746 lines
14 KiB
JSON
746 lines
14 KiB
JSON
|
|
{
|
||
|
|
"best_metric": null,
|
||
|
|
"best_model_checkpoint": null,
|
||
|
|
"epoch": 14.883720930232558,
|
||
|
|
"global_step": 120,
|
||
|
|
"is_hyper_param_search": false,
|
||
|
|
"is_local_process_zero": true,
|
||
|
|
"is_world_process_zero": true,
|
||
|
|
"log_history": [
|
||
|
|
{
|
||
|
|
"epoch": 0.12,
|
||
|
|
"learning_rate": 0.0,
|
||
|
|
"loss": 19.2168,
|
||
|
|
"step": 1
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.25,
|
||
|
|
"learning_rate": 5e-06,
|
||
|
|
"loss": 13.542,
|
||
|
|
"step": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.37,
|
||
|
|
"learning_rate": 7.924812503605782e-06,
|
||
|
|
"loss": 13.5635,
|
||
|
|
"step": 3
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.5,
|
||
|
|
"learning_rate": 1e-05,
|
||
|
|
"loss": 5.6401,
|
||
|
|
"step": 4
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.62,
|
||
|
|
"learning_rate": 1e-05,
|
||
|
|
"loss": 3.0406,
|
||
|
|
"step": 5
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.74,
|
||
|
|
"learning_rate": 9.913793103448277e-06,
|
||
|
|
"loss": 2.762,
|
||
|
|
"step": 6
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.87,
|
||
|
|
"learning_rate": 9.827586206896553e-06,
|
||
|
|
"loss": 2.2676,
|
||
|
|
"step": 7
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 0.99,
|
||
|
|
"learning_rate": 9.741379310344829e-06,
|
||
|
|
"loss": 2.1296,
|
||
|
|
"step": 8
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.12,
|
||
|
|
"learning_rate": 9.655172413793105e-06,
|
||
|
|
"loss": 2.0573,
|
||
|
|
"step": 9
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.24,
|
||
|
|
"learning_rate": 9.56896551724138e-06,
|
||
|
|
"loss": 1.8231,
|
||
|
|
"step": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.36,
|
||
|
|
"learning_rate": 9.482758620689655e-06,
|
||
|
|
"loss": 1.8406,
|
||
|
|
"step": 11
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.49,
|
||
|
|
"learning_rate": 9.396551724137931e-06,
|
||
|
|
"loss": 1.7724,
|
||
|
|
"step": 12
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.61,
|
||
|
|
"learning_rate": 9.310344827586207e-06,
|
||
|
|
"loss": 1.775,
|
||
|
|
"step": 13
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.74,
|
||
|
|
"learning_rate": 9.224137931034484e-06,
|
||
|
|
"loss": 1.812,
|
||
|
|
"step": 14
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.86,
|
||
|
|
"learning_rate": 9.13793103448276e-06,
|
||
|
|
"loss": 1.671,
|
||
|
|
"step": 15
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 1.98,
|
||
|
|
"learning_rate": 9.051724137931036e-06,
|
||
|
|
"loss": 1.7983,
|
||
|
|
"step": 16
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.11,
|
||
|
|
"learning_rate": 8.965517241379312e-06,
|
||
|
|
"loss": 1.7026,
|
||
|
|
"step": 17
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.23,
|
||
|
|
"learning_rate": 8.879310344827588e-06,
|
||
|
|
"loss": 1.7212,
|
||
|
|
"step": 18
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.36,
|
||
|
|
"learning_rate": 8.793103448275862e-06,
|
||
|
|
"loss": 1.6182,
|
||
|
|
"step": 19
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.48,
|
||
|
|
"learning_rate": 8.706896551724138e-06,
|
||
|
|
"loss": 1.5165,
|
||
|
|
"step": 20
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.6,
|
||
|
|
"learning_rate": 8.620689655172414e-06,
|
||
|
|
"loss": 1.6209,
|
||
|
|
"step": 21
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.73,
|
||
|
|
"learning_rate": 8.53448275862069e-06,
|
||
|
|
"loss": 1.6025,
|
||
|
|
"step": 22
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.85,
|
||
|
|
"learning_rate": 8.448275862068966e-06,
|
||
|
|
"loss": 1.5281,
|
||
|
|
"step": 23
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 2.98,
|
||
|
|
"learning_rate": 8.362068965517242e-06,
|
||
|
|
"loss": 1.5672,
|
||
|
|
"step": 24
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.1,
|
||
|
|
"learning_rate": 8.275862068965518e-06,
|
||
|
|
"loss": 1.4279,
|
||
|
|
"step": 25
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.22,
|
||
|
|
"learning_rate": 8.189655172413794e-06,
|
||
|
|
"loss": 1.4018,
|
||
|
|
"step": 26
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.35,
|
||
|
|
"learning_rate": 8.103448275862069e-06,
|
||
|
|
"loss": 1.3675,
|
||
|
|
"step": 27
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.47,
|
||
|
|
"learning_rate": 8.017241379310345e-06,
|
||
|
|
"loss": 1.3208,
|
||
|
|
"step": 28
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.6,
|
||
|
|
"learning_rate": 7.93103448275862e-06,
|
||
|
|
"loss": 1.3687,
|
||
|
|
"step": 29
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.72,
|
||
|
|
"learning_rate": 7.844827586206897e-06,
|
||
|
|
"loss": 1.4228,
|
||
|
|
"step": 30
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.84,
|
||
|
|
"learning_rate": 7.758620689655173e-06,
|
||
|
|
"loss": 1.3738,
|
||
|
|
"step": 31
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 3.97,
|
||
|
|
"learning_rate": 7.672413793103449e-06,
|
||
|
|
"loss": 1.4264,
|
||
|
|
"step": 32
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.09,
|
||
|
|
"learning_rate": 7.586206896551724e-06,
|
||
|
|
"loss": 1.1533,
|
||
|
|
"step": 33
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.22,
|
||
|
|
"learning_rate": 7.500000000000001e-06,
|
||
|
|
"loss": 1.2682,
|
||
|
|
"step": 34
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.34,
|
||
|
|
"learning_rate": 7.413793103448277e-06,
|
||
|
|
"loss": 1.1839,
|
||
|
|
"step": 35
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.47,
|
||
|
|
"learning_rate": 7.327586206896552e-06,
|
||
|
|
"loss": 1.172,
|
||
|
|
"step": 36
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.59,
|
||
|
|
"learning_rate": 7.241379310344828e-06,
|
||
|
|
"loss": 1.2039,
|
||
|
|
"step": 37
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.71,
|
||
|
|
"learning_rate": 7.155172413793104e-06,
|
||
|
|
"loss": 1.2039,
|
||
|
|
"step": 38
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.84,
|
||
|
|
"learning_rate": 7.0689655172413796e-06,
|
||
|
|
"loss": 1.1208,
|
||
|
|
"step": 39
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 4.96,
|
||
|
|
"learning_rate": 6.982758620689656e-06,
|
||
|
|
"loss": 1.1296,
|
||
|
|
"step": 40
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.09,
|
||
|
|
"learning_rate": 6.896551724137932e-06,
|
||
|
|
"loss": 1.0807,
|
||
|
|
"step": 41
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.21,
|
||
|
|
"learning_rate": 6.810344827586207e-06,
|
||
|
|
"loss": 1.0618,
|
||
|
|
"step": 42
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.33,
|
||
|
|
"learning_rate": 6.724137931034484e-06,
|
||
|
|
"loss": 0.9491,
|
||
|
|
"step": 43
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.46,
|
||
|
|
"learning_rate": 6.63793103448276e-06,
|
||
|
|
"loss": 0.9955,
|
||
|
|
"step": 44
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.58,
|
||
|
|
"learning_rate": 6.551724137931035e-06,
|
||
|
|
"loss": 0.9659,
|
||
|
|
"step": 45
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.71,
|
||
|
|
"learning_rate": 6.465517241379311e-06,
|
||
|
|
"loss": 0.9816,
|
||
|
|
"step": 46
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.83,
|
||
|
|
"learning_rate": 6.379310344827587e-06,
|
||
|
|
"loss": 0.9322,
|
||
|
|
"step": 47
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 5.95,
|
||
|
|
"learning_rate": 6.293103448275862e-06,
|
||
|
|
"loss": 0.93,
|
||
|
|
"step": 48
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.08,
|
||
|
|
"learning_rate": 6.206896551724138e-06,
|
||
|
|
"loss": 0.8545,
|
||
|
|
"step": 49
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.2,
|
||
|
|
"learning_rate": 6.1206896551724135e-06,
|
||
|
|
"loss": 0.8549,
|
||
|
|
"step": 50
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.33,
|
||
|
|
"learning_rate": 6.03448275862069e-06,
|
||
|
|
"loss": 0.7331,
|
||
|
|
"step": 51
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.45,
|
||
|
|
"learning_rate": 5.9482758620689665e-06,
|
||
|
|
"loss": 0.7704,
|
||
|
|
"step": 52
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.57,
|
||
|
|
"learning_rate": 5.862068965517242e-06,
|
||
|
|
"loss": 0.8404,
|
||
|
|
"step": 53
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.7,
|
||
|
|
"learning_rate": 5.775862068965518e-06,
|
||
|
|
"loss": 0.7831,
|
||
|
|
"step": 54
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.82,
|
||
|
|
"learning_rate": 5.689655172413794e-06,
|
||
|
|
"loss": 0.8036,
|
||
|
|
"step": 55
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 6.95,
|
||
|
|
"learning_rate": 5.603448275862069e-06,
|
||
|
|
"loss": 0.7164,
|
||
|
|
"step": 56
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.07,
|
||
|
|
"learning_rate": 5.517241379310345e-06,
|
||
|
|
"loss": 0.6884,
|
||
|
|
"step": 57
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.19,
|
||
|
|
"learning_rate": 5.431034482758621e-06,
|
||
|
|
"loss": 0.6529,
|
||
|
|
"step": 58
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.32,
|
||
|
|
"learning_rate": 5.344827586206896e-06,
|
||
|
|
"loss": 0.6389,
|
||
|
|
"step": 59
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.44,
|
||
|
|
"learning_rate": 5.258620689655173e-06,
|
||
|
|
"loss": 0.6364,
|
||
|
|
"step": 60
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.57,
|
||
|
|
"learning_rate": 5.172413793103449e-06,
|
||
|
|
"loss": 0.68,
|
||
|
|
"step": 61
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.69,
|
||
|
|
"learning_rate": 5.086206896551724e-06,
|
||
|
|
"loss": 0.6098,
|
||
|
|
"step": 62
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.81,
|
||
|
|
"learning_rate": 5e-06,
|
||
|
|
"loss": 0.5625,
|
||
|
|
"step": 63
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 7.94,
|
||
|
|
"learning_rate": 4.9137931034482765e-06,
|
||
|
|
"loss": 0.5627,
|
||
|
|
"step": 64
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.06,
|
||
|
|
"learning_rate": 4.8275862068965525e-06,
|
||
|
|
"loss": 0.5142,
|
||
|
|
"step": 65
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.19,
|
||
|
|
"learning_rate": 4.741379310344828e-06,
|
||
|
|
"loss": 0.5043,
|
||
|
|
"step": 66
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.31,
|
||
|
|
"learning_rate": 4.655172413793104e-06,
|
||
|
|
"loss": 0.5281,
|
||
|
|
"step": 67
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.43,
|
||
|
|
"learning_rate": 4.56896551724138e-06,
|
||
|
|
"loss": 0.4843,
|
||
|
|
"step": 68
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.56,
|
||
|
|
"learning_rate": 4.482758620689656e-06,
|
||
|
|
"loss": 0.4788,
|
||
|
|
"step": 69
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.68,
|
||
|
|
"learning_rate": 4.396551724137931e-06,
|
||
|
|
"loss": 0.4595,
|
||
|
|
"step": 70
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.81,
|
||
|
|
"learning_rate": 4.310344827586207e-06,
|
||
|
|
"loss": 0.3978,
|
||
|
|
"step": 71
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 8.93,
|
||
|
|
"learning_rate": 4.224137931034483e-06,
|
||
|
|
"loss": 0.4884,
|
||
|
|
"step": 72
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.05,
|
||
|
|
"learning_rate": 4.137931034482759e-06,
|
||
|
|
"loss": 0.466,
|
||
|
|
"step": 73
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.18,
|
||
|
|
"learning_rate": 4.051724137931034e-06,
|
||
|
|
"loss": 0.4035,
|
||
|
|
"step": 74
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.3,
|
||
|
|
"learning_rate": 3.96551724137931e-06,
|
||
|
|
"loss": 0.3379,
|
||
|
|
"step": 75
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.43,
|
||
|
|
"learning_rate": 3.8793103448275865e-06,
|
||
|
|
"loss": 0.362,
|
||
|
|
"step": 76
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.55,
|
||
|
|
"learning_rate": 3.793103448275862e-06,
|
||
|
|
"loss": 0.3672,
|
||
|
|
"step": 77
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.67,
|
||
|
|
"learning_rate": 3.7068965517241385e-06,
|
||
|
|
"loss": 0.3348,
|
||
|
|
"step": 78
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.8,
|
||
|
|
"learning_rate": 3.620689655172414e-06,
|
||
|
|
"loss": 0.3313,
|
||
|
|
"step": 79
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 9.92,
|
||
|
|
"learning_rate": 3.5344827586206898e-06,
|
||
|
|
"loss": 0.3441,
|
||
|
|
"step": 80
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.05,
|
||
|
|
"learning_rate": 3.448275862068966e-06,
|
||
|
|
"loss": 0.3408,
|
||
|
|
"step": 81
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.17,
|
||
|
|
"learning_rate": 3.362068965517242e-06,
|
||
|
|
"loss": 0.2847,
|
||
|
|
"step": 82
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.29,
|
||
|
|
"learning_rate": 3.2758620689655175e-06,
|
||
|
|
"loss": 0.2954,
|
||
|
|
"step": 83
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.42,
|
||
|
|
"learning_rate": 3.1896551724137935e-06,
|
||
|
|
"loss": 0.2464,
|
||
|
|
"step": 84
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.54,
|
||
|
|
"learning_rate": 3.103448275862069e-06,
|
||
|
|
"loss": 0.3356,
|
||
|
|
"step": 85
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.67,
|
||
|
|
"learning_rate": 3.017241379310345e-06,
|
||
|
|
"loss": 0.2178,
|
||
|
|
"step": 86
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.79,
|
||
|
|
"learning_rate": 2.931034482758621e-06,
|
||
|
|
"loss": 0.2901,
|
||
|
|
"step": 87
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 10.91,
|
||
|
|
"learning_rate": 2.844827586206897e-06,
|
||
|
|
"loss": 0.2761,
|
||
|
|
"step": 88
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.04,
|
||
|
|
"learning_rate": 2.7586206896551725e-06,
|
||
|
|
"loss": 0.2495,
|
||
|
|
"step": 89
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.16,
|
||
|
|
"learning_rate": 2.672413793103448e-06,
|
||
|
|
"loss": 0.2727,
|
||
|
|
"step": 90
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.29,
|
||
|
|
"learning_rate": 2.5862068965517246e-06,
|
||
|
|
"loss": 0.264,
|
||
|
|
"step": 91
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.41,
|
||
|
|
"learning_rate": 2.5e-06,
|
||
|
|
"loss": 0.2269,
|
||
|
|
"step": 92
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.53,
|
||
|
|
"learning_rate": 2.4137931034482762e-06,
|
||
|
|
"loss": 0.2134,
|
||
|
|
"step": 93
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.66,
|
||
|
|
"learning_rate": 2.327586206896552e-06,
|
||
|
|
"loss": 0.195,
|
||
|
|
"step": 94
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.78,
|
||
|
|
"learning_rate": 2.241379310344828e-06,
|
||
|
|
"loss": 0.1857,
|
||
|
|
"step": 95
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 11.91,
|
||
|
|
"learning_rate": 2.1551724137931035e-06,
|
||
|
|
"loss": 0.1687,
|
||
|
|
"step": 96
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.03,
|
||
|
|
"learning_rate": 2.0689655172413796e-06,
|
||
|
|
"loss": 0.1641,
|
||
|
|
"step": 97
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.16,
|
||
|
|
"learning_rate": 1.982758620689655e-06,
|
||
|
|
"loss": 0.1771,
|
||
|
|
"step": 98
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.28,
|
||
|
|
"learning_rate": 1.896551724137931e-06,
|
||
|
|
"loss": 0.1789,
|
||
|
|
"step": 99
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.4,
|
||
|
|
"learning_rate": 1.810344827586207e-06,
|
||
|
|
"loss": 0.1754,
|
||
|
|
"step": 100
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.53,
|
||
|
|
"learning_rate": 1.724137931034483e-06,
|
||
|
|
"loss": 0.1745,
|
||
|
|
"step": 101
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.65,
|
||
|
|
"learning_rate": 1.6379310344827587e-06,
|
||
|
|
"loss": 0.1543,
|
||
|
|
"step": 102
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.78,
|
||
|
|
"learning_rate": 1.5517241379310346e-06,
|
||
|
|
"loss": 0.1741,
|
||
|
|
"step": 103
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 12.9,
|
||
|
|
"learning_rate": 1.4655172413793104e-06,
|
||
|
|
"loss": 0.1693,
|
||
|
|
"step": 104
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.02,
|
||
|
|
"learning_rate": 1.3793103448275862e-06,
|
||
|
|
"loss": 0.1611,
|
||
|
|
"step": 105
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.15,
|
||
|
|
"learning_rate": 1.2931034482758623e-06,
|
||
|
|
"loss": 0.146,
|
||
|
|
"step": 106
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.27,
|
||
|
|
"learning_rate": 1.2068965517241381e-06,
|
||
|
|
"loss": 0.1475,
|
||
|
|
"step": 107
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.4,
|
||
|
|
"learning_rate": 1.120689655172414e-06,
|
||
|
|
"loss": 0.1351,
|
||
|
|
"step": 108
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.52,
|
||
|
|
"learning_rate": 1.0344827586206898e-06,
|
||
|
|
"loss": 0.1364,
|
||
|
|
"step": 109
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.64,
|
||
|
|
"learning_rate": 9.482758620689655e-07,
|
||
|
|
"loss": 0.1486,
|
||
|
|
"step": 110
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.77,
|
||
|
|
"learning_rate": 8.620689655172415e-07,
|
||
|
|
"loss": 0.1377,
|
||
|
|
"step": 111
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 13.89,
|
||
|
|
"learning_rate": 7.758620689655173e-07,
|
||
|
|
"loss": 0.1509,
|
||
|
|
"step": 112
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.02,
|
||
|
|
"learning_rate": 6.896551724137931e-07,
|
||
|
|
"loss": 0.1227,
|
||
|
|
"step": 113
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.14,
|
||
|
|
"learning_rate": 6.034482758620691e-07,
|
||
|
|
"loss": 0.1473,
|
||
|
|
"step": 114
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.26,
|
||
|
|
"learning_rate": 5.172413793103449e-07,
|
||
|
|
"loss": 0.1117,
|
||
|
|
"step": 115
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.39,
|
||
|
|
"learning_rate": 4.3103448275862073e-07,
|
||
|
|
"loss": 0.1206,
|
||
|
|
"step": 116
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.51,
|
||
|
|
"learning_rate": 3.4482758620689656e-07,
|
||
|
|
"loss": 0.1311,
|
||
|
|
"step": 117
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.64,
|
||
|
|
"learning_rate": 2.5862068965517245e-07,
|
||
|
|
"loss": 0.104,
|
||
|
|
"step": 118
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.76,
|
||
|
|
"learning_rate": 1.7241379310344828e-07,
|
||
|
|
"loss": 0.1136,
|
||
|
|
"step": 119
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.88,
|
||
|
|
"learning_rate": 8.620689655172414e-08,
|
||
|
|
"loss": 0.1168,
|
||
|
|
"step": 120
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"epoch": 14.88,
|
||
|
|
"step": 120,
|
||
|
|
"total_flos": 17854376017920.0,
|
||
|
|
"train_loss": 1.198190559943517,
|
||
|
|
"train_runtime": 4673.7359,
|
||
|
|
"train_samples_per_second": 3.306,
|
||
|
|
"train_steps_per_second": 0.026
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"max_steps": 120,
|
||
|
|
"num_train_epochs": 15,
|
||
|
|
"total_flos": 17854376017920.0,
|
||
|
|
"trial_name": null,
|
||
|
|
"trial_params": null
|
||
|
|
}
|