agent-7b-lora-1ep-llava-1.5-7b / trainer_state.json
tennant's picture
Upload folder using huggingface_hub
98a00ef verified
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 602,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.0526315789473685e-06,
"loss": 1.307,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 2.105263157894737e-06,
"loss": 1.2963,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 3.157894736842105e-06,
"loss": 1.3495,
"step": 3
},
{
"epoch": 0.01,
"learning_rate": 4.210526315789474e-06,
"loss": 1.2897,
"step": 4
},
{
"epoch": 0.01,
"learning_rate": 5.263157894736842e-06,
"loss": 1.2698,
"step": 5
},
{
"epoch": 0.01,
"learning_rate": 6.31578947368421e-06,
"loss": 1.3108,
"step": 6
},
{
"epoch": 0.01,
"learning_rate": 7.368421052631579e-06,
"loss": 1.2761,
"step": 7
},
{
"epoch": 0.01,
"learning_rate": 8.421052631578948e-06,
"loss": 1.2979,
"step": 8
},
{
"epoch": 0.01,
"learning_rate": 9.473684210526315e-06,
"loss": 1.2876,
"step": 9
},
{
"epoch": 0.02,
"learning_rate": 1.0526315789473684e-05,
"loss": 1.2871,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 1.1578947368421053e-05,
"loss": 1.2509,
"step": 11
},
{
"epoch": 0.02,
"learning_rate": 1.263157894736842e-05,
"loss": 1.2231,
"step": 12
},
{
"epoch": 0.02,
"learning_rate": 1.3684210526315791e-05,
"loss": 1.207,
"step": 13
},
{
"epoch": 0.02,
"learning_rate": 1.4736842105263159e-05,
"loss": 1.2046,
"step": 14
},
{
"epoch": 0.02,
"learning_rate": 1.578947368421053e-05,
"loss": 1.2459,
"step": 15
},
{
"epoch": 0.03,
"learning_rate": 1.6842105263157896e-05,
"loss": 1.1277,
"step": 16
},
{
"epoch": 0.03,
"learning_rate": 1.7894736842105264e-05,
"loss": 1.1285,
"step": 17
},
{
"epoch": 0.03,
"learning_rate": 1.894736842105263e-05,
"loss": 1.1319,
"step": 18
},
{
"epoch": 0.03,
"learning_rate": 2e-05,
"loss": 1.2113,
"step": 19
},
{
"epoch": 0.03,
"learning_rate": 1.9999854811710317e-05,
"loss": 1.0988,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 1.999941925105719e-05,
"loss": 1.0399,
"step": 21
},
{
"epoch": 0.04,
"learning_rate": 1.9998693330688283e-05,
"loss": 1.0854,
"step": 22
},
{
"epoch": 0.04,
"learning_rate": 1.9997677071682623e-05,
"loss": 1.1107,
"step": 23
},
{
"epoch": 0.04,
"learning_rate": 1.999637050354999e-05,
"loss": 1.0481,
"step": 24
},
{
"epoch": 0.04,
"learning_rate": 1.9994773664230064e-05,
"loss": 1.1414,
"step": 25
},
{
"epoch": 0.04,
"learning_rate": 1.9992886600091318e-05,
"loss": 1.0073,
"step": 26
},
{
"epoch": 0.04,
"learning_rate": 1.9990709365929678e-05,
"loss": 1.0046,
"step": 27
},
{
"epoch": 0.05,
"learning_rate": 1.9988242024966924e-05,
"loss": 0.998,
"step": 28
},
{
"epoch": 0.05,
"learning_rate": 1.9985484648848854e-05,
"loss": 0.982,
"step": 29
},
{
"epoch": 0.05,
"learning_rate": 1.9982437317643218e-05,
"loss": 1.0077,
"step": 30
},
{
"epoch": 0.05,
"learning_rate": 1.997910011983737e-05,
"loss": 0.987,
"step": 31
},
{
"epoch": 0.05,
"learning_rate": 1.9975473152335726e-05,
"loss": 0.9571,
"step": 32
},
{
"epoch": 0.05,
"learning_rate": 1.9971556520456928e-05,
"loss": 0.9772,
"step": 33
},
{
"epoch": 0.06,
"learning_rate": 1.996735033793079e-05,
"loss": 1.0445,
"step": 34
},
{
"epoch": 0.06,
"learning_rate": 1.9962854726894997e-05,
"loss": 0.9987,
"step": 35
},
{
"epoch": 0.06,
"learning_rate": 1.995806981789157e-05,
"loss": 0.9634,
"step": 36
},
{
"epoch": 0.06,
"learning_rate": 1.995299574986306e-05,
"loss": 1.0259,
"step": 37
},
{
"epoch": 0.06,
"learning_rate": 1.9947632670148517e-05,
"loss": 0.9271,
"step": 38
},
{
"epoch": 0.06,
"learning_rate": 1.9941980734479214e-05,
"loss": 0.9988,
"step": 39
},
{
"epoch": 0.07,
"learning_rate": 1.9936040106974132e-05,
"loss": 0.9891,
"step": 40
},
{
"epoch": 0.07,
"learning_rate": 1.992981096013517e-05,
"loss": 0.9391,
"step": 41
},
{
"epoch": 0.07,
"learning_rate": 1.9923293474842175e-05,
"loss": 0.9417,
"step": 42
},
{
"epoch": 0.07,
"learning_rate": 1.9916487840347644e-05,
"loss": 0.904,
"step": 43
},
{
"epoch": 0.07,
"learning_rate": 1.990939425427127e-05,
"loss": 0.9252,
"step": 44
},
{
"epoch": 0.07,
"learning_rate": 1.9902012922594178e-05,
"loss": 0.9581,
"step": 45
},
{
"epoch": 0.08,
"learning_rate": 1.9894344059652953e-05,
"loss": 0.8464,
"step": 46
},
{
"epoch": 0.08,
"learning_rate": 1.9886387888133413e-05,
"loss": 0.9173,
"step": 47
},
{
"epoch": 0.08,
"learning_rate": 1.9878144639064145e-05,
"loss": 0.915,
"step": 48
},
{
"epoch": 0.08,
"learning_rate": 1.9869614551809793e-05,
"loss": 0.9444,
"step": 49
},
{
"epoch": 0.08,
"learning_rate": 1.9860797874064123e-05,
"loss": 0.9006,
"step": 50
},
{
"epoch": 0.08,
"learning_rate": 1.9851694861842795e-05,
"loss": 0.895,
"step": 51
},
{
"epoch": 0.09,
"learning_rate": 1.984230577947597e-05,
"loss": 0.8678,
"step": 52
},
{
"epoch": 0.09,
"learning_rate": 1.9832630899600607e-05,
"loss": 0.9083,
"step": 53
},
{
"epoch": 0.09,
"learning_rate": 1.9822670503152567e-05,
"loss": 0.9079,
"step": 54
},
{
"epoch": 0.09,
"learning_rate": 1.9812424879358424e-05,
"loss": 0.85,
"step": 55
},
{
"epoch": 0.09,
"learning_rate": 1.9801894325727104e-05,
"loss": 0.8878,
"step": 56
},
{
"epoch": 0.09,
"learning_rate": 1.979107914804122e-05,
"loss": 0.8536,
"step": 57
},
{
"epoch": 0.1,
"learning_rate": 1.97799796603482e-05,
"loss": 0.8637,
"step": 58
},
{
"epoch": 0.1,
"learning_rate": 1.9768596184951174e-05,
"loss": 0.8916,
"step": 59
},
{
"epoch": 0.1,
"learning_rate": 1.9756929052399606e-05,
"loss": 0.8217,
"step": 60
},
{
"epoch": 0.1,
"learning_rate": 1.9744978601479693e-05,
"loss": 0.8361,
"step": 61
},
{
"epoch": 0.1,
"learning_rate": 1.9732745179204553e-05,
"loss": 0.8724,
"step": 62
},
{
"epoch": 0.1,
"learning_rate": 1.972022914080411e-05,
"loss": 0.8808,
"step": 63
},
{
"epoch": 0.11,
"learning_rate": 1.970743084971481e-05,
"loss": 0.8577,
"step": 64
},
{
"epoch": 0.11,
"learning_rate": 1.9694350677569043e-05,
"loss": 0.8291,
"step": 65
},
{
"epoch": 0.11,
"learning_rate": 1.9680989004184383e-05,
"loss": 0.8321,
"step": 66
},
{
"epoch": 0.11,
"learning_rate": 1.9667346217552528e-05,
"loss": 0.8632,
"step": 67
},
{
"epoch": 0.11,
"learning_rate": 1.965342271382805e-05,
"loss": 0.857,
"step": 68
},
{
"epoch": 0.11,
"learning_rate": 1.9639218897316885e-05,
"loss": 0.8949,
"step": 69
},
{
"epoch": 0.12,
"learning_rate": 1.9624735180464602e-05,
"loss": 0.8642,
"step": 70
},
{
"epoch": 0.12,
"learning_rate": 1.9609971983844412e-05,
"loss": 0.7951,
"step": 71
},
{
"epoch": 0.12,
"learning_rate": 1.9594929736144978e-05,
"loss": 0.8023,
"step": 72
},
{
"epoch": 0.12,
"learning_rate": 1.957960887415793e-05,
"loss": 0.8151,
"step": 73
},
{
"epoch": 0.12,
"learning_rate": 1.9564009842765225e-05,
"loss": 0.7973,
"step": 74
},
{
"epoch": 0.12,
"learning_rate": 1.9548133094926203e-05,
"loss": 0.8013,
"step": 75
},
{
"epoch": 0.13,
"learning_rate": 1.953197909166443e-05,
"loss": 0.8468,
"step": 76
},
{
"epoch": 0.13,
"learning_rate": 1.9515548302054335e-05,
"loss": 0.7396,
"step": 77
},
{
"epoch": 0.13,
"learning_rate": 1.949884120320756e-05,
"loss": 0.7952,
"step": 78
},
{
"epoch": 0.13,
"learning_rate": 1.948185828025913e-05,
"loss": 0.7813,
"step": 79
},
{
"epoch": 0.13,
"learning_rate": 1.946460002635335e-05,
"loss": 0.8011,
"step": 80
},
{
"epoch": 0.13,
"learning_rate": 1.9447066942629495e-05,
"loss": 0.7541,
"step": 81
},
{
"epoch": 0.14,
"learning_rate": 1.942925953820725e-05,
"loss": 0.8306,
"step": 82
},
{
"epoch": 0.14,
"learning_rate": 1.941117833017194e-05,
"loss": 0.7468,
"step": 83
},
{
"epoch": 0.14,
"learning_rate": 1.939282384355949e-05,
"loss": 0.7923,
"step": 84
},
{
"epoch": 0.14,
"learning_rate": 1.9374196611341212e-05,
"loss": 0.7392,
"step": 85
},
{
"epoch": 0.14,
"learning_rate": 1.9355297174408298e-05,
"loss": 0.7098,
"step": 86
},
{
"epoch": 0.14,
"learning_rate": 1.9336126081556134e-05,
"loss": 0.8436,
"step": 87
},
{
"epoch": 0.15,
"learning_rate": 1.931668388946836e-05,
"loss": 0.7865,
"step": 88
},
{
"epoch": 0.15,
"learning_rate": 1.9296971162700696e-05,
"loss": 0.7492,
"step": 89
},
{
"epoch": 0.15,
"learning_rate": 1.9276988473664557e-05,
"loss": 0.8131,
"step": 90
},
{
"epoch": 0.15,
"learning_rate": 1.9256736402610437e-05,
"loss": 0.8514,
"step": 91
},
{
"epoch": 0.15,
"learning_rate": 1.9236215537611044e-05,
"loss": 0.7131,
"step": 92
},
{
"epoch": 0.15,
"learning_rate": 1.9215426474544242e-05,
"loss": 0.7999,
"step": 93
},
{
"epoch": 0.16,
"learning_rate": 1.9194369817075725e-05,
"loss": 0.807,
"step": 94
},
{
"epoch": 0.16,
"learning_rate": 1.9173046176641515e-05,
"loss": 0.7894,
"step": 95
},
{
"epoch": 0.16,
"learning_rate": 1.9151456172430186e-05,
"loss": 0.7604,
"step": 96
},
{
"epoch": 0.16,
"learning_rate": 1.91296004313649e-05,
"loss": 0.711,
"step": 97
},
{
"epoch": 0.16,
"learning_rate": 1.9107479588085182e-05,
"loss": 0.7407,
"step": 98
},
{
"epoch": 0.16,
"learning_rate": 1.908509428492852e-05,
"loss": 0.7198,
"step": 99
},
{
"epoch": 0.17,
"learning_rate": 1.9062445171911688e-05,
"loss": 0.7989,
"step": 100
},
{
"epoch": 0.17,
"learning_rate": 1.903953290671188e-05,
"loss": 0.6815,
"step": 101
},
{
"epoch": 0.17,
"learning_rate": 1.9016358154647618e-05,
"loss": 0.8462,
"step": 102
},
{
"epoch": 0.17,
"learning_rate": 1.8992921588659424e-05,
"loss": 0.7353,
"step": 103
},
{
"epoch": 0.17,
"learning_rate": 1.8969223889290283e-05,
"loss": 0.741,
"step": 104
},
{
"epoch": 0.17,
"learning_rate": 1.8945265744665886e-05,
"loss": 0.6829,
"step": 105
},
{
"epoch": 0.18,
"learning_rate": 1.8921047850474645e-05,
"loss": 0.6901,
"step": 106
},
{
"epoch": 0.18,
"learning_rate": 1.8896570909947477e-05,
"loss": 0.7278,
"step": 107
},
{
"epoch": 0.18,
"learning_rate": 1.887183563383741e-05,
"loss": 0.7912,
"step": 108
},
{
"epoch": 0.18,
"learning_rate": 1.884684274039894e-05,
"loss": 0.778,
"step": 109
},
{
"epoch": 0.18,
"learning_rate": 1.8821592955367154e-05,
"loss": 0.7267,
"step": 110
},
{
"epoch": 0.18,
"learning_rate": 1.8796087011936665e-05,
"loss": 0.7756,
"step": 111
},
{
"epoch": 0.19,
"learning_rate": 1.8770325650740347e-05,
"loss": 0.6625,
"step": 112
},
{
"epoch": 0.19,
"learning_rate": 1.874430961982778e-05,
"loss": 0.7652,
"step": 113
},
{
"epoch": 0.19,
"learning_rate": 1.871803967464358e-05,
"loss": 0.6545,
"step": 114
},
{
"epoch": 0.19,
"learning_rate": 1.8691516578005426e-05,
"loss": 0.7228,
"step": 115
},
{
"epoch": 0.19,
"learning_rate": 1.866474110008193e-05,
"loss": 0.7667,
"step": 116
},
{
"epoch": 0.19,
"learning_rate": 1.8637714018370255e-05,
"loss": 0.7087,
"step": 117
},
{
"epoch": 0.2,
"learning_rate": 1.8610436117673557e-05,
"loss": 0.7187,
"step": 118
},
{
"epoch": 0.2,
"learning_rate": 1.8582908190078184e-05,
"loss": 0.6764,
"step": 119
},
{
"epoch": 0.2,
"learning_rate": 1.8555131034930686e-05,
"loss": 0.736,
"step": 120
},
{
"epoch": 0.2,
"learning_rate": 1.852710545881459e-05,
"loss": 0.6797,
"step": 121
},
{
"epoch": 0.2,
"learning_rate": 1.8498832275526988e-05,
"loss": 0.7409,
"step": 122
},
{
"epoch": 0.2,
"learning_rate": 1.8470312306054903e-05,
"loss": 0.7194,
"step": 123
},
{
"epoch": 0.21,
"learning_rate": 1.8441546378551457e-05,
"loss": 0.7724,
"step": 124
},
{
"epoch": 0.21,
"learning_rate": 1.8412535328311813e-05,
"loss": 0.7001,
"step": 125
},
{
"epoch": 0.21,
"learning_rate": 1.838327999774892e-05,
"loss": 0.702,
"step": 126
},
{
"epoch": 0.21,
"learning_rate": 1.8353781236369065e-05,
"loss": 0.7145,
"step": 127
},
{
"epoch": 0.21,
"learning_rate": 1.832403990074719e-05,
"loss": 0.7009,
"step": 128
},
{
"epoch": 0.21,
"learning_rate": 1.829405685450202e-05,
"loss": 0.6825,
"step": 129
},
{
"epoch": 0.22,
"learning_rate": 1.8263832968271e-05,
"loss": 0.6862,
"step": 130
},
{
"epoch": 0.22,
"learning_rate": 1.8233369119685e-05,
"loss": 0.714,
"step": 131
},
{
"epoch": 0.22,
"learning_rate": 1.8202666193342834e-05,
"loss": 0.7047,
"step": 132
},
{
"epoch": 0.22,
"learning_rate": 1.817172508078557e-05,
"loss": 0.689,
"step": 133
},
{
"epoch": 0.22,
"learning_rate": 1.814054668047066e-05,
"loss": 0.6849,
"step": 134
},
{
"epoch": 0.22,
"learning_rate": 1.8109131897745823e-05,
"loss": 0.6572,
"step": 135
},
{
"epoch": 0.23,
"learning_rate": 1.807748164482277e-05,
"loss": 0.6737,
"step": 136
},
{
"epoch": 0.23,
"learning_rate": 1.8045596840750722e-05,
"loss": 0.6582,
"step": 137
},
{
"epoch": 0.23,
"learning_rate": 1.801347841138972e-05,
"loss": 0.6704,
"step": 138
},
{
"epoch": 0.23,
"learning_rate": 1.7981127289383718e-05,
"loss": 0.6941,
"step": 139
},
{
"epoch": 0.23,
"learning_rate": 1.7948544414133534e-05,
"loss": 0.6395,
"step": 140
},
{
"epoch": 0.23,
"learning_rate": 1.7915730731769558e-05,
"loss": 0.693,
"step": 141
},
{
"epoch": 0.24,
"learning_rate": 1.788268719512427e-05,
"loss": 0.6509,
"step": 142
},
{
"epoch": 0.24,
"learning_rate": 1.7849414763704587e-05,
"loss": 0.6709,
"step": 143
},
{
"epoch": 0.24,
"learning_rate": 1.781591440366399e-05,
"loss": 0.5932,
"step": 144
},
{
"epoch": 0.24,
"learning_rate": 1.778218708777448e-05,
"loss": 0.6661,
"step": 145
},
{
"epoch": 0.24,
"learning_rate": 1.7748233795398308e-05,
"loss": 0.68,
"step": 146
},
{
"epoch": 0.24,
"learning_rate": 1.771405551245957e-05,
"loss": 0.6303,
"step": 147
},
{
"epoch": 0.25,
"learning_rate": 1.767965323141555e-05,
"loss": 0.6538,
"step": 148
},
{
"epoch": 0.25,
"learning_rate": 1.764502795122793e-05,
"loss": 0.6347,
"step": 149
},
{
"epoch": 0.25,
"learning_rate": 1.761018067733374e-05,
"loss": 0.6402,
"step": 150
},
{
"epoch": 0.25,
"learning_rate": 1.7575112421616203e-05,
"loss": 0.671,
"step": 151
},
{
"epoch": 0.25,
"learning_rate": 1.753982420237533e-05,
"loss": 0.6941,
"step": 152
},
{
"epoch": 0.25,
"learning_rate": 1.750431704429837e-05,
"loss": 0.6461,
"step": 153
},
{
"epoch": 0.26,
"learning_rate": 1.7468591978430024e-05,
"loss": 0.5665,
"step": 154
},
{
"epoch": 0.26,
"learning_rate": 1.7432650042142535e-05,
"loss": 0.6381,
"step": 155
},
{
"epoch": 0.26,
"learning_rate": 1.7396492279105562e-05,
"loss": 0.5423,
"step": 156
},
{
"epoch": 0.26,
"learning_rate": 1.736011973925585e-05,
"loss": 0.6849,
"step": 157
},
{
"epoch": 0.26,
"learning_rate": 1.7323533478766777e-05,
"loss": 0.6038,
"step": 158
},
{
"epoch": 0.26,
"learning_rate": 1.728673456001766e-05,
"loss": 0.6124,
"step": 159
},
{
"epoch": 0.27,
"learning_rate": 1.7249724051562905e-05,
"loss": 0.6959,
"step": 160
},
{
"epoch": 0.27,
"learning_rate": 1.7212503028101012e-05,
"loss": 0.5771,
"step": 161
},
{
"epoch": 0.27,
"learning_rate": 1.717507257044331e-05,
"loss": 0.6593,
"step": 162
},
{
"epoch": 0.27,
"learning_rate": 1.7137433765482644e-05,
"loss": 0.6619,
"step": 163
},
{
"epoch": 0.27,
"learning_rate": 1.709958770616174e-05,
"loss": 0.5744,
"step": 164
},
{
"epoch": 0.27,
"learning_rate": 1.7061535491441538e-05,
"loss": 0.6,
"step": 165
},
{
"epoch": 0.28,
"learning_rate": 1.7023278226269222e-05,
"loss": 0.6637,
"step": 166
},
{
"epoch": 0.28,
"learning_rate": 1.6984817021546177e-05,
"loss": 0.6429,
"step": 167
},
{
"epoch": 0.28,
"learning_rate": 1.6946152994095705e-05,
"loss": 0.5642,
"step": 168
},
{
"epoch": 0.28,
"learning_rate": 1.6907287266630614e-05,
"loss": 0.5787,
"step": 169
},
{
"epoch": 0.28,
"learning_rate": 1.6868220967720604e-05,
"loss": 0.6205,
"step": 170
},
{
"epoch": 0.28,
"learning_rate": 1.6828955231759495e-05,
"loss": 0.6477,
"step": 171
},
{
"epoch": 0.29,
"learning_rate": 1.6789491198932302e-05,
"loss": 0.6445,
"step": 172
},
{
"epoch": 0.29,
"learning_rate": 1.6749830015182106e-05,
"loss": 0.6598,
"step": 173
},
{
"epoch": 0.29,
"learning_rate": 1.6709972832176797e-05,
"loss": 0.6433,
"step": 174
},
{
"epoch": 0.29,
"learning_rate": 1.6669920807275622e-05,
"loss": 0.5778,
"step": 175
},
{
"epoch": 0.29,
"learning_rate": 1.662967510349558e-05,
"loss": 0.5938,
"step": 176
},
{
"epoch": 0.29,
"learning_rate": 1.658923688947765e-05,
"loss": 0.5571,
"step": 177
},
{
"epoch": 0.3,
"learning_rate": 1.6548607339452853e-05,
"loss": 0.6399,
"step": 178
},
{
"epoch": 0.3,
"learning_rate": 1.6507787633208173e-05,
"loss": 0.5792,
"step": 179
},
{
"epoch": 0.3,
"learning_rate": 1.646677895605227e-05,
"loss": 0.6683,
"step": 180
},
{
"epoch": 0.3,
"learning_rate": 1.642558249878109e-05,
"loss": 0.6226,
"step": 181
},
{
"epoch": 0.3,
"learning_rate": 1.6384199457643264e-05,
"loss": 0.5441,
"step": 182
},
{
"epoch": 0.3,
"learning_rate": 1.6342631034305386e-05,
"loss": 0.5371,
"step": 183
},
{
"epoch": 0.31,
"learning_rate": 1.6300878435817115e-05,
"loss": 0.6009,
"step": 184
},
{
"epoch": 0.31,
"learning_rate": 1.6258942874576117e-05,
"loss": 0.5632,
"step": 185
},
{
"epoch": 0.31,
"learning_rate": 1.6216825568292885e-05,
"loss": 0.4853,
"step": 186
},
{
"epoch": 0.31,
"learning_rate": 1.6174527739955345e-05,
"loss": 0.5988,
"step": 187
},
{
"epoch": 0.31,
"learning_rate": 1.613205061779337e-05,
"loss": 0.5738,
"step": 188
},
{
"epoch": 0.31,
"learning_rate": 1.6089395435243105e-05,
"loss": 0.5961,
"step": 189
},
{
"epoch": 0.32,
"learning_rate": 1.6046563430911148e-05,
"loss": 0.559,
"step": 190
},
{
"epoch": 0.32,
"learning_rate": 1.6003555848538586e-05,
"loss": 0.616,
"step": 191
},
{
"epoch": 0.32,
"learning_rate": 1.596037393696489e-05,
"loss": 0.5079,
"step": 192
},
{
"epoch": 0.32,
"learning_rate": 1.5917018950091642e-05,
"loss": 0.6197,
"step": 193
},
{
"epoch": 0.32,
"learning_rate": 1.587349214684611e-05,
"loss": 0.5498,
"step": 194
},
{
"epoch": 0.32,
"learning_rate": 1.5829794791144723e-05,
"loss": 0.5775,
"step": 195
},
{
"epoch": 0.33,
"learning_rate": 1.5785928151856345e-05,
"loss": 0.5465,
"step": 196
},
{
"epoch": 0.33,
"learning_rate": 1.5741893502765452e-05,
"loss": 0.6061,
"step": 197
},
{
"epoch": 0.33,
"learning_rate": 1.569769212253511e-05,
"loss": 0.6082,
"step": 198
},
{
"epoch": 0.33,
"learning_rate": 1.5653325294669884e-05,
"loss": 0.5346,
"step": 199
},
{
"epoch": 0.33,
"learning_rate": 1.5608794307478546e-05,
"loss": 0.5796,
"step": 200
},
{
"epoch": 0.33,
"learning_rate": 1.556410045403667e-05,
"loss": 0.576,
"step": 201
},
{
"epoch": 0.34,
"learning_rate": 1.5519245032149083e-05,
"loss": 0.5138,
"step": 202
},
{
"epoch": 0.34,
"learning_rate": 1.547422934431218e-05,
"loss": 0.569,
"step": 203
},
{
"epoch": 0.34,
"learning_rate": 1.542905469767611e-05,
"loss": 0.5156,
"step": 204
},
{
"epoch": 0.34,
"learning_rate": 1.5383722404006808e-05,
"loss": 0.5306,
"step": 205
},
{
"epoch": 0.34,
"learning_rate": 1.533823377964791e-05,
"loss": 0.5287,
"step": 206
},
{
"epoch": 0.34,
"learning_rate": 1.529259014548253e-05,
"loss": 0.6334,
"step": 207
},
{
"epoch": 0.35,
"learning_rate": 1.5246792826894906e-05,
"loss": 0.5342,
"step": 208
},
{
"epoch": 0.35,
"learning_rate": 1.5200843153731905e-05,
"loss": 0.4981,
"step": 209
},
{
"epoch": 0.35,
"learning_rate": 1.5154742460264426e-05,
"loss": 0.5076,
"step": 210
},
{
"epoch": 0.35,
"learning_rate": 1.5108492085148632e-05,
"loss": 0.504,
"step": 211
},
{
"epoch": 0.35,
"learning_rate": 1.5062093371387097e-05,
"loss": 0.5895,
"step": 212
},
{
"epoch": 0.35,
"learning_rate": 1.5015547666289798e-05,
"loss": 0.585,
"step": 213
},
{
"epoch": 0.36,
"learning_rate": 1.4968856321434997e-05,
"loss": 0.4688,
"step": 214
},
{
"epoch": 0.36,
"learning_rate": 1.492202069263e-05,
"loss": 0.4596,
"step": 215
},
{
"epoch": 0.36,
"learning_rate": 1.4875042139871768e-05,
"loss": 0.5631,
"step": 216
},
{
"epoch": 0.36,
"learning_rate": 1.482792202730745e-05,
"loss": 0.5341,
"step": 217
},
{
"epoch": 0.36,
"learning_rate": 1.4780661723194757e-05,
"loss": 0.5539,
"step": 218
},
{
"epoch": 0.36,
"learning_rate": 1.4733262599862234e-05,
"loss": 0.4678,
"step": 219
},
{
"epoch": 0.37,
"learning_rate": 1.4685726033669412e-05,
"loss": 0.5109,
"step": 220
},
{
"epoch": 0.37,
"learning_rate": 1.4638053404966836e-05,
"loss": 0.4716,
"step": 221
},
{
"epoch": 0.37,
"learning_rate": 1.4590246098055995e-05,
"loss": 0.6074,
"step": 222
},
{
"epoch": 0.37,
"learning_rate": 1.454230550114911e-05,
"loss": 0.5218,
"step": 223
},
{
"epoch": 0.37,
"learning_rate": 1.4494233006328837e-05,
"loss": 0.4455,
"step": 224
},
{
"epoch": 0.37,
"learning_rate": 1.444603000950784e-05,
"loss": 0.5235,
"step": 225
},
{
"epoch": 0.38,
"learning_rate": 1.4397697910388248e-05,
"loss": 0.5394,
"step": 226
},
{
"epoch": 0.38,
"learning_rate": 1.4349238112421025e-05,
"loss": 0.4862,
"step": 227
},
{
"epoch": 0.38,
"learning_rate": 1.4300652022765207e-05,
"loss": 0.5915,
"step": 228
},
{
"epoch": 0.38,
"learning_rate": 1.4251941052247044e-05,
"loss": 0.5277,
"step": 229
},
{
"epoch": 0.38,
"learning_rate": 1.420310661531904e-05,
"loss": 0.5018,
"step": 230
},
{
"epoch": 0.38,
"learning_rate": 1.4154150130018867e-05,
"loss": 0.5244,
"step": 231
},
{
"epoch": 0.39,
"learning_rate": 1.4105073017928199e-05,
"loss": 0.5861,
"step": 232
},
{
"epoch": 0.39,
"learning_rate": 1.405587670413143e-05,
"loss": 0.4707,
"step": 233
},
{
"epoch": 0.39,
"learning_rate": 1.4006562617174292e-05,
"loss": 0.5542,
"step": 234
},
{
"epoch": 0.39,
"learning_rate": 1.3957132189022373e-05,
"loss": 0.5076,
"step": 235
},
{
"epoch": 0.39,
"learning_rate": 1.3907586855019538e-05,
"loss": 0.4285,
"step": 236
},
{
"epoch": 0.39,
"learning_rate": 1.385792805384625e-05,
"loss": 0.5224,
"step": 237
},
{
"epoch": 0.4,
"learning_rate": 1.3808157227477788e-05,
"loss": 0.4926,
"step": 238
},
{
"epoch": 0.4,
"learning_rate": 1.3758275821142382e-05,
"loss": 0.4525,
"step": 239
},
{
"epoch": 0.4,
"learning_rate": 1.3708285283279252e-05,
"loss": 0.5444,
"step": 240
},
{
"epoch": 0.4,
"learning_rate": 1.3658187065496533e-05,
"loss": 0.466,
"step": 241
},
{
"epoch": 0.4,
"learning_rate": 1.3607982622529135e-05,
"loss": 0.4632,
"step": 242
},
{
"epoch": 0.4,
"learning_rate": 1.3557673412196504e-05,
"loss": 0.4275,
"step": 243
},
{
"epoch": 0.41,
"learning_rate": 1.3507260895360274e-05,
"loss": 0.446,
"step": 244
},
{
"epoch": 0.41,
"learning_rate": 1.3456746535881872e-05,
"loss": 0.3904,
"step": 245
},
{
"epoch": 0.41,
"learning_rate": 1.3406131800579985e-05,
"loss": 0.4532,
"step": 246
},
{
"epoch": 0.41,
"learning_rate": 1.3355418159187988e-05,
"loss": 0.4585,
"step": 247
},
{
"epoch": 0.41,
"learning_rate": 1.3304607084311246e-05,
"loss": 0.4695,
"step": 248
},
{
"epoch": 0.41,
"learning_rate": 1.3253700051384371e-05,
"loss": 0.5278,
"step": 249
},
{
"epoch": 0.42,
"learning_rate": 1.3202698538628376e-05,
"loss": 0.5221,
"step": 250
},
{
"epoch": 0.42,
"learning_rate": 1.3151604027007744e-05,
"loss": 0.4634,
"step": 251
},
{
"epoch": 0.42,
"learning_rate": 1.310041800018742e-05,
"loss": 0.5301,
"step": 252
},
{
"epoch": 0.42,
"learning_rate": 1.304914194448975e-05,
"loss": 0.426,
"step": 253
},
{
"epoch": 0.42,
"learning_rate": 1.2997777348851288e-05,
"loss": 0.4774,
"step": 254
},
{
"epoch": 0.42,
"learning_rate": 1.2946325704779602e-05,
"loss": 0.4463,
"step": 255
},
{
"epoch": 0.43,
"learning_rate": 1.289478850630993e-05,
"loss": 0.4482,
"step": 256
},
{
"epoch": 0.43,
"learning_rate": 1.284316724996181e-05,
"loss": 0.4746,
"step": 257
},
{
"epoch": 0.43,
"learning_rate": 1.279146343469563e-05,
"loss": 0.485,
"step": 258
},
{
"epoch": 0.43,
"learning_rate": 1.273967856186909e-05,
"loss": 0.4232,
"step": 259
},
{
"epoch": 0.43,
"learning_rate": 1.2687814135193613e-05,
"loss": 0.4519,
"step": 260
},
{
"epoch": 0.43,
"learning_rate": 1.2635871660690677e-05,
"loss": 0.4322,
"step": 261
},
{
"epoch": 0.44,
"learning_rate": 1.2583852646648097e-05,
"loss": 0.5083,
"step": 262
},
{
"epoch": 0.44,
"learning_rate": 1.25317586035762e-05,
"loss": 0.5181,
"step": 263
},
{
"epoch": 0.44,
"learning_rate": 1.2479591044163997e-05,
"loss": 0.4014,
"step": 264
},
{
"epoch": 0.44,
"learning_rate": 1.2427351483235224e-05,
"loss": 0.4485,
"step": 265
},
{
"epoch": 0.44,
"learning_rate": 1.2375041437704394e-05,
"loss": 0.458,
"step": 266
},
{
"epoch": 0.44,
"learning_rate": 1.232266242653271e-05,
"loss": 0.4355,
"step": 267
},
{
"epoch": 0.45,
"learning_rate": 1.2270215970683977e-05,
"loss": 0.3918,
"step": 268
},
{
"epoch": 0.45,
"learning_rate": 1.2217703593080445e-05,
"loss": 0.4261,
"step": 269
},
{
"epoch": 0.45,
"learning_rate": 1.2165126818558572e-05,
"loss": 0.3832,
"step": 270
},
{
"epoch": 0.45,
"learning_rate": 1.2112487173824755e-05,
"loss": 0.4464,
"step": 271
},
{
"epoch": 0.45,
"learning_rate": 1.2059786187410984e-05,
"loss": 0.4783,
"step": 272
},
{
"epoch": 0.45,
"learning_rate": 1.2007025389630484e-05,
"loss": 0.4119,
"step": 273
},
{
"epoch": 0.46,
"learning_rate": 1.1954206312533246e-05,
"loss": 0.4009,
"step": 274
},
{
"epoch": 0.46,
"learning_rate": 1.1901330489861564e-05,
"loss": 0.4197,
"step": 275
},
{
"epoch": 0.46,
"learning_rate": 1.1848399457005496e-05,
"loss": 0.4936,
"step": 276
},
{
"epoch": 0.46,
"learning_rate": 1.1795414750958265e-05,
"loss": 0.4746,
"step": 277
},
{
"epoch": 0.46,
"learning_rate": 1.1742377910271638e-05,
"loss": 0.4559,
"step": 278
},
{
"epoch": 0.46,
"learning_rate": 1.1689290475011258e-05,
"loss": 0.4041,
"step": 279
},
{
"epoch": 0.47,
"learning_rate": 1.1636153986711906e-05,
"loss": 0.4285,
"step": 280
},
{
"epoch": 0.47,
"learning_rate": 1.1582969988332757e-05,
"loss": 0.4191,
"step": 281
},
{
"epoch": 0.47,
"learning_rate": 1.1529740024212566e-05,
"loss": 0.3849,
"step": 282
},
{
"epoch": 0.47,
"learning_rate": 1.1476465640024814e-05,
"loss": 0.4019,
"step": 283
},
{
"epoch": 0.47,
"learning_rate": 1.1423148382732854e-05,
"loss": 0.4183,
"step": 284
},
{
"epoch": 0.47,
"learning_rate": 1.136978980054496e-05,
"loss": 0.4263,
"step": 285
},
{
"epoch": 0.48,
"learning_rate": 1.1316391442869394e-05,
"loss": 0.3615,
"step": 286
},
{
"epoch": 0.48,
"learning_rate": 1.12629548602694e-05,
"loss": 0.3791,
"step": 287
},
{
"epoch": 0.48,
"learning_rate": 1.1209481604418182e-05,
"loss": 0.4104,
"step": 288
},
{
"epoch": 0.48,
"learning_rate": 1.1155973228053854e-05,
"loss": 0.4351,
"step": 289
},
{
"epoch": 0.48,
"learning_rate": 1.1102431284934345e-05,
"loss": 0.4507,
"step": 290
},
{
"epoch": 0.48,
"learning_rate": 1.1048857329792284e-05,
"loss": 0.4237,
"step": 291
},
{
"epoch": 0.49,
"learning_rate": 1.099525291828986e-05,
"loss": 0.3851,
"step": 292
},
{
"epoch": 0.49,
"learning_rate": 1.0941619606973633e-05,
"loss": 0.4125,
"step": 293
},
{
"epoch": 0.49,
"learning_rate": 1.0887958953229349e-05,
"loss": 0.4317,
"step": 294
},
{
"epoch": 0.49,
"learning_rate": 1.083427251523672e-05,
"loss": 0.3784,
"step": 295
},
{
"epoch": 0.49,
"learning_rate": 1.0780561851924168e-05,
"loss": 0.4291,
"step": 296
},
{
"epoch": 0.49,
"learning_rate": 1.0726828522923563e-05,
"loss": 0.3963,
"step": 297
},
{
"epoch": 0.5,
"learning_rate": 1.0673074088524926e-05,
"loss": 0.4186,
"step": 298
},
{
"epoch": 0.5,
"learning_rate": 1.0619300109631146e-05,
"loss": 0.4471,
"step": 299
},
{
"epoch": 0.5,
"learning_rate": 1.0565508147712618e-05,
"loss": 0.3552,
"step": 300
},
{
"epoch": 0.5,
"learning_rate": 1.0511699764761935e-05,
"loss": 0.3688,
"step": 301
},
{
"epoch": 0.5,
"learning_rate": 1.0457876523248518e-05,
"loss": 0.3857,
"step": 302
},
{
"epoch": 0.5,
"learning_rate": 1.0404039986073244e-05,
"loss": 0.3482,
"step": 303
},
{
"epoch": 0.5,
"learning_rate": 1.035019171652306e-05,
"loss": 0.3513,
"step": 304
},
{
"epoch": 0.51,
"learning_rate": 1.0296333278225599e-05,
"loss": 0.3954,
"step": 305
},
{
"epoch": 0.51,
"learning_rate": 1.024246623510377e-05,
"loss": 0.3342,
"step": 306
},
{
"epoch": 0.51,
"learning_rate": 1.0188592151330343e-05,
"loss": 0.3874,
"step": 307
},
{
"epoch": 0.51,
"learning_rate": 1.0134712591282539e-05,
"loss": 0.3634,
"step": 308
},
{
"epoch": 0.51,
"learning_rate": 1.0080829119496587e-05,
"loss": 0.2762,
"step": 309
},
{
"epoch": 0.51,
"learning_rate": 1.0026943300622313e-05,
"loss": 0.3945,
"step": 310
},
{
"epoch": 0.52,
"learning_rate": 9.973056699377692e-06,
"loss": 0.329,
"step": 311
},
{
"epoch": 0.52,
"learning_rate": 9.919170880503416e-06,
"loss": 0.388,
"step": 312
},
{
"epoch": 0.52,
"learning_rate": 9.865287408717464e-06,
"loss": 0.3875,
"step": 313
},
{
"epoch": 0.52,
"learning_rate": 9.811407848669657e-06,
"loss": 0.3916,
"step": 314
},
{
"epoch": 0.52,
"learning_rate": 9.757533764896235e-06,
"loss": 0.3841,
"step": 315
},
{
"epoch": 0.52,
"learning_rate": 9.703666721774403e-06,
"loss": 0.3509,
"step": 316
},
{
"epoch": 0.53,
"learning_rate": 9.64980828347694e-06,
"loss": 0.3714,
"step": 317
},
{
"epoch": 0.53,
"learning_rate": 9.595960013926761e-06,
"loss": 0.3686,
"step": 318
},
{
"epoch": 0.53,
"learning_rate": 9.542123476751484e-06,
"loss": 0.3704,
"step": 319
},
{
"epoch": 0.53,
"learning_rate": 9.488300235238067e-06,
"loss": 0.325,
"step": 320
},
{
"epoch": 0.53,
"learning_rate": 9.434491852287385e-06,
"loss": 0.3366,
"step": 321
},
{
"epoch": 0.53,
"learning_rate": 9.38069989036886e-06,
"loss": 0.3826,
"step": 322
},
{
"epoch": 0.54,
"learning_rate": 9.326925911475075e-06,
"loss": 0.466,
"step": 323
},
{
"epoch": 0.54,
"learning_rate": 9.27317147707644e-06,
"loss": 0.3556,
"step": 324
},
{
"epoch": 0.54,
"learning_rate": 9.219438148075834e-06,
"loss": 0.3765,
"step": 325
},
{
"epoch": 0.54,
"learning_rate": 9.165727484763283e-06,
"loss": 0.3389,
"step": 326
},
{
"epoch": 0.54,
"learning_rate": 9.112041046770653e-06,
"loss": 0.297,
"step": 327
},
{
"epoch": 0.54,
"learning_rate": 9.058380393026369e-06,
"loss": 0.3571,
"step": 328
},
{
"epoch": 0.55,
"learning_rate": 9.00474708171014e-06,
"loss": 0.3293,
"step": 329
},
{
"epoch": 0.55,
"learning_rate": 8.951142670207718e-06,
"loss": 0.3474,
"step": 330
},
{
"epoch": 0.55,
"learning_rate": 8.897568715065658e-06,
"loss": 0.3039,
"step": 331
},
{
"epoch": 0.55,
"learning_rate": 8.844026771946148e-06,
"loss": 0.2916,
"step": 332
},
{
"epoch": 0.55,
"learning_rate": 8.790518395581823e-06,
"loss": 0.3709,
"step": 333
},
{
"epoch": 0.55,
"learning_rate": 8.737045139730605e-06,
"loss": 0.4029,
"step": 334
},
{
"epoch": 0.56,
"learning_rate": 8.683608557130608e-06,
"loss": 0.3154,
"step": 335
},
{
"epoch": 0.56,
"learning_rate": 8.63021019945504e-06,
"loss": 0.3733,
"step": 336
},
{
"epoch": 0.56,
"learning_rate": 8.576851617267151e-06,
"loss": 0.3529,
"step": 337
},
{
"epoch": 0.56,
"learning_rate": 8.52353435997519e-06,
"loss": 0.3987,
"step": 338
},
{
"epoch": 0.56,
"learning_rate": 8.470259975787438e-06,
"loss": 0.3061,
"step": 339
},
{
"epoch": 0.56,
"learning_rate": 8.417030011667241e-06,
"loss": 0.4173,
"step": 340
},
{
"epoch": 0.57,
"learning_rate": 8.363846013288096e-06,
"loss": 0.3643,
"step": 341
},
{
"epoch": 0.57,
"learning_rate": 8.310709524988743e-06,
"loss": 0.3373,
"step": 342
},
{
"epoch": 0.57,
"learning_rate": 8.257622089728362e-06,
"loss": 0.3179,
"step": 343
},
{
"epoch": 0.57,
"learning_rate": 8.20458524904174e-06,
"loss": 0.3175,
"step": 344
},
{
"epoch": 0.57,
"learning_rate": 8.151600542994506e-06,
"loss": 0.2896,
"step": 345
},
{
"epoch": 0.57,
"learning_rate": 8.098669510138438e-06,
"loss": 0.4014,
"step": 346
},
{
"epoch": 0.58,
"learning_rate": 8.045793687466757e-06,
"loss": 0.2838,
"step": 347
},
{
"epoch": 0.58,
"learning_rate": 7.992974610369521e-06,
"loss": 0.2978,
"step": 348
},
{
"epoch": 0.58,
"learning_rate": 7.940213812589018e-06,
"loss": 0.3782,
"step": 349
},
{
"epoch": 0.58,
"learning_rate": 7.887512826175247e-06,
"loss": 0.303,
"step": 350
},
{
"epoch": 0.58,
"learning_rate": 7.834873181441426e-06,
"loss": 0.2952,
"step": 351
},
{
"epoch": 0.58,
"learning_rate": 7.782296406919557e-06,
"loss": 0.3432,
"step": 352
},
{
"epoch": 0.59,
"learning_rate": 7.729784029316025e-06,
"loss": 0.358,
"step": 353
},
{
"epoch": 0.59,
"learning_rate": 7.677337573467294e-06,
"loss": 0.245,
"step": 354
},
{
"epoch": 0.59,
"learning_rate": 7.624958562295607e-06,
"loss": 0.3722,
"step": 355
},
{
"epoch": 0.59,
"learning_rate": 7.572648516764778e-06,
"loss": 0.2774,
"step": 356
},
{
"epoch": 0.59,
"learning_rate": 7.5204089558360076e-06,
"loss": 0.2663,
"step": 357
},
{
"epoch": 0.59,
"learning_rate": 7.468241396423801e-06,
"loss": 0.3424,
"step": 358
},
{
"epoch": 0.6,
"learning_rate": 7.416147353351909e-06,
"loss": 0.3345,
"step": 359
},
{
"epoch": 0.6,
"learning_rate": 7.364128339309326e-06,
"loss": 0.3148,
"step": 360
},
{
"epoch": 0.6,
"learning_rate": 7.312185864806391e-06,
"loss": 0.23,
"step": 361
},
{
"epoch": 0.6,
"learning_rate": 7.260321438130913e-06,
"loss": 0.2941,
"step": 362
},
{
"epoch": 0.6,
"learning_rate": 7.208536565304374e-06,
"loss": 0.3063,
"step": 363
},
{
"epoch": 0.6,
"learning_rate": 7.156832750038192e-06,
"loss": 0.3427,
"step": 364
},
{
"epoch": 0.61,
"learning_rate": 7.105211493690073e-06,
"loss": 0.2931,
"step": 365
},
{
"epoch": 0.61,
"learning_rate": 7.053674295220399e-06,
"loss": 0.3494,
"step": 366
},
{
"epoch": 0.61,
"learning_rate": 7.002222651148714e-06,
"loss": 0.3027,
"step": 367
},
{
"epoch": 0.61,
"learning_rate": 6.950858055510254e-06,
"loss": 0.4092,
"step": 368
},
{
"epoch": 0.61,
"learning_rate": 6.89958199981258e-06,
"loss": 0.2926,
"step": 369
},
{
"epoch": 0.61,
"learning_rate": 6.848395972992261e-06,
"loss": 0.3025,
"step": 370
},
{
"epoch": 0.62,
"learning_rate": 6.797301461371626e-06,
"loss": 0.3664,
"step": 371
},
{
"epoch": 0.62,
"learning_rate": 6.7462999486156315e-06,
"loss": 0.2993,
"step": 372
},
{
"epoch": 0.62,
"learning_rate": 6.695392915688759e-06,
"loss": 0.2819,
"step": 373
},
{
"epoch": 0.62,
"learning_rate": 6.644581840812019e-06,
"loss": 0.2939,
"step": 374
},
{
"epoch": 0.62,
"learning_rate": 6.593868199420017e-06,
"loss": 0.3519,
"step": 375
},
{
"epoch": 0.62,
"learning_rate": 6.543253464118131e-06,
"loss": 0.2812,
"step": 376
},
{
"epoch": 0.63,
"learning_rate": 6.492739104639727e-06,
"loss": 0.3164,
"step": 377
},
{
"epoch": 0.63,
"learning_rate": 6.4423265878035015e-06,
"loss": 0.2927,
"step": 378
},
{
"epoch": 0.63,
"learning_rate": 6.392017377470867e-06,
"loss": 0.3023,
"step": 379
},
{
"epoch": 0.63,
"learning_rate": 6.341812934503469e-06,
"loss": 0.3367,
"step": 380
},
{
"epoch": 0.63,
"learning_rate": 6.2917147167207495e-06,
"loss": 0.3241,
"step": 381
},
{
"epoch": 0.63,
"learning_rate": 6.241724178857621e-06,
"loss": 0.3172,
"step": 382
},
{
"epoch": 0.64,
"learning_rate": 6.191842772522214e-06,
"loss": 0.2406,
"step": 383
},
{
"epoch": 0.64,
"learning_rate": 6.142071946153751e-06,
"loss": 0.2409,
"step": 384
},
{
"epoch": 0.64,
"learning_rate": 6.092413144980465e-06,
"loss": 0.2456,
"step": 385
},
{
"epoch": 0.64,
"learning_rate": 6.04286781097763e-06,
"loss": 0.2985,
"step": 386
},
{
"epoch": 0.64,
"learning_rate": 5.993437382825711e-06,
"loss": 0.2831,
"step": 387
},
{
"epoch": 0.64,
"learning_rate": 5.944123295868574e-06,
"loss": 0.295,
"step": 388
},
{
"epoch": 0.65,
"learning_rate": 5.894926982071805e-06,
"loss": 0.2861,
"step": 389
},
{
"epoch": 0.65,
"learning_rate": 5.845849869981137e-06,
"loss": 0.2892,
"step": 390
},
{
"epoch": 0.65,
"learning_rate": 5.796893384680964e-06,
"loss": 0.2481,
"step": 391
},
{
"epoch": 0.65,
"learning_rate": 5.748058947752955e-06,
"loss": 0.2612,
"step": 392
},
{
"epoch": 0.65,
"learning_rate": 5.699347977234799e-06,
"loss": 0.3099,
"step": 393
},
{
"epoch": 0.65,
"learning_rate": 5.650761887578977e-06,
"loss": 0.3021,
"step": 394
},
{
"epoch": 0.66,
"learning_rate": 5.602302089611755e-06,
"loss": 0.2706,
"step": 395
},
{
"epoch": 0.66,
"learning_rate": 5.553969990492164e-06,
"loss": 0.2866,
"step": 396
},
{
"epoch": 0.66,
"learning_rate": 5.5057669936711625e-06,
"loss": 0.3077,
"step": 397
},
{
"epoch": 0.66,
"learning_rate": 5.457694498850892e-06,
"loss": 0.3154,
"step": 398
},
{
"epoch": 0.66,
"learning_rate": 5.409753901944006e-06,
"loss": 0.221,
"step": 399
},
{
"epoch": 0.66,
"learning_rate": 5.361946595033165e-06,
"loss": 0.299,
"step": 400
},
{
"epoch": 0.67,
"learning_rate": 5.314273966330591e-06,
"loss": 0.3202,
"step": 401
},
{
"epoch": 0.67,
"learning_rate": 5.266737400137765e-06,
"loss": 0.2917,
"step": 402
},
{
"epoch": 0.67,
"learning_rate": 5.219338276805243e-06,
"loss": 0.1962,
"step": 403
},
{
"epoch": 0.67,
"learning_rate": 5.172077972692553e-06,
"loss": 0.2724,
"step": 404
},
{
"epoch": 0.67,
"learning_rate": 5.124957860128237e-06,
"loss": 0.294,
"step": 405
},
{
"epoch": 0.67,
"learning_rate": 5.077979307370004e-06,
"loss": 0.2373,
"step": 406
},
{
"epoch": 0.68,
"learning_rate": 5.031143678565005e-06,
"loss": 0.257,
"step": 407
},
{
"epoch": 0.68,
"learning_rate": 4.984452333710207e-06,
"loss": 0.2973,
"step": 408
},
{
"epoch": 0.68,
"learning_rate": 4.937906628612905e-06,
"loss": 0.2349,
"step": 409
},
{
"epoch": 0.68,
"learning_rate": 4.89150791485137e-06,
"loss": 0.2586,
"step": 410
},
{
"epoch": 0.68,
"learning_rate": 4.845257539735577e-06,
"loss": 0.2942,
"step": 411
},
{
"epoch": 0.68,
"learning_rate": 4.7991568462680945e-06,
"loss": 0.3138,
"step": 412
},
{
"epoch": 0.69,
"learning_rate": 4.7532071731050975e-06,
"loss": 0.1886,
"step": 413
},
{
"epoch": 0.69,
"learning_rate": 4.707409854517471e-06,
"loss": 0.2804,
"step": 414
},
{
"epoch": 0.69,
"learning_rate": 4.661766220352098e-06,
"loss": 0.3557,
"step": 415
},
{
"epoch": 0.69,
"learning_rate": 4.616277595993196e-06,
"loss": 0.2548,
"step": 416
},
{
"epoch": 0.69,
"learning_rate": 4.57094530232389e-06,
"loss": 0.218,
"step": 417
},
{
"epoch": 0.69,
"learning_rate": 4.525770655687821e-06,
"loss": 0.2302,
"step": 418
},
{
"epoch": 0.7,
"learning_rate": 4.48075496785092e-06,
"loss": 0.2652,
"step": 419
},
{
"epoch": 0.7,
"learning_rate": 4.435899545963333e-06,
"loss": 0.1879,
"step": 420
},
{
"epoch": 0.7,
"learning_rate": 4.391205692521453e-06,
"loss": 0.2553,
"step": 421
},
{
"epoch": 0.7,
"learning_rate": 4.346674705330117e-06,
"loss": 0.2334,
"step": 422
},
{
"epoch": 0.7,
"learning_rate": 4.302307877464893e-06,
"loss": 0.1971,
"step": 423
},
{
"epoch": 0.7,
"learning_rate": 4.258106497234551e-06,
"loss": 0.2759,
"step": 424
},
{
"epoch": 0.71,
"learning_rate": 4.214071848143655e-06,
"loss": 0.2923,
"step": 425
},
{
"epoch": 0.71,
"learning_rate": 4.170205208855281e-06,
"loss": 0.2144,
"step": 426
},
{
"epoch": 0.71,
"learning_rate": 4.126507853153891e-06,
"loss": 0.2532,
"step": 427
},
{
"epoch": 0.71,
"learning_rate": 4.082981049908362e-06,
"loss": 0.2631,
"step": 428
},
{
"epoch": 0.71,
"learning_rate": 4.039626063035107e-06,
"loss": 0.3083,
"step": 429
},
{
"epoch": 0.71,
"learning_rate": 3.996444151461417e-06,
"loss": 0.2683,
"step": 430
},
{
"epoch": 0.72,
"learning_rate": 3.953436569088856e-06,
"loss": 0.3167,
"step": 431
},
{
"epoch": 0.72,
"learning_rate": 3.9106045647569005e-06,
"loss": 0.2403,
"step": 432
},
{
"epoch": 0.72,
"learning_rate": 3.867949382206632e-06,
"loss": 0.2376,
"step": 433
},
{
"epoch": 0.72,
"learning_rate": 3.825472260044658e-06,
"loss": 0.2971,
"step": 434
},
{
"epoch": 0.72,
"learning_rate": 3.7831744317071194e-06,
"loss": 0.2577,
"step": 435
},
{
"epoch": 0.72,
"learning_rate": 3.7410571254238835e-06,
"loss": 0.2485,
"step": 436
},
{
"epoch": 0.73,
"learning_rate": 3.6991215641828903e-06,
"loss": 0.2357,
"step": 437
},
{
"epoch": 0.73,
"learning_rate": 3.6573689656946177e-06,
"loss": 0.2413,
"step": 438
},
{
"epoch": 0.73,
"learning_rate": 3.615800542356738e-06,
"loss": 0.2159,
"step": 439
},
{
"epoch": 0.73,
"learning_rate": 3.574417501218913e-06,
"loss": 0.2191,
"step": 440
},
{
"epoch": 0.73,
"learning_rate": 3.5332210439477334e-06,
"loss": 0.274,
"step": 441
},
{
"epoch": 0.73,
"learning_rate": 3.4922123667918305e-06,
"loss": 0.1954,
"step": 442
},
{
"epoch": 0.74,
"learning_rate": 3.4513926605471504e-06,
"loss": 0.21,
"step": 443
},
{
"epoch": 0.74,
"learning_rate": 3.4107631105223528e-06,
"loss": 0.2699,
"step": 444
},
{
"epoch": 0.74,
"learning_rate": 3.3703248965044253e-06,
"loss": 0.2294,
"step": 445
},
{
"epoch": 0.74,
"learning_rate": 3.330079192724379e-06,
"loss": 0.2539,
"step": 446
},
{
"epoch": 0.74,
"learning_rate": 3.2900271678232045e-06,
"loss": 0.2823,
"step": 447
},
{
"epoch": 0.74,
"learning_rate": 3.250169984817897e-06,
"loss": 0.2871,
"step": 448
},
{
"epoch": 0.75,
"learning_rate": 3.2105088010677e-06,
"loss": 0.2456,
"step": 449
},
{
"epoch": 0.75,
"learning_rate": 3.171044768240508e-06,
"loss": 0.3168,
"step": 450
},
{
"epoch": 0.75,
"learning_rate": 3.131779032279397e-06,
"loss": 0.2354,
"step": 451
},
{
"epoch": 0.75,
"learning_rate": 3.0927127333693872e-06,
"loss": 0.182,
"step": 452
},
{
"epoch": 0.75,
"learning_rate": 3.053847005904298e-06,
"loss": 0.2304,
"step": 453
},
{
"epoch": 0.75,
"learning_rate": 3.0151829784538257e-06,
"loss": 0.2312,
"step": 454
},
{
"epoch": 0.76,
"learning_rate": 2.9767217737307805e-06,
"loss": 0.2553,
"step": 455
},
{
"epoch": 0.76,
"learning_rate": 2.938464508558466e-06,
"loss": 0.245,
"step": 456
},
{
"epoch": 0.76,
"learning_rate": 2.9004122938382617e-06,
"loss": 0.1771,
"step": 457
},
{
"epoch": 0.76,
"learning_rate": 2.86256623451736e-06,
"loss": 0.2863,
"step": 458
},
{
"epoch": 0.76,
"learning_rate": 2.8249274295566863e-06,
"loss": 0.255,
"step": 459
},
{
"epoch": 0.76,
"learning_rate": 2.7874969718989943e-06,
"loss": 0.2148,
"step": 460
},
{
"epoch": 0.77,
"learning_rate": 2.7502759484370946e-06,
"loss": 0.2717,
"step": 461
},
{
"epoch": 0.77,
"learning_rate": 2.7132654399823444e-06,
"loss": 0.2368,
"step": 462
},
{
"epoch": 0.77,
"learning_rate": 2.676466521233225e-06,
"loss": 0.1757,
"step": 463
},
{
"epoch": 0.77,
"learning_rate": 2.639880260744151e-06,
"loss": 0.2064,
"step": 464
},
{
"epoch": 0.77,
"learning_rate": 2.6035077208944416e-06,
"loss": 0.2223,
"step": 465
},
{
"epoch": 0.77,
"learning_rate": 2.5673499578574644e-06,
"loss": 0.2975,
"step": 466
},
{
"epoch": 0.78,
"learning_rate": 2.5314080215699822e-06,
"loss": 0.2544,
"step": 467
},
{
"epoch": 0.78,
"learning_rate": 2.4956829557016336e-06,
"loss": 0.249,
"step": 468
},
{
"epoch": 0.78,
"learning_rate": 2.4601757976246685e-06,
"loss": 0.2561,
"step": 469
},
{
"epoch": 0.78,
"learning_rate": 2.424887578383799e-06,
"loss": 0.2409,
"step": 470
},
{
"epoch": 0.78,
"learning_rate": 2.389819322666264e-06,
"loss": 0.2547,
"step": 471
},
{
"epoch": 0.78,
"learning_rate": 2.354972048772074e-06,
"loss": 0.2415,
"step": 472
},
{
"epoch": 0.79,
"learning_rate": 2.320346768584449e-06,
"loss": 0.2592,
"step": 473
},
{
"epoch": 0.79,
"learning_rate": 2.2859444875404347e-06,
"loss": 0.2757,
"step": 474
},
{
"epoch": 0.79,
"learning_rate": 2.2517662046016975e-06,
"loss": 0.1666,
"step": 475
},
{
"epoch": 0.79,
"learning_rate": 2.2178129122255255e-06,
"loss": 0.1428,
"step": 476
},
{
"epoch": 0.79,
"learning_rate": 2.184085596336011e-06,
"loss": 0.204,
"step": 477
},
{
"epoch": 0.79,
"learning_rate": 2.150585236295415e-06,
"loss": 0.2266,
"step": 478
},
{
"epoch": 0.8,
"learning_rate": 2.1173128048757307e-06,
"loss": 0.2414,
"step": 479
},
{
"epoch": 0.8,
"learning_rate": 2.0842692682304442e-06,
"loss": 0.2231,
"step": 480
},
{
"epoch": 0.8,
"learning_rate": 2.0514555858664663e-06,
"loss": 0.2148,
"step": 481
},
{
"epoch": 0.8,
"learning_rate": 2.0188727106162874e-06,
"loss": 0.2468,
"step": 482
},
{
"epoch": 0.8,
"learning_rate": 1.986521588610285e-06,
"loss": 0.1853,
"step": 483
},
{
"epoch": 0.8,
"learning_rate": 1.9544031592492763e-06,
"loss": 0.2459,
"step": 484
},
{
"epoch": 0.81,
"learning_rate": 1.922518355177232e-06,
"loss": 0.239,
"step": 485
},
{
"epoch": 0.81,
"learning_rate": 1.890868102254182e-06,
"loss": 0.1963,
"step": 486
},
{
"epoch": 0.81,
"learning_rate": 1.859453319529343e-06,
"loss": 0.1815,
"step": 487
},
{
"epoch": 0.81,
"learning_rate": 1.82827491921443e-06,
"loss": 0.2342,
"step": 488
},
{
"epoch": 0.81,
"learning_rate": 1.797333806657171e-06,
"loss": 0.2527,
"step": 489
},
{
"epoch": 0.81,
"learning_rate": 1.7666308803150045e-06,
"loss": 0.2313,
"step": 490
},
{
"epoch": 0.82,
"learning_rate": 1.7361670317290014e-06,
"loss": 0.2199,
"step": 491
},
{
"epoch": 0.82,
"learning_rate": 1.7059431454979825e-06,
"loss": 0.2708,
"step": 492
},
{
"epoch": 0.82,
"learning_rate": 1.6759600992528147e-06,
"loss": 0.2499,
"step": 493
},
{
"epoch": 0.82,
"learning_rate": 1.6462187636309345e-06,
"loss": 0.1915,
"step": 494
},
{
"epoch": 0.82,
"learning_rate": 1.6167200022510799e-06,
"loss": 0.2494,
"step": 495
},
{
"epoch": 0.82,
"learning_rate": 1.587464671688187e-06,
"loss": 0.2037,
"step": 496
},
{
"epoch": 0.83,
"learning_rate": 1.5584536214485457e-06,
"loss": 0.1655,
"step": 497
},
{
"epoch": 0.83,
"learning_rate": 1.5296876939450978e-06,
"loss": 0.2213,
"step": 498
},
{
"epoch": 0.83,
"learning_rate": 1.501167724473016e-06,
"loss": 0.2324,
"step": 499
},
{
"epoch": 0.83,
"learning_rate": 1.4728945411854135e-06,
"loss": 0.2378,
"step": 500
},
{
"epoch": 0.83,
"learning_rate": 1.444868965069315e-06,
"loss": 0.1851,
"step": 501
},
{
"epoch": 0.83,
"learning_rate": 1.4170918099218166e-06,
"loss": 0.3088,
"step": 502
},
{
"epoch": 0.84,
"learning_rate": 1.3895638823264447e-06,
"loss": 0.24,
"step": 503
},
{
"epoch": 0.84,
"learning_rate": 1.3622859816297473e-06,
"loss": 0.2054,
"step": 504
},
{
"epoch": 0.84,
"learning_rate": 1.3352588999180726e-06,
"loss": 0.2595,
"step": 505
},
{
"epoch": 0.84,
"learning_rate": 1.3084834219945731e-06,
"loss": 0.2321,
"step": 506
},
{
"epoch": 0.84,
"learning_rate": 1.2819603253564206e-06,
"loss": 0.2112,
"step": 507
},
{
"epoch": 0.84,
"learning_rate": 1.255690380172222e-06,
"loss": 0.222,
"step": 508
},
{
"epoch": 0.85,
"learning_rate": 1.2296743492596587e-06,
"loss": 0.3065,
"step": 509
},
{
"epoch": 0.85,
"learning_rate": 1.203912988063335e-06,
"loss": 0.2594,
"step": 510
},
{
"epoch": 0.85,
"learning_rate": 1.1784070446328477e-06,
"loss": 0.1897,
"step": 511
},
{
"epoch": 0.85,
"learning_rate": 1.153157259601062e-06,
"loss": 0.1806,
"step": 512
},
{
"epoch": 0.85,
"learning_rate": 1.1281643661625896e-06,
"loss": 0.2801,
"step": 513
},
{
"epoch": 0.85,
"learning_rate": 1.1034290900525279e-06,
"loss": 0.2051,
"step": 514
},
{
"epoch": 0.86,
"learning_rate": 1.078952149525362e-06,
"loss": 0.232,
"step": 515
},
{
"epoch": 0.86,
"learning_rate": 1.0547342553341144e-06,
"loss": 0.1894,
"step": 516
},
{
"epoch": 0.86,
"learning_rate": 1.030776110709718e-06,
"loss": 0.2261,
"step": 517
},
{
"epoch": 0.86,
"learning_rate": 1.0070784113405763e-06,
"loss": 0.2517,
"step": 518
},
{
"epoch": 0.86,
"learning_rate": 9.836418453523833e-07,
"loss": 0.2008,
"step": 519
},
{
"epoch": 0.86,
"learning_rate": 9.604670932881211e-07,
"loss": 0.2296,
"step": 520
},
{
"epoch": 0.87,
"learning_rate": 9.375548280883129e-07,
"loss": 0.2176,
"step": 521
},
{
"epoch": 0.87,
"learning_rate": 9.149057150714802e-07,
"loss": 0.2145,
"step": 522
},
{
"epoch": 0.87,
"learning_rate": 8.925204119148189e-07,
"loss": 0.2103,
"step": 523
},
{
"epoch": 0.87,
"learning_rate": 8.703995686351041e-07,
"loss": 0.2677,
"step": 524
},
{
"epoch": 0.87,
"learning_rate": 8.485438275698154e-07,
"loss": 0.2143,
"step": 525
},
{
"epoch": 0.87,
"learning_rate": 8.269538233584884e-07,
"loss": 0.1758,
"step": 526
},
{
"epoch": 0.88,
"learning_rate": 8.056301829242785e-07,
"loss": 0.2168,
"step": 527
},
{
"epoch": 0.88,
"learning_rate": 7.845735254557608e-07,
"loss": 0.2606,
"step": 528
},
{
"epoch": 0.88,
"learning_rate": 7.637844623889557e-07,
"loss": 0.1909,
"step": 529
},
{
"epoch": 0.88,
"learning_rate": 7.432635973895652e-07,
"loss": 0.198,
"step": 530
},
{
"epoch": 0.88,
"learning_rate": 7.230115263354431e-07,
"loss": 0.1598,
"step": 531
},
{
"epoch": 0.88,
"learning_rate": 7.030288372993066e-07,
"loss": 0.1778,
"step": 532
},
{
"epoch": 0.89,
"learning_rate": 6.833161105316421e-07,
"loss": 0.1455,
"step": 533
},
{
"epoch": 0.89,
"learning_rate": 6.638739184438681e-07,
"loss": 0.2497,
"step": 534
},
{
"epoch": 0.89,
"learning_rate": 6.447028255917054e-07,
"loss": 0.2006,
"step": 535
},
{
"epoch": 0.89,
"learning_rate": 6.258033886587911e-07,
"loss": 0.2817,
"step": 536
},
{
"epoch": 0.89,
"learning_rate": 6.071761564405121e-07,
"loss": 0.1753,
"step": 537
},
{
"epoch": 0.89,
"learning_rate": 5.888216698280646e-07,
"loss": 0.2139,
"step": 538
},
{
"epoch": 0.9,
"learning_rate": 5.707404617927526e-07,
"loss": 0.229,
"step": 539
},
{
"epoch": 0.9,
"learning_rate": 5.529330573705083e-07,
"loss": 0.2406,
"step": 540
},
{
"epoch": 0.9,
"learning_rate": 5.353999736466531e-07,
"loss": 0.197,
"step": 541
},
{
"epoch": 0.9,
"learning_rate": 5.181417197408733e-07,
"loss": 0.2249,
"step": 542
},
{
"epoch": 0.9,
"learning_rate": 5.011587967924414e-07,
"loss": 0.2058,
"step": 543
},
{
"epoch": 0.9,
"learning_rate": 4.844516979456671e-07,
"loss": 0.3333,
"step": 544
},
{
"epoch": 0.91,
"learning_rate": 4.6802090833557136e-07,
"loss": 0.2527,
"step": 545
},
{
"epoch": 0.91,
"learning_rate": 4.5186690507379894e-07,
"loss": 0.2259,
"step": 546
},
{
"epoch": 0.91,
"learning_rate": 4.359901572347758e-07,
"loss": 0.2049,
"step": 547
},
{
"epoch": 0.91,
"learning_rate": 4.203911258420712e-07,
"loss": 0.2262,
"step": 548
},
{
"epoch": 0.91,
"learning_rate": 4.0507026385502747e-07,
"loss": 0.1843,
"step": 549
},
{
"epoch": 0.91,
"learning_rate": 3.9002801615558805e-07,
"loss": 0.2131,
"step": 550
},
{
"epoch": 0.92,
"learning_rate": 3.7526481953539915e-07,
"loss": 0.2956,
"step": 551
},
{
"epoch": 0.92,
"learning_rate": 3.607811026831176e-07,
"loss": 0.2117,
"step": 552
},
{
"epoch": 0.92,
"learning_rate": 3.4657728617195295e-07,
"loss": 0.217,
"step": 553
},
{
"epoch": 0.92,
"learning_rate": 3.32653782447474e-07,
"loss": 0.2173,
"step": 554
},
{
"epoch": 0.92,
"learning_rate": 3.1901099581561846e-07,
"loss": 0.1826,
"step": 555
},
{
"epoch": 0.92,
"learning_rate": 3.056493224309587e-07,
"loss": 0.306,
"step": 556
},
{
"epoch": 0.93,
"learning_rate": 2.9256915028519575e-07,
"loss": 0.1949,
"step": 557
},
{
"epoch": 0.93,
"learning_rate": 2.7977085919589253e-07,
"loss": 0.1905,
"step": 558
},
{
"epoch": 0.93,
"learning_rate": 2.672548207954495e-07,
"loss": 0.2251,
"step": 559
},
{
"epoch": 0.93,
"learning_rate": 2.550213985203076e-07,
"loss": 0.2097,
"step": 560
},
{
"epoch": 0.93,
"learning_rate": 2.430709476003978e-07,
"loss": 0.2342,
"step": 561
},
{
"epoch": 0.93,
"learning_rate": 2.3140381504882736e-07,
"loss": 0.198,
"step": 562
},
{
"epoch": 0.94,
"learning_rate": 2.200203396517997e-07,
"loss": 0.1558,
"step": 563
},
{
"epoch": 0.94,
"learning_rate": 2.0892085195878154e-07,
"loss": 0.2013,
"step": 564
},
{
"epoch": 0.94,
"learning_rate": 1.9810567427289596e-07,
"loss": 0.2878,
"step": 565
},
{
"epoch": 0.94,
"learning_rate": 1.8757512064157658e-07,
"loss": 0.2805,
"step": 566
},
{
"epoch": 0.94,
"learning_rate": 1.7732949684743593e-07,
"loss": 0.1875,
"step": 567
},
{
"epoch": 0.94,
"learning_rate": 1.6736910039939159e-07,
"loss": 0.2391,
"step": 568
},
{
"epoch": 0.95,
"learning_rate": 1.5769422052403172e-07,
"loss": 0.2417,
"step": 569
},
{
"epoch": 0.95,
"learning_rate": 1.483051381572076e-07,
"loss": 0.1948,
"step": 570
},
{
"epoch": 0.95,
"learning_rate": 1.3920212593588113e-07,
"loss": 0.2577,
"step": 571
},
{
"epoch": 0.95,
"learning_rate": 1.303854481902067e-07,
"loss": 0.1837,
"step": 572
},
{
"epoch": 0.95,
"learning_rate": 1.218553609358575e-07,
"loss": 0.2045,
"step": 573
},
{
"epoch": 0.95,
"learning_rate": 1.1361211186658893e-07,
"loss": 0.1795,
"step": 574
},
{
"epoch": 0.96,
"learning_rate": 1.0565594034704918e-07,
"loss": 0.2652,
"step": 575
},
{
"epoch": 0.96,
"learning_rate": 9.798707740582447e-08,
"loss": 0.1932,
"step": 576
},
{
"epoch": 0.96,
"learning_rate": 9.060574572873238e-08,
"loss": 0.1601,
"step": 577
},
{
"epoch": 0.96,
"learning_rate": 8.351215965235915e-08,
"loss": 0.1875,
"step": 578
},
{
"epoch": 0.96,
"learning_rate": 7.670652515782917e-08,
"loss": 0.2563,
"step": 579
},
{
"epoch": 0.96,
"learning_rate": 7.018903986483083e-08,
"loss": 0.2188,
"step": 580
},
{
"epoch": 0.97,
"learning_rate": 6.395989302587113e-08,
"loss": 0.2162,
"step": 581
},
{
"epoch": 0.97,
"learning_rate": 5.801926552078563e-08,
"loss": 0.1879,
"step": 582
},
{
"epoch": 0.97,
"learning_rate": 5.236732985148374e-08,
"loss": 0.2181,
"step": 583
},
{
"epoch": 0.97,
"learning_rate": 4.7004250136940547e-08,
"loss": 0.1975,
"step": 584
},
{
"epoch": 0.97,
"learning_rate": 4.1930182108430584e-08,
"loss": 0.1885,
"step": 585
},
{
"epoch": 0.97,
"learning_rate": 3.714527310500371e-08,
"loss": 0.1943,
"step": 586
},
{
"epoch": 0.98,
"learning_rate": 3.264966206921294e-08,
"loss": 0.1817,
"step": 587
},
{
"epoch": 0.98,
"learning_rate": 2.8443479543073248e-08,
"loss": 0.295,
"step": 588
},
{
"epoch": 0.98,
"learning_rate": 2.4526847664273488e-08,
"loss": 0.2196,
"step": 589
},
{
"epoch": 0.98,
"learning_rate": 2.0899880162630336e-08,
"loss": 0.2263,
"step": 590
},
{
"epoch": 0.98,
"learning_rate": 1.7562682356786488e-08,
"loss": 0.2453,
"step": 591
},
{
"epoch": 0.98,
"learning_rate": 1.451535115114866e-08,
"loss": 0.2673,
"step": 592
},
{
"epoch": 0.99,
"learning_rate": 1.175797503307874e-08,
"loss": 0.1863,
"step": 593
},
{
"epoch": 0.99,
"learning_rate": 9.290634070322491e-09,
"loss": 0.2215,
"step": 594
},
{
"epoch": 0.99,
"learning_rate": 7.113399908681429e-09,
"loss": 0.2139,
"step": 595
},
{
"epoch": 0.99,
"learning_rate": 5.226335769936697e-09,
"loss": 0.1891,
"step": 596
},
{
"epoch": 0.99,
"learning_rate": 3.6294964500116492e-09,
"loss": 0.1882,
"step": 597
},
{
"epoch": 0.99,
"learning_rate": 2.322928317378681e-09,
"loss": 0.1911,
"step": 598
},
{
"epoch": 1.0,
"learning_rate": 1.3066693117191886e-09,
"loss": 0.225,
"step": 599
},
{
"epoch": 1.0,
"learning_rate": 5.807489428111268e-10,
"loss": 0.1494,
"step": 600
},
{
"epoch": 1.0,
"learning_rate": 1.4518828968523857e-10,
"loss": 0.1802,
"step": 601
},
{
"epoch": 1.0,
"learning_rate": 0.0,
"loss": 0.2316,
"step": 602
},
{
"epoch": 1.0,
"step": 602,
"total_flos": 58586457047040.0,
"train_loss": 0.4762434925747868,
"train_runtime": 4263.4552,
"train_samples_per_second": 4.516,
"train_steps_per_second": 0.141
}
],
"logging_steps": 1.0,
"max_steps": 602,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 5000,
"total_flos": 58586457047040.0,
"train_batch_size": 16,
"trial_name": null,
"trial_params": null
}