Commit
•
316f20a
1
Parent(s):
aa3347d
End of training
Browse files- all_results.json +12 -0
- eval_results.json +8 -0
- train_results.json +7 -0
- trainer_state.json +454 -0
all_results.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 1.0,
|
| 4 |
+
"eval_loss": 0.027025289833545685,
|
| 5 |
+
"eval_runtime": 60.4394,
|
| 6 |
+
"eval_samples_per_second": 16.876,
|
| 7 |
+
"eval_steps_per_second": 2.118,
|
| 8 |
+
"train_loss": 0.49404804557561877,
|
| 9 |
+
"train_runtime": 637.2078,
|
| 10 |
+
"train_samples_per_second": 8.004,
|
| 11 |
+
"train_steps_per_second": 1.004
|
| 12 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"eval_accuracy": 1.0,
|
| 4 |
+
"eval_loss": 0.027025289833545685,
|
| 5 |
+
"eval_runtime": 60.4394,
|
| 6 |
+
"eval_samples_per_second": 16.876,
|
| 7 |
+
"eval_steps_per_second": 2.118
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 5.0,
|
| 3 |
+
"train_loss": 0.49404804557561877,
|
| 4 |
+
"train_runtime": 637.2078,
|
| 5 |
+
"train_samples_per_second": 8.004,
|
| 6 |
+
"train_steps_per_second": 1.004
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,454 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.0,
|
| 3 |
+
"best_model_checkpoint": "whisper-tiny-ft-marathi-numbers/checkpoint-384",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"global_step": 640,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.08,
|
| 12 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 13 |
+
"loss": 3.0129,
|
| 14 |
+
"step": 10
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.16,
|
| 18 |
+
"learning_rate": 9.375000000000001e-06,
|
| 19 |
+
"loss": 2.9996,
|
| 20 |
+
"step": 20
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.23,
|
| 24 |
+
"learning_rate": 1.40625e-05,
|
| 25 |
+
"loss": 2.9805,
|
| 26 |
+
"step": 30
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.31,
|
| 30 |
+
"learning_rate": 1.8750000000000002e-05,
|
| 31 |
+
"loss": 2.8834,
|
| 32 |
+
"step": 40
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.39,
|
| 36 |
+
"learning_rate": 2.34375e-05,
|
| 37 |
+
"loss": 2.7423,
|
| 38 |
+
"step": 50
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.47,
|
| 42 |
+
"learning_rate": 2.8125e-05,
|
| 43 |
+
"loss": 2.4969,
|
| 44 |
+
"step": 60
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.55,
|
| 48 |
+
"learning_rate": 2.9687500000000003e-05,
|
| 49 |
+
"loss": 2.2685,
|
| 50 |
+
"step": 70
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.62,
|
| 54 |
+
"learning_rate": 2.9166666666666666e-05,
|
| 55 |
+
"loss": 1.878,
|
| 56 |
+
"step": 80
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.7,
|
| 60 |
+
"learning_rate": 2.8645833333333336e-05,
|
| 61 |
+
"loss": 1.6038,
|
| 62 |
+
"step": 90
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.78,
|
| 66 |
+
"learning_rate": 2.8125e-05,
|
| 67 |
+
"loss": 1.4139,
|
| 68 |
+
"step": 100
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.86,
|
| 72 |
+
"learning_rate": 2.7604166666666666e-05,
|
| 73 |
+
"loss": 1.1624,
|
| 74 |
+
"step": 110
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.94,
|
| 78 |
+
"learning_rate": 2.7083333333333335e-05,
|
| 79 |
+
"loss": 1.0002,
|
| 80 |
+
"step": 120
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.0,
|
| 84 |
+
"eval_accuracy": 0.9990196078431373,
|
| 85 |
+
"eval_loss": 0.7065549492835999,
|
| 86 |
+
"eval_runtime": 60.4233,
|
| 87 |
+
"eval_samples_per_second": 16.881,
|
| 88 |
+
"eval_steps_per_second": 2.118,
|
| 89 |
+
"step": 128
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.02,
|
| 93 |
+
"learning_rate": 2.6614583333333334e-05,
|
| 94 |
+
"loss": 0.7602,
|
| 95 |
+
"step": 130
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.09,
|
| 99 |
+
"learning_rate": 2.609375e-05,
|
| 100 |
+
"loss": 0.6346,
|
| 101 |
+
"step": 140
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 1.17,
|
| 105 |
+
"learning_rate": 2.5572916666666667e-05,
|
| 106 |
+
"loss": 0.5208,
|
| 107 |
+
"step": 150
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 1.25,
|
| 111 |
+
"learning_rate": 2.5052083333333334e-05,
|
| 112 |
+
"loss": 0.4167,
|
| 113 |
+
"step": 160
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 1.33,
|
| 117 |
+
"learning_rate": 2.4531250000000003e-05,
|
| 118 |
+
"loss": 0.3407,
|
| 119 |
+
"step": 170
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.41,
|
| 123 |
+
"learning_rate": 2.4010416666666666e-05,
|
| 124 |
+
"loss": 0.2763,
|
| 125 |
+
"step": 180
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 1.48,
|
| 129 |
+
"learning_rate": 2.3489583333333336e-05,
|
| 130 |
+
"loss": 0.221,
|
| 131 |
+
"step": 190
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.56,
|
| 135 |
+
"learning_rate": 2.296875e-05,
|
| 136 |
+
"loss": 0.2,
|
| 137 |
+
"step": 200
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.64,
|
| 141 |
+
"learning_rate": 2.2447916666666666e-05,
|
| 142 |
+
"loss": 0.1717,
|
| 143 |
+
"step": 210
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 1.72,
|
| 147 |
+
"learning_rate": 2.1927083333333336e-05,
|
| 148 |
+
"loss": 0.1375,
|
| 149 |
+
"step": 220
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.8,
|
| 153 |
+
"learning_rate": 2.140625e-05,
|
| 154 |
+
"loss": 0.1655,
|
| 155 |
+
"step": 230
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 1.88,
|
| 159 |
+
"learning_rate": 2.088541666666667e-05,
|
| 160 |
+
"loss": 0.1008,
|
| 161 |
+
"step": 240
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 1.95,
|
| 165 |
+
"learning_rate": 2.0364583333333335e-05,
|
| 166 |
+
"loss": 0.0895,
|
| 167 |
+
"step": 250
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 2.0,
|
| 171 |
+
"eval_accuracy": 0.9990196078431373,
|
| 172 |
+
"eval_loss": 0.07644587755203247,
|
| 173 |
+
"eval_runtime": 61.3641,
|
| 174 |
+
"eval_samples_per_second": 16.622,
|
| 175 |
+
"eval_steps_per_second": 2.086,
|
| 176 |
+
"step": 256
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.03,
|
| 180 |
+
"learning_rate": 1.984375e-05,
|
| 181 |
+
"loss": 0.0779,
|
| 182 |
+
"step": 260
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.11,
|
| 186 |
+
"learning_rate": 1.9322916666666668e-05,
|
| 187 |
+
"loss": 0.0674,
|
| 188 |
+
"step": 270
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.19,
|
| 192 |
+
"learning_rate": 1.8802083333333334e-05,
|
| 193 |
+
"loss": 0.0639,
|
| 194 |
+
"step": 280
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 2.27,
|
| 198 |
+
"learning_rate": 1.828125e-05,
|
| 199 |
+
"loss": 0.0591,
|
| 200 |
+
"step": 290
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.34,
|
| 204 |
+
"learning_rate": 1.7760416666666667e-05,
|
| 205 |
+
"loss": 0.0475,
|
| 206 |
+
"step": 300
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.42,
|
| 210 |
+
"learning_rate": 1.7239583333333333e-05,
|
| 211 |
+
"loss": 0.0448,
|
| 212 |
+
"step": 310
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.5,
|
| 216 |
+
"learning_rate": 1.671875e-05,
|
| 217 |
+
"loss": 0.0403,
|
| 218 |
+
"step": 320
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.58,
|
| 222 |
+
"learning_rate": 1.6197916666666666e-05,
|
| 223 |
+
"loss": 0.0629,
|
| 224 |
+
"step": 330
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.66,
|
| 228 |
+
"learning_rate": 1.5677083333333333e-05,
|
| 229 |
+
"loss": 0.0353,
|
| 230 |
+
"step": 340
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.73,
|
| 234 |
+
"learning_rate": 1.515625e-05,
|
| 235 |
+
"loss": 0.0347,
|
| 236 |
+
"step": 350
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.81,
|
| 240 |
+
"learning_rate": 1.4635416666666667e-05,
|
| 241 |
+
"loss": 0.0322,
|
| 242 |
+
"step": 360
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.89,
|
| 246 |
+
"learning_rate": 1.4114583333333334e-05,
|
| 247 |
+
"loss": 0.0296,
|
| 248 |
+
"step": 370
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.97,
|
| 252 |
+
"learning_rate": 1.359375e-05,
|
| 253 |
+
"loss": 0.0281,
|
| 254 |
+
"step": 380
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 3.0,
|
| 258 |
+
"eval_accuracy": 1.0,
|
| 259 |
+
"eval_loss": 0.027025289833545685,
|
| 260 |
+
"eval_runtime": 60.3432,
|
| 261 |
+
"eval_samples_per_second": 16.903,
|
| 262 |
+
"eval_steps_per_second": 2.121,
|
| 263 |
+
"step": 384
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 3.05,
|
| 267 |
+
"learning_rate": 1.3072916666666667e-05,
|
| 268 |
+
"loss": 0.0264,
|
| 269 |
+
"step": 390
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 3.12,
|
| 273 |
+
"learning_rate": 1.2552083333333335e-05,
|
| 274 |
+
"loss": 0.0267,
|
| 275 |
+
"step": 400
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 3.2,
|
| 279 |
+
"learning_rate": 1.2031250000000001e-05,
|
| 280 |
+
"loss": 0.0247,
|
| 281 |
+
"step": 410
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 3.28,
|
| 285 |
+
"learning_rate": 1.1510416666666667e-05,
|
| 286 |
+
"loss": 0.0232,
|
| 287 |
+
"step": 420
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 3.36,
|
| 291 |
+
"learning_rate": 1.0989583333333332e-05,
|
| 292 |
+
"loss": 0.023,
|
| 293 |
+
"step": 430
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 3.44,
|
| 297 |
+
"learning_rate": 1.046875e-05,
|
| 298 |
+
"loss": 0.0223,
|
| 299 |
+
"step": 440
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 3.52,
|
| 303 |
+
"learning_rate": 9.947916666666667e-06,
|
| 304 |
+
"loss": 0.0223,
|
| 305 |
+
"step": 450
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 3.59,
|
| 309 |
+
"learning_rate": 9.427083333333333e-06,
|
| 310 |
+
"loss": 0.0222,
|
| 311 |
+
"step": 460
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 3.67,
|
| 315 |
+
"learning_rate": 8.90625e-06,
|
| 316 |
+
"loss": 0.0225,
|
| 317 |
+
"step": 470
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.75,
|
| 321 |
+
"learning_rate": 8.385416666666668e-06,
|
| 322 |
+
"loss": 0.0204,
|
| 323 |
+
"step": 480
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 3.83,
|
| 327 |
+
"learning_rate": 7.864583333333334e-06,
|
| 328 |
+
"loss": 0.0204,
|
| 329 |
+
"step": 490
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.91,
|
| 333 |
+
"learning_rate": 7.34375e-06,
|
| 334 |
+
"loss": 0.019,
|
| 335 |
+
"step": 500
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 3.98,
|
| 339 |
+
"learning_rate": 6.822916666666667e-06,
|
| 340 |
+
"loss": 0.0191,
|
| 341 |
+
"step": 510
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 4.0,
|
| 345 |
+
"eval_accuracy": 1.0,
|
| 346 |
+
"eval_loss": 0.018605878576636314,
|
| 347 |
+
"eval_runtime": 59.6121,
|
| 348 |
+
"eval_samples_per_second": 17.111,
|
| 349 |
+
"eval_steps_per_second": 2.147,
|
| 350 |
+
"step": 512
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 4.06,
|
| 354 |
+
"learning_rate": 6.3020833333333335e-06,
|
| 355 |
+
"loss": 0.0187,
|
| 356 |
+
"step": 520
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 4.14,
|
| 360 |
+
"learning_rate": 5.781250000000001e-06,
|
| 361 |
+
"loss": 0.0175,
|
| 362 |
+
"step": 530
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 4.22,
|
| 366 |
+
"learning_rate": 5.260416666666666e-06,
|
| 367 |
+
"loss": 0.0178,
|
| 368 |
+
"step": 540
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 4.3,
|
| 372 |
+
"learning_rate": 4.739583333333334e-06,
|
| 373 |
+
"loss": 0.0175,
|
| 374 |
+
"step": 550
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 4.38,
|
| 378 |
+
"learning_rate": 4.21875e-06,
|
| 379 |
+
"loss": 0.0177,
|
| 380 |
+
"step": 560
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 4.45,
|
| 384 |
+
"learning_rate": 3.697916666666667e-06,
|
| 385 |
+
"loss": 0.017,
|
| 386 |
+
"step": 570
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 4.53,
|
| 390 |
+
"learning_rate": 3.1770833333333333e-06,
|
| 391 |
+
"loss": 0.0177,
|
| 392 |
+
"step": 580
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 4.61,
|
| 396 |
+
"learning_rate": 2.65625e-06,
|
| 397 |
+
"loss": 0.0174,
|
| 398 |
+
"step": 590
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 4.69,
|
| 402 |
+
"learning_rate": 2.1354166666666666e-06,
|
| 403 |
+
"loss": 0.0176,
|
| 404 |
+
"step": 600
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 4.77,
|
| 408 |
+
"learning_rate": 1.6145833333333335e-06,
|
| 409 |
+
"loss": 0.0168,
|
| 410 |
+
"step": 610
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 4.84,
|
| 414 |
+
"learning_rate": 1.0937500000000001e-06,
|
| 415 |
+
"loss": 0.0165,
|
| 416 |
+
"step": 620
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 4.92,
|
| 420 |
+
"learning_rate": 5.729166666666668e-07,
|
| 421 |
+
"loss": 0.0164,
|
| 422 |
+
"step": 630
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 5.0,
|
| 426 |
+
"learning_rate": 5.208333333333333e-08,
|
| 427 |
+
"loss": 0.017,
|
| 428 |
+
"step": 640
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 5.0,
|
| 432 |
+
"eval_accuracy": 1.0,
|
| 433 |
+
"eval_loss": 0.016615983098745346,
|
| 434 |
+
"eval_runtime": 60.2433,
|
| 435 |
+
"eval_samples_per_second": 16.931,
|
| 436 |
+
"eval_steps_per_second": 2.125,
|
| 437 |
+
"step": 640
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 5.0,
|
| 441 |
+
"step": 640,
|
| 442 |
+
"total_flos": 5.6813800896e+16,
|
| 443 |
+
"train_loss": 0.49404804557561877,
|
| 444 |
+
"train_runtime": 637.2078,
|
| 445 |
+
"train_samples_per_second": 8.004,
|
| 446 |
+
"train_steps_per_second": 1.004
|
| 447 |
+
}
|
| 448 |
+
],
|
| 449 |
+
"max_steps": 640,
|
| 450 |
+
"num_train_epochs": 5,
|
| 451 |
+
"total_flos": 5.6813800896e+16,
|
| 452 |
+
"trial_name": null,
|
| 453 |
+
"trial_params": null
|
| 454 |
+
}
|