Upload trainer_state.json
Browse files- trainer_state.json +136 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,136 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9050816297531128,
|
| 3 |
+
"best_model_checkpoint": "./mbert_ar_ur/checkpoint-5000",
|
| 4 |
+
"epoch": 4.0,
|
| 5 |
+
"global_step": 5000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.4,
|
| 12 |
+
"learning_rate": 1.885714285714286e-05,
|
| 13 |
+
"loss": 0.4454,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.8,
|
| 18 |
+
"learning_rate": 1.7714285714285717e-05,
|
| 19 |
+
"loss": 0.2955,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.0,
|
| 24 |
+
"eval_LOC_f1": 0.09510682288077188,
|
| 25 |
+
"eval_ORG_f1": 0.24583663758921492,
|
| 26 |
+
"eval_PER_f1": 0.3119353304802663,
|
| 27 |
+
"eval_loss": 1.4924763441085815,
|
| 28 |
+
"eval_overall_accuracy": 0.4886179434773416,
|
| 29 |
+
"eval_overall_f1": 0.1982125758059368,
|
| 30 |
+
"eval_overall_precision": 0.16568836712913554,
|
| 31 |
+
"eval_overall_recall": 0.2466243050039714,
|
| 32 |
+
"eval_runtime": 2.7088,
|
| 33 |
+
"eval_samples_per_second": 369.163,
|
| 34 |
+
"eval_steps_per_second": 23.257,
|
| 35 |
+
"step": 1250
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 1.2,
|
| 39 |
+
"learning_rate": 1.6571428571428574e-05,
|
| 40 |
+
"loss": 0.2182,
|
| 41 |
+
"step": 1500
|
| 42 |
+
},
|
| 43 |
+
{
|
| 44 |
+
"epoch": 1.6,
|
| 45 |
+
"learning_rate": 1.542857142857143e-05,
|
| 46 |
+
"loss": 0.1877,
|
| 47 |
+
"step": 2000
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 2.0,
|
| 51 |
+
"learning_rate": 1.4285714285714287e-05,
|
| 52 |
+
"loss": 0.181,
|
| 53 |
+
"step": 2500
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 2.0,
|
| 57 |
+
"eval_LOC_f1": 0.16472203157172272,
|
| 58 |
+
"eval_ORG_f1": 0.26800929512006194,
|
| 59 |
+
"eval_PER_f1": 0.6124661246612466,
|
| 60 |
+
"eval_loss": 1.1303032636642456,
|
| 61 |
+
"eval_overall_accuracy": 0.6376770737895553,
|
| 62 |
+
"eval_overall_f1": 0.3045238514346066,
|
| 63 |
+
"eval_overall_precision": 0.2734745494783433,
|
| 64 |
+
"eval_overall_recall": 0.3435266084193805,
|
| 65 |
+
"eval_runtime": 2.7853,
|
| 66 |
+
"eval_samples_per_second": 359.026,
|
| 67 |
+
"eval_steps_per_second": 22.619,
|
| 68 |
+
"step": 2500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 2.4,
|
| 72 |
+
"learning_rate": 1.3142857142857145e-05,
|
| 73 |
+
"loss": 0.1295,
|
| 74 |
+
"step": 3000
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 2.8,
|
| 78 |
+
"learning_rate": 1.2e-05,
|
| 79 |
+
"loss": 0.1253,
|
| 80 |
+
"step": 3500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 3.0,
|
| 84 |
+
"eval_LOC_f1": 0.15284677111196027,
|
| 85 |
+
"eval_ORG_f1": 0.3176020408163266,
|
| 86 |
+
"eval_PER_f1": 0.6216577540106952,
|
| 87 |
+
"eval_loss": 1.348933458328247,
|
| 88 |
+
"eval_overall_accuracy": 0.6324617661568821,
|
| 89 |
+
"eval_overall_f1": 0.3217743355043126,
|
| 90 |
+
"eval_overall_precision": 0.2889661713563073,
|
| 91 |
+
"eval_overall_recall": 0.36298649722001586,
|
| 92 |
+
"eval_runtime": 2.5633,
|
| 93 |
+
"eval_samples_per_second": 390.124,
|
| 94 |
+
"eval_steps_per_second": 24.578,
|
| 95 |
+
"step": 3750
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 3.2,
|
| 99 |
+
"learning_rate": 1.0857142857142858e-05,
|
| 100 |
+
"loss": 0.098,
|
| 101 |
+
"step": 4000
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 3.6,
|
| 105 |
+
"learning_rate": 9.714285714285715e-06,
|
| 106 |
+
"loss": 0.0916,
|
| 107 |
+
"step": 4500
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 4.0,
|
| 111 |
+
"learning_rate": 8.571428571428571e-06,
|
| 112 |
+
"loss": 0.0866,
|
| 113 |
+
"step": 5000
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 4.0,
|
| 117 |
+
"eval_LOC_f1": 0.42877697841726614,
|
| 118 |
+
"eval_ORG_f1": 0.4028497409326424,
|
| 119 |
+
"eval_PER_f1": 0.6109282422646478,
|
| 120 |
+
"eval_loss": 0.9050816297531128,
|
| 121 |
+
"eval_overall_accuracy": 0.7715836211149482,
|
| 122 |
+
"eval_overall_f1": 0.47474747474747475,
|
| 123 |
+
"eval_overall_precision": 0.464638783269962,
|
| 124 |
+
"eval_overall_recall": 0.4853057982525814,
|
| 125 |
+
"eval_runtime": 2.9178,
|
| 126 |
+
"eval_samples_per_second": 342.725,
|
| 127 |
+
"eval_steps_per_second": 21.592,
|
| 128 |
+
"step": 5000
|
| 129 |
+
}
|
| 130 |
+
],
|
| 131 |
+
"max_steps": 8750,
|
| 132 |
+
"num_train_epochs": 7,
|
| 133 |
+
"total_flos": 1363663262792160.0,
|
| 134 |
+
"trial_name": null,
|
| 135 |
+
"trial_params": null
|
| 136 |
+
}
|