7245078704831144e-05,
|
|
"loss": 0.6711,
|
|
"step": 313
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 1.7225417479412424e-05,
|
|
"loss": 0.6809,
|
|
"step": 314
|
|
},
|
|
{
|
|
"epoch": 0.82,
|
|
"learning_rate": 1.720569763864239e-05,
|
|
"loss": 0.667,
|
|
"step": 315
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 1.7185919342495942e-05,
|
|
"loss": 0.6692,
|
|
"step": 316
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 1.7166082751422178e-05,
|
|
"loss": 0.679,
|
|
"step": 317
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 1.714618802634312e-05,
|
|
"loss": 0.6648,
|
|
"step": 318
|
|
},
|
|
{
|
|
"epoch": 0.83,
|
|
"learning_rate": 1.7126235328652386e-05,
|
|
"loss": 0.666,
|
|
"step": 319
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1.71062248202139e-05,
|
|
"loss": 0.676,
|
|
"step": 320
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1.708615666336056e-05,
|
|
"loss": 0.6758,
|
|
"step": 321
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1.7066031020892935e-05,
|
|
"loss": 0.6777,
|
|
"step": 322
|
|
},
|
|
{
|
|
"epoch": 0.84,
|
|
"learning_rate": 1.704584805607793e-05,
|
|
"loss": 0.6736,
|
|
"step": 323
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1.702560793264749e-05,
|
|
"loss": 0.675,
|
|
"step": 324
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1.7005310814797233e-05,
|
|
"loss": 0.6816,
|
|
"step": 325
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1.698495686718515e-05,
|
|
"loss": 0.6741,
|
|
"step": 326
|
|
},
|
|
{
|
|
"epoch": 0.85,
|
|
"learning_rate": 1.6964546254930248e-05,
|
|
"loss": 0.6748,
|
|
"step": 327
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1.6944079143611226e-05,
|
|
"loss": 0.677,
|
|
"step": 328
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1.692355569926512e-05,
|
|
"loss": 0.6765,
|
|
"step": 329
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1.690297608838597e-05,
|
|
"loss": 0.6777,
|
|
"step": 330
|
|
},
|
|
{
|
|
"epoch": 0.86,
|
|
"learning_rate": 1.6882340477923445e-05,
|
|
"loss": 0.6648,
|
|
"step": 331
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1.686164903528152e-05,
|
|
"loss": 0.6602,
|
|
"step": 332
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1.684090192831709e-05,
|
|
"loss": 0.6738,
|
|
"step": 333
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1.682009932533863e-05,
|
|
"loss": 0.6711,
|
|
"step": 334
|
|
},
|
|
{
|
|
"epoch": 0.87,
|
|
"learning_rate": 1.6799241395104816e-05,
|
|
"loss": 0.6738,
|
|
"step": 335
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 1.677832830682315e-05,
|
|
"loss": 0.6777,
|
|
"step": 336
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 1.6757360230148617e-05,
|
|
"loss": 0.6719,
|
|
"step": 337
|
|
},
|
|
{
|
|
"epoch": 0.88,
|
|
"learning_rate": 1.6736337335182276e-05,
|
|
"loss": 0.6753,
|
|
"step": 338
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.671525979246989e-05,
|
|
"loss": 0.6704,
|
|
"step": 339
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.6694127773000554e-05,
|
|
"loss": 0.6714,
|
|
"step": 340
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.667294144820529e-05,
|
|
"loss": 0.6775,
|
|
"step": 341
|
|
},
|
|
{
|
|
"epoch": 0.89,
|
|
"learning_rate": 1.6651700989955685e-05,
|
|
"loss": 0.6729,
|
|
"step": 342
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 1.6630406570562453e-05,
|
|
"loss": 0.6726,
|
|
"step": 343
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 1.6609058362774072e-05,
|
|
"loss": 0.6714,
|
|
"step": 344
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 1.6587656539775374e-05,
|
|
"loss": 0.6675,
|
|
"step": 345
|
|
},
|
|
{
|
|
"epoch": 0.9,
|
|
"learning_rate": 1.656620127518614e-05,
|
|
"loss": 0.6726,
|
|
"step": 346
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 1.6544692743059686e-05,
|
|
"loss": 0.6807,
|
|
"step": 347
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 1.6523131117881446e-05,
|
|
"loss": 0.6719,
|
|
"step": 348
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 1.6501516574567583e-05,
|
|
"loss": 0.6719,
|
|
"step": 349
|
|
},
|
|
{
|
|
"epoch": 0.91,
|
|
"learning_rate": 1.6479849288463536e-05,
|
|
"loss": 0.6633,
|
|
"step": 350
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 1.6458129435342626e-05,
|
|
"loss": 0.676,
|
|
"step": 351
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 1.643635719140461e-05,
|
|
"loss": 0.6716,
|
|
"step": 352
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 1.6414532733274265e-05,
|
|
"loss": 0.6685,
|
|
"step": 353
|
|
},
|
|
{
|
|
"epoch": 0.92,
|
|
"learning_rate": 1.639265623799994e-05,
|
|
"loss": 0.6733,
|
|
"step": 354
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 1.637072788305214e-05,
|
|
"loss": 0.675,
|
|
"step": 355
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 1.634874784632207e-05,
|
|
"loss": 0.6685,
|
|
"step": 356
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 1.6326716306120195e-05,
|
|
"loss": 0.677,
|
|
"step": 357
|
|
},
|
|
{
|
|
"epoch": 0.93,
|
|
"learning_rate": 1.6304633441174803e-05,
|
|
"loss": 0.6685,
|
|
"step": 358
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 1.6282499430630543e-05,
|
|
"loss": 0.6638,
|
|
"step": 359
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 1.6260314454046982e-05,
|
|
"loss": 0.6726,
|
|
"step": 360
|
|
},
|
|
{
|
|
"epoch": 0.94,
|
|
"learning_rate": 1.6238078691397134e-05,
|
|
"loss": 0.6772,
|
|
"step": 361
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 1.6215792323066013e-05,
|
|
"loss": 0.6746,
|
|
"step": 362
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 1.6193455529849168e-05,
|
|
"loss": 0.6741,
|
|
"step": 363
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 1.6171068492951218e-05,
|
|
"loss": 0.6665,
|
|
"step": 364
|
|
},
|
|
{
|
|
"epoch": 0.95,
|
|
"learning_rate": 1.6148631393984363e-05,
|
|
"loss": 0.6658,
|
|
"step": 365
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 1.612614441496694e-05,
|
|
"loss": 0.6772,
|
|
"step": 366
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 1.610360773832193e-05,
|
|
"loss": 0.6714,
|
|
"step": 367
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 1.608102154687547e-05,
|
|
"loss": 0.6714,
|
|
"step": 368
|
|
},
|
|
{
|
|
"epoch": 0.96,
|
|
"learning_rate": 1.6058386023855392e-05,
|
|
"loss": 0.6743,
|
|
"step": 369
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 1.603570135288972e-05,
|
|
"loss": 0.6775,
|
|
"step": 370
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 1.6012967718005185e-05,
|
|
"loss": 0.6711,
|
|
"step": 371
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 1.599018530362573e-05,
|
|
"loss": 0.668,
|
|
"step": 372
|
|
},
|
|
{
|
|
"epoch": 0.97,
|
|
"learning_rate": 1.5967354294571018e-05,
|
|
"loss": 0.6738,
|
|
"step": 373
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 1.594447487605493e-05,
|
|
"loss": 0.6697,
|
|
"step": 374
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 1.5921547233684073e-05,
|
|
"loss": 0.6733,
|
|
"step": 375
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 1.589857155345624e-05,
|
|
"loss": 0.6738,
|
|
"step": 376
|
|
},
|
|
{
|
|
"epoch": 0.98,
|
|
"learning_rate": 1.587554802175895e-05,
|
|
"loss": 0.6702,
|
|
"step": 377
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 1.5852476825367896e-05,
|
|
"loss": 0.6724,
|
|
"step": 378
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 1.582935815144546e-05,
|
|
"loss": 0.6699,
|
|
"step": 379
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 1.580619218753916e-05,
|
|
"loss": 0.6772,
|
|
"step": 380
|
|
},
|
|
{
|
|
"epoch": 0.99,
|
|
"learning_rate": 1.578297912158017e-05,
|
|
"loss": 0.6685,
|
|
"step": 381
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.5759719141881752e-05,
|
|
"loss": 0.6724,
|
|
"step": 382
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.573641243713777e-05,
|
|
"loss": 0.6746,
|
|
"step": 383
|
|
},
|
|
{
|
|
"epoch": 1.0,
|
|
"learning_rate": 1.5713059196421125e-05,
|
|
"loss": 0.6211,
|
|
"step": 384
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5689659609182248e-05,
|
|
"loss": 0.624,
|
|
"step": 385
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5666213865247542e-05,
|
|
"loss": 0.6123,
|
|
"step": 386
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5642722154817848e-05,
|
|
"loss": 0.616,
|
|
"step": 387
|
|
},
|
|
{
|
|
"epoch": 1.01,
|
|
"learning_rate": 1.5619184668466917e-05,
|
|
"loss": 0.6174,
|
|
"step": 388
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.559560159713984e-05,
|
|
"loss": 0.6257,
|
|
"step": 389
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5571973132151518e-05,
|
|
"loss": 0.626,
|
|
"step": 390
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5548299465185096e-05,
|
|
"loss": 0.6157,
|
|
"step": 391
|
|
},
|
|
{
|
|
"epoch": 1.02,
|
|
"learning_rate": 1.5524580788290425e-05,
|
|
"loss": 0.6182,
|
|
"step": 392
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 1.550081729388248e-05,
|
|
"loss": 0.6167,
|
|
"step": 393
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 1.547700917473983e-05,
|
|
"loss": 0.6165,
|
|
"step": 394
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 1.545315662400304e-05,
|
|
"loss": 0.6162,
|
|
"step": 395
|
|
},
|
|
{
|
|
"epoch": 1.03,
|
|
"learning_rate": 1.5429259835173133e-05,
|
|
"loss": 0.616,
|
|
"step": 396
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 1.540531900211e-05,
|
|
"loss": 0.6072,
|
|
"step": 397
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 1.5381334319030844e-05,
|
|
"loss": 0.6152,
|
|
"step": 398
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 1.5357305980508582e-05,
|
|
"loss": 0.6121,
|
|
"step": 399
|
|
},
|
|
{
|
|
"epoch": 1.04,
|
|
"learning_rate": 1.53332341814703e-05,
|
|
"loss": 0.6255,
|
|
"step": 400
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 1.5309119117195642e-05,
|
|
"loss": 0.6152,
|
|
"step": 401
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 1.528496098331523e-05,
|
|
"loss": 0.6155,
|
|
"step": 402
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 1.52607599758091e-05,
|
|
"loss": 0.6208,
|
|
"step": 403
|
|
},
|
|
{
|
|
"epoch": 1.05,
|
|
"learning_rate": 1.5236516291005075e-05,
|
|
"loss": 0.6133,
|
|
"step": 404
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 1.5212230125577207e-05,
|
|
"loss": 0.6179,
|
|
"step": 405
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 1.5187901676544162e-05,
|
|
"loss": 0.614,
|
|
"step": 406
|
|
},
|
|
{
|
|
"epoch": 1.06,
|
|
"learning_rate": 1.516353114126763e-05,
|
|
"loss": 0.6204,
|
|
"step": 407
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 1.5139118717450714e-05,
|
|
"loss": 0.6255,
|
|
"step": 408
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 1.511466460313634e-05,
|
|
"loss": 0.6174,
|
|
"step": 409
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 1.5090168996705643e-05,
|
|
"loss": 0.613,
|
|
"step": 410
|
|
},
|
|
{
|
|
"epoch": 1.07,
|
|
"learning_rate": 1.506563209687635e-05,
|
|
"loss": 0.6179,
|
|
"step": 411
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 1.5041054102701184e-05,
|
|
"loss": 0.6179,
|
|
"step": 412
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 1.5016435213566239e-05,
|
|
"loss": 0.6184,
|
|
"step": 413
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 1.499177562918936e-05,
|
|
"loss": 0.6201,
|
|
"step": 414
|
|
},
|
|
{
|
|
"epoch": 1.08,
|
|
"learning_rate": 1.4967075549618535e-05,
|
|
"loss": 0.6184,
|
|
"step": 415
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 1.4942335175230258e-05,
|
|
"loss": 0.6182,
|
|
"step": 416
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 1.4917554706727915e-05,
|
|
"loss": 0.6162,
|
|
"step": 417
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 1.489273434514015e-05,
|
|
"loss": 0.6174,
|
|
"step": 418
|
|
},
|
|
{
|
|
"epoch": 1.09,
|
|
"learning_rate": 1.4867874291819226e-05,
|
|
"loss": 0.6152,
|
|
"step": 419
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 1.4842974748439416e-05,
|
|
"loss": 0.6223,
|
|
"step": 420
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 1.481803591699534e-05,
|
|
"loss": 0.6172,
|
|
"step": 421
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 1.4793057999800335e-05,
|
|
"loss": 0.613,
|
|
"step": 422
|
|
},
|
|
{
|
|
"epoch": 1.1,
|
|
"learning_rate": 1.4768041199484837e-05,
|
|
"loss": 0.6233,
|
|
"step": 423
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 1.4742985718994693e-05,
|
|
"loss": 0.6121,
|
|
"step": 424
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 1.471789176158955e-05,
|
|
"loss": 0.6194,
|
|
"step": 425
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 1.469275953084119e-05,
|
|
"loss": 0.6167,
|
|
"step": 426
|
|
},
|
|
{
|
|
"epoch": 1.11,
|
|
"learning_rate": 1.466758923063189e-05,
|
|
"loss": 0.6145,
|
|
"step": 427
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 1.4642381065152748e-05,
|
|
"loss": 0.6169,
|
|
"step": 428
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 1.461713523890206e-05,
|
|
"loss": 0.6194,
|
|
"step": 429
|
|
},
|
|
{
|
|
"epoch": 1.12,
|
|
"learning_rate": 1.4591851956683618e-05,
|
|
"loss": 0.6238,
|
|
"step": 430
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 1.456653142360509e-05,
|
|
"loss": 0.6196,
|
|
"step": 431
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 1.4541173845076323e-05,
|
|
"loss": 0.6179,
|
|
"step": 432
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 1.45157794268077e-05,
|
|
"loss": 0.6221,
|
|
"step": 433
|
|
},
|
|
{
|
|
"epoch": 1.13,
|
|
"learning_rate": 1.4490348374808454e-05,
|
|
"loss": 0.6179,
|
|
"step": 434
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 1.4464880895385018e-05,
|
|
"loss": 0.6182,
|
|
"step": 435
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 1.4439377195139321e-05,
|
|
"loss": 0.6218,
|
|
"step": 436
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 1.4413837480967146e-05,
|
|
"loss": 0.613,
|
|
"step": 437
|
|
},
|
|
{
|
|
"epoch": 1.14,
|
|
"learning_rate": 1.438826196005642e-05,
|
|
"loss": 0.6267,
|
|
"step": 438
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 1.4362650839885558e-05,
|
|
"loss": 0.6184,
|
|
"step": 439
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 1.4337004328221768e-05,
|
|
"loss": 0.6235,
|
|
"step": 440
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 1.4311322633119367e-05,
|
|
"loss": 0.616,
|
|
"step": 441
|
|
},
|
|
{
|
|
"epoch": 1.15,
|
|
"learning_rate": 1.4285605962918086e-05,
|
|
"loss": 0.6123,
|
|
"step": 442
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 1.4259854526241399e-05,
|
|
"loss": 0.6165,
|
|
"step": 443
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 1.4234068531994812e-05,
|
|
"loss": 0.6125,
|
|
"step": 444
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 1.420824818936418e-05,
|
|
"loss": 0.6128,
|
|
"step": 445
|
|
},
|
|
{
|
|
"epoch": 1.16,
|
|
"learning_rate": 1.4182393707813998e-05,
|
|
"loss": 0.6143,
|
|
"step": 446
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 1.4156505297085715e-05,
|
|
"loss": 0.6147,
|
|
"step": 447
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 1.4130583167196027e-05,
|
|
"loss": 0.6169,
|
|
"step": 448
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 1.4104627528435166e-05,
|
|
"loss": 0.6218,
|
|
"step": 449
|
|
},
|
|
{
|
|
"epoch": 1.17,
|
|
"learning_rate": 1.4078638591365205e-05,
|
|
"loss": 0.6152,
|
|
"step": 450
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 1.4052616566818355e-05,
|
|
"loss": 0.6145,
|
|
"step": 451
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 1.4026561665895225e-05,
|
|
"loss": 0.6162,
|
|
"step": 452
|
|
},
|
|
{
|
|
"epoch": 1.18,
|
|
"learning_rate": 1.4000474099963145e-05,
|
|
"loss": 0.6213,
|
|
"step": 453
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 1.3974354080654435e-05,
|
|
"loss": 0.6238,
|
|
"step": 454
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 1.3948201819864685e-05,
|
|
"loss": 0.6072,
|
|
"step": 455
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 1.3922017529751036e-05,
|
|
"loss": 0.6182,
|
|
"step": 456
|
|
},
|
|
{
|
|
"epoch": 1.19,
|
|
"learning_rate": 1.3895801422730472e-05,
|
|
"loss": 0.6194,
|
|
"step": 457
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 1.3869553711478087e-05,
|
|
"loss": 0.6084,
|
|
"step": 458
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 1.3843274608925356e-05,
|
|
"loss": 0.6172,
|
|
"step": 459
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 1.3816964328258418e-05,
|
|
"loss": 0.6199,
|
|
"step": 460
|
|
},
|
|
{
|
|
"epoch": 1.2,
|
|
"learning_rate": 1.3790623082916326e-05,
|
|
"loss": 0.6223,
|
|
"step": 461
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 1.3764251086589353e-05,
|
|
"loss": 0.6118,
|
|
"step": 462
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 1.3737848553217222e-05,
|
|
"loss": 0.626,
|
|
"step": 463
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 1.3711415696987378e-05,
|
|
"loss": 0.6187,
|
|
"step": 464
|
|
},
|
|
{
|
|
"epoch": 1.21,
|
|
"learning_rate": 1.3684952732333274e-05,
|
|
"loss": 0.6104,
|
|
"step": 465
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 1.3658459873932604e-05,
|
|
"loss": 0.6248,
|
|
"step": 466
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 1.3631937336705568e-05,
|
|
"loss": 0.6123,
|
|
"step": 467
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 1.3605385335813142e-05,
|
|
"loss": 0.6191,
|
|
"step": 468
|
|
},
|
|
{
|
|
"epoch": 1.22,
|
|
"learning_rate": 1.3578804086655311e-05,
|
|
"loss": 0.6213,
|
|
"step": 469
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 1.3552193804869343e-05,
|
|
"loss": 0.6204,
|
|
"step": 470
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 1.3525554706328026e-05,
|
|
"loss": 0.6094,
|
|
"step": 471
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 1.3498887007137918e-05,
|
|
"loss": 0.6128,
|
|
"step": 472
|
|
},
|
|
{
|
|
"epoch": 1.23,
|
|
"learning_rate": 1.3472190923637603e-05,
|
|
"loss": 0.6138,
|
|
"step": 473
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 1.3445466672395923e-05,
|
|
"loss": 0.6208,
|
|
"step": 474
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 1.3418714470210228e-05,
|
|
"loss": 0.6177,
|
|
"step": 475
|
|
},
|
|
{
|
|
"epoch": 1.24,
|
|
"learning_rate": 1.3391934534104625e-05,
|
|
"loss": 0.6216,
|
|
"step": 476
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 1.3365127081328193e-05,
|
|
"loss": 0.6272,
|
|
"step": 477
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 1.3338292329353253e-05,
|
|
"loss": 0.6157,
|
|
"step": 478
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 1.331143049587358e-05,
|
|
"loss": 0.6204,
|
|
"step": 479
|
|
},
|
|
{
|
|
"epoch": 1.25,
|
|
"learning_rate": 1.3284541798802644e-05,
|
|
"loss": 0.6204,
|
|
"step": 480
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1.3257626456271849e-05,
|
|
"loss": 0.616,
|
|
"step": 481
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1.3230684686628746e-05,
|
|
"loss": 0.6204,
|
|
"step": 482
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1.3203716708435281e-05,
|
|
"loss": 0.6235,
|
|
"step": 483
|
|
},
|
|
{
|
|
"epoch": 1.26,
|
|
"learning_rate": 1.3176722740466015e-05,
|
|
"loss": 0.6101,
|
|
"step": 484
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 1.3149703001706341e-05,
|
|
"loss": 0.6162,
|
|
"step": 485
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 1.3122657711350722e-05,
|
|
"loss": 0.625,
|
|
"step": 486
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 1.3095587088800903e-05,
|
|
"loss": 0.627,
|
|
"step": 487
|
|
},
|
|
{
|
|
"epoch": 1.27,
|
|
"learning_rate": 1.306849135366413e-05,
|
|
"loss": 0.6147,
|
|
"step": 488
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 1.304137072575137e-05,
|
|
"loss": 0.614,
|
|
"step": 489
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 1.3014225425075544e-05,
|
|
"loss": 0.625,
|
|
"step": 490
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 1.2987055671849708e-05,
|
|
"loss": 0.6082,
|
|
"step": 491
|
|
},
|
|
{
|
|
"epoch": 1.28,
|
|
"learning_rate": 1.2959861686485304e-05,
|
|
"loss": 0.614,
|
|
"step": 492
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 1.2932643689590342e-05,
|
|
"loss": 0.6165,
|
|
"step": 493
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 1.290540190196763e-05,
|
|
"loss": 0.6096,
|
|
"step": 494
|
|
},
|
|
{
|
|
"epoch": 1.29,
|
|
"learning_rate": 1.287813654461297e-05,
|
|
"loss": 0.6221,
|
|
"step": 495
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 1.2850847838713375e-05,
|
|
"loss": 0.6184,
|
|
"step": 496
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 1.2823536005645272e-05,
|
|
"loss": 0.6252,
|
|
"step": 497
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 1.2796201266972698e-05,
|
|
"loss": 0.6172,
|
|
"step": 498
|
|
},
|
|
{
|
|
"epoch": 1.3,
|
|
"learning_rate": 1.2768843844445514e-05,
|
|
"loss": 0.6218,
|
|
"step": 499
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1.2741463959997598e-05,
|
|
"loss": 0.6108,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"eval_loss": 0.6617740988731384,
|
|
"eval_runtime": 1480.2934,
|
|
"eval_samples_per_second": 1.352,
|
|
"eval_steps_per_second": 0.338,
|
|
"step": 500
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1.2714061835745048e-05,
|
|
"loss": 0.6133,
|
|
"step": 501
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1.2686637693984384e-05,
|
|
"loss": 0.615,
|
|
"step": 502
|
|
},
|
|
{
|
|
"epoch": 1.31,
|
|
"learning_rate": 1.2659191757190739e-05,
|
|
"loss": 0.6147,
|
|
"step": 503
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 1.263172424801605e-05,
|
|
"loss": 0.624,
|
|
"step": 504
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 1.260423538928726e-05,
|
|
"loss": 0.6165,
|
|
"step": 505
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 1.2576725404004517e-05,
|
|
"loss": 0.6169,
|
|
"step": 506
|
|
},
|
|
{
|
|
"epoch": 1.32,
|
|
"learning_rate": 1.2549194515339345e-05,
|
|
"loss": 0.6169,
|
|
"step": 507
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 1.2521642946632841e-05,
|
|
"loss": 0.6165,
|
|
"step": 508
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 1.2494070921393879e-05,
|
|
"loss": 0.6199,
|
|
"step": 509
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 1.2466478663297279e-05,
|
|
"loss": 0.6082,
|
|
"step": 510
|
|
},
|
|
{
|
|
"epoch": 1.33,
|
|
"learning_rate": 1.2438866396181988e-05,
|
|
"loss": 0.6216,
|
|
"step": 511
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 1.2411234344049293e-05,
|
|
"loss": 0.623,
|
|
"step": 512
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 1.2383582731060966e-05,
|
|
"loss": 0.6206,
|
|
"step": 513
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 1.2355911781537478e-05,
|
|
"loss": 0.6165,
|
|
"step": 514
|
|
},
|
|
{
|
|
"epoch": 1.34,
|
|
"learning_rate": 1.2328221719956154e-05,
|
|
"loss": 0.6128,
|
|
"step": 515
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 1.2300512770949373e-05,
|
|
"loss": 0.6157,
|
|
"step": 516
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 1.227278515930273e-05,
|
|
"loss": 0.613,
|
|
"step": 517
|
|
},
|
|
{
|
|
"epoch": 1.35,
|
|
"learning_rate": 1.2245039109953219e-05,
|
|
"loss": 0.6155,
|
|
"step": 518
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1.221727484798741e-05,
|
|
"loss": 0.6211,
|
|
"step": 519
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1.2189492598639623e-05,
|
|
"loss": 0.6067,
|
|
"step": 520
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1.2161692587290093e-05,
|
|
"loss": 0.6265,
|
|
"step": 521
|
|
},
|
|
{
|
|
"epoch": 1.36,
|
|
"learning_rate": 1.2133875039463147e-05,
|
|
"loss": 0.6128,
|
|
"step": 522
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 1.210604018082538e-05,
|
|
"loss": 0.6172,
|
|
"step": 523
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 1.2078188237183817e-05,
|
|
"loss": 0.6123,
|
|
"step": 524
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 1.2050319434484077e-05,
|
|
"loss": 0.6187,
|
|
"step": 525
|
|
},
|
|
{
|
|
"epoch": 1.37,
|
|
"learning_rate": 1.2022433998808557e-05,
|
|
"loss": 0.624,
|
|
"step": 526
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 1.1994532156374573e-05,
|
|
"loss": 0.6206,
|
|
"step": 527
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 1.1966614133532559e-05,
|
|
"loss": 0.6135,
|
|
"step": 528
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 1.1938680156764189e-05,
|
|
"loss": 0.6274,
|
|
"step": 529
|
|
},
|
|
{
|
|
"epoch": 1.38,
|
|
"learning_rate": 1.1910730452680575e-05,
|
|
"loss": 0.6267,
|
|
"step": 530
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 1.1882765248020412e-05,
|
|
"loss": 0.6157,
|
|
"step": 531
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 1.1854784769648137e-05,
|
|
"loss": 0.6128,
|
|
"step": 532
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 1.18267892445521e-05,
|
|
"loss": 0.6182,
|
|
"step": 533
|
|
},
|
|
{
|
|
"epoch": 1.39,
|
|
"learning_rate": 1.1798778899842712e-05,
|
|
"loss": 0.6189,
|
|
"step": 534
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 1.1770753962750608e-05,
|
|
"loss": 0.6199,
|
|
"step": 535
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 1.1742714660624799e-05,
|
|
"loss": 0.6133,
|
|
"step": 536
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 1.1714661220930834e-05,
|
|
"loss": 0.6194,
|
|
"step": 537
|
|
},
|
|
{
|
|
"epoch": 1.4,
|
|
"learning_rate": 1.1686593871248952e-05,
|
|
"loss": 0.6174,
|
|
"step": 538
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 1.1658512839272231e-05,
|
|
"loss": 0.6221,
|
|
"step": 539
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 1.163041835280475e-05,
|
|
"loss": 0.6135,
|
|
"step": 540
|
|
},
|
|
{
|
|
"epoch": 1.41,
|
|
"learning_rate": 1.1602310639759724e-05,
|
|
"loss": 0.6165,
|
|
"step": 541
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 1.1574189928157689e-05,
|
|
"loss": 0.6121,
|
|
"step": 542
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 1.1546056446124612e-05,
|
|
"loss": 0.6216,
|
|
"step": 543
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 1.151791042189006e-05,
|
|
"loss": 0.6162,
|
|
"step": 544
|
|
},
|
|
{
|
|
"epoch": 1.42,
|
|
"learning_rate": 1.148975208378536e-05,
|
|
"loss": 0.6121,
|
|
"step": 545
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 1.1461581660241717e-05,
|
|
"loss": 0.6123,
|
|
"step": 546
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 1.1433399379788388e-05,
|
|
"loss": 0.6396,
|
|
"step": 547
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 1.1405205471050822e-05,
|
|
"loss": 0.6155,
|
|
"step": 548
|
|
},
|
|
{
|
|
"epoch": 1.43,
|
|
"learning_rate": 1.1377000162748786e-05,
|
|
"loss": 0.6213,
|
|
"step": 549
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 1.1348783683694535e-05,
|
|
"loss": 0.6221,
|
|
"step": 550
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 1.1320556262790951e-05,
|
|
"loss": 0.6194,
|
|
"step": 551
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 1.1292318129029665e-05,
|
|
"loss": 0.6125,
|
|
"step": 552
|
|
},
|
|
{
|
|
"epoch": 1.44,
|
|
"learning_rate": 1.1264069511489225e-05,
|
|
"loss": 0.6138,
|
|
"step": 553
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 1.1235810639333234e-05,
|
|
"loss": 0.6179,
|
|
"step": 554
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 1.1207541741808461e-05,
|
|
"loss": 0.6187,
|
|
"step": 555
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 1.117926304824303e-05,
|
|
"loss": 0.6111,
|
|
"step": 556
|
|
},
|
|
{
|
|
"epoch": 1.45,
|
|
"learning_rate": 1.1150974788044522e-05,
|
|
"loss": 0.6235,
|
|
"step": 557
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 1.1122677190698123e-05,
|
|
"loss": 0.6152,
|
|
"step": 558
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 1.1094370485764775e-05,
|
|
"loss": 0.6116,
|
|
"step": 559
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 1.10660549028793e-05,
|
|
"loss": 0.6208,
|
|
"step": 560
|
|
},
|
|
{
|
|
"epoch": 1.46,
|
|
"learning_rate": 1.1037730671748538e-05,
|
|
"loss": 0.625,
|
|
"step": 561
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 1.1009398022149496e-05,
|
|
"loss": 0.6238,
|
|
"step": 562
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 1.098105718392746e-05,
|
|
"loss": 0.6128,
|
|
"step": 563
|
|
},
|
|
{
|
|
"epoch": 1.47,
|
|
"learning_rate": 1.0952708386994159e-05,
|
|
"loss": 0.6165,
|
|
"step": 564
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 1.0924351861325882e-05,
|
|
"loss": 0.6067,
|
|
"step": 565
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 1.0895987836961608e-05,
|
|
"loss": 0.6104,
|
|
"step": 566
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 1.0867616544001165e-05,
|
|
"loss": 0.6218,
|
|
"step": 567
|
|
},
|
|
{
|
|
"epoch": 1.48,
|
|
"learning_rate": 1.0839238212603335e-05,
|
|
"loss": 0.6133,
|
|
"step": 568
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 1.0810853072984e-05,
|
|
"loss": 0.6077,
|
|
"step": 569
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 1.0782461355414273e-05,
|
|
"loss": 0.6189,
|
|
"step": 570
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 1.0754063290218632e-05,
|
|
"loss": 0.6111,
|
|
"step": 571
|
|
},
|
|
{
|
|
"epoch": 1.49,
|
|
"learning_rate": 1.0725659107773046e-05,
|
|
"loss": 0.6108,
|
|
"step": 572
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 1.0697249038503115e-05,
|
|
"loss": 0.6118,
|
|
"step": 573
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 1.0668833312882187e-05,
|
|
"loss": 0.6238,
|
|
"step": 574
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 1.0640412161429506e-05,
|
|
"loss": 0.6113,
|
|
"step": 575
|
|
},
|
|
{
|
|
"epoch": 1.5,
|
|
"learning_rate": 1.0611985814708325e-05,
|
|
"loss": 0.6189,
|
|
"step": 576
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 1.0583554503324046e-05,
|
|
"loss": 0.6121,
|
|
"step": 577
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 1.0555118457922345e-05,
|
|
"loss": 0.6169,
|
|
"step": 578
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 1.0526677909187306e-05,
|
|
"loss": 0.6086,
|
|
"step": 579
|
|
},
|
|
{
|
|
"epoch": 1.51,
|
|
"learning_rate": 1.0498233087839538e-05,
|
|
"loss": 0.6121,
|
|
"step": 580
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1.046978422463432e-05,
|
|
"loss": 0.6099,
|
|
"step": 581
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1.0441331550359713e-05,
|
|
"loss": 0.6152,
|
|
"step": 582
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1.04128752958347e-05,
|
|
"loss": 0.6155,
|
|
"step": 583
|
|
},
|
|
{
|
|
"epoch": 1.52,
|
|
"learning_rate": 1.0384415691907304e-05,
|
|
"loss": 0.615,
|
|
"step": 584
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 1.0355952969452728e-05,
|
|
"loss": 0.6135,
|
|
"step": 585
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 1.0327487359371458e-05,
|
|
"loss": 0.6226,
|
|
"step": 586
|
|
},
|
|
{
|
|
"epoch": 1.53,
|
|
"learning_rate": 1.0299019092587421e-05,
|
|
"loss": 0.6133,
|
|
"step": 587
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 1.0270548400046087e-05,
|
|
"loss": 0.6157,
|
|
"step": 588
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 1.0242075512712612e-05,
|
|
"loss": 0.6125,
|
|
"step": 589
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 1.0213600661569952e-05,
|
|
"loss": 0.6167,
|
|
"step": 590
|
|
},
|
|
{
|
|
"epoch": 1.54,
|
|
"learning_rate": 1.018512407761699e-05,
|
|
"loss": 0.6196,
|
|
"step": 591
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 1.0156645991866679e-05,
|
|
"loss": 0.6167,
|
|
"step": 592
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 1.0128166635344141e-05,
|
|
"loss": 0.6182,
|
|
"step": 593
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 1.0099686239084813e-05,
|
|
"loss": 0.6206,
|
|
"step": 594
|
|
},
|
|
{
|
|
"epoch": 1.55,
|
|
"learning_rate": 1.0071205034132575e-05,
|
|
"loss": 0.6221,
|
|
"step": 595
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 1.0042723251537849e-05,
|
|
"loss": 0.6174,
|
|
"step": 596
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 1.0014241122355763e-05,
|
|
"loss": 0.615,
|
|
"step": 597
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 9.985758877644242e-06,
|
|
"loss": 0.6165,
|
|
"step": 598
|
|
},
|
|
{
|
|
"epoch": 1.56,
|
|
"learning_rate": 9.957276748462154e-06,
|
|
"loss": 0.6167,
|
|
"step": 599
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 9.92879496586743e-06,
|
|
"loss": 0.6099,
|
|
"step": 600
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 9.900313760915188e-06,
|
|
"loss": 0.624,
|
|
"step": 601
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 9.871833364655866e-06,
|
|
"loss": 0.6174,
|
|
"step": 602
|
|
},
|
|
{
|
|
"epoch": 1.57,
|
|
"learning_rate": 9.843354008133326e-06,
|
|
"loss": 0.6233,
|
|
"step": 603
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 9.814875922383013e-06,
|
|
"loss": 0.625,
|
|
"step": 604
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 9.786399338430051e-06,
|
|
"loss": 0.6113,
|
|
"step": 605
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 9.757924487287388e-06,
|
|
"loss": 0.6216,
|
|
"step": 606
|
|
},
|
|
{
|
|
"epoch": 1.58,
|
|
"learning_rate": 9.729451599953916e-06,
|
|
"loss": 0.615,
|
|
"step": 607
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 9.70098090741258e-06,
|
|
"loss": 0.6125,
|
|
"step": 608
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 9.672512640628547e-06,
|
|
"loss": 0.6116,
|
|
"step": 609
|
|
},
|
|
{
|
|
"epoch": 1.59,
|
|
"learning_rate": 9.644047030547277e-06,
|
|
"loss": 0.6055,
|
|
"step": 610
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 9.615584308092698e-06,
|
|
"loss": 0.6133,
|
|
"step": 611
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 9.587124704165302e-06,
|
|
"loss": 0.6145,
|
|
"step": 612
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 9.55866844964029e-06,
|
|
"loss": 0.6194,
|
|
"step": 613
|
|
},
|
|
{
|
|
"epoch": 1.6,
|
|
"learning_rate": 9.530215775365683e-06,
|
|
"loss": 0.6162,
|
|
"step": 614
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 9.501766912160468e-06,
|
|
"loss": 0.6023,
|
|
"step": 615
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 9.473322090812697e-06,
|
|
"loss": 0.6111,
|
|
"step": 616
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 9.44488154207766e-06,
|
|
"loss": 0.6079,
|
|
"step": 617
|
|
},
|
|
{
|
|
"epoch": 1.61,
|
|
"learning_rate": 9.416445496675958e-06,
|
|
"loss": 0.6072,
|
|
"step": 618
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 9.38801418529168e-06,
|
|
"loss": 0.6143,
|
|
"step": 619
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 9.359587838570496e-06,
|
|
"loss": 0.6135,
|
|
"step": 620
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 9.331166687117816e-06,
|
|
"loss": 0.6104,
|
|
"step": 621
|
|
},
|
|
{
|
|
"epoch": 1.62,
|
|
"learning_rate": 9.302750961496889e-06,
|
|
"loss": 0.6077,
|
|
"step": 622
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 9.274340892226954e-06,
|
|
"loss": 0.6084,
|
|
"step": 623
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 9.245936709781371e-06,
|
|
"loss": 0.6143,
|
|
"step": 624
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 9.217538644585727e-06,
|
|
"loss": 0.6128,
|
|
"step": 625
|
|
},
|
|
{
|
|
"epoch": 1.63,
|
|
"learning_rate": 9.189146927016004e-06,
|
|
"loss": 0.6223,
|
|
"step": 626
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 9.160761787396665e-06,
|
|
"loss": 0.6182,
|
|
"step": 627
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 9.132383455998836e-06,
|
|
"loss": 0.6099,
|
|
"step": 628
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 9.104012163038393e-06,
|
|
"loss": 0.6155,
|
|
"step": 629
|
|
},
|
|
{
|
|
"epoch": 1.64,
|
|
"learning_rate": 9.075648138674123e-06,
|
|
"loss": 0.6018,
|
|
"step": 630
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 9.047291613005844e-06,
|
|
"loss": 0.6104,
|
|
"step": 631
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 9.018942816072544e-06,
|
|
"loss": 0.6152,
|
|
"step": 632
|
|
},
|
|
{
|
|
"epoch": 1.65,
|
|
"learning_rate": 8.990601977850507e-06,
|
|
"loss": 0.6204,
|
|
"step": 633
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 8.962269328251465e-06,
|
|
"loss": 0.6125,
|
|
"step": 634
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 8.933945097120704e-06,
|
|
"loss": 0.6143,
|
|
"step": 635
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 8.90562951423523e-06,
|
|
"loss": 0.6108,
|
|
"step": 636
|
|
},
|
|
{
|
|
"epoch": 1.66,
|
|
"learning_rate": 8.87732280930188e-06,
|
|
"loss": 0.613,
|
|
"step": 637
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 8.849025211955482e-06,
|
|
"loss": 0.6135,
|
|
"step": 638
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 8.820736951756972e-06,
|
|
"loss": 0.6152,
|
|
"step": 639
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 8.79245825819154e-06,
|
|
"loss": 0.6169,
|
|
"step": 640
|
|
},
|
|
{
|
|
"epoch": 1.67,
|
|
"learning_rate": 8.764189360666771e-06,
|
|
"loss": 0.6162,
|
|
"step": 641
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 8.735930488510773e-06,
|
|
"loss": 0.6128,
|
|
"step": 642
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 8.70768187097034e-06,
|
|
"loss": 0.613,
|
|
"step": 643
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 8.67944373720905e-06,
|
|
"loss": 0.6089,
|
|
"step": 644
|
|
},
|
|
{
|
|
"epoch": 1.68,
|
|
"learning_rate": 8.651216316305467e-06,
|
|
"loss": 0.6182,
|
|
"step": 645
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 8.622999837251219e-06,
|
|
"loss": 0.6143,
|
|
"step": 646
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 8.594794528949183e-06,
|
|
"loss": 0.6128,
|
|
"step": 647
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 8.566600620211614e-06,
|
|
"loss": 0.6091,
|
|
"step": 648
|
|
},
|
|
{
|
|
"epoch": 1.69,
|
|
"learning_rate": 8.538418339758287e-06,
|
|
"loss": 0.615,
|
|
"step": 649
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 8.510247916214645e-06,
|
|
"loss": 0.6135,
|
|
"step": 650
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 8.482089578109945e-06,
|
|
"loss": 0.616,
|
|
"step": 651
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 8.453943553875393e-06,
|
|
"loss": 0.6099,
|
|
"step": 652
|
|
},
|
|
{
|
|
"epoch": 1.7,
|
|
"learning_rate": 8.425810071842316e-06,
|
|
"loss": 0.6206,
|
|
"step": 653
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 8.397689360240279e-06,
|
|
"loss": 0.615,
|
|
"step": 654
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 8.369581647195252e-06,
|
|
"loss": 0.6025,
|
|
"step": 655
|
|
},
|
|
{
|
|
"epoch": 1.71,
|
|
"learning_rate": 8.341487160727772e-06,
|
|
"loss": 0.6204,
|
|
"step": 656
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 8.31340612875105e-06,
|
|
"loss": 0.6084,
|
|
"step": 657
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 8.285338779069167e-06,
|
|
"loss": 0.6072,
|
|
"step": 658
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 8.257285339375203e-06,
|
|
"loss": 0.606,
|
|
"step": 659
|
|
},
|
|
{
|
|
"epoch": 1.72,
|
|
"learning_rate": 8.229246037249395e-06,
|
|
"loss": 0.6106,
|
|
"step": 660
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 8.20122110015729e-06,
|
|
"loss": 0.6006,
|
|
"step": 661
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 8.173210755447905e-06,
|
|
"loss": 0.6106,
|
|
"step": 662
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 8.145215230351864e-06,
|
|
"loss": 0.6079,
|
|
"step": 663
|
|
},
|
|
{
|
|
"epoch": 1.73,
|
|
"learning_rate": 8.117234751979595e-06,
|
|
"loss": 0.6099,
|
|
"step": 664
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 8.089269547319428e-06,
|
|
"loss": 0.6074,
|
|
"step": 665
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 8.061319843235816e-06,
|
|
"loss": 0.6113,
|
|
"step": 666
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 8.033385866467445e-06,
|
|
"loss": 0.6125,
|
|
"step": 667
|
|
},
|
|
{
|
|
"epoch": 1.74,
|
|
"learning_rate": 8.005467843625428e-06,
|
|
"loss": 0.6138,
|
|
"step": 668
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 7.977566001191447e-06,
|
|
"loss": 0.6113,
|
|
"step": 669
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 7.949680565515924e-06,
|
|
"loss": 0.6179,
|
|
"step": 670
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 7.921811762816186e-06,
|
|
"loss": 0.6106,
|
|
"step": 671
|
|
},
|
|
{
|
|
"epoch": 1.75,
|
|
"learning_rate": 7.893959819174619e-06,
|
|
"loss": 0.6108,
|
|
"step": 672
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 7.866124960536856e-06,
|
|
"loss": 0.6082,
|
|
"step": 673
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 7.838307412709909e-06,
|
|
"loss": 0.6143,
|
|
"step": 674
|
|
},
|
|
{
|
|
"epoch": 1.76,
|
|
"learning_rate": 7.810507401360379e-06,
|
|
"loss": 0.6184,
|
|
"step": 675
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 7.782725152012591e-06,
|
|
"loss": 0.6123,
|
|
"step": 676
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 7.754960890046785e-06,
|
|
"loss": 0.6147,
|
|
"step": 677
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 7.727214840697272e-06,
|
|
"loss": 0.6121,
|
|
"step": 678
|
|
},
|
|
{
|
|
"epoch": 1.77,
|
|
"learning_rate": 7.69948722905063e-06,
|
|
"loss": 0.6104,
|
|
"step": 679
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 7.671778280043847e-06,
|
|
"loss": 0.6116,
|
|
"step": 680
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 7.644088218462527e-06,
|
|
"loss": 0.6091,
|
|
"step": 681
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 7.616417268939037e-06,
|
|
"loss": 0.6074,
|
|
"step": 682
|
|
},
|
|
{
|
|
"epoch": 1.78,
|
|
"learning_rate": 7.588765655950712e-06,
|
|
"loss": 0.6179,
|
|
"step": 683
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 7.561133603818014e-06,
|
|
"loss": 0.6072,
|
|
"step": 684
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 7.533521336702727e-06,
|
|
"loss": 0.6069,
|
|
"step": 685
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 7.505929078606123e-06,
|
|
"loss": 0.6169,
|
|
"step": 686
|
|
},
|
|
{
|
|
"epoch": 1.79,
|
|
"learning_rate": 7.47835705336716e-06,
|
|
"loss": 0.6091,
|
|
"step": 687
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 7.450805484660659e-06,
|
|
"loss": 0.6077,
|
|
"step": 688
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 7.423274595995482e-06,
|
|
"loss": 0.6167,
|
|
"step": 689
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 7.3957646107127415e-06,
|
|
"loss": 0.6096,
|
|
"step": 690
|
|
},
|
|
{
|
|
"epoch": 1.8,
|
|
"learning_rate": 7.368275751983951e-06,
|
|
"loss": 0.6155,
|
|
"step": 691
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 7.3408082428092656e-06,
|
|
"loss": 0.6226,
|
|
"step": 692
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 7.313362306015618e-06,
|
|
"loss": 0.6062,
|
|
"step": 693
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 7.285938164254955e-06,
|
|
"loss": 0.6104,
|
|
"step": 694
|
|
},
|
|
{
|
|
"epoch": 1.81,
|
|
"learning_rate": 7.258536040002406e-06,
|
|
"loss": 0.6111,
|
|
"step": 695
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 7.23115615555449e-06,
|
|
"loss": 0.614,
|
|
"step": 696
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 7.203798733027304e-06,
|
|
"loss": 0.6033,
|
|
"step": 697
|
|
},
|
|
{
|
|
"epoch": 1.82,
|
|
"learning_rate": 7.176463994354732e-06,
|
|
"loss": 0.6128,
|
|
"step": 698
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 7.1491521612866255e-06,
|
|
"loss": 0.6094,
|
|
"step": 699
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 7.121863455387034e-06,
|
|
"loss": 0.6069,
|
|
"step": 700
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 7.094598098032375e-06,
|
|
"loss": 0.6118,
|
|
"step": 701
|
|
},
|
|
{
|
|
"epoch": 1.83,
|
|
"learning_rate": 7.067356310409659e-06,
|
|
"loss": 0.6118,
|
|
"step": 702
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 7.0401383135147e-06,
|
|
"loss": 0.6069,
|
|
"step": 703
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 7.0129443281502925e-06,
|
|
"loss": 0.6125,
|
|
"step": 704
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 6.98577457492446e-06,
|
|
"loss": 0.6118,
|
|
"step": 705
|
|
},
|
|
{
|
|
"epoch": 1.84,
|
|
"learning_rate": 6.9586292742486294e-06,
|
|
"loss": 0.6067,
|
|
"step": 706
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 6.931508646335874e-06,
|
|
"loss": 0.6133,
|
|
"step": 707
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 6.904412911199099e-06,
|
|
"loss": 0.6074,
|
|
"step": 708
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 6.87734228864928e-06,
|
|
"loss": 0.6003,
|
|
"step": 709
|
|
},
|
|
{
|
|
"epoch": 1.85,
|
|
"learning_rate": 6.850296998293661e-06,
|
|
"loss": 0.6108,
|
|
"step": 710
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 6.823277259533989e-06,
|
|
"loss": 0.6094,
|
|
"step": 711
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 6.796283291564723e-06,
|
|
"loss": 0.6079,
|
|
"step": 712
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 6.7693153133712585e-06,
|
|
"loss": 0.6074,
|
|
"step": 713
|
|
},
|
|
{
|
|
"epoch": 1.86,
|
|
"learning_rate": 6.742373543728155e-06,
|
|
"loss": 0.6096,
|
|
"step": 714
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 6.715458201197358e-06,
|
|
"loss": 0.6174,
|
|
"step": 715
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 6.688569504126421e-06,
|
|
"loss": 0.6006,
|
|
"step": 716
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 6.661707670646751e-06,
|
|
"loss": 0.6074,
|
|
"step": 717
|
|
},
|
|
{
|
|
"epoch": 1.87,
|
|
"learning_rate": 6.634872918671811e-06,
|
|
"loss": 0.6116,
|
|
"step": 718
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 6.608065465895377e-06,
|
|
"loss": 0.6096,
|
|
"step": 719
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 6.581285529789774e-06,
|
|
"loss": 0.6028,
|
|
"step": 720
|
|
},
|
|
{
|
|
"epoch": 1.88,
|
|
"learning_rate": 6.554533327604079e-06,
|
|
"loss": 0.615,
|
|
"step": 721
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 6.5278090763623995e-06,
|
|
"loss": 0.6108,
|
|
"step": 722
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 6.501112992862082e-06,
|
|
"loss": 0.6104,
|
|
"step": 723
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 6.474445293671977e-06,
|
|
"loss": 0.605,
|
|
"step": 724
|
|
},
|
|
{
|
|
"epoch": 1.89,
|
|
"learning_rate": 6.447806195130659e-06,
|
|
"loss": 0.6145,
|
|
"step": 725
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 6.4211959133446945e-06,
|
|
"loss": 0.6091,
|
|
"step": 726
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 6.394614664186863e-06,
|
|
"loss": 0.6042,
|
|
"step": 727
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 6.368062663294436e-06,
|
|
"loss": 0.6074,
|
|
"step": 728
|
|
},
|
|
{
|
|
"epoch": 1.9,
|
|
"learning_rate": 6.3415401260674e-06,
|
|
"loss": 0.6086,
|
|
"step": 729
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 6.315047267666729e-06,
|
|
"loss": 0.6025,
|
|
"step": 730
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 6.288584303012624e-06,
|
|
"loss": 0.6052,
|
|
"step": 731
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 6.262151446782785e-06,
|
|
"loss": 0.6028,
|
|
"step": 732
|
|
},
|
|
{
|
|
"epoch": 1.91,
|
|
"learning_rate": 6.2357489134106485e-06,
|
|
"loss": 0.6111,
|
|
"step": 733
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 6.2093769170836735e-06,
|
|
"loss": 0.6123,
|
|
"step": 734
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 6.183035671741586e-06,
|
|
"loss": 0.604,
|
|
"step": 735
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 6.156725391074642e-06,
|
|
"loss": 0.613,
|
|
"step": 736
|
|
},
|
|
{
|
|
"epoch": 1.92,
|
|
"learning_rate": 6.1304462885219154e-06,
|
|
"loss": 0.6116,
|
|
"step": 737
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 6.104198577269527e-06,
|
|
"loss": 0.614,
|
|
"step": 738
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 6.077982470248968e-06,
|
|
"loss": 0.6064,
|
|
"step": 739
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 6.0517981801353194e-06,
|
|
"loss": 0.6072,
|
|
"step": 740
|
|
},
|
|
{
|
|
"epoch": 1.93,
|
|
"learning_rate": 6.025645919345569e-06,
|
|
"loss": 0.6038,
|
|
"step": 741
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5.9995259000368555e-06,
|
|
"loss": 0.5999,
|
|
"step": 742
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5.973438334104779e-06,
|
|
"loss": 0.6211,
|
|
"step": 743
|
|
},
|
|
{
|
|
"epoch": 1.94,
|
|
"learning_rate": 5.947383433181648e-06,
|
|
"loss": 0.6106,
|
|
"step": 744
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5.921361408634798e-06,
|
|
"loss": 0.6079,
|
|
"step": 745
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5.895372471564837e-06,
|
|
"loss": 0.6035,
|
|
"step": 746
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5.86941683280398e-06,
|
|
"loss": 0.6099,
|
|
"step": 747
|
|
},
|
|
{
|
|
"epoch": 1.95,
|
|
"learning_rate": 5.843494702914289e-06,
|
|
"loss": 0.6067,
|
|
"step": 748
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5.817606292186006e-06,
|
|
"loss": 0.6135,
|
|
"step": 749
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5.791751810635824e-06,
|
|
"loss": 0.6008,
|
|
"step": 750
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5.765931468005189e-06,
|
|
"loss": 0.6047,
|
|
"step": 751
|
|
},
|
|
{
|
|
"epoch": 1.96,
|
|
"learning_rate": 5.740145473758606e-06,
|
|
"loss": 0.5994,
|
|
"step": 752
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5.714394037081918e-06,
|
|
"loss": 0.606,
|
|
"step": 753
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5.688677366880637e-06,
|
|
"loss": 0.6047,
|
|
"step": 754
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5.6629956717782334e-06,
|
|
"loss": 0.6028,
|
|
"step": 755
|
|
},
|
|
{
|
|
"epoch": 1.97,
|
|
"learning_rate": 5.637349160114445e-06,
|
|
"loss": 0.6145,
|
|
"step": 756
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5.6117380399435824e-06,
|
|
"loss": 0.6035,
|
|
"step": 757
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5.586162519032857e-06,
|
|
"loss": 0.6091,
|
|
"step": 758
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5.560622804860679e-06,
|
|
"loss": 0.6096,
|
|
"step": 759
|
|
},
|
|
{
|
|
"epoch": 1.98,
|
|
"learning_rate": 5.535119104614988e-06,
|
|
"loss": 0.6089,
|
|
"step": 760
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5.509651625191547e-06,
|
|
"loss": 0.6035,
|
|
"step": 761
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5.484220573192307e-06,
|
|
"loss": 0.6091,
|
|
"step": 762
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5.458826154923682e-06,
|
|
"loss": 0.6147,
|
|
"step": 763
|
|
},
|
|
{
|
|
"epoch": 1.99,
|
|
"learning_rate": 5.433468576394914e-06,
|
|
"loss": 0.5999,
|
|
"step": 764
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 5.408148043316381e-06,
|
|
"loss": 0.6072,
|
|
"step": 765
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 5.382864761097941e-06,
|
|
"loss": 0.6077,
|
|
"step": 766
|
|
},
|
|
{
|
|
"epoch": 2.0,
|
|
"learning_rate": 5.357618934847253e-06,
|
|
"loss": 0.5457,
|
|
"step": 767
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 5.332410769368115e-06,
|
|
"loss": 0.5583,
|
|
"step": 768
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 5.307240469158815e-06,
|
|
"loss": 0.5388,
|
|
"step": 769
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 5.28210823841045e-06,
|
|
"loss": 0.5428,
|
|
"step": 770
|
|
},
|
|
{
|
|
"epoch": 2.01,
|
|
"learning_rate": 5.25701428100531e-06,
|
|
"loss": 0.5518,
|
|
"step": 771
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 5.231958800515164e-06,
|
|
"loss": 0.5564,
|
|
"step": 772
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 5.206942000199666e-06,
|
|
"loss": 0.5483,
|
|
"step": 773
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 5.1819640830046646e-06,
|
|
"loss": 0.5444,
|
|
"step": 774
|
|
},
|
|
{
|
|
"epoch": 2.02,
|
|
"learning_rate": 5.157025251560588e-06,
|
|
"loss": 0.5471,
|
|
"step": 775
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 5.1321257081807755e-06,
|
|
"loss": 0.5417,
|
|
"step": 776
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 5.107265654859856e-06,
|
|
"loss": 0.5383,
|
|
"step": 777
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 5.082445293272087e-06,
|
|
"loss": 0.5405,
|
|
"step": 778
|
|
},
|
|
{
|
|
"epoch": 2.03,
|
|
"learning_rate": 5.057664824769746e-06,
|
|
"loss": 0.5403,
|
|
"step": 779
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 5.032924450381469e-06,
|
|
"loss": 0.55,
|
|
"step": 780
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 5.0082243708106426e-06,
|
|
"loss": 0.551,
|
|
"step": 781
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 4.983564786433763e-06,
|
|
"loss": 0.5391,
|
|
"step": 782
|
|
},
|
|
{
|
|
"epoch": 2.04,
|
|
"learning_rate": 4.9589458972988155e-06,
|
|
"loss": 0.5371,
|
|
"step": 783
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 4.934367903123652e-06,
|
|
"loss": 0.5425,
|
|
"step": 784
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 4.90983100329436e-06,
|
|
"loss": 0.5359,
|
|
"step": 785
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 4.88533539686366e-06,
|
|
"loss": 0.5388,
|
|
"step": 786
|
|
},
|
|
{
|
|
"epoch": 2.05,
|
|
"learning_rate": 4.860881282549286e-06,
|
|
"loss": 0.5376,
|
|
"step": 787
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 4.836468858732374e-06,
|
|
"loss": 0.5471,
|
|
"step": 788
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 4.812098323455839e-06,
|
|
"loss": 0.5386,
|
|
"step": 789
|
|
},
|
|
{
|
|
"epoch": 2.06,
|
|
"learning_rate": 4.787769874422798e-06,
|
|
"loss": 0.5466,
|
|
"step": 790
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 4.763483708994931e-06,
|
|
"loss": 0.5557,
|
|
"step": 791
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 4.739240024190904e-06,
|
|
"loss": 0.5388,
|
|
"step": 792
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 4.715039016684769e-06,
|
|
"loss": 0.5437,
|
|
"step": 793
|
|
},
|
|
{
|
|
"epoch": 2.07,
|
|
"learning_rate": 4.690880882804363e-06,
|
|
"loss": 0.5414,
|
|
"step": 794
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 4.666765818529702e-06,
|
|
"loss": 0.5447,
|
|
"step": 795
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 4.642694019491423e-06,
|
|
"loss": 0.5474,
|
|
"step": 796
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 4.618665680969164e-06,
|
|
"loss": 0.5298,
|
|
"step": 797
|
|
},
|
|
{
|
|
"epoch": 2.08,
|
|
"learning_rate": 4.5946809978899985e-06,
|
|
"loss": 0.5376,
|
|
"step": 798
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 4.570740164826869e-06,
|
|
"loss": 0.5371,
|
|
"step": 799
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 4.546843375996959e-06,
|
|
"loss": 0.5391,
|
|
"step": 800
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 4.522990825260172e-06,
|
|
"loss": 0.5391,
|
|
"step": 801
|
|
},
|
|
{
|
|
"epoch": 2.09,
|
|
"learning_rate": 4.49918270611752e-06,
|
|
"loss": 0.5427,
|
|
"step": 802
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 4.475419211709576e-06,
|
|
"loss": 0.5339,
|
|
"step": 803
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 4.451700534814903e-06,
|
|
"loss": 0.5422,
|
|
"step": 804
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 4.428026867848486e-06,
|
|
"loss": 0.538,
|
|
"step": 805
|
|
},
|
|
{
|
|
"epoch": 2.1,
|
|
"learning_rate": 4.4043984028601624e-06,
|
|
"loss": 0.5356,
|
|
"step": 806
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 4.380815331533088e-06,
|
|
"loss": 0.5402,
|
|
"step": 807
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 4.357277845182155e-06,
|
|
"loss": 0.5361,
|
|
"step": 808
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 4.333786134752462e-06,
|
|
"loss": 0.5393,
|
|
"step": 809
|
|
},
|
|
{
|
|
"epoch": 2.11,
|
|
"learning_rate": 4.310340390817753e-06,
|
|
"loss": 0.5483,
|
|
"step": 810
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 4.286940803578879e-06,
|
|
"loss": 0.541,
|
|
"step": 811
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 4.263587562862234e-06,
|
|
"loss": 0.5415,
|
|
"step": 812
|
|
},
|
|
{
|
|
"epoch": 2.12,
|
|
"learning_rate": 4.240280858118254e-06,
|
|
"loss": 0.5332,
|
|
"step": 813
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 4.217020878419834e-06,
|
|
"loss": 0.5363,
|
|
"step": 814
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 4.19380781246084e-06,
|
|
"loss": 0.5449,
|
|
"step": 815
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 4.170641848554543e-06,
|
|
"loss": 0.5378,
|
|
"step": 816
|
|
},
|
|
{
|
|
"epoch": 2.13,
|
|
"learning_rate": 4.1475231746321035e-06,
|
|
"loss": 0.5435,
|
|
"step": 817
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 4.124451978241054e-06,
|
|
"loss": 0.5366,
|
|
"step": 818
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 4.101428446543762e-06,
|
|
"loss": 0.5398,
|
|
"step": 819
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 4.07845276631593e-06,
|
|
"loss": 0.5415,
|
|
"step": 820
|
|
},
|
|
{
|
|
"epoch": 2.14,
|
|
"learning_rate": 4.055525123945068e-06,
|
|
"loss": 0.5474,
|
|
"step": 821
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 4.032645705428984e-06,
|
|
"loss": 0.5457,
|
|
"step": 822
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 4.0098146963742734e-06,
|
|
"loss": 0.5332,
|
|
"step": 823
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 3.987032281994821e-06,
|
|
"loss": 0.537,
|
|
"step": 824
|
|
},
|
|
{
|
|
"epoch": 2.15,
|
|
"learning_rate": 3.964298647110282e-06,
|
|
"loss": 0.5282,
|
|
"step": 825
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 3.941613976144612e-06,
|
|
"loss": 0.5381,
|
|
"step": 826
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 3.918978453124533e-06,
|
|
"loss": 0.5432,
|
|
"step": 827
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 3.896392261678078e-06,
|
|
"loss": 0.5344,
|
|
"step": 828
|
|
},
|
|
{
|
|
"epoch": 2.16,
|
|
"learning_rate": 3.8738555850330635e-06,
|
|
"loss": 0.5398,
|
|
"step": 829
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 3.85136860601564e-06,
|
|
"loss": 0.5405,
|
|
"step": 830
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 3.828931507048786e-06,
|
|
"loss": 0.5496,
|
|
"step": 831
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 3.8065444701508312e-06,
|
|
"loss": 0.5352,
|
|
"step": 832
|
|
},
|
|
{
|
|
"epoch": 2.17,
|
|
"learning_rate": 3.78420767693399e-06,
|
|
"loss": 0.5417,
|
|
"step": 833
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 3.7619213086028695e-06,
|
|
"loss": 0.5347,
|
|
"step": 834
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 3.7396855459530225e-06,
|
|
"loss": 0.5376,
|
|
"step": 835
|
|
},
|
|
{
|
|
"epoch": 2.18,
|
|
"learning_rate": 3.7175005693694587e-06,
|
|
"loss": 0.5376,
|
|
"step": 836
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 3.6953665588251984e-06,
|
|
"loss": 0.5442,
|
|
"step": 837
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 3.6732836938798077e-06,
|
|
"loss": 0.5393,
|
|
"step": 838
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 3.6512521536779364e-06,
|
|
"loss": 0.5449,
|
|
"step": 839
|
|
},
|
|
{
|
|
"epoch": 2.19,
|
|
"learning_rate": 3.6292721169478652e-06,
|
|
"loss": 0.5454,
|
|
"step": 840
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 3.607343762000063e-06,
|
|
"loss": 0.5354,
|
|
"step": 841
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 3.5854672667257373e-06,
|
|
"loss": 0.5315,
|
|
"step": 842
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 3.563642808595392e-06,
|
|
"loss": 0.5435,
|
|
"step": 843
|
|
},
|
|
{
|
|
"epoch": 2.2,
|
|
"learning_rate": 3.5418705646573758e-06,
|
|
"loss": 0.5408,
|
|
"step": 844
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 3.5201507115364685e-06,
|
|
"loss": 0.5417,
|
|
"step": 845
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 3.4984834254324217e-06,
|
|
"loss": 0.5415,
|
|
"step": 846
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 3.476868882118557e-06,
|
|
"loss": 0.5391,
|
|
"step": 847
|
|
},
|
|
{
|
|
"epoch": 2.21,
|
|
"learning_rate": 3.4553072569403178e-06,
|
|
"loss": 0.5391,
|
|
"step": 848
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 3.4337987248138594e-06,
|
|
"loss": 0.5388,
|
|
"step": 849
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 3.4123434602246277e-06,
|
|
"loss": 0.5496,
|
|
"step": 850
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 3.3909416372259305e-06,
|
|
"loss": 0.5391,
|
|
"step": 851
|
|
},
|
|
{
|
|
"epoch": 2.22,
|
|
"learning_rate": 3.3695934294375544e-06,
|
|
"loss": 0.5425,
|
|
"step": 852
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 3.3482990100443157e-06,
|
|
"loss": 0.5317,
|
|
"step": 853
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 3.3270585517947096e-06,
|
|
"loss": 0.5396,
|
|
"step": 854
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 3.305872226999449e-06,
|
|
"loss": 0.5432,
|
|
"step": 855
|
|
},
|
|
{
|
|
"epoch": 2.23,
|
|
"learning_rate": 3.284740207530115e-06,
|
|
"loss": 0.5432,
|
|
"step": 856
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 3.2636626648177284e-06,
|
|
"loss": 0.5337,
|
|
"step": 857
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 3.2426397698513836e-06,
|
|
"loss": 0.5427,
|
|
"step": 858
|
|
},
|
|
{
|
|
"epoch": 2.24,
|
|
"learning_rate": 3.2216716931768487e-06,
|
|
"loss": 0.5336,
|
|
"step": 859
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 3.200758604895189e-06,
|
|
"loss": 0.5388,
|
|
"step": 860
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 3.1799006746613713e-06,
|
|
"loss": 0.5374,
|
|
"step": 861
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 3.15909807168291e-06,
|
|
"loss": 0.5433,
|
|
"step": 862
|
|
},
|
|
{
|
|
"epoch": 2.25,
|
|
"learning_rate": 3.138350964718485e-06,
|
|
"loss": 0.5374,
|
|
"step": 863
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 3.1176595220765584e-06,
|
|
"loss": 0.5417,
|
|
"step": 864
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 3.0970239116140343e-06,
|
|
"loss": 0.5349,
|
|
"step": 865
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 3.0764443007348808e-06,
|
|
"loss": 0.5417,
|
|
"step": 866
|
|
},
|
|
{
|
|
"epoch": 2.26,
|
|
"learning_rate": 3.055920856388779e-06,
|
|
"loss": 0.5437,
|
|
"step": 867
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 3.035453745069755e-06,
|
|
"loss": 0.532,
|
|
"step": 868
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 3.0150431328148534e-06,
|
|
"loss": 0.5386,
|
|
"step": 869
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 2.994689185202767e-06,
|
|
"loss": 0.5422,
|
|
"step": 870
|
|
},
|
|
{
|
|
"epoch": 2.27,
|
|
"learning_rate": 2.9743920673525126e-06,
|
|
"loss": 0.5334,
|
|
"step": 871
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 2.95415194392207e-06,
|
|
"loss": 0.5461,
|
|
"step": 872
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 2.9339689791070713e-06,
|
|
"loss": 0.5369,
|
|
"step": 873
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 2.913843336639444e-06,
|
|
"loss": 0.5342,
|
|
"step": 874
|
|
},
|
|
{
|
|
"epoch": 2.28,
|
|
"learning_rate": 2.8937751797861025e-06,
|
|
"loss": 0.542,
|
|
"step": 875
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 2.873764671347614e-06,
|
|
"loss": 0.5408,
|
|
"step": 876
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 2.853811973656885e-06,
|
|
"loss": 0.5381,
|
|
"step": 877
|
|
},
|
|
{
|
|
"epoch": 2.29,
|
|
"learning_rate": 2.8339172485778243e-06,
|
|
"loss": 0.5337,
|
|
"step": 878
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 2.8140806575040603e-06,
|
|
"loss": 0.532,
|
|
"step": 879
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 2.7943023613576127e-06,
|
|
"loss": 0.5393,
|
|
"step": 880
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 2.7745825205875764e-06,
|
|
"loss": 0.5354,
|
|
"step": 881
|
|
},
|
|
{
|
|
"epoch": 2.3,
|
|
"learning_rate": 2.7549212951688598e-06,
|
|
"loss": 0.5376,
|
|
"step": 882
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 2.7353188446008374e-06,
|
|
"loss": 0.5371,
|
|
"step": 883
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 2.7157753279060926e-06,
|
|
"loss": 0.5386,
|
|
"step": 884
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 2.6962909036291075e-06,
|
|
"loss": 0.54,
|
|
"step": 885
|
|
},
|
|
{
|
|
"epoch": 2.31,
|
|
"learning_rate": 2.6768657298349897e-06,
|
|
"loss": 0.5359,
|
|
"step": 886
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 2.657499964108181e-06,
|
|
"loss": 0.5375,
|
|
"step": 887
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 2.6381937635511854e-06,
|
|
"loss": 0.5403,
|
|
"step": 888
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 2.6189472847832853e-06,
|
|
"loss": 0.5383,
|
|
"step": 889
|
|
},
|
|
{
|
|
"epoch": 2.32,
|
|
"learning_rate": 2.5997606839392866e-06,
|
|
"loss": 0.5403,
|
|
"step": 890
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 2.5806341166682294e-06,
|
|
"loss": 0.5364,
|
|
"step": 891
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 2.5615677381321492e-06,
|
|
"loss": 0.5388,
|
|
"step": 892
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 2.5425617030048045e-06,
|
|
"loss": 0.5431,
|
|
"step": 893
|
|
},
|
|
{
|
|
"epoch": 2.33,
|
|
"learning_rate": 2.523616165470425e-06,
|
|
"loss": 0.541,
|
|
"step": 894
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 2.504731279222463e-06,
|
|
"loss": 0.5425,
|
|
"step": 895
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 2.485907197462337e-06,
|
|
"loss": 0.5386,
|
|
"step": 896
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 2.467144072898202e-06,
|
|
"loss": 0.5361,
|
|
"step": 897
|
|
},
|
|
{
|
|
"epoch": 2.34,
|
|
"learning_rate": 2.448442057743706e-06,
|
|
"loss": 0.5432,
|
|
"step": 898
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 2.4298013037167544e-06,
|
|
"loss": 0.5339,
|
|
"step": 899
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 2.411221962038269e-06,
|
|
"loss": 0.5369,
|
|
"step": 900
|
|
},
|
|
{
|
|
"epoch": 2.35,
|
|
"learning_rate": 2.3927041834309873e-06,
|
|
"loss": 0.54,
|
|
"step": 901
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 2.3742481181182063e-06,
|
|
"loss": 0.5459,
|
|
"step": 902
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 2.3558539158225956e-06,
|
|
"loss": 0.5319,
|
|
"step": 903
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 2.3375217257649605e-06,
|
|
"loss": 0.5393,
|
|
"step": 904
|
|
},
|
|
{
|
|
"epoch": 2.36,
|
|
"learning_rate": 2.319251696663044e-06,
|
|
"loss": 0.5376,
|
|
"step": 905
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 2.3010439767303084e-06,
|
|
"loss": 0.5425,
|
|
"step": 906
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 2.2828987136747503e-06,
|
|
"loss": 0.5461,
|
|
"step": 907
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 2.2648160546976785e-06,
|
|
"loss": 0.5332,
|
|
"step": 908
|
|
},
|
|
{
|
|
"epoch": 2.37,
|
|
"learning_rate": 2.2467961464925514e-06,
|
|
"loss": 0.5408,
|
|
"step": 909
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 2.228839135243753e-06,
|
|
"loss": 0.5457,
|
|
"step": 910
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 2.210945166625431e-06,
|
|
"loss": 0.5295,
|
|
"step": 911
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 2.193114385800309e-06,
|
|
"loss": 0.5393,
|
|
"step": 912
|
|
},
|
|
{
|
|
"epoch": 2.38,
|
|
"learning_rate": 2.1753469374184987e-06,
|
|
"loss": 0.543,
|
|
"step": 913
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 2.157642965616341e-06,
|
|
"loss": 0.5393,
|
|
"step": 914
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 2.140002614015232e-06,
|
|
"loss": 0.5374,
|
|
"step": 915
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 2.122426025720454e-06,
|
|
"loss": 0.5288,
|
|
"step": 916
|
|
},
|
|
{
|
|
"epoch": 2.39,
|
|
"learning_rate": 2.104913343320013e-06,
|
|
"loss": 0.54,
|
|
"step": 917
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 2.087464708883494e-06,
|
|
"loss": 0.5505,
|
|
"step": 918
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 2.070080263960892e-06,
|
|
"loss": 0.542,
|
|
"step": 919
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 2.0527601495814773e-06,
|
|
"loss": 0.5391,
|
|
"step": 920
|
|
},
|
|
{
|
|
"epoch": 2.4,
|
|
"learning_rate": 2.0355045062526457e-06,
|
|
"loss": 0.5432,
|
|
"step": 921
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 2.0183134739587806e-06,
|
|
"loss": 0.5347,
|
|
"step": 922
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 2.00118719216011e-06,
|
|
"loss": 0.5427,
|
|
"step": 923
|
|
},
|
|
{
|
|
"epoch": 2.41,
|
|
"learning_rate": 1.9841257997915865e-06,
|
|
"loss": 0.5359,
|
|
"step": 924
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 1.9671294352617543e-06,
|
|
"loss": 0.5356,
|
|
"step": 925
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 1.9501982364516248e-06,
|
|
"loss": 0.5396,
|
|
"step": 926
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 1.9333323407135653e-06,
|
|
"loss": 0.5355,
|
|
"step": 927
|
|
},
|
|
{
|
|
"epoch": 2.42,
|
|
"learning_rate": 1.9165318848701686e-06,
|
|
"loss": 0.5374,
|
|
"step": 928
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 1.8997970052131665e-06,
|
|
"loss": 0.5319,
|
|
"step": 929
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 1.883127837502302e-06,
|
|
"loss": 0.5459,
|
|
"step": 930
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 1.8665245169642443e-06,
|
|
"loss": 0.5337,
|
|
"step": 931
|
|
},
|
|
{
|
|
"epoch": 2.43,
|
|
"learning_rate": 1.8499871782914824e-06,
|
|
"loss": 0.5417,
|
|
"step": 932
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 1.8335159556412385e-06,
|
|
"loss": 0.5432,
|
|
"step": 933
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 1.8171109826343669e-06,
|
|
"loss": 0.5322,
|
|
"step": 934
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 1.8007723923542919e-06,
|
|
"loss": 0.5404,
|
|
"step": 935
|
|
},
|
|
{
|
|
"epoch": 2.44,
|
|
"learning_rate": 1.7845003173459007e-06,
|
|
"loss": 0.5425,
|
|
"step": 936
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 1.7682948896145036e-06,
|
|
"loss": 0.5364,
|
|
"step": 937
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 1.7521562406247239e-06,
|
|
"loss": 0.5386,
|
|
"step": 938
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 1.7360845012994609e-06,
|
|
"loss": 0.527,
|
|
"step": 939
|
|
},
|
|
{
|
|
"epoch": 2.45,
|
|
"learning_rate": 1.7200798020188103e-06,
|
|
"loss": 0.53,
|
|
"step": 940
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 1.7041422726190182e-06,
|
|
"loss": 0.5404,
|
|
"step": 941
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 1.6882720423914211e-06,
|
|
"loss": 0.5405,
|
|
"step": 942
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 1.6724692400814002e-06,
|
|
"loss": 0.5321,
|
|
"step": 943
|
|
},
|
|
{
|
|
"epoch": 2.46,
|
|
"learning_rate": 1.6567339938873362e-06,
|
|
"loss": 0.5371,
|
|
"step": 944
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 1.6410664314595625e-06,
|
|
"loss": 0.5396,
|
|
"step": 945
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 1.6254666798993458e-06,
|
|
"loss": 0.5427,
|
|
"step": 946
|
|
},
|
|
{
|
|
"epoch": 2.47,
|
|
"learning_rate": 1.609934865757835e-06,
|
|
"loss": 0.5437,
|
|
"step": 947
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 1.59447111503505e-06,
|
|
"loss": 0.5386,
|
|
"step": 948
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 1.5790755531788571e-06,
|
|
"loss": 0.533,
|
|
"step": 949
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 1.563748305083944e-06,
|
|
"loss": 0.5442,
|
|
"step": 950
|
|
},
|
|
{
|
|
"epoch": 2.48,
|
|
"learning_rate": 1.5484894950908091e-06,
|
|
"loss": 0.5319,
|
|
"step": 951
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 1.5332992469847597e-06,
|
|
"loss": 0.543,
|
|
"step": 952
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 1.5181776839949014e-06,
|
|
"loss": 0.54,
|
|
"step": 953
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 1.5031249287931403e-06,
|
|
"loss": 0.5305,
|
|
"step": 954
|
|
},
|
|
{
|
|
"epoch": 2.49,
|
|
"learning_rate": 1.4881411034931825e-06,
|
|
"loss": 0.5337,
|
|
"step": 955
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 1.4732263296495564e-06,
|
|
"loss": 0.54,
|
|
"step": 956
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 1.458380728256611e-06,
|
|
"loss": 0.5398,
|
|
"step": 957
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 1.4436044197475463e-06,
|
|
"loss": 0.5378,
|
|
"step": 958
|
|
},
|
|
{
|
|
"epoch": 2.5,
|
|
"learning_rate": 1.4288975239934312e-06,
|
|
"loss": 0.5408,
|
|
"step": 959
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 1.414260160302231e-06,
|
|
"loss": 0.5342,
|
|
"step": 960
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 1.3996924474178443e-06,
|
|
"loss": 0.5342,
|
|
"step": 961
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 1.3851945035191272e-06,
|
|
"loss": 0.543,
|
|
"step": 962
|
|
},
|
|
{
|
|
"epoch": 2.51,
|
|
"learning_rate": 1.3707664462189518e-06,
|
|
"loss": 0.5444,
|
|
"step": 963
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 1.3564083925632333e-06,
|
|
"loss": 0.5354,
|
|
"step": 964
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 1.3421204590300064e-06,
|
|
"loss": 0.5305,
|
|
"step": 965
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 1.327902761528448e-06,
|
|
"loss": 0.5383,
|
|
"step": 966
|
|
},
|
|
{
|
|
"epoch": 2.52,
|
|
"learning_rate": 1.3137554153979648e-06,
|
|
"loss": 0.5308,
|
|
"step": 967
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 1.2996785354072371e-06,
|
|
"loss": 0.5364,
|
|
"step": 968
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 1.2856722357533058e-06,
|
|
"loss": 0.5376,
|
|
"step": 969
|
|
},
|
|
{
|
|
"epoch": 2.53,
|
|
"learning_rate": 1.2717366300606303e-06,
|
|
"loss": 0.5359,
|
|
"step": 970
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 1.2578718313801796e-06,
|
|
"loss": 0.5442,
|
|
"step": 971
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 1.2440779521885027e-06,
|
|
"loss": 0.5391,
|
|
"step": 972
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 1.2303551043868278e-06,
|
|
"loss": 0.5452,
|
|
"step": 973
|
|
},
|
|
{
|
|
"epoch": 2.54,
|
|
"learning_rate": 1.2167033993001443e-06,
|
|
"loss": 0.5374,
|
|
"step": 974
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 1.2031229476763074e-06,
|
|
"loss": 0.5282,
|
|
"step": 975
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 1.1896138596851436e-06,
|
|
"loss": 0.5332,
|
|
"step": 976
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 1.1761762449175362e-06,
|
|
"loss": 0.5336,
|
|
"step": 977
|
|
},
|
|
{
|
|
"epoch": 2.55,
|
|
"learning_rate": 1.162810212384564e-06,
|
|
"loss": 0.5388,
|
|
"step": 978
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 1.149515870516592e-06,
|
|
"loss": 0.5327,
|
|
"step": 979
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 1.13629332716241e-06,
|
|
"loss": 0.5338,
|
|
"step": 980
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 1.1231426895883502e-06,
|
|
"loss": 0.5291,
|
|
"step": 981
|
|
},
|
|
{
|
|
"epoch": 2.56,
|
|
"learning_rate": 1.1100640644774175e-06,
|
|
"loss": 0.5415,
|
|
"step": 982
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 1.0970575579284192e-06,
|
|
"loss": 0.5347,
|
|
"step": 983
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 1.084123275455119e-06,
|
|
"loss": 0.5454,
|
|
"step": 984
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 1.0712613219853585e-06,
|
|
"loss": 0.5417,
|
|
"step": 985
|
|
},
|
|
{
|
|
"epoch": 2.57,
|
|
"learning_rate": 1.0584718018602314e-06,
|
|
"loss": 0.5349,
|
|
"step": 986
|
|
},
|
|
{
|
|
"epoch": 2.58,
|
|
"learning_rate": 1.0457548188332155e-06,
|
|
"loss": 0.5374,
|
|
"step": 987
|
|
},
|
|
{
|
|
"epoch": 2.58,
|
|
"learning_rate": 1.0331104760693455e-06,
|
|
"loss": 0.5398,
|
|
"step": 988
|
|
},
|
|
{
|
|
"epoch": 2.58,
|
|
"learning_rate": 1.0205388761443668e-06,
|
|
"loss": 0.5337,
|
|
"step": 989
|
|
},
|
|
{
|
|
"epoch": 2.58,
|
|
"learning_rate": 1.0080401210439073e-06,
|
|
"loss": 0.5405,
|
|
"step": 990
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 9.956143121626526e-07,
|
|
"loss": 0.5408,
|
|
"step": 991
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 9.832615503035182e-07,
|
|
"loss": 0.5322,
|
|
"step": 992
|
|
},
|
|
{
|
|
"epoch": 2.59,
|
|
"learning_rate": 9.709819356768368e-07,
|
|
"loss": 0.5327,
|
|
"step": 993
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 9.58775567899538e-07,
|
|
"loss": 0.5417,
|
|
"step": 994
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 9.466425459943496e-07,
|
|
"loss": 0.5328,
|
|
"step": 995
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 9.345829683889873e-07,
|
|
"loss": 0.5278,
|
|
"step": 996
|
|
},
|
|
{
|
|
"epoch": 2.6,
|
|
"learning_rate": 9.225969329153572e-07,
|
|
"loss": 0.5342,
|
|
"step": 997
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 9.106845368087691e-07,
|
|
"loss": 0.5371,
|
|
"step": 998
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 8.988458767071373e-07,
|
|
"loss": 0.5391,
|
|
"step": 999
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 8.870810486502002e-07,
|
|
"loss": 0.5388,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"eval_loss": 0.665318489074707,
|
|
"eval_runtime": 1461.0767,
|
|
"eval_samples_per_second": 1.37,
|
|
"eval_steps_per_second": 0.343,
|
|
"step": 1000
|
|
},
|
|
{
|
|
"epoch": 2.61,
|
|
"learning_rate": 8.753901480787474e-07,
|
|
"loss": 0.5403,
|
|
"step": 1001
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 8.637732698338352e-07,
|
|
"loss": 0.5374,
|
|
"step": 1002
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 8.522305081560267e-07,
|
|
"loss": 0.5359,
|
|
"step": 1003
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 8.407619566846203e-07,
|
|
"loss": 0.543,
|
|
"step": 1004
|
|
},
|
|
{
|
|
"epoch": 2.62,
|
|
"learning_rate": 8.293677084568963e-07,
|
|
"loss": 0.541,
|
|
"step": 1005
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 8.180478559073524e-07,
|
|
"loss": 0.5311,
|
|
"step": 1006
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 8.068024908669658e-07,
|
|
"loss": 0.5315,
|
|
"step": 1007
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 7.956317045624406e-07,
|
|
"loss": 0.5267,
|
|
"step": 1008
|
|
},
|
|
{
|
|
"epoch": 2.63,
|
|
"learning_rate": 7.845355876154703e-07,
|
|
"loss": 0.5383,
|
|
"step": 1009
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 7.735142300420018e-07,
|
|
"loss": 0.5354,
|
|
"step": 1010
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 7.625677212515025e-07,
|
|
"loss": 0.5297,
|
|
"step": 1011
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 7.516961500462438e-07,
|
|
"loss": 0.5334,
|
|
"step": 1012
|
|
},
|
|
{
|
|
"epoch": 2.64,
|
|
"learning_rate": 7.408996046205674e-07,
|
|
"loss": 0.54,
|
|
"step": 1013
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 7.30178172560182e-07,
|
|
"loss": 0.5383,
|
|
"step": 1014
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 7.19531940841447e-07,
|
|
"loss": 0.5286,
|
|
"step": 1015
|
|
},
|
|
{
|
|
"epoch": 2.65,
|
|
"learning_rate": 7.089609958306709e-07,
|
|
"loss": 0.5405,
|
|
"step": 1016
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 6.984654232833999e-07,
|
|
"loss": 0.5369,
|
|
"step": 1017
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 6.880453083437388e-07,
|
|
"loss": 0.5283,
|
|
"step": 1018
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 6.777007355436405e-07,
|
|
"loss": 0.5361,
|
|
"step": 1019
|
|
},
|
|
{
|
|
"epoch": 2.66,
|
|
"learning_rate": 6.674317888022452e-07,
|
|
"loss": 0.5425,
|
|
"step": 1020
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 6.572385514251744e-07,
|
|
"loss": 0.5308,
|
|
"step": 1021
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 6.471211061038696e-07,
|
|
"loss": 0.5466,
|
|
"step": 1022
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 6.370795349149228e-07,
|
|
"loss": 0.5336,
|
|
"step": 1023
|
|
},
|
|
{
|
|
"epoch": 2.67,
|
|
"learning_rate": 6.271139193193954e-07,
|
|
"loss": 0.5396,
|
|
"step": 1024
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 6.17224340162178e-07,
|
|
"loss": 0.5293,
|
|
"step": 1025
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 6.074108776713206e-07,
|
|
"loss": 0.5312,
|
|
"step": 1026
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 5.976736114573867e-07,
|
|
"loss": 0.531,
|
|
"step": 1027
|
|
},
|
|
{
|
|
"epoch": 2.68,
|
|
"learning_rate": 5.880126205128033e-07,
|
|
"loss": 0.5315,
|
|
"step": 1028
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 5.784279832112294e-07,
|
|
"loss": 0.5396,
|
|
"step": 1029
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 5.689197773069055e-07,
|
|
"loss": 0.5396,
|
|
"step": 1030
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 5.594880799340385e-07,
|
|
"loss": 0.5352,
|
|
"step": 1031
|
|
},
|
|
{
|
|
"epoch": 2.69,
|
|
"learning_rate": 5.501329676061662e-07,
|
|
"loss": 0.5317,
|
|
"step": 1032
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 5.408545162155421e-07,
|
|
"loss": 0.5403,
|
|
"step": 1033
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 5.316528010325084e-07,
|
|
"loss": 0.5308,
|
|
"step": 1034
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 5.225278967049042e-07,
|
|
"loss": 0.5337,
|
|
"step": 1035
|
|
},
|
|
{
|
|
"epoch": 2.7,
|
|
"learning_rate": 5.134798772574434e-07,
|
|
"loss": 0.5327,
|
|
"step": 1036
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 5.045088160911226e-07,
|
|
"loss": 0.531,
|
|
"step": 1037
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 4.95614785982621e-07,
|
|
"loss": 0.5396,
|
|
"step": 1038
|
|
},
|
|
{
|
|
"epoch": 2.71,
|
|
"learning_rate": 4.867978590837152e-07,
|
|
"loss": 0.5283,
|
|
"step": 1039
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 4.780581069206913e-07,
|
|
"loss": 0.5283,
|
|
"step": 1040
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 4.693956003937605e-07,
|
|
"loss": 0.538,
|
|
"step": 1041
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 4.6081040977649094e-07,
|
|
"loss": 0.5322,
|
|
"step": 1042
|
|
},
|
|
{
|
|
"epoch": 2.72,
|
|
"learning_rate": 4.523026047152346e-07,
|
|
"loss": 0.5371,
|
|
"step": 1043
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 4.438722542285623e-07,
|
|
"loss": 0.5396,
|
|
"step": 1044
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 4.355194267067009e-07,
|
|
"loss": 0.5374,
|
|
"step": 1045
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 4.272441899109858e-07,
|
|
"loss": 0.535,
|
|
"step": 1046
|
|
},
|
|
{
|
|
"epoch": 2.73,
|
|
"learning_rate": 4.1904661097330043e-07,
|
|
"loss": 0.5374,
|
|
"step": 1047
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 4.1092675639554657e-07,
|
|
"loss": 0.5374,
|
|
"step": 1048
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 4.028846920490892e-07,
|
|
"loss": 0.5381,
|
|
"step": 1049
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 3.949204831742315e-07,
|
|
"loss": 0.5386,
|
|
"step": 1050
|
|
},
|
|
{
|
|
"epoch": 2.74,
|
|
"learning_rate": 3.870341943796829e-07,
|
|
"loss": 0.5337,
|
|
"step": 1051
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 3.792258896420353e-07,
|
|
"loss": 0.5211,
|
|
"step": 1052
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 3.7149563230524653e-07,
|
|
"loss": 0.5365,
|
|
"step": 1053
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 3.638434850801198e-07,
|
|
"loss": 0.5303,
|
|
"step": 1054
|
|
},
|
|
{
|
|
"epoch": 2.75,
|
|
"learning_rate": 3.5626951004380407e-07,
|
|
"loss": 0.5356,
|
|
"step": 1055
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 3.4877376863928004e-07,
|
|
"loss": 0.5344,
|
|
"step": 1056
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 3.4135632167487274e-07,
|
|
"loss": 0.5334,
|
|
"step": 1057
|
|
},
|
|
{
|
|
"epoch": 2.76,
|
|
"learning_rate": 3.3401722932374534e-07,
|
|
"loss": 0.5325,
|
|
"step": 1058
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 3.267565511234305e-07,
|
|
"loss": 0.5341,
|
|
"step": 1059
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 3.19574345975322e-07,
|
|
"loss": 0.5299,
|
|
"step": 1060
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 3.1247067214422054e-07,
|
|
"loss": 0.5371,
|
|
"step": 1061
|
|
},
|
|
{
|
|
"epoch": 2.77,
|
|
"learning_rate": 3.0544558725784213e-07,
|
|
"loss": 0.5334,
|
|
"step": 1062
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 2.9849914830636595e-07,
|
|
"loss": 0.5398,
|
|
"step": 1063
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 2.916314116419616e-07,
|
|
"loss": 0.5356,
|
|
"step": 1064
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 2.8484243297833945e-07,
|
|
"loss": 0.5311,
|
|
"step": 1065
|
|
},
|
|
{
|
|
"epoch": 2.78,
|
|
"learning_rate": 2.781322673902886e-07,
|
|
"loss": 0.5342,
|
|
"step": 1066
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 2.71500969313242e-07,
|
|
"loss": 0.5334,
|
|
"step": 1067
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 2.649485925428252e-07,
|
|
"loss": 0.5366,
|
|
"step": 1068
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 2.584751902344285e-07,
|
|
"loss": 0.531,
|
|
"step": 1069
|
|
},
|
|
{
|
|
"epoch": 2.79,
|
|
"learning_rate": 2.520808149027654e-07,
|
|
"loss": 0.5288,
|
|
"step": 1070
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 2.457655184214569e-07,
|
|
"loss": 0.5283,
|
|
"step": 1071
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 2.395293520226061e-07,
|
|
"loss": 0.5291,
|
|
"step": 1072
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 2.3337236629637827e-07,
|
|
"loss": 0.5378,
|
|
"step": 1073
|
|
},
|
|
{
|
|
"epoch": 2.8,
|
|
"learning_rate": 2.2729461119059826e-07,
|
|
"loss": 0.5383,
|
|
"step": 1074
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 2.212961360103416e-07,
|
|
"loss": 0.542,
|
|
"step": 1075
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 2.1537698941753504e-07,
|
|
"loss": 0.5293,
|
|
"step": 1076
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 2.0953721943055784e-07,
|
|
"loss": 0.5338,
|
|
"step": 1077
|
|
},
|
|
{
|
|
"epoch": 2.81,
|
|
"learning_rate": 2.037768734238621e-07,
|
|
"loss": 0.5354,
|
|
"step": 1078
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 1.9809599812757652e-07,
|
|
"loss": 0.542,
|
|
"step": 1079
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 1.9249463962713432e-07,
|
|
"loss": 0.5297,
|
|
"step": 1080
|
|
},
|
|
{
|
|
"epoch": 2.82,
|
|
"learning_rate": 1.8697284336290145e-07,
|
|
"loss": 0.5391,
|
|
"step": 1081
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1.8153065412980232e-07,
|
|
"loss": 0.528,
|
|
"step": 1082
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1.761681160769557e-07,
|
|
"loss": 0.5343,
|
|
"step": 1083
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1.70885272707324e-07,
|
|
"loss": 0.5398,
|
|
"step": 1084
|
|
},
|
|
{
|
|
"epoch": 2.83,
|
|
"learning_rate": 1.656821668773534e-07,
|
|
"loss": 0.5413,
|
|
"step": 1085
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 1.6055884079662742e-07,
|
|
"loss": 0.531,
|
|
"step": 1086
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 1.55515336027533e-07,
|
|
"loss": 0.5347,
|
|
"step": 1087
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 1.5055169348490496e-07,
|
|
"loss": 0.5304,
|
|
"step": 1088
|
|
},
|
|
{
|
|
"epoch": 2.84,
|
|
"learning_rate": 1.4566795343571528e-07,
|
|
"loss": 0.5341,
|
|
"step": 1089
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 1.4086415549872777e-07,
|
|
"loss": 0.5349,
|
|
"step": 1090
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 1.3614033864419284e-07,
|
|
"loss": 0.5342,
|
|
"step": 1091
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 1.3149654119351874e-07,
|
|
"loss": 0.5422,
|
|
"step": 1092
|
|
},
|
|
{
|
|
"epoch": 2.85,
|
|
"learning_rate": 1.2693280081896853e-07,
|
|
"loss": 0.5282,
|
|
"step": 1093
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 1.2244915454334815e-07,
|
|
"loss": 0.5374,
|
|
"step": 1094
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 1.1804563873971442e-07,
|
|
"loss": 0.5371,
|
|
"step": 1095
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 1.1372228913106965e-07,
|
|
"loss": 0.5408,
|
|
"step": 1096
|
|
},
|
|
{
|
|
"epoch": 2.86,
|
|
"learning_rate": 1.0947914079008192e-07,
|
|
"loss": 0.5325,
|
|
"step": 1097
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 1.0531622813879427e-07,
|
|
"loss": 0.5341,
|
|
"step": 1098
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 1.0123358494834701e-07,
|
|
"loss": 0.5413,
|
|
"step": 1099
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 9.723124433870357e-08,
|
|
"loss": 0.5359,
|
|
"step": 1100
|
|
},
|
|
{
|
|
"epoch": 2.87,
|
|
"learning_rate": 9.330923877838182e-08,
|
|
"loss": 0.5376,
|
|
"step": 1101
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 8.946760008419208e-08,
|
|
"loss": 0.5361,
|
|
"step": 1102
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 8.570635942097727e-08,
|
|
"loss": 0.5333,
|
|
"step": 1103
|
|
},
|
|
{
|
|
"epoch": 2.88,
|
|
"learning_rate": 8.202554730136203e-08,
|
|
"loss": 0.5343,
|
|
"step": 1104
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 7.842519358550071e-08,
|
|
"loss": 0.5403,
|
|
"step": 1105
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 7.4905327480842e-08,
|
|
"loss": 0.5342,
|
|
"step": 1106
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 7.146597754188578e-08,
|
|
"loss": 0.5447,
|
|
"step": 1107
|
|
},
|
|
{
|
|
"epoch": 2.89,
|
|
"learning_rate": 6.81071716699544e-08,
|
|
"loss": 0.5352,
|
|
"step": 1108
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 6.482893711296623e-08,
|
|
"loss": 0.5325,
|
|
"step": 1109
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 6.163130046521582e-08,
|
|
"loss": 0.5371,
|
|
"step": 1110
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 5.851428766715406e-08,
|
|
"loss": 0.5393,
|
|
"step": 1111
|
|
},
|
|
{
|
|
"epoch": 2.9,
|
|
"learning_rate": 5.54779240051817e-08,
|
|
"loss": 0.5356,
|
|
"step": 1112
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 5.252223411144286e-08,
|
|
"loss": 0.5287,
|
|
"step": 1113
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 4.9647241963624023e-08,
|
|
"loss": 0.543,
|
|
"step": 1114
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 4.685297088476204e-08,
|
|
"loss": 0.5322,
|
|
"step": 1115
|
|
},
|
|
{
|
|
"epoch": 2.91,
|
|
"learning_rate": 4.4139443543053106e-08,
|
|
"loss": 0.5308,
|
|
"step": 1116
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 4.1506681951669624e-08,
|
|
"loss": 0.5344,
|
|
"step": 1117
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 3.8954707468581435e-08,
|
|
"loss": 0.5417,
|
|
"step": 1118
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 3.648354079638039e-08,
|
|
"loss": 0.5361,
|
|
"step": 1119
|
|
},
|
|
{
|
|
"epoch": 2.92,
|
|
"learning_rate": 3.40932019821183e-08,
|
|
"loss": 0.5364,
|
|
"step": 1120
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 3.17837104171359e-08,
|
|
"loss": 0.537,
|
|
"step": 1121
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 2.9555084836914162e-08,
|
|
"loss": 0.5383,
|
|
"step": 1122
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 2.7407343320916546e-08,
|
|
"loss": 0.5295,
|
|
"step": 1123
|
|
},
|
|
{
|
|
"epoch": 2.93,
|
|
"learning_rate": 2.534050329244253e-08,
|
|
"loss": 0.5365,
|
|
"step": 1124
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 2.3354581518491014e-08,
|
|
"loss": 0.5374,
|
|
"step": 1125
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 2.1449594109618223e-08,
|
|
"loss": 0.5376,
|
|
"step": 1126
|
|
},
|
|
{
|
|
"epoch": 2.94,
|
|
"learning_rate": 1.962555651981113e-08,
|
|
"loss": 0.5356,
|
|
"step": 1127
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 1.7882483546360907e-08,
|
|
"loss": 0.5333,
|
|
"step": 1128
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 1.6220389329741903e-08,
|
|
"loss": 0.5383,
|
|
"step": 1129
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 1.463928735349951e-08,
|
|
"loss": 0.5312,
|
|
"step": 1130
|
|
},
|
|
{
|
|
"epoch": 2.95,
|
|
"learning_rate": 1.3139190444136917e-08,
|
|
"loss": 0.5378,
|
|
"step": 1131
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 1.1720110771015203e-08,
|
|
"loss": 0.5457,
|
|
"step": 1132
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 1.038205984625007e-08,
|
|
"loss": 0.5366,
|
|
"step": 1133
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 9.125048524624147e-09,
|
|
"loss": 0.5347,
|
|
"step": 1134
|
|
},
|
|
{
|
|
"epoch": 2.96,
|
|
"learning_rate": 7.949087003493727e-09,
|
|
"loss": 0.5354,
|
|
"step": 1135
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 6.8541848227077215e-09,
|
|
"loss": 0.5244,
|
|
"step": 1136
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 5.840350864533273e-09,
|
|
"loss": 0.5415,
|
|
"step": 1137
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 4.9075933535791545e-09,
|
|
"loss": 0.5347,
|
|
"step": 1138
|
|
},
|
|
{
|
|
"epoch": 2.97,
|
|
"learning_rate": 4.055919856730262e-09,
|
|
"loss": 0.5393,
|
|
"step": 1139
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 3.285337283087664e-09,
|
|
"loss": 0.5388,
|
|
"step": 1140
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 2.595851883913092e-09,
|
|
"loss": 0.5315,
|
|
"step": 1141
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 1.9874692525745365e-09,
|
|
"loss": 0.5347,
|
|
"step": 1142
|
|
},
|
|
{
|
|
"epoch": 2.98,
|
|
"learning_rate": 1.4601943245062812e-09,
|
|
"loss": 0.538,
|
|
"step": 1143
|
|
},
|
|
{
|
|
"epoch": 2.99,
|
|
"learning_rate": 1.0140313771644927e-09,
|
|
"loss": 0.5327,
|
|
"step": 1144
|
|
},
|
|
{
|
|
"epoch": 2.99,
|
|
"learning_rate": 6.489840299916949e-10,
|
|
"loss": 0.5393,
|
|
"step": 1145
|
|
},
|
|
{
|
|
"epoch": 2.99,
|
|
"learning_rate": 3.6505524439345295e-10,
|
|
"loss": 0.5325,
|
|
"step": 1146
|
|
},
|
|
{
|
|
"epoch": 2.99,
|
|
"learning_rate": 1.6224732370728747e-10,
|
|
"loss": 0.531,
|
|
"step": 1147
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"learning_rate": 4.056191319046221e-11,
|
|
"loss": 0.5312,
|
|
"step": 1148
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"learning_rate": 0.0,
|
|
"loss": 0.5336,
|
|
"step": 1149
|
|
},
|
|
{
|
|
"epoch": 3.0,
|
|
"step": 1149,
|
|
"total_flos": 2909638496354304.0,
|
|
"train_loss": 0.6237030162097476,
|
|
"train_runtime": 269199.3228,
|
|
"train_samples_per_second": 1.092,
|
|
"train_steps_per_second": 0.004
|
|
}
|
|
],
|
|
"max_steps": 1149,
|
|
"num_train_epochs": 3,
|
|
"total_flos": 2909638496354304.0,
|
|
"trial_name": null,
|
|
"trial_params": null
|
|
}
|