Upload trainer_state.json
Browse files- trainer_state.json +850 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,850 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.11314285546541214,
|
| 3 |
+
"best_model_checkpoint": "/content/drive/MyDrive/Emotionen/multilingual-e5-large/multilingual_e5_large_2024-05-28/checkpoint-10126",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"eval_steps": 1,
|
| 6 |
+
"global_step": 50630,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.09875567845151097,
|
| 13 |
+
"grad_norm": 0.31362369656562805,
|
| 14 |
+
"learning_rate": 1.980248864309698e-05,
|
| 15 |
+
"loss": 0.1658,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.19751135690302193,
|
| 20 |
+
"grad_norm": 0.3713989555835724,
|
| 21 |
+
"learning_rate": 1.9604977286193957e-05,
|
| 22 |
+
"loss": 0.1306,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.29626703535453286,
|
| 27 |
+
"grad_norm": 0.33911290764808655,
|
| 28 |
+
"learning_rate": 1.9407465929290937e-05,
|
| 29 |
+
"loss": 0.1229,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.39502271380604387,
|
| 34 |
+
"grad_norm": 0.41552165150642395,
|
| 35 |
+
"learning_rate": 1.9209954572387913e-05,
|
| 36 |
+
"loss": 0.1199,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.4937783922575548,
|
| 41 |
+
"grad_norm": 0.5132314562797546,
|
| 42 |
+
"learning_rate": 1.9012443215484893e-05,
|
| 43 |
+
"loss": 0.1189,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.5925340707090657,
|
| 48 |
+
"grad_norm": 0.460574209690094,
|
| 49 |
+
"learning_rate": 1.8814931858581872e-05,
|
| 50 |
+
"loss": 0.118,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.6912897491605767,
|
| 55 |
+
"grad_norm": 0.4870443642139435,
|
| 56 |
+
"learning_rate": 1.8617420501678848e-05,
|
| 57 |
+
"loss": 0.117,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.7900454276120877,
|
| 62 |
+
"grad_norm": 0.39810240268707275,
|
| 63 |
+
"learning_rate": 1.8419909144775828e-05,
|
| 64 |
+
"loss": 0.1164,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.8888011060635986,
|
| 69 |
+
"grad_norm": 0.3674732744693756,
|
| 70 |
+
"learning_rate": 1.8222397787872804e-05,
|
| 71 |
+
"loss": 0.1153,
|
| 72 |
+
"step": 4500
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.9875567845151096,
|
| 76 |
+
"grad_norm": 0.419662743806839,
|
| 77 |
+
"learning_rate": 1.802488643096978e-05,
|
| 78 |
+
"loss": 0.1158,
|
| 79 |
+
"step": 5000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 1.0,
|
| 83 |
+
"eval_accuracy": 0.18166666666666667,
|
| 84 |
+
"eval_f1_micro": 0.30011781927237674,
|
| 85 |
+
"eval_loss": 0.11470416933298111,
|
| 86 |
+
"eval_roc_auc": 0.5961780161730722,
|
| 87 |
+
"eval_runtime": 39.4359,
|
| 88 |
+
"eval_samples_per_second": 456.437,
|
| 89 |
+
"eval_steps_per_second": 14.276,
|
| 90 |
+
"step": 5063
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 1.0863124629666205,
|
| 94 |
+
"grad_norm": 0.41265037655830383,
|
| 95 |
+
"learning_rate": 1.782737507406676e-05,
|
| 96 |
+
"loss": 0.1126,
|
| 97 |
+
"step": 5500
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 1.1850681414181317,
|
| 101 |
+
"grad_norm": 0.38658422231674194,
|
| 102 |
+
"learning_rate": 1.762986371716374e-05,
|
| 103 |
+
"loss": 0.1106,
|
| 104 |
+
"step": 6000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.2838238198696426,
|
| 108 |
+
"grad_norm": 0.45577454566955566,
|
| 109 |
+
"learning_rate": 1.7432352360260715e-05,
|
| 110 |
+
"loss": 0.1114,
|
| 111 |
+
"step": 6500
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 1.3825794983211535,
|
| 115 |
+
"grad_norm": 0.5785081386566162,
|
| 116 |
+
"learning_rate": 1.7234841003357695e-05,
|
| 117 |
+
"loss": 0.1117,
|
| 118 |
+
"step": 7000
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 1.4813351767726644,
|
| 122 |
+
"grad_norm": 0.42792847752571106,
|
| 123 |
+
"learning_rate": 1.703732964645467e-05,
|
| 124 |
+
"loss": 0.1113,
|
| 125 |
+
"step": 7500
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.5800908552241753,
|
| 129 |
+
"grad_norm": 0.5515544414520264,
|
| 130 |
+
"learning_rate": 1.683981828955165e-05,
|
| 131 |
+
"loss": 0.1107,
|
| 132 |
+
"step": 8000
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 1.6788465336756864,
|
| 136 |
+
"grad_norm": 0.4049302637577057,
|
| 137 |
+
"learning_rate": 1.664230693264863e-05,
|
| 138 |
+
"loss": 0.1113,
|
| 139 |
+
"step": 8500
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 1.7776022121271973,
|
| 143 |
+
"grad_norm": 0.4182845652103424,
|
| 144 |
+
"learning_rate": 1.6444795575745606e-05,
|
| 145 |
+
"loss": 0.1112,
|
| 146 |
+
"step": 9000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.8763578905787082,
|
| 150 |
+
"grad_norm": 0.4281316101551056,
|
| 151 |
+
"learning_rate": 1.6247284218842586e-05,
|
| 152 |
+
"loss": 0.1109,
|
| 153 |
+
"step": 9500
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 1.9751135690302193,
|
| 157 |
+
"grad_norm": 0.4670790433883667,
|
| 158 |
+
"learning_rate": 1.6049772861939562e-05,
|
| 159 |
+
"loss": 0.1102,
|
| 160 |
+
"step": 10000
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 2.0,
|
| 164 |
+
"eval_accuracy": 0.21488888888888888,
|
| 165 |
+
"eval_f1_micro": 0.3304371622083875,
|
| 166 |
+
"eval_loss": 0.11314285546541214,
|
| 167 |
+
"eval_roc_auc": 0.6102983856871789,
|
| 168 |
+
"eval_runtime": 39.4597,
|
| 169 |
+
"eval_samples_per_second": 456.162,
|
| 170 |
+
"eval_steps_per_second": 14.268,
|
| 171 |
+
"step": 10126
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 2.0738692474817304,
|
| 175 |
+
"grad_norm": 0.46797189116477966,
|
| 176 |
+
"learning_rate": 1.585226150503654e-05,
|
| 177 |
+
"loss": 0.1071,
|
| 178 |
+
"step": 10500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 2.172624925933241,
|
| 182 |
+
"grad_norm": 0.5314794778823853,
|
| 183 |
+
"learning_rate": 1.565475014813352e-05,
|
| 184 |
+
"loss": 0.1066,
|
| 185 |
+
"step": 11000
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 2.271380604384752,
|
| 189 |
+
"grad_norm": 0.6094779372215271,
|
| 190 |
+
"learning_rate": 1.5457238791230497e-05,
|
| 191 |
+
"loss": 0.1048,
|
| 192 |
+
"step": 11500
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 2.3701362828362633,
|
| 196 |
+
"grad_norm": 0.46292218565940857,
|
| 197 |
+
"learning_rate": 1.5259727434327476e-05,
|
| 198 |
+
"loss": 0.1063,
|
| 199 |
+
"step": 12000
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 2.468891961287774,
|
| 203 |
+
"grad_norm": 0.4565483033657074,
|
| 204 |
+
"learning_rate": 1.5062216077424454e-05,
|
| 205 |
+
"loss": 0.1052,
|
| 206 |
+
"step": 12500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.567647639739285,
|
| 210 |
+
"grad_norm": 0.5294711589813232,
|
| 211 |
+
"learning_rate": 1.486470472052143e-05,
|
| 212 |
+
"loss": 0.107,
|
| 213 |
+
"step": 13000
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 2.6664033181907962,
|
| 217 |
+
"grad_norm": 0.5409520864486694,
|
| 218 |
+
"learning_rate": 1.4667193363618408e-05,
|
| 219 |
+
"loss": 0.1061,
|
| 220 |
+
"step": 13500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 2.765158996642307,
|
| 224 |
+
"grad_norm": 0.572907567024231,
|
| 225 |
+
"learning_rate": 1.4469682006715386e-05,
|
| 226 |
+
"loss": 0.1062,
|
| 227 |
+
"step": 14000
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 2.863914675093818,
|
| 231 |
+
"grad_norm": 0.4891163110733032,
|
| 232 |
+
"learning_rate": 1.4272170649812364e-05,
|
| 233 |
+
"loss": 0.1065,
|
| 234 |
+
"step": 14500
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 2.9626703535453287,
|
| 238 |
+
"grad_norm": 0.49211814999580383,
|
| 239 |
+
"learning_rate": 1.4074659292909343e-05,
|
| 240 |
+
"loss": 0.1059,
|
| 241 |
+
"step": 15000
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 3.0,
|
| 245 |
+
"eval_accuracy": 0.2378888888888889,
|
| 246 |
+
"eval_f1_micro": 0.35730189914865756,
|
| 247 |
+
"eval_loss": 0.11334217339754105,
|
| 248 |
+
"eval_roc_auc": 0.6243103133779553,
|
| 249 |
+
"eval_runtime": 39.4216,
|
| 250 |
+
"eval_samples_per_second": 456.603,
|
| 251 |
+
"eval_steps_per_second": 14.282,
|
| 252 |
+
"step": 15189
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 3.06142603199684,
|
| 256 |
+
"grad_norm": 0.7181198596954346,
|
| 257 |
+
"learning_rate": 1.3877147936006321e-05,
|
| 258 |
+
"loss": 0.1024,
|
| 259 |
+
"step": 15500
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 3.160181710448351,
|
| 263 |
+
"grad_norm": 0.57438725233078,
|
| 264 |
+
"learning_rate": 1.3679636579103299e-05,
|
| 265 |
+
"loss": 0.0994,
|
| 266 |
+
"step": 16000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 3.2589373888998616,
|
| 270 |
+
"grad_norm": 0.7449280619621277,
|
| 271 |
+
"learning_rate": 1.3482125222200277e-05,
|
| 272 |
+
"loss": 0.1003,
|
| 273 |
+
"step": 16500
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 3.3576930673513727,
|
| 277 |
+
"grad_norm": 0.6857854723930359,
|
| 278 |
+
"learning_rate": 1.3284613865297255e-05,
|
| 279 |
+
"loss": 0.1019,
|
| 280 |
+
"step": 17000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 3.456448745802884,
|
| 284 |
+
"grad_norm": 0.6422079801559448,
|
| 285 |
+
"learning_rate": 1.3087102508394234e-05,
|
| 286 |
+
"loss": 0.1011,
|
| 287 |
+
"step": 17500
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 3.5552044242543945,
|
| 291 |
+
"grad_norm": 0.6514602899551392,
|
| 292 |
+
"learning_rate": 1.2889591151491212e-05,
|
| 293 |
+
"loss": 0.1012,
|
| 294 |
+
"step": 18000
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 3.6539601027059057,
|
| 298 |
+
"grad_norm": 0.5365302562713623,
|
| 299 |
+
"learning_rate": 1.269207979458819e-05,
|
| 300 |
+
"loss": 0.1017,
|
| 301 |
+
"step": 18500
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 3.7527157811574163,
|
| 305 |
+
"grad_norm": 0.6385504603385925,
|
| 306 |
+
"learning_rate": 1.2494568437685168e-05,
|
| 307 |
+
"loss": 0.1009,
|
| 308 |
+
"step": 19000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 3.8514714596089275,
|
| 312 |
+
"grad_norm": 0.6212201714515686,
|
| 313 |
+
"learning_rate": 1.2297057080782147e-05,
|
| 314 |
+
"loss": 0.1014,
|
| 315 |
+
"step": 19500
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 3.9502271380604386,
|
| 319 |
+
"grad_norm": 0.7167210578918457,
|
| 320 |
+
"learning_rate": 1.2099545723879125e-05,
|
| 321 |
+
"loss": 0.102,
|
| 322 |
+
"step": 20000
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 4.0,
|
| 326 |
+
"eval_accuracy": 0.24166666666666667,
|
| 327 |
+
"eval_f1_micro": 0.3643195453110199,
|
| 328 |
+
"eval_loss": 0.11566581577062607,
|
| 329 |
+
"eval_roc_auc": 0.63082221324154,
|
| 330 |
+
"eval_runtime": 39.4388,
|
| 331 |
+
"eval_samples_per_second": 456.403,
|
| 332 |
+
"eval_steps_per_second": 14.275,
|
| 333 |
+
"step": 20252
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 4.048982816511949,
|
| 337 |
+
"grad_norm": 0.7785666584968567,
|
| 338 |
+
"learning_rate": 1.1902034366976103e-05,
|
| 339 |
+
"loss": 0.099,
|
| 340 |
+
"step": 20500
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 4.147738494963461,
|
| 344 |
+
"grad_norm": 0.6057512760162354,
|
| 345 |
+
"learning_rate": 1.1704523010073081e-05,
|
| 346 |
+
"loss": 0.0951,
|
| 347 |
+
"step": 21000
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 4.2464941734149715,
|
| 351 |
+
"grad_norm": 0.7943652868270874,
|
| 352 |
+
"learning_rate": 1.1507011653170059e-05,
|
| 353 |
+
"loss": 0.0956,
|
| 354 |
+
"step": 21500
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 4.345249851866482,
|
| 358 |
+
"grad_norm": 0.8373062014579773,
|
| 359 |
+
"learning_rate": 1.1309500296267035e-05,
|
| 360 |
+
"loss": 0.0949,
|
| 361 |
+
"step": 22000
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 4.444005530317993,
|
| 365 |
+
"grad_norm": 0.6966761946678162,
|
| 366 |
+
"learning_rate": 1.1111988939364013e-05,
|
| 367 |
+
"loss": 0.0962,
|
| 368 |
+
"step": 22500
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 4.542761208769504,
|
| 372 |
+
"grad_norm": 0.7649087309837341,
|
| 373 |
+
"learning_rate": 1.0914477582460992e-05,
|
| 374 |
+
"loss": 0.0964,
|
| 375 |
+
"step": 23000
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 4.641516887221015,
|
| 379 |
+
"grad_norm": 0.695787787437439,
|
| 380 |
+
"learning_rate": 1.071696622555797e-05,
|
| 381 |
+
"loss": 0.0967,
|
| 382 |
+
"step": 23500
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 4.740272565672527,
|
| 386 |
+
"grad_norm": 0.6631536483764648,
|
| 387 |
+
"learning_rate": 1.0519454868654948e-05,
|
| 388 |
+
"loss": 0.0968,
|
| 389 |
+
"step": 24000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 4.839028244124037,
|
| 393 |
+
"grad_norm": 0.6649767160415649,
|
| 394 |
+
"learning_rate": 1.0321943511751926e-05,
|
| 395 |
+
"loss": 0.0966,
|
| 396 |
+
"step": 24500
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 4.937783922575548,
|
| 400 |
+
"grad_norm": 0.6164096593856812,
|
| 401 |
+
"learning_rate": 1.0124432154848904e-05,
|
| 402 |
+
"loss": 0.0969,
|
| 403 |
+
"step": 25000
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 5.0,
|
| 407 |
+
"eval_accuracy": 0.25194444444444447,
|
| 408 |
+
"eval_f1_micro": 0.37071333765472114,
|
| 409 |
+
"eval_loss": 0.11759943515062332,
|
| 410 |
+
"eval_roc_auc": 0.6358616290591664,
|
| 411 |
+
"eval_runtime": 39.4749,
|
| 412 |
+
"eval_samples_per_second": 455.985,
|
| 413 |
+
"eval_steps_per_second": 14.262,
|
| 414 |
+
"step": 25315
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 5.036539601027059,
|
| 418 |
+
"grad_norm": 0.8620449304580688,
|
| 419 |
+
"learning_rate": 9.926920797945883e-06,
|
| 420 |
+
"loss": 0.0955,
|
| 421 |
+
"step": 25500
|
| 422 |
+
},
|
| 423 |
+
{
|
| 424 |
+
"epoch": 5.13529527947857,
|
| 425 |
+
"grad_norm": 0.7160885334014893,
|
| 426 |
+
"learning_rate": 9.729409441042861e-06,
|
| 427 |
+
"loss": 0.0897,
|
| 428 |
+
"step": 26000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 5.234050957930081,
|
| 432 |
+
"grad_norm": 0.8834371566772461,
|
| 433 |
+
"learning_rate": 9.531898084139839e-06,
|
| 434 |
+
"loss": 0.0897,
|
| 435 |
+
"step": 26500
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 5.332806636381592,
|
| 439 |
+
"grad_norm": 0.7659549117088318,
|
| 440 |
+
"learning_rate": 9.334386727236817e-06,
|
| 441 |
+
"loss": 0.0915,
|
| 442 |
+
"step": 27000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 5.431562314833103,
|
| 446 |
+
"grad_norm": 0.9961947202682495,
|
| 447 |
+
"learning_rate": 9.136875370333796e-06,
|
| 448 |
+
"loss": 0.0917,
|
| 449 |
+
"step": 27500
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 5.530317993284614,
|
| 453 |
+
"grad_norm": 0.8744510412216187,
|
| 454 |
+
"learning_rate": 8.939364013430774e-06,
|
| 455 |
+
"loss": 0.0911,
|
| 456 |
+
"step": 28000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 5.6290736717361245,
|
| 460 |
+
"grad_norm": 0.8435235023498535,
|
| 461 |
+
"learning_rate": 8.74185265652775e-06,
|
| 462 |
+
"loss": 0.0935,
|
| 463 |
+
"step": 28500
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 5.727829350187636,
|
| 467 |
+
"grad_norm": 0.848759114742279,
|
| 468 |
+
"learning_rate": 8.544341299624728e-06,
|
| 469 |
+
"loss": 0.0924,
|
| 470 |
+
"step": 29000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 5.826585028639147,
|
| 474 |
+
"grad_norm": 0.7850985527038574,
|
| 475 |
+
"learning_rate": 8.346829942721707e-06,
|
| 476 |
+
"loss": 0.0935,
|
| 477 |
+
"step": 29500
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 5.925340707090657,
|
| 481 |
+
"grad_norm": 0.8016532063484192,
|
| 482 |
+
"learning_rate": 8.149318585818685e-06,
|
| 483 |
+
"loss": 0.0917,
|
| 484 |
+
"step": 30000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 6.0,
|
| 488 |
+
"eval_accuracy": 0.25655555555555554,
|
| 489 |
+
"eval_f1_micro": 0.3737782371179684,
|
| 490 |
+
"eval_loss": 0.12158573418855667,
|
| 491 |
+
"eval_roc_auc": 0.6422934232904187,
|
| 492 |
+
"eval_runtime": 39.5217,
|
| 493 |
+
"eval_samples_per_second": 455.446,
|
| 494 |
+
"eval_steps_per_second": 14.245,
|
| 495 |
+
"step": 30378
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 6.024096385542169,
|
| 499 |
+
"grad_norm": 0.782853364944458,
|
| 500 |
+
"learning_rate": 7.951807228915663e-06,
|
| 501 |
+
"loss": 0.0914,
|
| 502 |
+
"step": 30500
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 6.12285206399368,
|
| 506 |
+
"grad_norm": 0.8463727235794067,
|
| 507 |
+
"learning_rate": 7.754295872012641e-06,
|
| 508 |
+
"loss": 0.0862,
|
| 509 |
+
"step": 31000
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 6.22160774244519,
|
| 513 |
+
"grad_norm": 0.9679527282714844,
|
| 514 |
+
"learning_rate": 7.55678451510962e-06,
|
| 515 |
+
"loss": 0.0868,
|
| 516 |
+
"step": 31500
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 6.320363420896702,
|
| 520 |
+
"grad_norm": 0.8114763498306274,
|
| 521 |
+
"learning_rate": 7.3592731582065975e-06,
|
| 522 |
+
"loss": 0.0865,
|
| 523 |
+
"step": 32000
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 6.419119099348213,
|
| 527 |
+
"grad_norm": 0.8869866132736206,
|
| 528 |
+
"learning_rate": 7.161761801303575e-06,
|
| 529 |
+
"loss": 0.0881,
|
| 530 |
+
"step": 32500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 6.517874777799723,
|
| 534 |
+
"grad_norm": 0.9731593132019043,
|
| 535 |
+
"learning_rate": 6.964250444400553e-06,
|
| 536 |
+
"loss": 0.0866,
|
| 537 |
+
"step": 33000
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 6.616630456251235,
|
| 541 |
+
"grad_norm": 0.8720415830612183,
|
| 542 |
+
"learning_rate": 6.766739087497531e-06,
|
| 543 |
+
"loss": 0.0881,
|
| 544 |
+
"step": 33500
|
| 545 |
+
},
|
| 546 |
+
{
|
| 547 |
+
"epoch": 6.7153861347027455,
|
| 548 |
+
"grad_norm": 0.8622246980667114,
|
| 549 |
+
"learning_rate": 6.56922773059451e-06,
|
| 550 |
+
"loss": 0.0883,
|
| 551 |
+
"step": 34000
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 6.814141813154256,
|
| 555 |
+
"grad_norm": 0.8367873430252075,
|
| 556 |
+
"learning_rate": 6.3717163736914875e-06,
|
| 557 |
+
"loss": 0.089,
|
| 558 |
+
"step": 34500
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 6.912897491605768,
|
| 562 |
+
"grad_norm": 0.8973580598831177,
|
| 563 |
+
"learning_rate": 6.174205016788466e-06,
|
| 564 |
+
"loss": 0.088,
|
| 565 |
+
"step": 35000
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 7.0,
|
| 569 |
+
"eval_accuracy": 0.2574444444444444,
|
| 570 |
+
"eval_f1_micro": 0.37320491873763867,
|
| 571 |
+
"eval_loss": 0.12590783834457397,
|
| 572 |
+
"eval_roc_auc": 0.6456773567793932,
|
| 573 |
+
"eval_runtime": 39.5146,
|
| 574 |
+
"eval_samples_per_second": 455.528,
|
| 575 |
+
"eval_steps_per_second": 14.248,
|
| 576 |
+
"step": 35441
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 7.011653170057278,
|
| 580 |
+
"grad_norm": 0.9826680421829224,
|
| 581 |
+
"learning_rate": 5.976693659885444e-06,
|
| 582 |
+
"loss": 0.0876,
|
| 583 |
+
"step": 35500
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 7.110408848508789,
|
| 587 |
+
"grad_norm": 1.1457393169403076,
|
| 588 |
+
"learning_rate": 5.779182302982423e-06,
|
| 589 |
+
"loss": 0.0813,
|
| 590 |
+
"step": 36000
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 7.209164526960301,
|
| 594 |
+
"grad_norm": 0.9791042804718018,
|
| 595 |
+
"learning_rate": 5.5816709460794006e-06,
|
| 596 |
+
"loss": 0.0834,
|
| 597 |
+
"step": 36500
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 7.307920205411811,
|
| 601 |
+
"grad_norm": 0.996573805809021,
|
| 602 |
+
"learning_rate": 5.3841595891763776e-06,
|
| 603 |
+
"loss": 0.0834,
|
| 604 |
+
"step": 37000
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 7.406675883863322,
|
| 608 |
+
"grad_norm": 1.0137947797775269,
|
| 609 |
+
"learning_rate": 5.186648232273355e-06,
|
| 610 |
+
"loss": 0.0836,
|
| 611 |
+
"step": 37500
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 7.505431562314833,
|
| 615 |
+
"grad_norm": 0.9656222462654114,
|
| 616 |
+
"learning_rate": 4.989136875370334e-06,
|
| 617 |
+
"loss": 0.0838,
|
| 618 |
+
"step": 38000
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 7.604187240766344,
|
| 622 |
+
"grad_norm": 0.9735883474349976,
|
| 623 |
+
"learning_rate": 4.791625518467312e-06,
|
| 624 |
+
"loss": 0.0832,
|
| 625 |
+
"step": 38500
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 7.702942919217855,
|
| 629 |
+
"grad_norm": 0.9456019997596741,
|
| 630 |
+
"learning_rate": 4.594114161564291e-06,
|
| 631 |
+
"loss": 0.0845,
|
| 632 |
+
"step": 39000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 7.8016985976693665,
|
| 636 |
+
"grad_norm": 0.98697429895401,
|
| 637 |
+
"learning_rate": 4.3966028046612684e-06,
|
| 638 |
+
"loss": 0.0842,
|
| 639 |
+
"step": 39500
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 7.900454276120877,
|
| 643 |
+
"grad_norm": 0.9538049697875977,
|
| 644 |
+
"learning_rate": 4.199091447758246e-06,
|
| 645 |
+
"loss": 0.0843,
|
| 646 |
+
"step": 40000
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 7.999209954572388,
|
| 650 |
+
"grad_norm": 0.9173344373703003,
|
| 651 |
+
"learning_rate": 4.001580090855224e-06,
|
| 652 |
+
"loss": 0.0848,
|
| 653 |
+
"step": 40500
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 8.0,
|
| 657 |
+
"eval_accuracy": 0.25755555555555554,
|
| 658 |
+
"eval_f1_micro": 0.370994086172909,
|
| 659 |
+
"eval_loss": 0.12952594459056854,
|
| 660 |
+
"eval_roc_auc": 0.6469219706450647,
|
| 661 |
+
"eval_runtime": 39.7032,
|
| 662 |
+
"eval_samples_per_second": 453.364,
|
| 663 |
+
"eval_steps_per_second": 14.18,
|
| 664 |
+
"step": 40504
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 8.097965633023898,
|
| 668 |
+
"grad_norm": 0.9694753885269165,
|
| 669 |
+
"learning_rate": 3.8040687339522024e-06,
|
| 670 |
+
"loss": 0.0784,
|
| 671 |
+
"step": 41000
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 8.19672131147541,
|
| 675 |
+
"grad_norm": 1.1527094841003418,
|
| 676 |
+
"learning_rate": 3.6065573770491806e-06,
|
| 677 |
+
"loss": 0.0798,
|
| 678 |
+
"step": 41500
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 8.295476989926922,
|
| 682 |
+
"grad_norm": 1.3963054418563843,
|
| 683 |
+
"learning_rate": 3.409046020146159e-06,
|
| 684 |
+
"loss": 0.0791,
|
| 685 |
+
"step": 42000
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 8.394232668378432,
|
| 689 |
+
"grad_norm": 1.226542353630066,
|
| 690 |
+
"learning_rate": 3.2115346632431367e-06,
|
| 691 |
+
"loss": 0.0808,
|
| 692 |
+
"step": 42500
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 8.492988346829943,
|
| 696 |
+
"grad_norm": 1.0286864042282104,
|
| 697 |
+
"learning_rate": 3.0140233063401146e-06,
|
| 698 |
+
"loss": 0.0803,
|
| 699 |
+
"step": 43000
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 8.591744025281454,
|
| 703 |
+
"grad_norm": 1.1959236860275269,
|
| 704 |
+
"learning_rate": 2.816511949437093e-06,
|
| 705 |
+
"loss": 0.0808,
|
| 706 |
+
"step": 43500
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 8.690499703732964,
|
| 710 |
+
"grad_norm": 0.9853003025054932,
|
| 711 |
+
"learning_rate": 2.619000592534071e-06,
|
| 712 |
+
"loss": 0.0809,
|
| 713 |
+
"step": 44000
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 8.789255382184475,
|
| 717 |
+
"grad_norm": 0.8722342252731323,
|
| 718 |
+
"learning_rate": 2.421489235631049e-06,
|
| 719 |
+
"loss": 0.0812,
|
| 720 |
+
"step": 44500
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 8.888011060635986,
|
| 724 |
+
"grad_norm": 0.9437873959541321,
|
| 725 |
+
"learning_rate": 2.2239778787280267e-06,
|
| 726 |
+
"loss": 0.0806,
|
| 727 |
+
"step": 45000
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 8.986766739087498,
|
| 731 |
+
"grad_norm": 1.0116231441497803,
|
| 732 |
+
"learning_rate": 2.026466521825005e-06,
|
| 733 |
+
"loss": 0.0805,
|
| 734 |
+
"step": 45500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 9.0,
|
| 738 |
+
"eval_accuracy": 0.2628333333333333,
|
| 739 |
+
"eval_f1_micro": 0.3753562815172111,
|
| 740 |
+
"eval_loss": 0.13347476720809937,
|
| 741 |
+
"eval_roc_auc": 0.6523150114535268,
|
| 742 |
+
"eval_runtime": 39.603,
|
| 743 |
+
"eval_samples_per_second": 454.511,
|
| 744 |
+
"eval_steps_per_second": 14.216,
|
| 745 |
+
"step": 45567
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 9.085522417539009,
|
| 749 |
+
"grad_norm": 0.9861115217208862,
|
| 750 |
+
"learning_rate": 1.828955164921983e-06,
|
| 751 |
+
"loss": 0.0768,
|
| 752 |
+
"step": 46000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 9.18427809599052,
|
| 756 |
+
"grad_norm": 0.7694704532623291,
|
| 757 |
+
"learning_rate": 1.6314438080189613e-06,
|
| 758 |
+
"loss": 0.0771,
|
| 759 |
+
"step": 46500
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 9.28303377444203,
|
| 763 |
+
"grad_norm": 1.1088521480560303,
|
| 764 |
+
"learning_rate": 1.4339324511159391e-06,
|
| 765 |
+
"loss": 0.0775,
|
| 766 |
+
"step": 47000
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 9.38178945289354,
|
| 770 |
+
"grad_norm": 1.0436288118362427,
|
| 771 |
+
"learning_rate": 1.2364210942129174e-06,
|
| 772 |
+
"loss": 0.0769,
|
| 773 |
+
"step": 47500
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 9.480545131345052,
|
| 777 |
+
"grad_norm": 1.055021047592163,
|
| 778 |
+
"learning_rate": 1.0389097373098954e-06,
|
| 779 |
+
"loss": 0.0781,
|
| 780 |
+
"step": 48000
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 9.579300809796564,
|
| 784 |
+
"grad_norm": 1.1790095567703247,
|
| 785 |
+
"learning_rate": 8.413983804068734e-07,
|
| 786 |
+
"loss": 0.077,
|
| 787 |
+
"step": 48500
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 9.678056488248075,
|
| 791 |
+
"grad_norm": 0.8629137873649597,
|
| 792 |
+
"learning_rate": 6.438870235038514e-07,
|
| 793 |
+
"loss": 0.0783,
|
| 794 |
+
"step": 49000
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 9.776812166699585,
|
| 798 |
+
"grad_norm": 1.006084680557251,
|
| 799 |
+
"learning_rate": 4.463756666008296e-07,
|
| 800 |
+
"loss": 0.0766,
|
| 801 |
+
"step": 49500
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 9.875567845151096,
|
| 805 |
+
"grad_norm": 1.0732314586639404,
|
| 806 |
+
"learning_rate": 2.4886430969780764e-07,
|
| 807 |
+
"loss": 0.0774,
|
| 808 |
+
"step": 50000
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 9.974323523602607,
|
| 812 |
+
"grad_norm": 1.122566819190979,
|
| 813 |
+
"learning_rate": 5.1352952794785706e-08,
|
| 814 |
+
"loss": 0.0778,
|
| 815 |
+
"step": 50500
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 10.0,
|
| 819 |
+
"eval_accuracy": 0.2629444444444444,
|
| 820 |
+
"eval_f1_micro": 0.3750875491622218,
|
| 821 |
+
"eval_loss": 0.13676531612873077,
|
| 822 |
+
"eval_roc_auc": 0.6544563659537705,
|
| 823 |
+
"eval_runtime": 39.7687,
|
| 824 |
+
"eval_samples_per_second": 452.617,
|
| 825 |
+
"eval_steps_per_second": 14.157,
|
| 826 |
+
"step": 50630
|
| 827 |
+
}
|
| 828 |
+
],
|
| 829 |
+
"logging_steps": 500,
|
| 830 |
+
"max_steps": 50630,
|
| 831 |
+
"num_input_tokens_seen": 0,
|
| 832 |
+
"num_train_epochs": 10,
|
| 833 |
+
"save_steps": 500,
|
| 834 |
+
"stateful_callbacks": {
|
| 835 |
+
"TrainerControl": {
|
| 836 |
+
"args": {
|
| 837 |
+
"should_epoch_stop": false,
|
| 838 |
+
"should_evaluate": false,
|
| 839 |
+
"should_log": false,
|
| 840 |
+
"should_save": true,
|
| 841 |
+
"should_training_stop": true
|
| 842 |
+
},
|
| 843 |
+
"attributes": {}
|
| 844 |
+
}
|
| 845 |
+
},
|
| 846 |
+
"total_flos": 1.3843807083022118e+17,
|
| 847 |
+
"train_batch_size": 32,
|
| 848 |
+
"trial_name": null,
|
| 849 |
+
"trial_params": null
|
| 850 |
+
}
|