End of training
Browse files- all_results.json +17 -0
- eval_results.json +12 -0
- train_results.json +8 -0
- trainer_state.json +661 -0
all_results.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9862154900510105,
|
| 4 |
+
"eval_f1": 0.9181643400484828,
|
| 5 |
+
"eval_loss": 0.0572221502661705,
|
| 6 |
+
"eval_precision": 0.9121408403919614,
|
| 7 |
+
"eval_recall": 0.9242679232581622,
|
| 8 |
+
"eval_runtime": 20.7303,
|
| 9 |
+
"eval_samples": 3250,
|
| 10 |
+
"eval_samples_per_second": 156.775,
|
| 11 |
+
"eval_steps_per_second": 4.92,
|
| 12 |
+
"train_loss": 0.07309725869308391,
|
| 13 |
+
"train_runtime": 1743.2547,
|
| 14 |
+
"train_samples": 14041,
|
| 15 |
+
"train_samples_per_second": 24.163,
|
| 16 |
+
"train_steps_per_second": 0.755
|
| 17 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9862154900510105,
|
| 4 |
+
"eval_f1": 0.9181643400484828,
|
| 5 |
+
"eval_loss": 0.0572221502661705,
|
| 6 |
+
"eval_precision": 0.9121408403919614,
|
| 7 |
+
"eval_recall": 0.9242679232581622,
|
| 8 |
+
"eval_runtime": 20.7303,
|
| 9 |
+
"eval_samples": 3250,
|
| 10 |
+
"eval_samples_per_second": 156.775,
|
| 11 |
+
"eval_steps_per_second": 4.92
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.07309725869308391,
|
| 4 |
+
"train_runtime": 1743.2547,
|
| 5 |
+
"train_samples": 14041,
|
| 6 |
+
"train_samples_per_second": 24.163,
|
| 7 |
+
"train_steps_per_second": 0.755
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,661 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9181643400484828,
|
| 3 |
+
"best_model_checkpoint": "hmBERT-CoNLL-cp3/checkpoint-1000",
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"global_step": 1317,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"eval_accuracy": 0.9007242708617266,
|
| 13 |
+
"eval_f1": 0.3684910996506405,
|
| 14 |
+
"eval_loss": 0.4115295708179474,
|
| 15 |
+
"eval_precision": 0.36430921052631576,
|
| 16 |
+
"eval_recall": 0.37277011107371255,
|
| 17 |
+
"eval_runtime": 21.3063,
|
| 18 |
+
"eval_samples_per_second": 152.537,
|
| 19 |
+
"eval_steps_per_second": 4.787,
|
| 20 |
+
"step": 25
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.11,
|
| 24 |
+
"eval_accuracy": 0.9460301390132783,
|
| 25 |
+
"eval_f1": 0.6640782981476988,
|
| 26 |
+
"eval_loss": 0.22425369918346405,
|
| 27 |
+
"eval_precision": 0.6393085189222862,
|
| 28 |
+
"eval_recall": 0.6908448333894311,
|
| 29 |
+
"eval_runtime": 21.4481,
|
| 30 |
+
"eval_samples_per_second": 151.529,
|
| 31 |
+
"eval_steps_per_second": 4.756,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.17,
|
| 36 |
+
"eval_accuracy": 0.958023441454772,
|
| 37 |
+
"eval_f1": 0.7474880579805633,
|
| 38 |
+
"eval_loss": 0.16170601546764374,
|
| 39 |
+
"eval_precision": 0.7319354838709677,
|
| 40 |
+
"eval_recall": 0.7637159205654662,
|
| 41 |
+
"eval_runtime": 21.6197,
|
| 42 |
+
"eval_samples_per_second": 150.326,
|
| 43 |
+
"eval_steps_per_second": 4.718,
|
| 44 |
+
"step": 75
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.23,
|
| 48 |
+
"eval_accuracy": 0.9585101826252872,
|
| 49 |
+
"eval_f1": 0.7455232462625266,
|
| 50 |
+
"eval_loss": 0.15437664091587067,
|
| 51 |
+
"eval_precision": 0.7281771501925546,
|
| 52 |
+
"eval_recall": 0.7637159205654662,
|
| 53 |
+
"eval_runtime": 21.768,
|
| 54 |
+
"eval_samples_per_second": 149.302,
|
| 55 |
+
"eval_steps_per_second": 4.686,
|
| 56 |
+
"step": 100
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.28,
|
| 60 |
+
"eval_accuracy": 0.964409485611931,
|
| 61 |
+
"eval_f1": 0.784738041002278,
|
| 62 |
+
"eval_loss": 0.13407838344573975,
|
| 63 |
+
"eval_precision": 0.7595275590551182,
|
| 64 |
+
"eval_recall": 0.81167956916863,
|
| 65 |
+
"eval_runtime": 21.8222,
|
| 66 |
+
"eval_samples_per_second": 148.931,
|
| 67 |
+
"eval_steps_per_second": 4.674,
|
| 68 |
+
"step": 125
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.34,
|
| 72 |
+
"eval_accuracy": 0.9692963669639033,
|
| 73 |
+
"eval_f1": 0.811351977494622,
|
| 74 |
+
"eval_loss": 0.12208627909421921,
|
| 75 |
+
"eval_precision": 0.7980143229166666,
|
| 76 |
+
"eval_recall": 0.8251430494782901,
|
| 77 |
+
"eval_runtime": 21.2673,
|
| 78 |
+
"eval_samples_per_second": 152.817,
|
| 79 |
+
"eval_steps_per_second": 4.796,
|
| 80 |
+
"step": 150
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.4,
|
| 84 |
+
"eval_accuracy": 0.9718663603442234,
|
| 85 |
+
"eval_f1": 0.8151923709306149,
|
| 86 |
+
"eval_loss": 0.10127394646406174,
|
| 87 |
+
"eval_precision": 0.796849887495982,
|
| 88 |
+
"eval_recall": 0.8343991921911814,
|
| 89 |
+
"eval_runtime": 21.1083,
|
| 90 |
+
"eval_samples_per_second": 153.968,
|
| 91 |
+
"eval_steps_per_second": 4.832,
|
| 92 |
+
"step": 175
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.46,
|
| 96 |
+
"eval_accuracy": 0.9731513570343834,
|
| 97 |
+
"eval_f1": 0.8333472419260619,
|
| 98 |
+
"eval_loss": 0.10758877545595169,
|
| 99 |
+
"eval_precision": 0.8265187882800861,
|
| 100 |
+
"eval_recall": 0.8402894648266577,
|
| 101 |
+
"eval_runtime": 20.8538,
|
| 102 |
+
"eval_samples_per_second": 155.847,
|
| 103 |
+
"eval_steps_per_second": 4.891,
|
| 104 |
+
"step": 200
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.51,
|
| 108 |
+
"eval_accuracy": 0.9762859701725011,
|
| 109 |
+
"eval_f1": 0.8543123543123543,
|
| 110 |
+
"eval_loss": 0.08832722157239914,
|
| 111 |
+
"eval_precision": 0.8453047775947282,
|
| 112 |
+
"eval_recall": 0.8635139683608213,
|
| 113 |
+
"eval_runtime": 21.5789,
|
| 114 |
+
"eval_samples_per_second": 150.61,
|
| 115 |
+
"eval_steps_per_second": 4.727,
|
| 116 |
+
"step": 225
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.57,
|
| 120 |
+
"eval_accuracy": 0.9762665005256804,
|
| 121 |
+
"eval_f1": 0.8535063638632394,
|
| 122 |
+
"eval_loss": 0.0972680076956749,
|
| 123 |
+
"eval_precision": 0.8438887974995888,
|
| 124 |
+
"eval_recall": 0.8633456748569506,
|
| 125 |
+
"eval_runtime": 21.3825,
|
| 126 |
+
"eval_samples_per_second": 151.993,
|
| 127 |
+
"eval_steps_per_second": 4.77,
|
| 128 |
+
"step": 250
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.63,
|
| 132 |
+
"eval_accuracy": 0.9765390755811689,
|
| 133 |
+
"eval_f1": 0.8575239683201333,
|
| 134 |
+
"eval_loss": 0.0882844626903534,
|
| 135 |
+
"eval_precision": 0.8496613249628283,
|
| 136 |
+
"eval_recall": 0.8655334904072702,
|
| 137 |
+
"eval_runtime": 21.4561,
|
| 138 |
+
"eval_samples_per_second": 151.472,
|
| 139 |
+
"eval_steps_per_second": 4.754,
|
| 140 |
+
"step": 275
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.68,
|
| 144 |
+
"eval_accuracy": 0.9766364238152719,
|
| 145 |
+
"eval_f1": 0.8551207327227309,
|
| 146 |
+
"eval_loss": 0.08785738795995712,
|
| 147 |
+
"eval_precision": 0.8462425840474621,
|
| 148 |
+
"eval_recall": 0.8641871423763042,
|
| 149 |
+
"eval_runtime": 21.5171,
|
| 150 |
+
"eval_samples_per_second": 151.043,
|
| 151 |
+
"eval_steps_per_second": 4.74,
|
| 152 |
+
"step": 300
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.74,
|
| 156 |
+
"eval_accuracy": 0.9786612670846151,
|
| 157 |
+
"eval_f1": 0.8711310264708322,
|
| 158 |
+
"eval_loss": 0.07805178314447403,
|
| 159 |
+
"eval_precision": 0.8592240955966607,
|
| 160 |
+
"eval_recall": 0.8833726018175698,
|
| 161 |
+
"eval_runtime": 21.5676,
|
| 162 |
+
"eval_samples_per_second": 150.689,
|
| 163 |
+
"eval_steps_per_second": 4.729,
|
| 164 |
+
"step": 325
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.8,
|
| 168 |
+
"eval_accuracy": 0.9803161870643666,
|
| 169 |
+
"eval_f1": 0.8810828765985717,
|
| 170 |
+
"eval_loss": 0.07252655178308487,
|
| 171 |
+
"eval_precision": 0.8696721311475409,
|
| 172 |
+
"eval_recall": 0.8927970380343319,
|
| 173 |
+
"eval_runtime": 21.8654,
|
| 174 |
+
"eval_samples_per_second": 148.637,
|
| 175 |
+
"eval_steps_per_second": 4.665,
|
| 176 |
+
"step": 350
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.85,
|
| 180 |
+
"eval_accuracy": 0.9807250496475994,
|
| 181 |
+
"eval_f1": 0.881333443900821,
|
| 182 |
+
"eval_loss": 0.07547631114721298,
|
| 183 |
+
"eval_precision": 0.8687264999182606,
|
| 184 |
+
"eval_recall": 0.8943116795691687,
|
| 185 |
+
"eval_runtime": 21.5439,
|
| 186 |
+
"eval_samples_per_second": 150.855,
|
| 187 |
+
"eval_steps_per_second": 4.735,
|
| 188 |
+
"step": 375
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.91,
|
| 192 |
+
"eval_accuracy": 0.9822242124527861,
|
| 193 |
+
"eval_f1": 0.8890735355213959,
|
| 194 |
+
"eval_loss": 0.06660941988229752,
|
| 195 |
+
"eval_precision": 0.8780567864762843,
|
| 196 |
+
"eval_recall": 0.9003702457085156,
|
| 197 |
+
"eval_runtime": 21.2792,
|
| 198 |
+
"eval_samples_per_second": 152.731,
|
| 199 |
+
"eval_steps_per_second": 4.793,
|
| 200 |
+
"step": 400
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.97,
|
| 204 |
+
"eval_accuracy": 0.982282621393248,
|
| 205 |
+
"eval_f1": 0.8935885647412856,
|
| 206 |
+
"eval_loss": 0.06582659482955933,
|
| 207 |
+
"eval_precision": 0.8877262913137353,
|
| 208 |
+
"eval_recall": 0.8995287781891619,
|
| 209 |
+
"eval_runtime": 21.0258,
|
| 210 |
+
"eval_samples_per_second": 154.572,
|
| 211 |
+
"eval_steps_per_second": 4.851,
|
| 212 |
+
"step": 425
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.03,
|
| 216 |
+
"eval_accuracy": 0.983664966317511,
|
| 217 |
+
"eval_f1": 0.8993299832495812,
|
| 218 |
+
"eval_loss": 0.0645442008972168,
|
| 219 |
+
"eval_precision": 0.89513171057019,
|
| 220 |
+
"eval_recall": 0.9035678222820599,
|
| 221 |
+
"eval_runtime": 21.8717,
|
| 222 |
+
"eval_samples_per_second": 148.594,
|
| 223 |
+
"eval_steps_per_second": 4.664,
|
| 224 |
+
"step": 450
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.08,
|
| 228 |
+
"eval_accuracy": 0.9831003465597135,
|
| 229 |
+
"eval_f1": 0.8950920756603616,
|
| 230 |
+
"eval_loss": 0.06972972303628922,
|
| 231 |
+
"eval_precision": 0.8864499092259449,
|
| 232 |
+
"eval_recall": 0.9039044092898014,
|
| 233 |
+
"eval_runtime": 22.0283,
|
| 234 |
+
"eval_samples_per_second": 147.537,
|
| 235 |
+
"eval_steps_per_second": 4.63,
|
| 236 |
+
"step": 475
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.14,
|
| 240 |
+
"learning_rate": 3.1017463933181475e-05,
|
| 241 |
+
"loss": 0.1392,
|
| 242 |
+
"step": 500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.14,
|
| 246 |
+
"eval_accuracy": 0.9824189089209922,
|
| 247 |
+
"eval_f1": 0.8908151358559759,
|
| 248 |
+
"eval_loss": 0.06883182376623154,
|
| 249 |
+
"eval_precision": 0.8824306472919419,
|
| 250 |
+
"eval_recall": 0.8993604846852912,
|
| 251 |
+
"eval_runtime": 21.1436,
|
| 252 |
+
"eval_samples_per_second": 153.711,
|
| 253 |
+
"eval_steps_per_second": 4.824,
|
| 254 |
+
"step": 500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.2,
|
| 258 |
+
"eval_accuracy": 0.9827498929169425,
|
| 259 |
+
"eval_f1": 0.8999163179916319,
|
| 260 |
+
"eval_loss": 0.06805375963449478,
|
| 261 |
+
"eval_precision": 0.8949733688415447,
|
| 262 |
+
"eval_recall": 0.9049141703130259,
|
| 263 |
+
"eval_runtime": 21.618,
|
| 264 |
+
"eval_samples_per_second": 150.337,
|
| 265 |
+
"eval_steps_per_second": 4.718,
|
| 266 |
+
"step": 525
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.25,
|
| 270 |
+
"eval_accuracy": 0.9823215606868891,
|
| 271 |
+
"eval_f1": 0.8915259903058667,
|
| 272 |
+
"eval_loss": 0.06759364902973175,
|
| 273 |
+
"eval_precision": 0.8854581673306773,
|
| 274 |
+
"eval_recall": 0.8976775496465836,
|
| 275 |
+
"eval_runtime": 21.6174,
|
| 276 |
+
"eval_samples_per_second": 150.342,
|
| 277 |
+
"eval_steps_per_second": 4.718,
|
| 278 |
+
"step": 550
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.31,
|
| 282 |
+
"eval_accuracy": 0.9841906467816673,
|
| 283 |
+
"eval_f1": 0.9013520280420632,
|
| 284 |
+
"eval_loss": 0.06176742911338806,
|
| 285 |
+
"eval_precision": 0.8940397350993378,
|
| 286 |
+
"eval_recall": 0.9087849209020532,
|
| 287 |
+
"eval_runtime": 21.5727,
|
| 288 |
+
"eval_samples_per_second": 150.653,
|
| 289 |
+
"eval_steps_per_second": 4.728,
|
| 290 |
+
"step": 575
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.37,
|
| 294 |
+
"eval_accuracy": 0.984015419960282,
|
| 295 |
+
"eval_f1": 0.9010107760421018,
|
| 296 |
+
"eval_loss": 0.0644294023513794,
|
| 297 |
+
"eval_precision": 0.8945098689666612,
|
| 298 |
+
"eval_recall": 0.9076068663749579,
|
| 299 |
+
"eval_runtime": 22.0692,
|
| 300 |
+
"eval_samples_per_second": 147.264,
|
| 301 |
+
"eval_steps_per_second": 4.622,
|
| 302 |
+
"step": 600
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.42,
|
| 306 |
+
"eval_accuracy": 0.9837233752579728,
|
| 307 |
+
"eval_f1": 0.9010347129506008,
|
| 308 |
+
"eval_loss": 0.0641368106007576,
|
| 309 |
+
"eval_precision": 0.8935782853359815,
|
| 310 |
+
"eval_recall": 0.9086166273981824,
|
| 311 |
+
"eval_runtime": 21.6306,
|
| 312 |
+
"eval_samples_per_second": 150.25,
|
| 313 |
+
"eval_steps_per_second": 4.716,
|
| 314 |
+
"step": 625
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.48,
|
| 318 |
+
"eval_accuracy": 0.9845800397180795,
|
| 319 |
+
"eval_f1": 0.9043724966622163,
|
| 320 |
+
"eval_loss": 0.06189671903848648,
|
| 321 |
+
"eval_precision": 0.8968884475339292,
|
| 322 |
+
"eval_recall": 0.9119824974755975,
|
| 323 |
+
"eval_runtime": 20.932,
|
| 324 |
+
"eval_samples_per_second": 155.265,
|
| 325 |
+
"eval_steps_per_second": 4.873,
|
| 326 |
+
"step": 650
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.54,
|
| 330 |
+
"eval_accuracy": 0.9848136754799268,
|
| 331 |
+
"eval_f1": 0.9074897257401661,
|
| 332 |
+
"eval_loss": 0.06076198071241379,
|
| 333 |
+
"eval_precision": 0.9045310148804547,
|
| 334 |
+
"eval_recall": 0.9104678559407607,
|
| 335 |
+
"eval_runtime": 21.5144,
|
| 336 |
+
"eval_samples_per_second": 151.062,
|
| 337 |
+
"eval_steps_per_second": 4.741,
|
| 338 |
+
"step": 675
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.59,
|
| 342 |
+
"eval_accuracy": 0.9851057201822359,
|
| 343 |
+
"eval_f1": 0.9090604869070525,
|
| 344 |
+
"eval_loss": 0.06238234415650368,
|
| 345 |
+
"eval_precision": 0.903842954583264,
|
| 346 |
+
"eval_recall": 0.9143386065297879,
|
| 347 |
+
"eval_runtime": 21.3845,
|
| 348 |
+
"eval_samples_per_second": 151.979,
|
| 349 |
+
"eval_steps_per_second": 4.77,
|
| 350 |
+
"step": 700
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.65,
|
| 354 |
+
"eval_accuracy": 0.9852420077099802,
|
| 355 |
+
"eval_f1": 0.9115851108322878,
|
| 356 |
+
"eval_loss": 0.05955210700631142,
|
| 357 |
+
"eval_precision": 0.9062032263429237,
|
| 358 |
+
"eval_recall": 0.9170313025917199,
|
| 359 |
+
"eval_runtime": 21.1164,
|
| 360 |
+
"eval_samples_per_second": 153.909,
|
| 361 |
+
"eval_steps_per_second": 4.83,
|
| 362 |
+
"step": 725
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 1.71,
|
| 366 |
+
"eval_accuracy": 0.9848331451267474,
|
| 367 |
+
"eval_f1": 0.9068602904356534,
|
| 368 |
+
"eval_loss": 0.0579909048974514,
|
| 369 |
+
"eval_precision": 0.8995033112582781,
|
| 370 |
+
"eval_recall": 0.9143386065297879,
|
| 371 |
+
"eval_runtime": 21.1108,
|
| 372 |
+
"eval_samples_per_second": 153.95,
|
| 373 |
+
"eval_steps_per_second": 4.832,
|
| 374 |
+
"step": 750
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.77,
|
| 378 |
+
"eval_accuracy": 0.9857676881741365,
|
| 379 |
+
"eval_f1": 0.9126685087498954,
|
| 380 |
+
"eval_loss": 0.05822019651532173,
|
| 381 |
+
"eval_precision": 0.9081819696717214,
|
| 382 |
+
"eval_recall": 0.9171995960955908,
|
| 383 |
+
"eval_runtime": 22.2525,
|
| 384 |
+
"eval_samples_per_second": 146.051,
|
| 385 |
+
"eval_steps_per_second": 4.584,
|
| 386 |
+
"step": 775
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.82,
|
| 390 |
+
"eval_accuracy": 0.9852420077099802,
|
| 391 |
+
"eval_f1": 0.9100617386951443,
|
| 392 |
+
"eval_loss": 0.0588374100625515,
|
| 393 |
+
"eval_precision": 0.9023825281270682,
|
| 394 |
+
"eval_recall": 0.9178727701110737,
|
| 395 |
+
"eval_runtime": 21.7438,
|
| 396 |
+
"eval_samples_per_second": 149.468,
|
| 397 |
+
"eval_steps_per_second": 4.691,
|
| 398 |
+
"step": 800
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.88,
|
| 402 |
+
"eval_accuracy": 0.9855535220591098,
|
| 403 |
+
"eval_f1": 0.911860174781523,
|
| 404 |
+
"eval_loss": 0.0592406764626503,
|
| 405 |
+
"eval_precision": 0.9020253581425984,
|
| 406 |
+
"eval_recall": 0.9219118142039717,
|
| 407 |
+
"eval_runtime": 21.1485,
|
| 408 |
+
"eval_samples_per_second": 153.675,
|
| 409 |
+
"eval_steps_per_second": 4.823,
|
| 410 |
+
"step": 825
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.94,
|
| 414 |
+
"eval_accuracy": 0.9852225380631595,
|
| 415 |
+
"eval_f1": 0.911764705882353,
|
| 416 |
+
"eval_loss": 0.06002892181277275,
|
| 417 |
+
"eval_precision": 0.9054098904746101,
|
| 418 |
+
"eval_recall": 0.9182093571188152,
|
| 419 |
+
"eval_runtime": 22.0668,
|
| 420 |
+
"eval_samples_per_second": 147.28,
|
| 421 |
+
"eval_steps_per_second": 4.622,
|
| 422 |
+
"step": 850
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.99,
|
| 426 |
+
"eval_accuracy": 0.9860792025232662,
|
| 427 |
+
"eval_f1": 0.913464751085867,
|
| 428 |
+
"eval_loss": 0.05682874843478203,
|
| 429 |
+
"eval_precision": 0.9067993366500829,
|
| 430 |
+
"eval_recall": 0.9202288791652642,
|
| 431 |
+
"eval_runtime": 20.9827,
|
| 432 |
+
"eval_samples_per_second": 154.89,
|
| 433 |
+
"eval_steps_per_second": 4.861,
|
| 434 |
+
"step": 875
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.05,
|
| 438 |
+
"eval_accuracy": 0.9860792025232662,
|
| 439 |
+
"eval_f1": 0.9171483622350675,
|
| 440 |
+
"eval_loss": 0.0570731945335865,
|
| 441 |
+
"eval_precision": 0.9130942452043369,
|
| 442 |
+
"eval_recall": 0.9212386401884888,
|
| 443 |
+
"eval_runtime": 21.3364,
|
| 444 |
+
"eval_samples_per_second": 152.322,
|
| 445 |
+
"eval_steps_per_second": 4.781,
|
| 446 |
+
"step": 900
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.11,
|
| 450 |
+
"eval_accuracy": 0.9857676881741365,
|
| 451 |
+
"eval_f1": 0.9156969443281708,
|
| 452 |
+
"eval_loss": 0.0577477402985096,
|
| 453 |
+
"eval_precision": 0.9110444777611194,
|
| 454 |
+
"eval_recall": 0.920397172669135,
|
| 455 |
+
"eval_runtime": 21.9855,
|
| 456 |
+
"eval_samples_per_second": 147.825,
|
| 457 |
+
"eval_steps_per_second": 4.639,
|
| 458 |
+
"step": 925
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.16,
|
| 462 |
+
"eval_accuracy": 0.986040263229625,
|
| 463 |
+
"eval_f1": 0.9184714441006773,
|
| 464 |
+
"eval_loss": 0.06048833578824997,
|
| 465 |
+
"eval_precision": 0.912747216220708,
|
| 466 |
+
"eval_recall": 0.9242679232581622,
|
| 467 |
+
"eval_runtime": 21.8831,
|
| 468 |
+
"eval_samples_per_second": 148.516,
|
| 469 |
+
"eval_steps_per_second": 4.661,
|
| 470 |
+
"step": 950
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.22,
|
| 474 |
+
"eval_accuracy": 0.986682761574705,
|
| 475 |
+
"eval_f1": 0.9166318254034618,
|
| 476 |
+
"eval_loss": 0.05745207890868187,
|
| 477 |
+
"eval_precision": 0.9109190626558086,
|
| 478 |
+
"eval_recall": 0.922416694715584,
|
| 479 |
+
"eval_runtime": 21.8724,
|
| 480 |
+
"eval_samples_per_second": 148.589,
|
| 481 |
+
"eval_steps_per_second": 4.663,
|
| 482 |
+
"step": 975
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.28,
|
| 486 |
+
"learning_rate": 1.2034927866362947e-05,
|
| 487 |
+
"loss": 0.0392,
|
| 488 |
+
"step": 1000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.28,
|
| 492 |
+
"eval_accuracy": 0.9862154900510105,
|
| 493 |
+
"eval_f1": 0.9181643400484828,
|
| 494 |
+
"eval_loss": 0.0572221502661705,
|
| 495 |
+
"eval_precision": 0.9121408403919614,
|
| 496 |
+
"eval_recall": 0.9242679232581622,
|
| 497 |
+
"eval_runtime": 21.5878,
|
| 498 |
+
"eval_samples_per_second": 150.548,
|
| 499 |
+
"eval_steps_per_second": 4.725,
|
| 500 |
+
"step": 1000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.33,
|
| 504 |
+
"eval_accuracy": 0.9869748062770142,
|
| 505 |
+
"eval_f1": 0.9211694730669348,
|
| 506 |
+
"eval_loss": 0.056650131940841675,
|
| 507 |
+
"eval_precision": 0.9170975813177648,
|
| 508 |
+
"eval_recall": 0.9252776842813868,
|
| 509 |
+
"eval_runtime": 21.5295,
|
| 510 |
+
"eval_samples_per_second": 150.956,
|
| 511 |
+
"eval_steps_per_second": 4.738,
|
| 512 |
+
"step": 1025
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.39,
|
| 516 |
+
"eval_accuracy": 0.9870916241579378,
|
| 517 |
+
"eval_f1": 0.9243514644351465,
|
| 518 |
+
"eval_loss": 0.0570232979953289,
|
| 519 |
+
"eval_precision": 0.9192743009320905,
|
| 520 |
+
"eval_recall": 0.9294850218781555,
|
| 521 |
+
"eval_runtime": 21.5481,
|
| 522 |
+
"eval_samples_per_second": 150.825,
|
| 523 |
+
"eval_steps_per_second": 4.734,
|
| 524 |
+
"step": 1050
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.45,
|
| 528 |
+
"eval_accuracy": 0.986682761574705,
|
| 529 |
+
"eval_f1": 0.9215079829474212,
|
| 530 |
+
"eval_loss": 0.05837239325046539,
|
| 531 |
+
"eval_precision": 0.9154625477495433,
|
| 532 |
+
"eval_recall": 0.9276337933355773,
|
| 533 |
+
"eval_runtime": 21.2288,
|
| 534 |
+
"eval_samples_per_second": 153.094,
|
| 535 |
+
"eval_steps_per_second": 4.805,
|
| 536 |
+
"step": 1075
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.51,
|
| 540 |
+
"eval_accuracy": 0.9867022312215257,
|
| 541 |
+
"eval_f1": 0.9226653289858708,
|
| 542 |
+
"eval_loss": 0.059072330594062805,
|
| 543 |
+
"eval_precision": 0.9167635819903639,
|
| 544 |
+
"eval_recall": 0.9286435543588017,
|
| 545 |
+
"eval_runtime": 22.3224,
|
| 546 |
+
"eval_samples_per_second": 145.594,
|
| 547 |
+
"eval_steps_per_second": 4.569,
|
| 548 |
+
"step": 1100
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.56,
|
| 552 |
+
"eval_accuracy": 0.9874031385070675,
|
| 553 |
+
"eval_f1": 0.9246323529411764,
|
| 554 |
+
"eval_loss": 0.057747144252061844,
|
| 555 |
+
"eval_precision": 0.9181878526385662,
|
| 556 |
+
"eval_recall": 0.931167956916863,
|
| 557 |
+
"eval_runtime": 21.9739,
|
| 558 |
+
"eval_samples_per_second": 147.903,
|
| 559 |
+
"eval_steps_per_second": 4.642,
|
| 560 |
+
"step": 1125
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.62,
|
| 564 |
+
"eval_accuracy": 0.987033215217476,
|
| 565 |
+
"eval_f1": 0.9233344492802142,
|
| 566 |
+
"eval_loss": 0.057037804275751114,
|
| 567 |
+
"eval_precision": 0.9184149184149184,
|
| 568 |
+
"eval_recall": 0.9283069673510602,
|
| 569 |
+
"eval_runtime": 21.3838,
|
| 570 |
+
"eval_samples_per_second": 151.984,
|
| 571 |
+
"eval_steps_per_second": 4.77,
|
| 572 |
+
"step": 1150
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.68,
|
| 576 |
+
"eval_accuracy": 0.987227911685682,
|
| 577 |
+
"eval_f1": 0.9244541119384256,
|
| 578 |
+
"eval_loss": 0.05629764869809151,
|
| 579 |
+
"eval_precision": 0.9191482282482116,
|
| 580 |
+
"eval_recall": 0.929821608885897,
|
| 581 |
+
"eval_runtime": 21.8946,
|
| 582 |
+
"eval_samples_per_second": 148.438,
|
| 583 |
+
"eval_steps_per_second": 4.659,
|
| 584 |
+
"step": 1175
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.73,
|
| 588 |
+
"eval_accuracy": 0.9871500330983995,
|
| 589 |
+
"eval_f1": 0.9246449456975774,
|
| 590 |
+
"eval_loss": 0.05648628994822502,
|
| 591 |
+
"eval_precision": 0.918049104180491,
|
| 592 |
+
"eval_recall": 0.9313362504207338,
|
| 593 |
+
"eval_runtime": 21.2748,
|
| 594 |
+
"eval_samples_per_second": 152.763,
|
| 595 |
+
"eval_steps_per_second": 4.794,
|
| 596 |
+
"step": 1200
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.79,
|
| 600 |
+
"eval_accuracy": 0.9873447295666057,
|
| 601 |
+
"eval_f1": 0.9243767776476494,
|
| 602 |
+
"eval_loss": 0.055928729474544525,
|
| 603 |
+
"eval_precision": 0.9189953426480373,
|
| 604 |
+
"eval_recall": 0.929821608885897,
|
| 605 |
+
"eval_runtime": 21.5883,
|
| 606 |
+
"eval_samples_per_second": 150.545,
|
| 607 |
+
"eval_steps_per_second": 4.725,
|
| 608 |
+
"step": 1225
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.85,
|
| 612 |
+
"eval_accuracy": 0.9873057902729645,
|
| 613 |
+
"eval_f1": 0.9238748536054878,
|
| 614 |
+
"eval_loss": 0.05620228126645088,
|
| 615 |
+
"eval_precision": 0.9184963406520292,
|
| 616 |
+
"eval_recall": 0.9293167283742848,
|
| 617 |
+
"eval_runtime": 21.3558,
|
| 618 |
+
"eval_samples_per_second": 152.184,
|
| 619 |
+
"eval_steps_per_second": 4.776,
|
| 620 |
+
"step": 1250
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 2.9,
|
| 624 |
+
"eval_accuracy": 0.9871500330983995,
|
| 625 |
+
"eval_f1": 0.922961104140527,
|
| 626 |
+
"eval_loss": 0.05639491230249405,
|
| 627 |
+
"eval_precision": 0.9175120572093797,
|
| 628 |
+
"eval_recall": 0.928475260854931,
|
| 629 |
+
"eval_runtime": 21.3856,
|
| 630 |
+
"eval_samples_per_second": 151.972,
|
| 631 |
+
"eval_steps_per_second": 4.77,
|
| 632 |
+
"step": 1275
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 2.96,
|
| 636 |
+
"eval_accuracy": 0.9873057902729645,
|
| 637 |
+
"eval_f1": 0.9237330657300552,
|
| 638 |
+
"eval_loss": 0.05629625543951988,
|
| 639 |
+
"eval_precision": 0.9180518617021277,
|
| 640 |
+
"eval_recall": 0.9294850218781555,
|
| 641 |
+
"eval_runtime": 21.6582,
|
| 642 |
+
"eval_samples_per_second": 150.059,
|
| 643 |
+
"eval_steps_per_second": 4.71,
|
| 644 |
+
"step": 1300
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 3.0,
|
| 648 |
+
"step": 1317,
|
| 649 |
+
"total_flos": 1368370122693444.0,
|
| 650 |
+
"train_loss": 0.07309725869308391,
|
| 651 |
+
"train_runtime": 1743.2547,
|
| 652 |
+
"train_samples_per_second": 24.163,
|
| 653 |
+
"train_steps_per_second": 0.755
|
| 654 |
+
}
|
| 655 |
+
],
|
| 656 |
+
"max_steps": 1317,
|
| 657 |
+
"num_train_epochs": 3,
|
| 658 |
+
"total_flos": 1368370122693444.0,
|
| 659 |
+
"trial_name": null,
|
| 660 |
+
"trial_params": null
|
| 661 |
+
}
|