Commit
•
477abc4
1
Parent(s):
9ebe1dc
End of training
Browse files- all_results.json +13 -0
- test_results.json +13 -0
- trainer_state.json +581 -0
all_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.193650793650794,
|
| 3 |
+
"eval_accuracy": 0.9253112033195021,
|
| 4 |
+
"eval_accuracy_hold": 1.0,
|
| 5 |
+
"eval_accuracy_preparation": 1.0,
|
| 6 |
+
"eval_accuracy_recovery": 0.7894736842105263,
|
| 7 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 8 |
+
"eval_accuracy_unknown": 0.6428571428571429,
|
| 9 |
+
"eval_loss": 0.31221458315849304,
|
| 10 |
+
"eval_runtime": 17.9771,
|
| 11 |
+
"eval_samples_per_second": 13.406,
|
| 12 |
+
"eval_steps_per_second": 1.724
|
| 13 |
+
}
|
test_results.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 4.193650793650794,
|
| 3 |
+
"eval_accuracy": 0.9253112033195021,
|
| 4 |
+
"eval_accuracy_hold": 1.0,
|
| 5 |
+
"eval_accuracy_preparation": 1.0,
|
| 6 |
+
"eval_accuracy_recovery": 0.7894736842105263,
|
| 7 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 8 |
+
"eval_accuracy_unknown": 0.6428571428571429,
|
| 9 |
+
"eval_loss": 0.31221458315849304,
|
| 10 |
+
"eval_runtime": 17.9771,
|
| 11 |
+
"eval_samples_per_second": 13.406,
|
| 12 |
+
"eval_steps_per_second": 1.724
|
| 13 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,581 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.9475982532751092,
|
| 3 |
+
"best_model_checkpoint": "videomae-base-finetuned-good-gesturePhaseV10/checkpoint-630",
|
| 4 |
+
"epoch": 4.193650793650794,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 630,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.015873015873015872,
|
| 13 |
+
"grad_norm": 10.510711669921875,
|
| 14 |
+
"learning_rate": 7.936507936507936e-06,
|
| 15 |
+
"loss": 1.5865,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.031746031746031744,
|
| 20 |
+
"grad_norm": 5.081316947937012,
|
| 21 |
+
"learning_rate": 1.5873015873015872e-05,
|
| 22 |
+
"loss": 1.19,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.047619047619047616,
|
| 27 |
+
"grad_norm": 4.605384349822998,
|
| 28 |
+
"learning_rate": 2.380952380952381e-05,
|
| 29 |
+
"loss": 0.9454,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.06349206349206349,
|
| 34 |
+
"grad_norm": 6.820161819458008,
|
| 35 |
+
"learning_rate": 3.1746031746031745e-05,
|
| 36 |
+
"loss": 0.905,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.07936507936507936,
|
| 41 |
+
"grad_norm": 3.678572654724121,
|
| 42 |
+
"learning_rate": 3.968253968253968e-05,
|
| 43 |
+
"loss": 0.9165,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.09523809523809523,
|
| 48 |
+
"grad_norm": 6.76840353012085,
|
| 49 |
+
"learning_rate": 4.761904761904762e-05,
|
| 50 |
+
"loss": 1.061,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.1111111111111111,
|
| 55 |
+
"grad_norm": 3.580310344696045,
|
| 56 |
+
"learning_rate": 4.938271604938271e-05,
|
| 57 |
+
"loss": 0.835,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.12698412698412698,
|
| 62 |
+
"grad_norm": 6.999443054199219,
|
| 63 |
+
"learning_rate": 4.850088183421517e-05,
|
| 64 |
+
"loss": 0.9638,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.14285714285714285,
|
| 69 |
+
"grad_norm": 3.5926501750946045,
|
| 70 |
+
"learning_rate": 4.761904761904762e-05,
|
| 71 |
+
"loss": 0.9482,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.15873015873015872,
|
| 76 |
+
"grad_norm": 3.4658665657043457,
|
| 77 |
+
"learning_rate": 4.673721340388007e-05,
|
| 78 |
+
"loss": 0.852,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.1746031746031746,
|
| 83 |
+
"grad_norm": 5.933653354644775,
|
| 84 |
+
"learning_rate": 4.585537918871252e-05,
|
| 85 |
+
"loss": 0.9297,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.19047619047619047,
|
| 90 |
+
"grad_norm": 3.1261682510375977,
|
| 91 |
+
"learning_rate": 4.4973544973544974e-05,
|
| 92 |
+
"loss": 1.1344,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.2015873015873016,
|
| 97 |
+
"eval_accuracy": 0.6899563318777293,
|
| 98 |
+
"eval_accuracy_hold": 0.0,
|
| 99 |
+
"eval_accuracy_preparation": 1.0,
|
| 100 |
+
"eval_accuracy_recovery": 0.0,
|
| 101 |
+
"eval_accuracy_stroke": 0.0,
|
| 102 |
+
"eval_accuracy_unknown": 0.0,
|
| 103 |
+
"eval_loss": 1.0020831823349,
|
| 104 |
+
"eval_runtime": 21.4274,
|
| 105 |
+
"eval_samples_per_second": 10.687,
|
| 106 |
+
"eval_steps_per_second": 1.353,
|
| 107 |
+
"step": 127
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.0047619047619047,
|
| 111 |
+
"grad_norm": 1.5987485647201538,
|
| 112 |
+
"learning_rate": 4.409171075837743e-05,
|
| 113 |
+
"loss": 0.9775,
|
| 114 |
+
"step": 130
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 1.0206349206349206,
|
| 118 |
+
"grad_norm": 6.478776931762695,
|
| 119 |
+
"learning_rate": 4.3209876543209875e-05,
|
| 120 |
+
"loss": 0.9226,
|
| 121 |
+
"step": 140
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 1.0365079365079366,
|
| 125 |
+
"grad_norm": 2.5407347679138184,
|
| 126 |
+
"learning_rate": 4.232804232804233e-05,
|
| 127 |
+
"loss": 0.9558,
|
| 128 |
+
"step": 150
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.0523809523809524,
|
| 132 |
+
"grad_norm": 2.95082426071167,
|
| 133 |
+
"learning_rate": 4.144620811287478e-05,
|
| 134 |
+
"loss": 0.8218,
|
| 135 |
+
"step": 160
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.0682539682539682,
|
| 139 |
+
"grad_norm": 8.649039268493652,
|
| 140 |
+
"learning_rate": 4.056437389770723e-05,
|
| 141 |
+
"loss": 0.9146,
|
| 142 |
+
"step": 170
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 1.084126984126984,
|
| 146 |
+
"grad_norm": 5.632199764251709,
|
| 147 |
+
"learning_rate": 3.968253968253968e-05,
|
| 148 |
+
"loss": 0.7657,
|
| 149 |
+
"step": 180
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.1,
|
| 153 |
+
"grad_norm": 3.284162998199463,
|
| 154 |
+
"learning_rate": 3.8800705467372136e-05,
|
| 155 |
+
"loss": 0.979,
|
| 156 |
+
"step": 190
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 1.115873015873016,
|
| 160 |
+
"grad_norm": 3.5784647464752197,
|
| 161 |
+
"learning_rate": 3.791887125220458e-05,
|
| 162 |
+
"loss": 0.74,
|
| 163 |
+
"step": 200
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 1.1317460317460317,
|
| 167 |
+
"grad_norm": 9.924901008605957,
|
| 168 |
+
"learning_rate": 3.7037037037037037e-05,
|
| 169 |
+
"loss": 0.8304,
|
| 170 |
+
"step": 210
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.1476190476190475,
|
| 174 |
+
"grad_norm": 7.721616744995117,
|
| 175 |
+
"learning_rate": 3.615520282186949e-05,
|
| 176 |
+
"loss": 0.7795,
|
| 177 |
+
"step": 220
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 1.1634920634920636,
|
| 181 |
+
"grad_norm": 6.16989278793335,
|
| 182 |
+
"learning_rate": 3.527336860670194e-05,
|
| 183 |
+
"loss": 0.6725,
|
| 184 |
+
"step": 230
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 1.1793650793650794,
|
| 188 |
+
"grad_norm": 5.634149074554443,
|
| 189 |
+
"learning_rate": 3.439153439153439e-05,
|
| 190 |
+
"loss": 0.5051,
|
| 191 |
+
"step": 240
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 1.1952380952380952,
|
| 195 |
+
"grad_norm": 10.187247276306152,
|
| 196 |
+
"learning_rate": 3.3509700176366844e-05,
|
| 197 |
+
"loss": 0.5961,
|
| 198 |
+
"step": 250
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.2015873015873015,
|
| 202 |
+
"eval_accuracy": 0.7947598253275109,
|
| 203 |
+
"eval_accuracy_hold": 0.2692307692307692,
|
| 204 |
+
"eval_accuracy_preparation": 0.9873417721518988,
|
| 205 |
+
"eval_accuracy_recovery": 0.058823529411764705,
|
| 206 |
+
"eval_accuracy_stroke": 0.0,
|
| 207 |
+
"eval_accuracy_unknown": 0.8181818181818182,
|
| 208 |
+
"eval_loss": 0.6022052764892578,
|
| 209 |
+
"eval_runtime": 18.9231,
|
| 210 |
+
"eval_samples_per_second": 12.102,
|
| 211 |
+
"eval_steps_per_second": 1.533,
|
| 212 |
+
"step": 254
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.0095238095238095,
|
| 216 |
+
"grad_norm": 2.109996795654297,
|
| 217 |
+
"learning_rate": 3.262786596119929e-05,
|
| 218 |
+
"loss": 0.4873,
|
| 219 |
+
"step": 260
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 2.0253968253968253,
|
| 223 |
+
"grad_norm": 13.33204460144043,
|
| 224 |
+
"learning_rate": 3.1746031746031745e-05,
|
| 225 |
+
"loss": 0.5243,
|
| 226 |
+
"step": 270
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 2.041269841269841,
|
| 230 |
+
"grad_norm": 6.260931491851807,
|
| 231 |
+
"learning_rate": 3.08641975308642e-05,
|
| 232 |
+
"loss": 0.4785,
|
| 233 |
+
"step": 280
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 2.057142857142857,
|
| 237 |
+
"grad_norm": 26.649112701416016,
|
| 238 |
+
"learning_rate": 2.998236331569665e-05,
|
| 239 |
+
"loss": 0.4677,
|
| 240 |
+
"step": 290
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 2.073015873015873,
|
| 244 |
+
"grad_norm": 7.0665283203125,
|
| 245 |
+
"learning_rate": 2.91005291005291e-05,
|
| 246 |
+
"loss": 0.3441,
|
| 247 |
+
"step": 300
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 2.088888888888889,
|
| 251 |
+
"grad_norm": 9.628989219665527,
|
| 252 |
+
"learning_rate": 2.8218694885361552e-05,
|
| 253 |
+
"loss": 0.7396,
|
| 254 |
+
"step": 310
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 2.104761904761905,
|
| 258 |
+
"grad_norm": 1.708268165588379,
|
| 259 |
+
"learning_rate": 2.7336860670194003e-05,
|
| 260 |
+
"loss": 0.5972,
|
| 261 |
+
"step": 320
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 2.1206349206349207,
|
| 265 |
+
"grad_norm": 8.236438751220703,
|
| 266 |
+
"learning_rate": 2.6455026455026456e-05,
|
| 267 |
+
"loss": 0.4421,
|
| 268 |
+
"step": 330
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 2.1365079365079365,
|
| 272 |
+
"grad_norm": 3.947949171066284,
|
| 273 |
+
"learning_rate": 2.5573192239858906e-05,
|
| 274 |
+
"loss": 0.3602,
|
| 275 |
+
"step": 340
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 2.1523809523809523,
|
| 279 |
+
"grad_norm": 4.807738304138184,
|
| 280 |
+
"learning_rate": 2.4691358024691357e-05,
|
| 281 |
+
"loss": 0.3214,
|
| 282 |
+
"step": 350
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 2.168253968253968,
|
| 286 |
+
"grad_norm": 9.600074768066406,
|
| 287 |
+
"learning_rate": 2.380952380952381e-05,
|
| 288 |
+
"loss": 0.4783,
|
| 289 |
+
"step": 360
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 2.1841269841269844,
|
| 293 |
+
"grad_norm": 19.80223846435547,
|
| 294 |
+
"learning_rate": 2.292768959435626e-05,
|
| 295 |
+
"loss": 0.2683,
|
| 296 |
+
"step": 370
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 2.2,
|
| 300 |
+
"grad_norm": 0.9186673164367676,
|
| 301 |
+
"learning_rate": 2.2045855379188714e-05,
|
| 302 |
+
"loss": 0.3453,
|
| 303 |
+
"step": 380
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 2.2015873015873018,
|
| 307 |
+
"eval_accuracy": 0.8777292576419214,
|
| 308 |
+
"eval_accuracy_hold": 0.8076923076923077,
|
| 309 |
+
"eval_accuracy_preparation": 0.9746835443037974,
|
| 310 |
+
"eval_accuracy_recovery": 0.4117647058823529,
|
| 311 |
+
"eval_accuracy_stroke": 0.0,
|
| 312 |
+
"eval_accuracy_unknown": 0.8636363636363636,
|
| 313 |
+
"eval_loss": 0.3925057649612427,
|
| 314 |
+
"eval_runtime": 17.6116,
|
| 315 |
+
"eval_samples_per_second": 13.003,
|
| 316 |
+
"eval_steps_per_second": 1.647,
|
| 317 |
+
"step": 381
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.0142857142857142,
|
| 321 |
+
"grad_norm": 11.65360164642334,
|
| 322 |
+
"learning_rate": 2.1164021164021164e-05,
|
| 323 |
+
"loss": 0.3585,
|
| 324 |
+
"step": 390
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 3.03015873015873,
|
| 328 |
+
"grad_norm": 4.8670454025268555,
|
| 329 |
+
"learning_rate": 2.0282186948853614e-05,
|
| 330 |
+
"loss": 0.4365,
|
| 331 |
+
"step": 400
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 3.046031746031746,
|
| 335 |
+
"grad_norm": 7.012415409088135,
|
| 336 |
+
"learning_rate": 1.9400352733686068e-05,
|
| 337 |
+
"loss": 0.183,
|
| 338 |
+
"step": 410
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 3.0619047619047617,
|
| 342 |
+
"grad_norm": 0.5204718112945557,
|
| 343 |
+
"learning_rate": 1.8518518518518518e-05,
|
| 344 |
+
"loss": 0.3634,
|
| 345 |
+
"step": 420
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 3.077777777777778,
|
| 349 |
+
"grad_norm": 13.590949058532715,
|
| 350 |
+
"learning_rate": 1.763668430335097e-05,
|
| 351 |
+
"loss": 0.1989,
|
| 352 |
+
"step": 430
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 3.0936507936507938,
|
| 356 |
+
"grad_norm": 7.4863691329956055,
|
| 357 |
+
"learning_rate": 1.6754850088183422e-05,
|
| 358 |
+
"loss": 0.1979,
|
| 359 |
+
"step": 440
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.1095238095238096,
|
| 363 |
+
"grad_norm": 4.570287704467773,
|
| 364 |
+
"learning_rate": 1.5873015873015872e-05,
|
| 365 |
+
"loss": 0.3445,
|
| 366 |
+
"step": 450
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 3.1253968253968254,
|
| 370 |
+
"grad_norm": 2.023911952972412,
|
| 371 |
+
"learning_rate": 1.4991181657848324e-05,
|
| 372 |
+
"loss": 0.1738,
|
| 373 |
+
"step": 460
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 3.141269841269841,
|
| 377 |
+
"grad_norm": 0.15230882167816162,
|
| 378 |
+
"learning_rate": 1.4109347442680776e-05,
|
| 379 |
+
"loss": 0.1889,
|
| 380 |
+
"step": 470
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 3.157142857142857,
|
| 384 |
+
"grad_norm": 0.4252125322818756,
|
| 385 |
+
"learning_rate": 1.3227513227513228e-05,
|
| 386 |
+
"loss": 0.1824,
|
| 387 |
+
"step": 480
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 3.173015873015873,
|
| 391 |
+
"grad_norm": 23.363100051879883,
|
| 392 |
+
"learning_rate": 1.2345679012345678e-05,
|
| 393 |
+
"loss": 0.2642,
|
| 394 |
+
"step": 490
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 3.188888888888889,
|
| 398 |
+
"grad_norm": 2.36000394821167,
|
| 399 |
+
"learning_rate": 1.146384479717813e-05,
|
| 400 |
+
"loss": 0.1551,
|
| 401 |
+
"step": 500
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 3.2015873015873018,
|
| 405 |
+
"eval_accuracy": 0.9432314410480349,
|
| 406 |
+
"eval_accuracy_hold": 0.9615384615384616,
|
| 407 |
+
"eval_accuracy_preparation": 0.9936708860759493,
|
| 408 |
+
"eval_accuracy_recovery": 0.7058823529411765,
|
| 409 |
+
"eval_accuracy_stroke": 0.16666666666666666,
|
| 410 |
+
"eval_accuracy_unknown": 0.9545454545454546,
|
| 411 |
+
"eval_loss": 0.2178172767162323,
|
| 412 |
+
"eval_runtime": 17.2686,
|
| 413 |
+
"eval_samples_per_second": 13.261,
|
| 414 |
+
"eval_steps_per_second": 1.679,
|
| 415 |
+
"step": 508
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 4.003174603174603,
|
| 419 |
+
"grad_norm": 0.14277657866477966,
|
| 420 |
+
"learning_rate": 1.0582010582010582e-05,
|
| 421 |
+
"loss": 0.1695,
|
| 422 |
+
"step": 510
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 4.019047619047619,
|
| 426 |
+
"grad_norm": 0.6571791768074036,
|
| 427 |
+
"learning_rate": 9.700176366843034e-06,
|
| 428 |
+
"loss": 0.0663,
|
| 429 |
+
"step": 520
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 4.034920634920635,
|
| 433 |
+
"grad_norm": 40.71855926513672,
|
| 434 |
+
"learning_rate": 8.818342151675484e-06,
|
| 435 |
+
"loss": 0.2735,
|
| 436 |
+
"step": 530
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 4.050793650793651,
|
| 440 |
+
"grad_norm": 12.299385070800781,
|
| 441 |
+
"learning_rate": 7.936507936507936e-06,
|
| 442 |
+
"loss": 0.0465,
|
| 443 |
+
"step": 540
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 4.066666666666666,
|
| 447 |
+
"grad_norm": 0.19348235428333282,
|
| 448 |
+
"learning_rate": 7.054673721340388e-06,
|
| 449 |
+
"loss": 0.0676,
|
| 450 |
+
"step": 550
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 4.082539682539682,
|
| 454 |
+
"grad_norm": 0.1367979794740677,
|
| 455 |
+
"learning_rate": 6.172839506172839e-06,
|
| 456 |
+
"loss": 0.0643,
|
| 457 |
+
"step": 560
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 4.098412698412698,
|
| 461 |
+
"grad_norm": 10.41691780090332,
|
| 462 |
+
"learning_rate": 5.291005291005291e-06,
|
| 463 |
+
"loss": 0.0992,
|
| 464 |
+
"step": 570
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 4.114285714285714,
|
| 468 |
+
"grad_norm": 21.60604476928711,
|
| 469 |
+
"learning_rate": 4.409171075837742e-06,
|
| 470 |
+
"loss": 0.2483,
|
| 471 |
+
"step": 580
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 4.130158730158731,
|
| 475 |
+
"grad_norm": 0.17013657093048096,
|
| 476 |
+
"learning_rate": 3.527336860670194e-06,
|
| 477 |
+
"loss": 0.0683,
|
| 478 |
+
"step": 590
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 4.146031746031746,
|
| 482 |
+
"grad_norm": 0.12723946571350098,
|
| 483 |
+
"learning_rate": 2.6455026455026455e-06,
|
| 484 |
+
"loss": 0.0116,
|
| 485 |
+
"step": 600
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 4.161904761904762,
|
| 489 |
+
"grad_norm": 0.1051618903875351,
|
| 490 |
+
"learning_rate": 1.763668430335097e-06,
|
| 491 |
+
"loss": 0.1836,
|
| 492 |
+
"step": 610
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 4.177777777777778,
|
| 496 |
+
"grad_norm": 8.298240661621094,
|
| 497 |
+
"learning_rate": 8.818342151675485e-07,
|
| 498 |
+
"loss": 0.0978,
|
| 499 |
+
"step": 620
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 4.193650793650794,
|
| 503 |
+
"grad_norm": 0.7603585124015808,
|
| 504 |
+
"learning_rate": 0.0,
|
| 505 |
+
"loss": 0.1213,
|
| 506 |
+
"step": 630
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.193650793650794,
|
| 510 |
+
"eval_accuracy": 0.9475982532751092,
|
| 511 |
+
"eval_accuracy_hold": 0.9615384615384616,
|
| 512 |
+
"eval_accuracy_preparation": 1.0,
|
| 513 |
+
"eval_accuracy_recovery": 0.7058823529411765,
|
| 514 |
+
"eval_accuracy_stroke": 0.6666666666666666,
|
| 515 |
+
"eval_accuracy_unknown": 0.8181818181818182,
|
| 516 |
+
"eval_loss": 0.2031552791595459,
|
| 517 |
+
"eval_runtime": 17.2621,
|
| 518 |
+
"eval_samples_per_second": 13.266,
|
| 519 |
+
"eval_steps_per_second": 1.68,
|
| 520 |
+
"step": 630
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 4.193650793650794,
|
| 524 |
+
"step": 630,
|
| 525 |
+
"total_flos": 6.275349161424323e+18,
|
| 526 |
+
"train_loss": 0.5261434914100738,
|
| 527 |
+
"train_runtime": 1007.5783,
|
| 528 |
+
"train_samples_per_second": 5.002,
|
| 529 |
+
"train_steps_per_second": 0.625
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 4.193650793650794,
|
| 533 |
+
"eval_accuracy": 0.9253112033195021,
|
| 534 |
+
"eval_accuracy_hold": 1.0,
|
| 535 |
+
"eval_accuracy_preparation": 1.0,
|
| 536 |
+
"eval_accuracy_recovery": 0.7894736842105263,
|
| 537 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 538 |
+
"eval_accuracy_unknown": 0.6428571428571429,
|
| 539 |
+
"eval_loss": 0.31221458315849304,
|
| 540 |
+
"eval_runtime": 24.503,
|
| 541 |
+
"eval_samples_per_second": 9.836,
|
| 542 |
+
"eval_steps_per_second": 1.265,
|
| 543 |
+
"step": 630
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 4.193650793650794,
|
| 547 |
+
"eval_accuracy": 0.9253112033195021,
|
| 548 |
+
"eval_accuracy_hold": 1.0,
|
| 549 |
+
"eval_accuracy_preparation": 1.0,
|
| 550 |
+
"eval_accuracy_recovery": 0.7894736842105263,
|
| 551 |
+
"eval_accuracy_stroke": 0.42857142857142855,
|
| 552 |
+
"eval_accuracy_unknown": 0.6428571428571429,
|
| 553 |
+
"eval_loss": 0.31221458315849304,
|
| 554 |
+
"eval_runtime": 17.9771,
|
| 555 |
+
"eval_samples_per_second": 13.406,
|
| 556 |
+
"eval_steps_per_second": 1.724,
|
| 557 |
+
"step": 630
|
| 558 |
+
}
|
| 559 |
+
],
|
| 560 |
+
"logging_steps": 10,
|
| 561 |
+
"max_steps": 630,
|
| 562 |
+
"num_input_tokens_seen": 0,
|
| 563 |
+
"num_train_epochs": 9223372036854775807,
|
| 564 |
+
"save_steps": 500,
|
| 565 |
+
"stateful_callbacks": {
|
| 566 |
+
"TrainerControl": {
|
| 567 |
+
"args": {
|
| 568 |
+
"should_epoch_stop": false,
|
| 569 |
+
"should_evaluate": false,
|
| 570 |
+
"should_log": false,
|
| 571 |
+
"should_save": true,
|
| 572 |
+
"should_training_stop": true
|
| 573 |
+
},
|
| 574 |
+
"attributes": {}
|
| 575 |
+
}
|
| 576 |
+
},
|
| 577 |
+
"total_flos": 6.275349161424323e+18,
|
| 578 |
+
"train_batch_size": 8,
|
| 579 |
+
"trial_name": null,
|
| 580 |
+
"trial_params": null
|
| 581 |
+
}
|