WizardLM-30B-Uncensored / trainer_state.json
ehartford's picture
Upload folder using huggingface_hub
7617837
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9965075669383,
"global_step": 1287,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 2e-05,
"loss": 0.905,
"step": 2
},
{
"epoch": 0.01,
"learning_rate": 1.9999880457421163e-05,
"loss": 0.6497,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 1.9999521832542736e-05,
"loss": 0.6121,
"step": 6
},
{
"epoch": 0.02,
"learning_rate": 1.9998924133938902e-05,
"loss": 0.7236,
"step": 8
},
{
"epoch": 0.02,
"learning_rate": 1.9998087375899756e-05,
"loss": 0.7515,
"step": 10
},
{
"epoch": 0.03,
"learning_rate": 1.9997011578430938e-05,
"loss": 0.7073,
"step": 12
},
{
"epoch": 0.03,
"learning_rate": 1.9995696767253165e-05,
"loss": 0.6146,
"step": 14
},
{
"epoch": 0.04,
"learning_rate": 1.9994142973801627e-05,
"loss": 0.5923,
"step": 16
},
{
"epoch": 0.04,
"learning_rate": 1.9992350235225215e-05,
"loss": 0.5629,
"step": 18
},
{
"epoch": 0.05,
"learning_rate": 1.999031859438565e-05,
"loss": 0.5383,
"step": 20
},
{
"epoch": 0.05,
"learning_rate": 1.9988048099856443e-05,
"loss": 0.516,
"step": 22
},
{
"epoch": 0.06,
"learning_rate": 1.9985538805921757e-05,
"loss": 0.5035,
"step": 24
},
{
"epoch": 0.06,
"learning_rate": 1.998279077257508e-05,
"loss": 0.5244,
"step": 26
},
{
"epoch": 0.07,
"learning_rate": 1.9979804065517808e-05,
"loss": 0.486,
"step": 28
},
{
"epoch": 0.07,
"learning_rate": 1.9976578756157684e-05,
"loss": 0.4945,
"step": 30
},
{
"epoch": 0.07,
"learning_rate": 1.9973114921607055e-05,
"loss": 0.4966,
"step": 32
},
{
"epoch": 0.08,
"learning_rate": 1.9969412644681077e-05,
"loss": 0.4935,
"step": 34
},
{
"epoch": 0.08,
"learning_rate": 1.9965472013895685e-05,
"loss": 0.4739,
"step": 36
},
{
"epoch": 0.09,
"learning_rate": 1.996129312346552e-05,
"loss": 0.4913,
"step": 38
},
{
"epoch": 0.09,
"learning_rate": 1.9956876073301645e-05,
"loss": 0.4641,
"step": 40
},
{
"epoch": 0.1,
"learning_rate": 1.9952220969009175e-05,
"loss": 0.4691,
"step": 42
},
{
"epoch": 0.1,
"learning_rate": 1.9947327921884746e-05,
"loss": 0.4666,
"step": 44
},
{
"epoch": 0.11,
"learning_rate": 1.994219704891385e-05,
"loss": 0.4501,
"step": 46
},
{
"epoch": 0.11,
"learning_rate": 1.9936828472768043e-05,
"loss": 0.4558,
"step": 48
},
{
"epoch": 0.12,
"learning_rate": 1.9931222321802016e-05,
"loss": 0.4712,
"step": 50
},
{
"epoch": 0.12,
"learning_rate": 1.9925378730050518e-05,
"loss": 0.4661,
"step": 52
},
{
"epoch": 0.13,
"learning_rate": 1.9919297837225152e-05,
"loss": 0.4735,
"step": 54
},
{
"epoch": 0.13,
"learning_rate": 1.9912979788711042e-05,
"loss": 0.4526,
"step": 56
},
{
"epoch": 0.14,
"learning_rate": 1.990642473556335e-05,
"loss": 0.4453,
"step": 58
},
{
"epoch": 0.14,
"learning_rate": 1.9899632834503662e-05,
"loss": 0.4713,
"step": 60
},
{
"epoch": 0.14,
"learning_rate": 1.989260424791626e-05,
"loss": 0.4622,
"step": 62
},
{
"epoch": 0.15,
"learning_rate": 1.9885339143844217e-05,
"loss": 0.4585,
"step": 64
},
{
"epoch": 0.15,
"learning_rate": 1.987783769598538e-05,
"loss": 0.4576,
"step": 66
},
{
"epoch": 0.16,
"learning_rate": 1.9870100083688242e-05,
"loss": 0.4353,
"step": 68
},
{
"epoch": 0.16,
"learning_rate": 1.9862126491947624e-05,
"loss": 0.4509,
"step": 70
},
{
"epoch": 0.17,
"learning_rate": 1.985391711140027e-05,
"loss": 0.4402,
"step": 72
},
{
"epoch": 0.17,
"learning_rate": 1.9845472138320282e-05,
"loss": 0.437,
"step": 74
},
{
"epoch": 0.18,
"learning_rate": 1.9836791774614437e-05,
"loss": 0.4613,
"step": 76
},
{
"epoch": 0.18,
"learning_rate": 1.982787622781735e-05,
"loss": 0.4567,
"step": 78
},
{
"epoch": 0.19,
"learning_rate": 1.9818725711086506e-05,
"loss": 0.4541,
"step": 80
},
{
"epoch": 0.19,
"learning_rate": 1.980934044319718e-05,
"loss": 0.4398,
"step": 82
},
{
"epoch": 0.2,
"learning_rate": 1.9799720648537197e-05,
"loss": 0.4283,
"step": 84
},
{
"epoch": 0.2,
"learning_rate": 1.978986655710157e-05,
"loss": 0.4443,
"step": 86
},
{
"epoch": 0.2,
"learning_rate": 1.9779778404487e-05,
"loss": 0.4457,
"step": 88
},
{
"epoch": 0.21,
"learning_rate": 1.9769456431886244e-05,
"loss": 0.4326,
"step": 90
},
{
"epoch": 0.21,
"learning_rate": 1.9758900886082343e-05,
"loss": 0.4557,
"step": 92
},
{
"epoch": 0.22,
"learning_rate": 1.9748112019442734e-05,
"loss": 0.4402,
"step": 94
},
{
"epoch": 0.22,
"learning_rate": 1.9737090089913205e-05,
"loss": 0.465,
"step": 96
},
{
"epoch": 0.23,
"learning_rate": 1.9725835361011726e-05,
"loss": 0.4387,
"step": 98
},
{
"epoch": 0.23,
"learning_rate": 1.971434810182217e-05,
"loss": 0.4479,
"step": 100
},
{
"epoch": 0.24,
"learning_rate": 1.9702628586987846e-05,
"loss": 0.4344,
"step": 102
},
{
"epoch": 0.24,
"learning_rate": 1.9690677096704964e-05,
"loss": 0.4302,
"step": 104
},
{
"epoch": 0.25,
"learning_rate": 1.9678493916715914e-05,
"loss": 0.4331,
"step": 106
},
{
"epoch": 0.25,
"learning_rate": 1.966607933830245e-05,
"loss": 0.4224,
"step": 108
},
{
"epoch": 0.26,
"learning_rate": 1.9653433658278717e-05,
"loss": 0.4225,
"step": 110
},
{
"epoch": 0.26,
"learning_rate": 1.9640557178984152e-05,
"loss": 0.4177,
"step": 112
},
{
"epoch": 0.27,
"learning_rate": 1.9627450208276265e-05,
"loss": 0.4546,
"step": 114
},
{
"epoch": 0.27,
"learning_rate": 1.9614113059523273e-05,
"loss": 0.4257,
"step": 116
},
{
"epoch": 0.27,
"learning_rate": 1.9600546051596604e-05,
"loss": 0.4453,
"step": 118
},
{
"epoch": 0.28,
"learning_rate": 1.9586749508863284e-05,
"loss": 0.458,
"step": 120
},
{
"epoch": 0.28,
"learning_rate": 1.9572723761178168e-05,
"loss": 0.4287,
"step": 122
},
{
"epoch": 0.29,
"learning_rate": 1.955846914387607e-05,
"loss": 0.4581,
"step": 124
},
{
"epoch": 0.29,
"learning_rate": 1.954398599776373e-05,
"loss": 0.4343,
"step": 126
},
{
"epoch": 0.3,
"learning_rate": 1.952927466911168e-05,
"loss": 0.4431,
"step": 128
},
{
"epoch": 0.3,
"learning_rate": 1.9514335509645948e-05,
"loss": 0.4332,
"step": 130
},
{
"epoch": 0.31,
"learning_rate": 1.9499168876539666e-05,
"loss": 0.4315,
"step": 132
},
{
"epoch": 0.31,
"learning_rate": 1.9483775132404517e-05,
"loss": 0.4403,
"step": 134
},
{
"epoch": 0.32,
"learning_rate": 1.946815464528208e-05,
"loss": 0.4618,
"step": 136
},
{
"epoch": 0.32,
"learning_rate": 1.9452307788635015e-05,
"loss": 0.4292,
"step": 138
},
{
"epoch": 0.33,
"learning_rate": 1.9436234941338145e-05,
"loss": 0.4333,
"step": 140
},
{
"epoch": 0.33,
"learning_rate": 1.9419936487669396e-05,
"loss": 0.4557,
"step": 142
},
{
"epoch": 0.34,
"learning_rate": 1.94034128173006e-05,
"loss": 0.4575,
"step": 144
},
{
"epoch": 0.34,
"learning_rate": 1.938666432528819e-05,
"loss": 0.4012,
"step": 146
},
{
"epoch": 0.34,
"learning_rate": 1.9369691412063755e-05,
"loss": 0.4579,
"step": 148
},
{
"epoch": 0.35,
"learning_rate": 1.9352494483424456e-05,
"loss": 0.4337,
"step": 150
},
{
"epoch": 0.35,
"learning_rate": 1.9335073950523335e-05,
"loss": 0.4142,
"step": 152
},
{
"epoch": 0.36,
"learning_rate": 1.9317430229859474e-05,
"loss": 0.4545,
"step": 154
},
{
"epoch": 0.36,
"learning_rate": 1.929956374326805e-05,
"loss": 0.4679,
"step": 156
},
{
"epoch": 0.37,
"learning_rate": 1.928147491791024e-05,
"loss": 0.4178,
"step": 158
},
{
"epoch": 0.37,
"learning_rate": 1.9263164186263003e-05,
"loss": 0.4474,
"step": 160
},
{
"epoch": 0.38,
"learning_rate": 1.9244631986108768e-05,
"loss": 0.4237,
"step": 162
},
{
"epoch": 0.38,
"learning_rate": 1.922587876052492e-05,
"loss": 0.4456,
"step": 164
},
{
"epoch": 0.39,
"learning_rate": 1.920690495787326e-05,
"loss": 0.412,
"step": 166
},
{
"epoch": 0.39,
"learning_rate": 1.918771103178924e-05,
"loss": 0.4279,
"step": 168
},
{
"epoch": 0.4,
"learning_rate": 1.916829744117115e-05,
"loss": 0.413,
"step": 170
},
{
"epoch": 0.4,
"learning_rate": 1.9148664650169128e-05,
"loss": 0.4508,
"step": 172
},
{
"epoch": 0.41,
"learning_rate": 1.9128813128174063e-05,
"loss": 0.4054,
"step": 174
},
{
"epoch": 0.41,
"learning_rate": 1.9108743349806382e-05,
"loss": 0.4021,
"step": 176
},
{
"epoch": 0.41,
"learning_rate": 1.90884557949047e-05,
"loss": 0.4392,
"step": 178
},
{
"epoch": 0.42,
"learning_rate": 1.9067950948514343e-05,
"loss": 0.4414,
"step": 180
},
{
"epoch": 0.42,
"learning_rate&qu