Upload trainer_state.json
Browse files- trainer_state.json +562 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,562 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 7.0,
|
| 5 |
+
"global_step": 26299,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.27,
|
| 12 |
+
"learning_rate": 1.923951481044907e-05,
|
| 13 |
+
"loss": 0.0205,
|
| 14 |
+
"step": 1000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.27,
|
| 18 |
+
"eval_EG_f1": 0.9960332310455804,
|
| 19 |
+
"eval_ET_f1": 0.9988771614641814,
|
| 20 |
+
"eval_TE_f1": 0.9952833720146739,
|
| 21 |
+
"eval_loss": 0.0021067976485937834,
|
| 22 |
+
"eval_overall_accuracy": 0.9995180550845189,
|
| 23 |
+
"eval_overall_f1": 0.9967312922623949,
|
| 24 |
+
"eval_overall_precision": 0.9966566866267466,
|
| 25 |
+
"eval_overall_recall": 0.9968059090682237,
|
| 26 |
+
"eval_runtime": 16.2281,
|
| 27 |
+
"eval_samples_per_second": 411.569,
|
| 28 |
+
"eval_steps_per_second": 25.758,
|
| 29 |
+
"step": 1000
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.53,
|
| 33 |
+
"learning_rate": 1.8479029620898135e-05,
|
| 34 |
+
"loss": 0.0027,
|
| 35 |
+
"step": 2000
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.53,
|
| 39 |
+
"eval_EG_f1": 0.9955841628620612,
|
| 40 |
+
"eval_ET_f1": 0.9992513849378649,
|
| 41 |
+
"eval_TE_f1": 0.995433789954338,
|
| 42 |
+
"eval_loss": 0.0024818568490445614,
|
| 43 |
+
"eval_overall_accuracy": 0.9993931064027275,
|
| 44 |
+
"eval_overall_f1": 0.9967563251659264,
|
| 45 |
+
"eval_overall_precision": 0.9966568534504266,
|
| 46 |
+
"eval_overall_recall": 0.9968558167390328,
|
| 47 |
+
"eval_runtime": 16.4446,
|
| 48 |
+
"eval_samples_per_second": 406.152,
|
| 49 |
+
"eval_steps_per_second": 25.419,
|
| 50 |
+
"step": 2000
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.8,
|
| 54 |
+
"learning_rate": 1.77185444313472e-05,
|
| 55 |
+
"loss": 0.0018,
|
| 56 |
+
"step": 3000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.8,
|
| 60 |
+
"eval_EG_f1": 0.9987274496594056,
|
| 61 |
+
"eval_ET_f1": 0.999700553975146,
|
| 62 |
+
"eval_TE_f1": 0.9982783142450782,
|
| 63 |
+
"eval_loss": 0.0007141608512029052,
|
| 64 |
+
"eval_overall_accuracy": 0.9998393516948396,
|
| 65 |
+
"eval_overall_f1": 0.9989020860365306,
|
| 66 |
+
"eval_overall_precision": 0.9988522381356355,
|
| 67 |
+
"eval_overall_recall": 0.9989519389130109,
|
| 68 |
+
"eval_runtime": 16.3629,
|
| 69 |
+
"eval_samples_per_second": 408.181,
|
| 70 |
+
"eval_steps_per_second": 25.546,
|
| 71 |
+
"step": 3000
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 1.06,
|
| 75 |
+
"learning_rate": 1.695805924179627e-05,
|
| 76 |
+
"loss": 0.0015,
|
| 77 |
+
"step": 4000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.06,
|
| 81 |
+
"eval_EG_f1": 0.9991765850737331,
|
| 82 |
+
"eval_ET_f1": 0.9992513849378649,
|
| 83 |
+
"eval_TE_f1": 0.9983530468633028,
|
| 84 |
+
"eval_loss": 0.0010974227916449308,
|
| 85 |
+
"eval_overall_accuracy": 0.9997858022597862,
|
| 86 |
+
"eval_overall_f1": 0.9989270118527761,
|
| 87 |
+
"eval_overall_precision": 0.9989020860365306,
|
| 88 |
+
"eval_overall_recall": 0.9989519389130109,
|
| 89 |
+
"eval_runtime": 16.7722,
|
| 90 |
+
"eval_samples_per_second": 398.219,
|
| 91 |
+
"eval_steps_per_second": 24.922,
|
| 92 |
+
"step": 4000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.33,
|
| 96 |
+
"learning_rate": 1.6197574052245332e-05,
|
| 97 |
+
"loss": 0.0005,
|
| 98 |
+
"step": 5000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.33,
|
| 102 |
+
"eval_EG_f1": 0.9982030548068285,
|
| 103 |
+
"eval_ET_f1": 0.999850276987573,
|
| 104 |
+
"eval_TE_f1": 0.9979038778260219,
|
| 105 |
+
"eval_loss": 0.0007138435612432659,
|
| 106 |
+
"eval_overall_accuracy": 0.9998125769773129,
|
| 107 |
+
"eval_overall_f1": 0.998652425633859,
|
| 108 |
+
"eval_overall_precision": 0.998702271025705,
|
| 109 |
+
"eval_overall_recall": 0.9986025852173479,
|
| 110 |
+
"eval_runtime": 16.5358,
|
| 111 |
+
"eval_samples_per_second": 403.912,
|
| 112 |
+
"eval_steps_per_second": 25.279,
|
| 113 |
+
"step": 5000
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.6,
|
| 117 |
+
"learning_rate": 1.54370888626944e-05,
|
| 118 |
+
"loss": 0.0008,
|
| 119 |
+
"step": 6000
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.6,
|
| 123 |
+
"eval_EG_f1": 0.9983530468633028,
|
| 124 |
+
"eval_ET_f1": 0.999550830962719,
|
| 125 |
+
"eval_TE_f1": 0.9979038778260219,
|
| 126 |
+
"eval_loss": 0.0009068112121894956,
|
| 127 |
+
"eval_overall_accuracy": 0.9997947271656284,
|
| 128 |
+
"eval_overall_f1": 0.9986025852173479,
|
| 129 |
+
"eval_overall_precision": 0.9986025852173479,
|
| 130 |
+
"eval_overall_recall": 0.9986025852173479,
|
| 131 |
+
"eval_runtime": 16.3597,
|
| 132 |
+
"eval_samples_per_second": 408.259,
|
| 133 |
+
"eval_steps_per_second": 25.551,
|
| 134 |
+
"step": 6000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.86,
|
| 138 |
+
"learning_rate": 1.4676603673143466e-05,
|
| 139 |
+
"loss": 0.0009,
|
| 140 |
+
"step": 7000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.86,
|
| 144 |
+
"eval_EG_f1": 0.9989519389130109,
|
| 145 |
+
"eval_ET_f1": 0.999850276987573,
|
| 146 |
+
"eval_TE_f1": 0.9988022159005839,
|
| 147 |
+
"eval_loss": 0.0010577181819826365,
|
| 148 |
+
"eval_overall_accuracy": 0.9998482766006819,
|
| 149 |
+
"eval_overall_f1": 0.999201477267056,
|
| 150 |
+
"eval_overall_precision": 0.999201477267056,
|
| 151 |
+
"eval_overall_recall": 0.999201477267056,
|
| 152 |
+
"eval_runtime": 16.3447,
|
| 153 |
+
"eval_samples_per_second": 408.634,
|
| 154 |
+
"eval_steps_per_second": 25.574,
|
| 155 |
+
"step": 7000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 2.13,
|
| 159 |
+
"learning_rate": 1.3916118483592534e-05,
|
| 160 |
+
"loss": 0.0005,
|
| 161 |
+
"step": 8000
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 2.13,
|
| 165 |
+
"eval_EG_f1": 0.9991765850737331,
|
| 166 |
+
"eval_ET_f1": 0.999850276987573,
|
| 167 |
+
"eval_TE_f1": 0.9991765850737331,
|
| 168 |
+
"eval_loss": 0.0005780701176263392,
|
| 169 |
+
"eval_overall_accuracy": 0.9998839762240508,
|
| 170 |
+
"eval_overall_f1": 0.9994011378381076,
|
| 171 |
+
"eval_overall_precision": 0.9993512650331853,
|
| 172 |
+
"eval_overall_recall": 0.999451015621101,
|
| 173 |
+
"eval_runtime": 16.6712,
|
| 174 |
+
"eval_samples_per_second": 400.631,
|
| 175 |
+
"eval_steps_per_second": 25.073,
|
| 176 |
+
"step": 8000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.4,
|
| 180 |
+
"learning_rate": 1.31556332940416e-05,
|
| 181 |
+
"loss": 0.0002,
|
| 182 |
+
"step": 9000
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.4,
|
| 186 |
+
"eval_EG_f1": 0.9988771614641814,
|
| 187 |
+
"eval_ET_f1": 0.999850276987573,
|
| 188 |
+
"eval_TE_f1": 0.9985777378546299,
|
| 189 |
+
"eval_loss": 0.0005696099251508713,
|
| 190 |
+
"eval_overall_accuracy": 0.999892901129893,
|
| 191 |
+
"eval_overall_f1": 0.9991017067571614,
|
| 192 |
+
"eval_overall_precision": 0.9990518488946554,
|
| 193 |
+
"eval_overall_recall": 0.999151569596247,
|
| 194 |
+
"eval_runtime": 16.4448,
|
| 195 |
+
"eval_samples_per_second": 406.147,
|
| 196 |
+
"eval_steps_per_second": 25.418,
|
| 197 |
+
"step": 9000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 2.66,
|
| 201 |
+
"learning_rate": 1.2395148104490666e-05,
|
| 202 |
+
"loss": 0.0005,
|
| 203 |
+
"step": 10000
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 2.66,
|
| 207 |
+
"eval_EG_f1": 0.9989519389130109,
|
| 208 |
+
"eval_ET_f1": 0.999850276987573,
|
| 209 |
+
"eval_TE_f1": 0.9988771614641814,
|
| 210 |
+
"eval_loss": 0.0004689108463935554,
|
| 211 |
+
"eval_overall_accuracy": 0.999892901129893,
|
| 212 |
+
"eval_overall_f1": 0.9992264504054896,
|
| 213 |
+
"eval_overall_precision": 0.9992015171174768,
|
| 214 |
+
"eval_overall_recall": 0.9992513849378649,
|
| 215 |
+
"eval_runtime": 16.3326,
|
| 216 |
+
"eval_samples_per_second": 408.938,
|
| 217 |
+
"eval_steps_per_second": 25.593,
|
| 218 |
+
"step": 10000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.93,
|
| 222 |
+
"learning_rate": 1.1634662914939731e-05,
|
| 223 |
+
"loss": 0.0002,
|
| 224 |
+
"step": 11000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.93,
|
| 228 |
+
"eval_EG_f1": 0.9988771614641814,
|
| 229 |
+
"eval_ET_f1": 1.0,
|
| 230 |
+
"eval_TE_f1": 0.9990268732689572,
|
| 231 |
+
"eval_loss": 0.0005850127199664712,
|
| 232 |
+
"eval_overall_accuracy": 0.9999286007532621,
|
| 233 |
+
"eval_overall_f1": 0.9993013274777922,
|
| 234 |
+
"eval_overall_precision": 0.9992514596536753,
|
| 235 |
+
"eval_overall_recall": 0.999351200279483,
|
| 236 |
+
"eval_runtime": 16.4437,
|
| 237 |
+
"eval_samples_per_second": 406.175,
|
| 238 |
+
"eval_steps_per_second": 25.42,
|
| 239 |
+
"step": 11000
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 3.19,
|
| 243 |
+
"learning_rate": 1.08741777253888e-05,
|
| 244 |
+
"loss": 0.0008,
|
| 245 |
+
"step": 12000
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 3.19,
|
| 249 |
+
"eval_EG_f1": 0.9992513849378649,
|
| 250 |
+
"eval_ET_f1": 1.0,
|
| 251 |
+
"eval_TE_f1": 0.9992513849378649,
|
| 252 |
+
"eval_loss": 0.0004198316019028425,
|
| 253 |
+
"eval_overall_accuracy": 0.9999464505649466,
|
| 254 |
+
"eval_overall_f1": 0.99950092329191,
|
| 255 |
+
"eval_overall_precision": 0.99950092329191,
|
| 256 |
+
"eval_overall_recall": 0.99950092329191,
|
| 257 |
+
"eval_runtime": 16.4892,
|
| 258 |
+
"eval_samples_per_second": 405.052,
|
| 259 |
+
"eval_steps_per_second": 25.35,
|
| 260 |
+
"step": 12000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 3.46,
|
| 264 |
+
"learning_rate": 1.0113692535837865e-05,
|
| 265 |
+
"loss": 0.0002,
|
| 266 |
+
"step": 13000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 3.46,
|
| 270 |
+
"eval_EG_f1": 0.9991016619254379,
|
| 271 |
+
"eval_ET_f1": 0.999700553975146,
|
| 272 |
+
"eval_TE_f1": 0.9989519389130109,
|
| 273 |
+
"eval_loss": 0.0011158857960253954,
|
| 274 |
+
"eval_overall_accuracy": 0.9998482766006819,
|
| 275 |
+
"eval_overall_f1": 0.9992513849378649,
|
| 276 |
+
"eval_overall_precision": 0.9992513849378649,
|
| 277 |
+
"eval_overall_recall": 0.9992513849378649,
|
| 278 |
+
"eval_runtime": 16.7537,
|
| 279 |
+
"eval_samples_per_second": 398.659,
|
| 280 |
+
"eval_steps_per_second": 24.95,
|
| 281 |
+
"step": 13000
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 3.73,
|
| 285 |
+
"learning_rate": 9.353207346286931e-06,
|
| 286 |
+
"loss": 0.0003,
|
| 287 |
+
"step": 14000
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 3.73,
|
| 291 |
+
"eval_EG_f1": 0.9988022159005839,
|
| 292 |
+
"eval_ET_f1": 0.999850276987573,
|
| 293 |
+
"eval_TE_f1": 0.9986524928881569,
|
| 294 |
+
"eval_loss": 0.0010174426715821028,
|
| 295 |
+
"eval_overall_accuracy": 0.9998393516948396,
|
| 296 |
+
"eval_overall_f1": 0.9991016619254379,
|
| 297 |
+
"eval_overall_precision": 0.9991016619254379,
|
| 298 |
+
"eval_overall_recall": 0.9991016619254379,
|
| 299 |
+
"eval_runtime": 16.5225,
|
| 300 |
+
"eval_samples_per_second": 404.237,
|
| 301 |
+
"eval_steps_per_second": 25.299,
|
| 302 |
+
"step": 14000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 3.99,
|
| 306 |
+
"learning_rate": 8.592722156735998e-06,
|
| 307 |
+
"loss": 0.0,
|
| 308 |
+
"step": 15000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 3.99,
|
| 312 |
+
"eval_EG_f1": 0.9992513849378649,
|
| 313 |
+
"eval_ET_f1": 1.0,
|
| 314 |
+
"eval_TE_f1": 0.9992513849378649,
|
| 315 |
+
"eval_loss": 0.001382750691846013,
|
| 316 |
+
"eval_overall_accuracy": 0.9999018260357353,
|
| 317 |
+
"eval_overall_f1": 0.99950092329191,
|
| 318 |
+
"eval_overall_precision": 0.99950092329191,
|
| 319 |
+
"eval_overall_recall": 0.99950092329191,
|
| 320 |
+
"eval_runtime": 16.3741,
|
| 321 |
+
"eval_samples_per_second": 407.899,
|
| 322 |
+
"eval_steps_per_second": 25.528,
|
| 323 |
+
"step": 15000
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 4.26,
|
| 327 |
+
"learning_rate": 7.832236967185065e-06,
|
| 328 |
+
"loss": 0.0002,
|
| 329 |
+
"step": 16000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 4.26,
|
| 333 |
+
"eval_EG_f1": 0.9990268732689572,
|
| 334 |
+
"eval_ET_f1": 0.9997754322928364,
|
| 335 |
+
"eval_TE_f1": 0.9991016619254379,
|
| 336 |
+
"eval_loss": 0.0008425627020187676,
|
| 337 |
+
"eval_overall_accuracy": 0.9998839762240508,
|
| 338 |
+
"eval_overall_f1": 0.9993013274777922,
|
| 339 |
+
"eval_overall_precision": 0.9992514596536753,
|
| 340 |
+
"eval_overall_recall": 0.999351200279483,
|
| 341 |
+
"eval_runtime": 16.4088,
|
| 342 |
+
"eval_samples_per_second": 407.037,
|
| 343 |
+
"eval_steps_per_second": 25.474,
|
| 344 |
+
"step": 16000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.52,
|
| 348 |
+
"learning_rate": 7.071751777634132e-06,
|
| 349 |
+
"loss": 0.0001,
|
| 350 |
+
"step": 17000
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 4.52,
|
| 354 |
+
"eval_EG_f1": 0.9994011079502919,
|
| 355 |
+
"eval_ET_f1": 0.999700553975146,
|
| 356 |
+
"eval_TE_f1": 0.9992513849378649,
|
| 357 |
+
"eval_loss": 0.000884207256603986,
|
| 358 |
+
"eval_overall_accuracy": 0.9999196758474198,
|
| 359 |
+
"eval_overall_f1": 0.999451015621101,
|
| 360 |
+
"eval_overall_precision": 0.999451015621101,
|
| 361 |
+
"eval_overall_recall": 0.999451015621101,
|
| 362 |
+
"eval_runtime": 16.6905,
|
| 363 |
+
"eval_samples_per_second": 400.169,
|
| 364 |
+
"eval_steps_per_second": 25.044,
|
| 365 |
+
"step": 17000
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 4.79,
|
| 369 |
+
"learning_rate": 6.311266588083198e-06,
|
| 370 |
+
"loss": 0.0005,
|
| 371 |
+
"step": 18000
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 4.79,
|
| 375 |
+
"eval_EG_f1": 0.9992513849378649,
|
| 376 |
+
"eval_ET_f1": 1.0,
|
| 377 |
+
"eval_TE_f1": 0.9992513849378649,
|
| 378 |
+
"eval_loss": 0.0005574871320277452,
|
| 379 |
+
"eval_overall_accuracy": 0.9999375256591043,
|
| 380 |
+
"eval_overall_f1": 0.99950092329191,
|
| 381 |
+
"eval_overall_precision": 0.99950092329191,
|
| 382 |
+
"eval_overall_recall": 0.99950092329191,
|
| 383 |
+
"eval_runtime": 16.5015,
|
| 384 |
+
"eval_samples_per_second": 404.752,
|
| 385 |
+
"eval_steps_per_second": 25.331,
|
| 386 |
+
"step": 18000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 5.06,
|
| 390 |
+
"learning_rate": 5.5507813985322644e-06,
|
| 391 |
+
"loss": 0.0,
|
| 392 |
+
"step": 19000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 5.06,
|
| 396 |
+
"eval_EG_f1": 0.9989519389130109,
|
| 397 |
+
"eval_ET_f1": 1.0,
|
| 398 |
+
"eval_TE_f1": 0.9989519389130109,
|
| 399 |
+
"eval_loss": 0.0005674651474691927,
|
| 400 |
+
"eval_overall_accuracy": 0.9999018260357353,
|
| 401 |
+
"eval_overall_f1": 0.999301292608674,
|
| 402 |
+
"eval_overall_precision": 0.999301292608674,
|
| 403 |
+
"eval_overall_recall": 0.999301292608674,
|
| 404 |
+
"eval_runtime": 16.4647,
|
| 405 |
+
"eval_samples_per_second": 405.657,
|
| 406 |
+
"eval_steps_per_second": 25.388,
|
| 407 |
+
"step": 19000
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 5.32,
|
| 411 |
+
"learning_rate": 4.79029620898133e-06,
|
| 412 |
+
"loss": 0.0001,
|
| 413 |
+
"step": 20000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 5.32,
|
| 417 |
+
"eval_EG_f1": 0.9992513849378649,
|
| 418 |
+
"eval_ET_f1": 1.0,
|
| 419 |
+
"eval_TE_f1": 0.9992513849378649,
|
| 420 |
+
"eval_loss": 0.0006091786199249327,
|
| 421 |
+
"eval_overall_accuracy": 0.9999375256591043,
|
| 422 |
+
"eval_overall_f1": 0.99950092329191,
|
| 423 |
+
"eval_overall_precision": 0.99950092329191,
|
| 424 |
+
"eval_overall_recall": 0.99950092329191,
|
| 425 |
+
"eval_runtime": 16.4349,
|
| 426 |
+
"eval_samples_per_second": 406.391,
|
| 427 |
+
"eval_steps_per_second": 25.434,
|
| 428 |
+
"step": 20000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 5.59,
|
| 432 |
+
"learning_rate": 4.029811019430397e-06,
|
| 433 |
+
"loss": 0.0,
|
| 434 |
+
"step": 21000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 5.59,
|
| 438 |
+
"eval_EG_f1": 0.9994011079502919,
|
| 439 |
+
"eval_ET_f1": 1.0,
|
| 440 |
+
"eval_TE_f1": 0.9994011079502919,
|
| 441 |
+
"eval_loss": 0.0006266526179388165,
|
| 442 |
+
"eval_overall_accuracy": 0.9999553754707888,
|
| 443 |
+
"eval_overall_f1": 0.9996007386335279,
|
| 444 |
+
"eval_overall_precision": 0.9996007386335279,
|
| 445 |
+
"eval_overall_recall": 0.9996007386335279,
|
| 446 |
+
"eval_runtime": 16.6877,
|
| 447 |
+
"eval_samples_per_second": 400.236,
|
| 448 |
+
"eval_steps_per_second": 25.048,
|
| 449 |
+
"step": 21000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 5.86,
|
| 453 |
+
"learning_rate": 3.2693258298794635e-06,
|
| 454 |
+
"loss": 0.0,
|
| 455 |
+
"step": 22000
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 5.86,
|
| 459 |
+
"eval_EG_f1": 0.9992513849378649,
|
| 460 |
+
"eval_ET_f1": 1.0,
|
| 461 |
+
"eval_TE_f1": 0.9992513849378649,
|
| 462 |
+
"eval_loss": 0.0006635423633269966,
|
| 463 |
+
"eval_overall_accuracy": 0.9999375256591043,
|
| 464 |
+
"eval_overall_f1": 0.99950092329191,
|
| 465 |
+
"eval_overall_precision": 0.99950092329191,
|
| 466 |
+
"eval_overall_recall": 0.99950092329191,
|
| 467 |
+
"eval_runtime": 16.716,
|
| 468 |
+
"eval_samples_per_second": 399.557,
|
| 469 |
+
"eval_steps_per_second": 25.006,
|
| 470 |
+
"step": 22000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 6.12,
|
| 474 |
+
"learning_rate": 2.50884064032853e-06,
|
| 475 |
+
"loss": 0.0,
|
| 476 |
+
"step": 23000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 6.12,
|
| 480 |
+
"eval_EG_f1": 0.9994011079502919,
|
| 481 |
+
"eval_ET_f1": 1.0,
|
| 482 |
+
"eval_TE_f1": 0.9994011079502919,
|
| 483 |
+
"eval_loss": 0.0006512191030196846,
|
| 484 |
+
"eval_overall_accuracy": 0.9999464505649466,
|
| 485 |
+
"eval_overall_f1": 0.9996007386335279,
|
| 486 |
+
"eval_overall_precision": 0.9996007386335279,
|
| 487 |
+
"eval_overall_recall": 0.9996007386335279,
|
| 488 |
+
"eval_runtime": 16.3818,
|
| 489 |
+
"eval_samples_per_second": 407.709,
|
| 490 |
+
"eval_steps_per_second": 25.516,
|
| 491 |
+
"step": 23000
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 6.39,
|
| 495 |
+
"learning_rate": 1.7483554507775962e-06,
|
| 496 |
+
"loss": 0.0,
|
| 497 |
+
"step": 24000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 6.39,
|
| 501 |
+
"eval_EG_f1": 0.9994011079502919,
|
| 502 |
+
"eval_ET_f1": 1.0,
|
| 503 |
+
"eval_TE_f1": 0.9994011079502919,
|
| 504 |
+
"eval_loss": 0.000729710329324007,
|
| 505 |
+
"eval_overall_accuracy": 0.9999464505649466,
|
| 506 |
+
"eval_overall_f1": 0.9996007386335279,
|
| 507 |
+
"eval_overall_precision": 0.9996007386335279,
|
| 508 |
+
"eval_overall_recall": 0.9996007386335279,
|
| 509 |
+
"eval_runtime": 16.5641,
|
| 510 |
+
"eval_samples_per_second": 403.222,
|
| 511 |
+
"eval_steps_per_second": 25.235,
|
| 512 |
+
"step": 24000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 6.65,
|
| 516 |
+
"learning_rate": 9.878702612266627e-07,
|
| 517 |
+
"loss": 0.0001,
|
| 518 |
+
"step": 25000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 6.65,
|
| 522 |
+
"eval_EG_f1": 0.9994011079502919,
|
| 523 |
+
"eval_ET_f1": 1.0,
|
| 524 |
+
"eval_TE_f1": 0.9994011079502919,
|
| 525 |
+
"eval_loss": 0.0005905419238843024,
|
| 526 |
+
"eval_overall_accuracy": 0.9999464505649466,
|
| 527 |
+
"eval_overall_f1": 0.9996007386335279,
|
| 528 |
+
"eval_overall_precision": 0.9996007386335279,
|
| 529 |
+
"eval_overall_recall": 0.9996007386335279,
|
| 530 |
+
"eval_runtime": 16.5058,
|
| 531 |
+
"eval_samples_per_second": 404.647,
|
| 532 |
+
"eval_steps_per_second": 25.325,
|
| 533 |
+
"step": 25000
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 6.92,
|
| 537 |
+
"learning_rate": 2.2738507167572915e-07,
|
| 538 |
+
"loss": 0.0001,
|
| 539 |
+
"step": 26000
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 6.92,
|
| 543 |
+
"eval_EG_f1": 0.9994011079502919,
|
| 544 |
+
"eval_ET_f1": 1.0,
|
| 545 |
+
"eval_TE_f1": 0.9994011079502919,
|
| 546 |
+
"eval_loss": 0.0006466144695878029,
|
| 547 |
+
"eval_overall_accuracy": 0.9999464505649466,
|
| 548 |
+
"eval_overall_f1": 0.9996007386335279,
|
| 549 |
+
"eval_overall_precision": 0.9996007386335279,
|
| 550 |
+
"eval_overall_recall": 0.9996007386335279,
|
| 551 |
+
"eval_runtime": 16.8702,
|
| 552 |
+
"eval_samples_per_second": 395.906,
|
| 553 |
+
"eval_steps_per_second": 24.777,
|
| 554 |
+
"step": 26000
|
| 555 |
+
}
|
| 556 |
+
],
|
| 557 |
+
"max_steps": 26299,
|
| 558 |
+
"num_train_epochs": 7,
|
| 559 |
+
"total_flos": 6405072201643752.0,
|
| 560 |
+
"trial_name": null,
|
| 561 |
+
"trial_params": null
|
| 562 |
+
}
|