End of training
Browse files- all_results.json +6 -0
- test_results.json +6 -0
- trainer_state.json +740 -0
all_results.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"accuracy": 0.9153605015673981,
|
| 3 |
+
"f1": 0.913735921309191,
|
| 4 |
+
"precision": 0.9325578764839285,
|
| 5 |
+
"recall": 0.9153605015673981
|
| 6 |
+
}
|
test_results.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"accuracy": 0.9153605015673981,
|
| 3 |
+
"f1": 0.913735921309191,
|
| 4 |
+
"precision": 0.9325578764839285,
|
| 5 |
+
"recall": 0.9153605015673981
|
| 6 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,740 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9766666666666667,
|
| 3 |
+
"best_model_checkpoint": "/media/cse/HDD/Shawon/shawon/MY DATA/VideoMAE_BdSLW60_SR_8_kineticsFinetuned_withoutAug/checkpoint-3716",
|
| 4 |
+
"epoch": 8.050053879310346,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 8361,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.005387931034482759,
|
| 13 |
+
"grad_norm": 20.11014747619629,
|
| 14 |
+
"learning_rate": 2.6670258620689655e-06,
|
| 15 |
+
"loss": 16.4554,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.010775862068965518,
|
| 20 |
+
"grad_norm": 26.971817016601562,
|
| 21 |
+
"learning_rate": 5.360991379310345e-06,
|
| 22 |
+
"loss": 16.1745,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.016163793103448277,
|
| 27 |
+
"grad_norm": 41.51228713989258,
|
| 28 |
+
"learning_rate": 8.028017241379311e-06,
|
| 29 |
+
"loss": 15.3342,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.021551724137931036,
|
| 34 |
+
"grad_norm": 51.878631591796875,
|
| 35 |
+
"learning_rate": 1.072198275862069e-05,
|
| 36 |
+
"loss": 13.6057,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.02693965517241379,
|
| 41 |
+
"grad_norm": 52.984981536865234,
|
| 42 |
+
"learning_rate": 1.3415948275862069e-05,
|
| 43 |
+
"loss": 11.1741,
|
| 44 |
+
"step": 500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.032327586206896554,
|
| 48 |
+
"grad_norm": 50.443359375,
|
| 49 |
+
"learning_rate": 1.610991379310345e-05,
|
| 50 |
+
"loss": 8.8003,
|
| 51 |
+
"step": 600
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.03771551724137931,
|
| 55 |
+
"grad_norm": 50.93900680541992,
|
| 56 |
+
"learning_rate": 1.880387931034483e-05,
|
| 57 |
+
"loss": 6.2116,
|
| 58 |
+
"step": 700
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.04310344827586207,
|
| 62 |
+
"grad_norm": 40.679100036621094,
|
| 63 |
+
"learning_rate": 2.149784482758621e-05,
|
| 64 |
+
"loss": 4.3287,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.04849137931034483,
|
| 69 |
+
"grad_norm": 37.3469123840332,
|
| 70 |
+
"learning_rate": 2.4191810344827586e-05,
|
| 71 |
+
"loss": 2.859,
|
| 72 |
+
"step": 900
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.05005387931034483,
|
| 76 |
+
"eval_accuracy": 0.9183333333333333,
|
| 77 |
+
"eval_f1": 0.9087385884819826,
|
| 78 |
+
"eval_loss": 0.8475048542022705,
|
| 79 |
+
"eval_precision": 0.9240348890875206,
|
| 80 |
+
"eval_recall": 0.9183333333333333,
|
| 81 |
+
"eval_runtime": 192.0614,
|
| 82 |
+
"eval_samples_per_second": 3.124,
|
| 83 |
+
"eval_steps_per_second": 1.562,
|
| 84 |
+
"step": 929
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 1.0038254310344827,
|
| 88 |
+
"grad_norm": 19.420085906982422,
|
| 89 |
+
"learning_rate": 2.688577586206897e-05,
|
| 90 |
+
"loss": 1.5587,
|
| 91 |
+
"step": 1000
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 1.0092133620689656,
|
| 95 |
+
"grad_norm": 17.20052146911621,
|
| 96 |
+
"learning_rate": 2.9579741379310345e-05,
|
| 97 |
+
"loss": 0.9491,
|
| 98 |
+
"step": 1100
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.0146012931034483,
|
| 102 |
+
"grad_norm": 24.49296760559082,
|
| 103 |
+
"learning_rate": 3.2273706896551725e-05,
|
| 104 |
+
"loss": 0.6366,
|
| 105 |
+
"step": 1200
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.0199892241379311,
|
| 109 |
+
"grad_norm": 43.13790512084961,
|
| 110 |
+
"learning_rate": 3.496767241379311e-05,
|
| 111 |
+
"loss": 0.4445,
|
| 112 |
+
"step": 1300
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.0253771551724138,
|
| 116 |
+
"grad_norm": 1.7655223608016968,
|
| 117 |
+
"learning_rate": 3.7661637931034485e-05,
|
| 118 |
+
"loss": 0.2924,
|
| 119 |
+
"step": 1400
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.0307650862068964,
|
| 123 |
+
"grad_norm": 2.0372657775878906,
|
| 124 |
+
"learning_rate": 4.035560344827586e-05,
|
| 125 |
+
"loss": 0.1758,
|
| 126 |
+
"step": 1500
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 1.0361530172413793,
|
| 130 |
+
"grad_norm": 1.5986464023590088,
|
| 131 |
+
"learning_rate": 4.3049568965517245e-05,
|
| 132 |
+
"loss": 0.2277,
|
| 133 |
+
"step": 1600
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 1.041540948275862,
|
| 137 |
+
"grad_norm": 1.0040245056152344,
|
| 138 |
+
"learning_rate": 4.574353448275862e-05,
|
| 139 |
+
"loss": 0.1491,
|
| 140 |
+
"step": 1700
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.0469288793103448,
|
| 144 |
+
"grad_norm": 2.6459920406341553,
|
| 145 |
+
"learning_rate": 4.8437500000000005e-05,
|
| 146 |
+
"loss": 0.0555,
|
| 147 |
+
"step": 1800
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 1.0500538793103449,
|
| 151 |
+
"eval_accuracy": 0.9216666666666666,
|
| 152 |
+
"eval_f1": 0.9093368000230129,
|
| 153 |
+
"eval_loss": 0.30231231451034546,
|
| 154 |
+
"eval_precision": 0.9317002164502166,
|
| 155 |
+
"eval_recall": 0.9216666666666666,
|
| 156 |
+
"eval_runtime": 187.0802,
|
| 157 |
+
"eval_samples_per_second": 3.207,
|
| 158 |
+
"eval_steps_per_second": 1.604,
|
| 159 |
+
"step": 1858
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 2.002262931034483,
|
| 163 |
+
"grad_norm": 0.5840566158294678,
|
| 164 |
+
"learning_rate": 4.9874281609195405e-05,
|
| 165 |
+
"loss": 0.0647,
|
| 166 |
+
"step": 1900
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 2.0076508620689655,
|
| 170 |
+
"grad_norm": 0.16274438798427582,
|
| 171 |
+
"learning_rate": 4.95749521072797e-05,
|
| 172 |
+
"loss": 0.0725,
|
| 173 |
+
"step": 2000
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 2.013038793103448,
|
| 177 |
+
"grad_norm": 0.10402801632881165,
|
| 178 |
+
"learning_rate": 4.9275622605363985e-05,
|
| 179 |
+
"loss": 0.0139,
|
| 180 |
+
"step": 2100
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 2.0184267241379312,
|
| 184 |
+
"grad_norm": 0.0963917076587677,
|
| 185 |
+
"learning_rate": 4.897629310344828e-05,
|
| 186 |
+
"loss": 0.0206,
|
| 187 |
+
"step": 2200
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 2.023814655172414,
|
| 191 |
+
"grad_norm": 0.08095109462738037,
|
| 192 |
+
"learning_rate": 4.867696360153257e-05,
|
| 193 |
+
"loss": 0.0499,
|
| 194 |
+
"step": 2300
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.0292025862068965,
|
| 198 |
+
"grad_norm": 1.0376495122909546,
|
| 199 |
+
"learning_rate": 4.837763409961686e-05,
|
| 200 |
+
"loss": 0.0105,
|
| 201 |
+
"step": 2400
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.034590517241379,
|
| 205 |
+
"grad_norm": 0.027790505439043045,
|
| 206 |
+
"learning_rate": 4.808129789272031e-05,
|
| 207 |
+
"loss": 0.0226,
|
| 208 |
+
"step": 2500
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 2.0399784482758623,
|
| 212 |
+
"grad_norm": 0.031076578423380852,
|
| 213 |
+
"learning_rate": 4.778496168582376e-05,
|
| 214 |
+
"loss": 0.0359,
|
| 215 |
+
"step": 2600
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 2.045366379310345,
|
| 219 |
+
"grad_norm": 0.02143680490553379,
|
| 220 |
+
"learning_rate": 4.748563218390804e-05,
|
| 221 |
+
"loss": 0.0025,
|
| 222 |
+
"step": 2700
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 2.0500538793103447,
|
| 226 |
+
"eval_accuracy": 0.935,
|
| 227 |
+
"eval_f1": 0.927261710739568,
|
| 228 |
+
"eval_loss": 0.2999415695667267,
|
| 229 |
+
"eval_precision": 0.9378208874458874,
|
| 230 |
+
"eval_recall": 0.935,
|
| 231 |
+
"eval_runtime": 186.5359,
|
| 232 |
+
"eval_samples_per_second": 3.217,
|
| 233 |
+
"eval_steps_per_second": 1.608,
|
| 234 |
+
"step": 2787
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 3.000700431034483,
|
| 238 |
+
"grad_norm": 0.044046301394701004,
|
| 239 |
+
"learning_rate": 4.7186302681992336e-05,
|
| 240 |
+
"loss": 0.2036,
|
| 241 |
+
"step": 2800
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 3.0060883620689656,
|
| 245 |
+
"grad_norm": 0.02674109861254692,
|
| 246 |
+
"learning_rate": 4.688697318007663e-05,
|
| 247 |
+
"loss": 0.0227,
|
| 248 |
+
"step": 2900
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.011476293103448,
|
| 252 |
+
"grad_norm": 0.604284405708313,
|
| 253 |
+
"learning_rate": 4.658764367816092e-05,
|
| 254 |
+
"loss": 0.0018,
|
| 255 |
+
"step": 3000
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.016864224137931,
|
| 259 |
+
"grad_norm": 0.012123900465667248,
|
| 260 |
+
"learning_rate": 4.6288314176245215e-05,
|
| 261 |
+
"loss": 0.0013,
|
| 262 |
+
"step": 3100
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 3.022252155172414,
|
| 266 |
+
"grad_norm": 0.17276327311992645,
|
| 267 |
+
"learning_rate": 4.598898467432951e-05,
|
| 268 |
+
"loss": 0.0304,
|
| 269 |
+
"step": 3200
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.0276400862068966,
|
| 273 |
+
"grad_norm": 0.01406328845769167,
|
| 274 |
+
"learning_rate": 4.5689655172413794e-05,
|
| 275 |
+
"loss": 0.0529,
|
| 276 |
+
"step": 3300
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 3.0330280172413793,
|
| 280 |
+
"grad_norm": 0.026818769052624702,
|
| 281 |
+
"learning_rate": 4.539032567049809e-05,
|
| 282 |
+
"loss": 0.0059,
|
| 283 |
+
"step": 3400
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 3.038415948275862,
|
| 287 |
+
"grad_norm": 0.022273056209087372,
|
| 288 |
+
"learning_rate": 4.509099616858238e-05,
|
| 289 |
+
"loss": 0.0431,
|
| 290 |
+
"step": 3500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 3.043803879310345,
|
| 294 |
+
"grad_norm": 0.008690926246345043,
|
| 295 |
+
"learning_rate": 4.4791666666666673e-05,
|
| 296 |
+
"loss": 0.001,
|
| 297 |
+
"step": 3600
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 3.0491918103448277,
|
| 301 |
+
"grad_norm": 0.009022524580359459,
|
| 302 |
+
"learning_rate": 4.449233716475096e-05,
|
| 303 |
+
"loss": 0.0009,
|
| 304 |
+
"step": 3700
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 3.0500538793103447,
|
| 308 |
+
"eval_accuracy": 0.9766666666666667,
|
| 309 |
+
"eval_f1": 0.9761724029000916,
|
| 310 |
+
"eval_loss": 0.08761004358530045,
|
| 311 |
+
"eval_precision": 0.9809054834054833,
|
| 312 |
+
"eval_recall": 0.9766666666666667,
|
| 313 |
+
"eval_runtime": 184.6188,
|
| 314 |
+
"eval_samples_per_second": 3.25,
|
| 315 |
+
"eval_steps_per_second": 1.625,
|
| 316 |
+
"step": 3716
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 4.004525862068966,
|
| 320 |
+
"grad_norm": 0.007579015102237463,
|
| 321 |
+
"learning_rate": 4.419300766283525e-05,
|
| 322 |
+
"loss": 0.0007,
|
| 323 |
+
"step": 3800
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 4.009913793103448,
|
| 327 |
+
"grad_norm": 0.010904570110142231,
|
| 328 |
+
"learning_rate": 4.3893678160919546e-05,
|
| 329 |
+
"loss": 0.0395,
|
| 330 |
+
"step": 3900
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 4.015301724137931,
|
| 334 |
+
"grad_norm": 0.006146453786641359,
|
| 335 |
+
"learning_rate": 4.359434865900383e-05,
|
| 336 |
+
"loss": 0.0006,
|
| 337 |
+
"step": 4000
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 4.020689655172414,
|
| 341 |
+
"grad_norm": 0.013002044521272182,
|
| 342 |
+
"learning_rate": 4.3295019157088125e-05,
|
| 343 |
+
"loss": 0.0011,
|
| 344 |
+
"step": 4100
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 4.026077586206896,
|
| 348 |
+
"grad_norm": 0.12733735144138336,
|
| 349 |
+
"learning_rate": 4.299568965517242e-05,
|
| 350 |
+
"loss": 0.0451,
|
| 351 |
+
"step": 4200
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 4.031465517241379,
|
| 355 |
+
"grad_norm": 11.122708320617676,
|
| 356 |
+
"learning_rate": 4.269636015325671e-05,
|
| 357 |
+
"loss": 0.305,
|
| 358 |
+
"step": 4300
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 4.0368534482758625,
|
| 362 |
+
"grad_norm": 0.13978660106658936,
|
| 363 |
+
"learning_rate": 4.2397030651341e-05,
|
| 364 |
+
"loss": 0.2215,
|
| 365 |
+
"step": 4400
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 4.042241379310345,
|
| 369 |
+
"grad_norm": 0.37221765518188477,
|
| 370 |
+
"learning_rate": 4.209770114942529e-05,
|
| 371 |
+
"loss": 0.1446,
|
| 372 |
+
"step": 4500
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 4.047629310344828,
|
| 376 |
+
"grad_norm": 20.41011619567871,
|
| 377 |
+
"learning_rate": 4.1801364942528733e-05,
|
| 378 |
+
"loss": 0.1355,
|
| 379 |
+
"step": 4600
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 4.050053879310345,
|
| 383 |
+
"eval_accuracy": 0.9266666666666666,
|
| 384 |
+
"eval_f1": 0.9146410798064111,
|
| 385 |
+
"eval_loss": 0.25398045778274536,
|
| 386 |
+
"eval_precision": 0.9182419154919156,
|
| 387 |
+
"eval_recall": 0.9266666666666666,
|
| 388 |
+
"eval_runtime": 176.3808,
|
| 389 |
+
"eval_samples_per_second": 3.402,
|
| 390 |
+
"eval_steps_per_second": 1.701,
|
| 391 |
+
"step": 4645
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 5.002963362068965,
|
| 395 |
+
"grad_norm": 0.017676187679171562,
|
| 396 |
+
"learning_rate": 4.1502035440613026e-05,
|
| 397 |
+
"loss": 0.1491,
|
| 398 |
+
"step": 4700
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 5.008351293103448,
|
| 402 |
+
"grad_norm": 1.5707539319992065,
|
| 403 |
+
"learning_rate": 4.120270593869732e-05,
|
| 404 |
+
"loss": 0.0298,
|
| 405 |
+
"step": 4800
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 5.013739224137931,
|
| 409 |
+
"grad_norm": 0.08283871412277222,
|
| 410 |
+
"learning_rate": 4.090337643678161e-05,
|
| 411 |
+
"loss": 0.1471,
|
| 412 |
+
"step": 4900
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 5.019127155172414,
|
| 416 |
+
"grad_norm": 0.0916813537478447,
|
| 417 |
+
"learning_rate": 4.0604046934865905e-05,
|
| 418 |
+
"loss": 0.1646,
|
| 419 |
+
"step": 5000
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 5.024515086206897,
|
| 423 |
+
"grad_norm": 0.012382575310766697,
|
| 424 |
+
"learning_rate": 4.03047174329502e-05,
|
| 425 |
+
"loss": 0.015,
|
| 426 |
+
"step": 5100
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 5.029903017241379,
|
| 430 |
+
"grad_norm": 0.008144189603626728,
|
| 431 |
+
"learning_rate": 4.000838122605364e-05,
|
| 432 |
+
"loss": 0.1365,
|
| 433 |
+
"step": 5200
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 5.035290948275862,
|
| 437 |
+
"grad_norm": 0.021491670981049538,
|
| 438 |
+
"learning_rate": 3.9709051724137935e-05,
|
| 439 |
+
"loss": 0.0302,
|
| 440 |
+
"step": 5300
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 5.040678879310345,
|
| 444 |
+
"grad_norm": 0.03770313411951065,
|
| 445 |
+
"learning_rate": 3.940972222222222e-05,
|
| 446 |
+
"loss": 0.1047,
|
| 447 |
+
"step": 5400
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 5.046066810344827,
|
| 451 |
+
"grad_norm": 0.0966072753071785,
|
| 452 |
+
"learning_rate": 3.9110392720306514e-05,
|
| 453 |
+
"loss": 0.099,
|
| 454 |
+
"step": 5500
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 5.050053879310345,
|
| 458 |
+
"eval_accuracy": 0.9633333333333334,
|
| 459 |
+
"eval_f1": 0.9578282624655623,
|
| 460 |
+
"eval_loss": 0.26155975461006165,
|
| 461 |
+
"eval_precision": 0.956979842282474,
|
| 462 |
+
"eval_recall": 0.9633333333333334,
|
| 463 |
+
"eval_runtime": 178.9094,
|
| 464 |
+
"eval_samples_per_second": 3.354,
|
| 465 |
+
"eval_steps_per_second": 1.677,
|
| 466 |
+
"step": 5574
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 6.001400862068966,
|
| 470 |
+
"grad_norm": 0.0034083151258528233,
|
| 471 |
+
"learning_rate": 3.881106321839081e-05,
|
| 472 |
+
"loss": 0.0444,
|
| 473 |
+
"step": 5600
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 6.006788793103448,
|
| 477 |
+
"grad_norm": 0.0034083956852555275,
|
| 478 |
+
"learning_rate": 3.851173371647509e-05,
|
| 479 |
+
"loss": 0.0439,
|
| 480 |
+
"step": 5700
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 6.012176724137931,
|
| 484 |
+
"grad_norm": 0.004689768888056278,
|
| 485 |
+
"learning_rate": 3.8212404214559386e-05,
|
| 486 |
+
"loss": 0.0007,
|
| 487 |
+
"step": 5800
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 6.017564655172414,
|
| 491 |
+
"grad_norm": 0.004528137389570475,
|
| 492 |
+
"learning_rate": 3.791307471264368e-05,
|
| 493 |
+
"loss": 0.0529,
|
| 494 |
+
"step": 5900
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 6.022952586206896,
|
| 498 |
+
"grad_norm": 0.005246564745903015,
|
| 499 |
+
"learning_rate": 3.7613745210727965e-05,
|
| 500 |
+
"loss": 0.0016,
|
| 501 |
+
"step": 6000
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 6.0283405172413795,
|
| 505 |
+
"grad_norm": 0.003063632408156991,
|
| 506 |
+
"learning_rate": 3.731441570881226e-05,
|
| 507 |
+
"loss": 0.0002,
|
| 508 |
+
"step": 6100
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 6.033728448275862,
|
| 512 |
+
"grad_norm": 0.002391215180978179,
|
| 513 |
+
"learning_rate": 3.701508620689655e-05,
|
| 514 |
+
"loss": 0.0002,
|
| 515 |
+
"step": 6200
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 6.039116379310345,
|
| 519 |
+
"grad_norm": 0.0016633198829367757,
|
| 520 |
+
"learning_rate": 3.6715756704980844e-05,
|
| 521 |
+
"loss": 0.0002,
|
| 522 |
+
"step": 6300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 6.044504310344828,
|
| 526 |
+
"grad_norm": 0.013642443343997002,
|
| 527 |
+
"learning_rate": 3.641642720306514e-05,
|
| 528 |
+
"loss": 0.0002,
|
| 529 |
+
"step": 6400
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 6.04989224137931,
|
| 533 |
+
"grad_norm": 0.0030818418599665165,
|
| 534 |
+
"learning_rate": 3.611709770114943e-05,
|
| 535 |
+
"loss": 0.0656,
|
| 536 |
+
"step": 6500
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 6.050053879310345,
|
| 540 |
+
"eval_accuracy": 0.9616666666666667,
|
| 541 |
+
"eval_f1": 0.9597682298955686,
|
| 542 |
+
"eval_loss": 0.20607294142246246,
|
| 543 |
+
"eval_precision": 0.9674627687127687,
|
| 544 |
+
"eval_recall": 0.9616666666666667,
|
| 545 |
+
"eval_runtime": 184.33,
|
| 546 |
+
"eval_samples_per_second": 3.255,
|
| 547 |
+
"eval_steps_per_second": 1.628,
|
| 548 |
+
"step": 6503
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 7.005226293103449,
|
| 552 |
+
"grad_norm": 0.003281041979789734,
|
| 553 |
+
"learning_rate": 3.5817768199233717e-05,
|
| 554 |
+
"loss": 0.0108,
|
| 555 |
+
"step": 6600
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 7.010614224137931,
|
| 559 |
+
"grad_norm": 0.06187641620635986,
|
| 560 |
+
"learning_rate": 3.551843869731801e-05,
|
| 561 |
+
"loss": 0.0458,
|
| 562 |
+
"step": 6700
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 7.016002155172414,
|
| 566 |
+
"grad_norm": 0.6124621033668518,
|
| 567 |
+
"learning_rate": 3.52191091954023e-05,
|
| 568 |
+
"loss": 0.0931,
|
| 569 |
+
"step": 6800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 7.021390086206897,
|
| 573 |
+
"grad_norm": 0.013154719024896622,
|
| 574 |
+
"learning_rate": 3.4919779693486596e-05,
|
| 575 |
+
"loss": 0.2289,
|
| 576 |
+
"step": 6900
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 7.026778017241379,
|
| 580 |
+
"grad_norm": 0.33358174562454224,
|
| 581 |
+
"learning_rate": 3.462045019157089e-05,
|
| 582 |
+
"loss": 0.0804,
|
| 583 |
+
"step": 7000
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 7.032165948275862,
|
| 587 |
+
"grad_norm": 0.010887747630476952,
|
| 588 |
+
"learning_rate": 3.4321120689655175e-05,
|
| 589 |
+
"loss": 0.0032,
|
| 590 |
+
"step": 7100
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 7.0375538793103445,
|
| 594 |
+
"grad_norm": 0.26187142729759216,
|
| 595 |
+
"learning_rate": 3.402179118773947e-05,
|
| 596 |
+
"loss": 0.0134,
|
| 597 |
+
"step": 7200
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 7.042941810344828,
|
| 601 |
+
"grad_norm": 0.01975095644593239,
|
| 602 |
+
"learning_rate": 3.372246168582376e-05,
|
| 603 |
+
"loss": 0.0507,
|
| 604 |
+
"step": 7300
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 7.048329741379311,
|
| 608 |
+
"grad_norm": 0.004032758995890617,
|
| 609 |
+
"learning_rate": 3.342313218390805e-05,
|
| 610 |
+
"loss": 0.0314,
|
| 611 |
+
"step": 7400
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 7.050053879310345,
|
| 615 |
+
"eval_accuracy": 0.9533333333333334,
|
| 616 |
+
"eval_f1": 0.9526073505957408,
|
| 617 |
+
"eval_loss": 0.20366963744163513,
|
| 618 |
+
"eval_precision": 0.9623829642579644,
|
| 619 |
+
"eval_recall": 0.9533333333333334,
|
| 620 |
+
"eval_runtime": 186.0707,
|
| 621 |
+
"eval_samples_per_second": 3.225,
|
| 622 |
+
"eval_steps_per_second": 1.612,
|
| 623 |
+
"step": 7432
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 8.003663793103449,
|
| 627 |
+
"grad_norm": 0.003311087377369404,
|
| 628 |
+
"learning_rate": 3.312380268199234e-05,
|
| 629 |
+
"loss": 0.0496,
|
| 630 |
+
"step": 7500
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 8.009051724137931,
|
| 634 |
+
"grad_norm": 0.03303457051515579,
|
| 635 |
+
"learning_rate": 3.282447318007663e-05,
|
| 636 |
+
"loss": 0.0004,
|
| 637 |
+
"step": 7600
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 8.014439655172414,
|
| 641 |
+
"grad_norm": 0.0018716267077252269,
|
| 642 |
+
"learning_rate": 3.252514367816092e-05,
|
| 643 |
+
"loss": 0.0002,
|
| 644 |
+
"step": 7700
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 8.019827586206896,
|
| 648 |
+
"grad_norm": 0.0035127492155879736,
|
| 649 |
+
"learning_rate": 3.222581417624521e-05,
|
| 650 |
+
"loss": 0.1371,
|
| 651 |
+
"step": 7800
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 8.02521551724138,
|
| 655 |
+
"grad_norm": 0.009622437879443169,
|
| 656 |
+
"learning_rate": 3.1926484674329505e-05,
|
| 657 |
+
"loss": 0.0445,
|
| 658 |
+
"step": 7900
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 8.030603448275862,
|
| 662 |
+
"grad_norm": 0.0022457086015492678,
|
| 663 |
+
"learning_rate": 3.163014846743295e-05,
|
| 664 |
+
"loss": 0.0073,
|
| 665 |
+
"step": 8000
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 8.035991379310344,
|
| 669 |
+
"grad_norm": 0.002641331171616912,
|
| 670 |
+
"learning_rate": 3.133081896551724e-05,
|
| 671 |
+
"loss": 0.0007,
|
| 672 |
+
"step": 8100
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 8.041379310344828,
|
| 676 |
+
"grad_norm": 0.004722919315099716,
|
| 677 |
+
"learning_rate": 3.1031489463601535e-05,
|
| 678 |
+
"loss": 0.0721,
|
| 679 |
+
"step": 8200
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 8.04676724137931,
|
| 683 |
+
"grad_norm": 0.0021432852372527122,
|
| 684 |
+
"learning_rate": 3.073215996168583e-05,
|
| 685 |
+
"loss": 0.0874,
|
| 686 |
+
"step": 8300
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 8.050053879310346,
|
| 690 |
+
"eval_accuracy": 0.95,
|
| 691 |
+
"eval_f1": 0.9457895216400736,
|
| 692 |
+
"eval_loss": 0.219703808426857,
|
| 693 |
+
"eval_precision": 0.9611800144300143,
|
| 694 |
+
"eval_recall": 0.95,
|
| 695 |
+
"eval_runtime": 186.2504,
|
| 696 |
+
"eval_samples_per_second": 3.221,
|
| 697 |
+
"eval_steps_per_second": 1.611,
|
| 698 |
+
"step": 8361
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 8.050053879310346,
|
| 702 |
+
"step": 8361,
|
| 703 |
+
"total_flos": 8.337990588259369e+19,
|
| 704 |
+
"train_loss": 1.2332628082536137,
|
| 705 |
+
"train_runtime": 25272.533,
|
| 706 |
+
"train_samples_per_second": 5.875,
|
| 707 |
+
"train_steps_per_second": 0.734
|
| 708 |
+
}
|
| 709 |
+
],
|
| 710 |
+
"logging_steps": 100,
|
| 711 |
+
"max_steps": 18560,
|
| 712 |
+
"num_input_tokens_seen": 0,
|
| 713 |
+
"num_train_epochs": 9223372036854775807,
|
| 714 |
+
"save_steps": 500,
|
| 715 |
+
"stateful_callbacks": {
|
| 716 |
+
"EarlyStoppingCallback": {
|
| 717 |
+
"args": {
|
| 718 |
+
"early_stopping_patience": 5,
|
| 719 |
+
"early_stopping_threshold": 0.0
|
| 720 |
+
},
|
| 721 |
+
"attributes": {
|
| 722 |
+
"early_stopping_patience_counter": 5
|
| 723 |
+
}
|
| 724 |
+
},
|
| 725 |
+
"TrainerControl": {
|
| 726 |
+
"args": {
|
| 727 |
+
"should_epoch_stop": false,
|
| 728 |
+
"should_evaluate": false,
|
| 729 |
+
"should_log": false,
|
| 730 |
+
"should_save": true,
|
| 731 |
+
"should_training_stop": true
|
| 732 |
+
},
|
| 733 |
+
"attributes": {}
|
| 734 |
+
}
|
| 735 |
+
},
|
| 736 |
+
"total_flos": 8.337990588259369e+19,
|
| 737 |
+
"train_batch_size": 2,
|
| 738 |
+
"trial_name": null,
|
| 739 |
+
"trial_params": null
|
| 740 |
+
}
|