Upload trainer_state.json
Browse files- trainer_state.json +475 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,475 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 2500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.4,
|
| 12 |
+
"learning_rate": 2.88e-05,
|
| 13 |
+
"loss": 0.3202,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.4,
|
| 18 |
+
"eval_accuracy": 0.8813009961455675,
|
| 19 |
+
"eval_f1": 0.16261325703385787,
|
| 20 |
+
"eval_loss": 0.24328218400478363,
|
| 21 |
+
"eval_precision": 0.2519394163280384,
|
| 22 |
+
"eval_recall": 0.12004928709734201,
|
| 23 |
+
"eval_runtime": 4.054,
|
| 24 |
+
"eval_samples_per_second": 123.334,
|
| 25 |
+
"eval_steps_per_second": 30.833,
|
| 26 |
+
"step": 100
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.8,
|
| 30 |
+
"learning_rate": 2.7600000000000003e-05,
|
| 31 |
+
"loss": 0.23,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.8,
|
| 36 |
+
"eval_accuracy": 0.9034639835811182,
|
| 37 |
+
"eval_f1": 0.49651100375738055,
|
| 38 |
+
"eval_loss": 0.21028906106948853,
|
| 39 |
+
"eval_precision": 0.5048208113516464,
|
| 40 |
+
"eval_recall": 0.48847033972892095,
|
| 41 |
+
"eval_runtime": 4.2572,
|
| 42 |
+
"eval_samples_per_second": 117.449,
|
| 43 |
+
"eval_steps_per_second": 29.362,
|
| 44 |
+
"step": 200
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 1.2,
|
| 48 |
+
"learning_rate": 2.64e-05,
|
| 49 |
+
"loss": 0.2013,
|
| 50 |
+
"step": 300
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.2,
|
| 54 |
+
"eval_accuracy": 0.8792611503228713,
|
| 55 |
+
"eval_f1": 0.5627305035874741,
|
| 56 |
+
"eval_loss": 0.2621181607246399,
|
| 57 |
+
"eval_precision": 0.4545060658578856,
|
| 58 |
+
"eval_recall": 0.7386023587396585,
|
| 59 |
+
"eval_runtime": 4.1357,
|
| 60 |
+
"eval_samples_per_second": 120.897,
|
| 61 |
+
"eval_steps_per_second": 30.224,
|
| 62 |
+
"step": 300
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.6,
|
| 66 |
+
"learning_rate": 2.52e-05,
|
| 67 |
+
"loss": 0.1874,
|
| 68 |
+
"step": 400
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.6,
|
| 72 |
+
"eval_accuracy": 0.893452470340892,
|
| 73 |
+
"eval_f1": 0.5619149696320114,
|
| 74 |
+
"eval_loss": 0.2326020449399948,
|
| 75 |
+
"eval_precision": 0.47293721433726243,
|
| 76 |
+
"eval_recall": 0.6921316669600422,
|
| 77 |
+
"eval_runtime": 4.241,
|
| 78 |
+
"eval_samples_per_second": 117.897,
|
| 79 |
+
"eval_steps_per_second": 29.474,
|
| 80 |
+
"step": 400
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 2.0,
|
| 84 |
+
"learning_rate": 2.4e-05,
|
| 85 |
+
"loss": 0.1847,
|
| 86 |
+
"step": 500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 2.0,
|
| 90 |
+
"eval_accuracy": 0.9060920058066777,
|
| 91 |
+
"eval_f1": 0.5619039721369932,
|
| 92 |
+
"eval_loss": 0.20794397592544556,
|
| 93 |
+
"eval_precision": 0.5312010034493572,
|
| 94 |
+
"eval_recall": 0.5963738778384088,
|
| 95 |
+
"eval_runtime": 4.124,
|
| 96 |
+
"eval_samples_per_second": 121.24,
|
| 97 |
+
"eval_steps_per_second": 30.31,
|
| 98 |
+
"step": 500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 2.4,
|
| 102 |
+
"learning_rate": 2.2800000000000002e-05,
|
| 103 |
+
"loss": 0.1567,
|
| 104 |
+
"step": 600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 2.4,
|
| 108 |
+
"eval_accuracy": 0.9071932722631025,
|
| 109 |
+
"eval_f1": 0.503008186211658,
|
| 110 |
+
"eval_loss": 0.23015139997005463,
|
| 111 |
+
"eval_precision": 0.5720053835800808,
|
| 112 |
+
"eval_recall": 0.4488646365076571,
|
| 113 |
+
"eval_runtime": 4.1201,
|
| 114 |
+
"eval_samples_per_second": 121.355,
|
| 115 |
+
"eval_steps_per_second": 30.339,
|
| 116 |
+
"step": 600
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 2.8,
|
| 120 |
+
"learning_rate": 2.16e-05,
|
| 121 |
+
"loss": 0.1484,
|
| 122 |
+
"step": 700
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 2.8,
|
| 126 |
+
"eval_accuracy": 0.9038269009360765,
|
| 127 |
+
"eval_f1": 0.5580642412882338,
|
| 128 |
+
"eval_loss": 0.22997109591960907,
|
| 129 |
+
"eval_precision": 0.540785997357992,
|
| 130 |
+
"eval_recall": 0.5764830135539518,
|
| 131 |
+
"eval_runtime": 4.134,
|
| 132 |
+
"eval_samples_per_second": 120.948,
|
| 133 |
+
"eval_steps_per_second": 30.237,
|
| 134 |
+
"step": 700
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 3.2,
|
| 138 |
+
"learning_rate": 2.04e-05,
|
| 139 |
+
"loss": 0.1388,
|
| 140 |
+
"step": 800
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 3.2,
|
| 144 |
+
"eval_accuracy": 0.9075561896180607,
|
| 145 |
+
"eval_f1": 0.5364304509572634,
|
| 146 |
+
"eval_loss": 0.2365296632051468,
|
| 147 |
+
"eval_precision": 0.5535580524344569,
|
| 148 |
+
"eval_recall": 0.5203309276535821,
|
| 149 |
+
"eval_runtime": 4.1587,
|
| 150 |
+
"eval_samples_per_second": 120.229,
|
| 151 |
+
"eval_steps_per_second": 30.057,
|
| 152 |
+
"step": 800
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 3.6,
|
| 156 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 157 |
+
"loss": 0.1191,
|
| 158 |
+
"step": 900
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 3.6,
|
| 162 |
+
"eval_accuracy": 0.9067302397757421,
|
| 163 |
+
"eval_f1": 0.574726200505476,
|
| 164 |
+
"eval_loss": 0.26086461544036865,
|
| 165 |
+
"eval_precision": 0.5511391177896268,
|
| 166 |
+
"eval_recall": 0.6004224608343601,
|
| 167 |
+
"eval_runtime": 4.1544,
|
| 168 |
+
"eval_samples_per_second": 120.354,
|
| 169 |
+
"eval_steps_per_second": 30.088,
|
| 170 |
+
"step": 900
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 4.0,
|
| 174 |
+
"learning_rate": 1.8e-05,
|
| 175 |
+
"loss": 0.1193,
|
| 176 |
+
"step": 1000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 4.0,
|
| 180 |
+
"eval_accuracy": 0.9059543474996246,
|
| 181 |
+
"eval_f1": 0.5809305373525557,
|
| 182 |
+
"eval_loss": 0.25283825397491455,
|
| 183 |
+
"eval_precision": 0.543281752719473,
|
| 184 |
+
"eval_recall": 0.6241858827671185,
|
| 185 |
+
"eval_runtime": 4.1474,
|
| 186 |
+
"eval_samples_per_second": 120.557,
|
| 187 |
+
"eval_steps_per_second": 30.139,
|
| 188 |
+
"step": 1000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 4.4,
|
| 192 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 193 |
+
"loss": 0.088,
|
| 194 |
+
"step": 1100
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 4.4,
|
| 198 |
+
"eval_accuracy": 0.9037142714121239,
|
| 199 |
+
"eval_f1": 0.5845009103142563,
|
| 200 |
+
"eval_loss": 0.2839806079864502,
|
| 201 |
+
"eval_precision": 0.5310701956271576,
|
| 202 |
+
"eval_recall": 0.6498855835240275,
|
| 203 |
+
"eval_runtime": 4.1556,
|
| 204 |
+
"eval_samples_per_second": 120.318,
|
| 205 |
+
"eval_steps_per_second": 30.08,
|
| 206 |
+
"step": 1100
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 4.8,
|
| 210 |
+
"learning_rate": 1.56e-05,
|
| 211 |
+
"loss": 0.0924,
|
| 212 |
+
"step": 1200
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 4.8,
|
| 216 |
+
"eval_accuracy": 0.9085197977674325,
|
| 217 |
+
"eval_f1": 0.5776627856834843,
|
| 218 |
+
"eval_loss": 0.27629220485687256,
|
| 219 |
+
"eval_precision": 0.5662833953331079,
|
| 220 |
+
"eval_recall": 0.5895088892800563,
|
| 221 |
+
"eval_runtime": 4.1675,
|
| 222 |
+
"eval_samples_per_second": 119.975,
|
| 223 |
+
"eval_steps_per_second": 29.994,
|
| 224 |
+
"step": 1200
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 5.2,
|
| 228 |
+
"learning_rate": 1.44e-05,
|
| 229 |
+
"loss": 0.0834,
|
| 230 |
+
"step": 1300
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 5.2,
|
| 234 |
+
"eval_accuracy": 0.9037267858036743,
|
| 235 |
+
"eval_f1": 0.5866475003992974,
|
| 236 |
+
"eval_loss": 0.332010954618454,
|
| 237 |
+
"eval_precision": 0.5369098085075281,
|
| 238 |
+
"eval_recall": 0.6465411019186763,
|
| 239 |
+
"eval_runtime": 4.1738,
|
| 240 |
+
"eval_samples_per_second": 119.795,
|
| 241 |
+
"eval_steps_per_second": 29.949,
|
| 242 |
+
"step": 1300
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 5.6,
|
| 246 |
+
"learning_rate": 1.32e-05,
|
| 247 |
+
"loss": 0.0654,
|
| 248 |
+
"step": 1400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 5.6,
|
| 252 |
+
"eval_accuracy": 0.9057416028432698,
|
| 253 |
+
"eval_f1": 0.574710687542546,
|
| 254 |
+
"eval_loss": 0.32423922419548035,
|
| 255 |
+
"eval_precision": 0.5562510294844342,
|
| 256 |
+
"eval_recall": 0.5944375990142581,
|
| 257 |
+
"eval_runtime": 4.1539,
|
| 258 |
+
"eval_samples_per_second": 120.368,
|
| 259 |
+
"eval_steps_per_second": 30.092,
|
| 260 |
+
"step": 1400
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 6.0,
|
| 264 |
+
"learning_rate": 1.2e-05,
|
| 265 |
+
"loss": 0.0689,
|
| 266 |
+
"step": 1500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 6.0,
|
| 270 |
+
"eval_accuracy": 0.9046403363868448,
|
| 271 |
+
"eval_f1": 0.5581112750629285,
|
| 272 |
+
"eval_loss": 0.31789475679397583,
|
| 273 |
+
"eval_precision": 0.550513698630137,
|
| 274 |
+
"eval_recall": 0.5659214926949481,
|
| 275 |
+
"eval_runtime": 4.1716,
|
| 276 |
+
"eval_samples_per_second": 119.859,
|
| 277 |
+
"eval_steps_per_second": 29.965,
|
| 278 |
+
"step": 1500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 6.4,
|
| 282 |
+
"learning_rate": 1.08e-05,
|
| 283 |
+
"loss": 0.0498,
|
| 284 |
+
"step": 1600
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 6.4,
|
| 288 |
+
"eval_accuracy": 0.9053661710967613,
|
| 289 |
+
"eval_f1": 0.5820808768579258,
|
| 290 |
+
"eval_loss": 0.38915345072746277,
|
| 291 |
+
"eval_precision": 0.5509273813266269,
|
| 292 |
+
"eval_recall": 0.6169688435134659,
|
| 293 |
+
"eval_runtime": 4.1814,
|
| 294 |
+
"eval_samples_per_second": 119.577,
|
| 295 |
+
"eval_steps_per_second": 29.894,
|
| 296 |
+
"step": 1600
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 6.8,
|
| 300 |
+
"learning_rate": 9.600000000000001e-06,
|
| 301 |
+
"loss": 0.0528,
|
| 302 |
+
"step": 1700
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 6.8,
|
| 306 |
+
"eval_accuracy": 0.9048155378685488,
|
| 307 |
+
"eval_f1": 0.5776866283839212,
|
| 308 |
+
"eval_loss": 0.3601633608341217,
|
| 309 |
+
"eval_precision": 0.5409433092640958,
|
| 310 |
+
"eval_recall": 0.619785249075867,
|
| 311 |
+
"eval_runtime": 4.1473,
|
| 312 |
+
"eval_samples_per_second": 120.56,
|
| 313 |
+
"eval_steps_per_second": 30.14,
|
| 314 |
+
"step": 1700
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 7.2,
|
| 318 |
+
"learning_rate": 8.400000000000001e-06,
|
| 319 |
+
"loss": 0.0474,
|
| 320 |
+
"step": 1800
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 7.2,
|
| 324 |
+
"eval_accuracy": 0.9040396455924313,
|
| 325 |
+
"eval_f1": 0.5793253173012691,
|
| 326 |
+
"eval_loss": 0.39758625626564026,
|
| 327 |
+
"eval_precision": 0.5510722795869738,
|
| 328 |
+
"eval_recall": 0.6106319309980637,
|
| 329 |
+
"eval_runtime": 4.1737,
|
| 330 |
+
"eval_samples_per_second": 119.798,
|
| 331 |
+
"eval_steps_per_second": 29.949,
|
| 332 |
+
"step": 1800
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 7.6,
|
| 336 |
+
"learning_rate": 7.2e-06,
|
| 337 |
+
"loss": 0.039,
|
| 338 |
+
"step": 1900
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 7.6,
|
| 342 |
+
"eval_accuracy": 0.9035766131050709,
|
| 343 |
+
"eval_f1": 0.5778368499750789,
|
| 344 |
+
"eval_loss": 0.4138449728488922,
|
| 345 |
+
"eval_precision": 0.5471134182790625,
|
| 346 |
+
"eval_recall": 0.6122161591269143,
|
| 347 |
+
"eval_runtime": 4.1525,
|
| 348 |
+
"eval_samples_per_second": 120.408,
|
| 349 |
+
"eval_steps_per_second": 30.102,
|
| 350 |
+
"step": 1900
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 8.0,
|
| 354 |
+
"learning_rate": 6e-06,
|
| 355 |
+
"loss": 0.0446,
|
| 356 |
+
"step": 2000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 8.0,
|
| 360 |
+
"eval_accuracy": 0.9039520448515793,
|
| 361 |
+
"eval_f1": 0.5882447535579319,
|
| 362 |
+
"eval_loss": 0.408151775598526,
|
| 363 |
+
"eval_precision": 0.5414446417998816,
|
| 364 |
+
"eval_recall": 0.6439007217039253,
|
| 365 |
+
"eval_runtime": 4.1562,
|
| 366 |
+
"eval_samples_per_second": 120.303,
|
| 367 |
+
"eval_steps_per_second": 30.076,
|
| 368 |
+
"step": 2000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 8.4,
|
| 372 |
+
"learning_rate": 4.800000000000001e-06,
|
| 373 |
+
"loss": 0.0333,
|
| 374 |
+
"step": 2100
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 8.4,
|
| 378 |
+
"eval_accuracy": 0.9046528507783952,
|
| 379 |
+
"eval_f1": 0.5720617062984743,
|
| 380 |
+
"eval_loss": 0.4318484365940094,
|
| 381 |
+
"eval_precision": 0.5545274289491078,
|
| 382 |
+
"eval_recall": 0.5907410667136067,
|
| 383 |
+
"eval_runtime": 4.1724,
|
| 384 |
+
"eval_samples_per_second": 119.834,
|
| 385 |
+
"eval_steps_per_second": 29.959,
|
| 386 |
+
"step": 2100
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 8.8,
|
| 390 |
+
"learning_rate": 3.6e-06,
|
| 391 |
+
"loss": 0.0327,
|
| 392 |
+
"step": 2200
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 8.8,
|
| 396 |
+
"eval_accuracy": 0.9054913150122641,
|
| 397 |
+
"eval_f1": 0.5734657499363381,
|
| 398 |
+
"eval_loss": 0.4232546091079712,
|
| 399 |
+
"eval_precision": 0.5537704918032786,
|
| 400 |
+
"eval_recall": 0.5946136243619081,
|
| 401 |
+
"eval_runtime": 4.1536,
|
| 402 |
+
"eval_samples_per_second": 120.378,
|
| 403 |
+
"eval_steps_per_second": 30.095,
|
| 404 |
+
"step": 2200
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 9.2,
|
| 408 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 409 |
+
"loss": 0.03,
|
| 410 |
+
"step": 2300
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 9.2,
|
| 414 |
+
"eval_accuracy": 0.9049782249587025,
|
| 415 |
+
"eval_f1": 0.5769523005487548,
|
| 416 |
+
"eval_loss": 0.44003215432167053,
|
| 417 |
+
"eval_precision": 0.5543478260869565,
|
| 418 |
+
"eval_recall": 0.6014786129202605,
|
| 419 |
+
"eval_runtime": 4.2605,
|
| 420 |
+
"eval_samples_per_second": 117.358,
|
| 421 |
+
"eval_steps_per_second": 29.339,
|
| 422 |
+
"step": 2300
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 9.6,
|
| 426 |
+
"learning_rate": 1.2000000000000002e-06,
|
| 427 |
+
"loss": 0.0286,
|
| 428 |
+
"step": 2400
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 9.6,
|
| 432 |
+
"eval_accuracy": 0.9048280522600991,
|
| 433 |
+
"eval_f1": 0.5807528586929305,
|
| 434 |
+
"eval_loss": 0.4442707598209381,
|
| 435 |
+
"eval_precision": 0.5522222222222222,
|
| 436 |
+
"eval_recall": 0.6123921844745643,
|
| 437 |
+
"eval_runtime": 4.3652,
|
| 438 |
+
"eval_samples_per_second": 114.542,
|
| 439 |
+
"eval_steps_per_second": 28.636,
|
| 440 |
+
"step": 2400
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 10.0,
|
| 444 |
+
"learning_rate": 0.0,
|
| 445 |
+
"loss": 0.0261,
|
| 446 |
+
"step": 2500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 10.0,
|
| 450 |
+
"eval_accuracy": 0.9050407969164539,
|
| 451 |
+
"eval_f1": 0.5811535881958416,
|
| 452 |
+
"eval_loss": 0.4490407407283783,
|
| 453 |
+
"eval_precision": 0.5548263166319833,
|
| 454 |
+
"eval_recall": 0.6101038549551135,
|
| 455 |
+
"eval_runtime": 4.2364,
|
| 456 |
+
"eval_samples_per_second": 118.025,
|
| 457 |
+
"eval_steps_per_second": 29.506,
|
| 458 |
+
"step": 2500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 10.0,
|
| 462 |
+
"step": 2500,
|
| 463 |
+
"total_flos": 2612991191040000.0,
|
| 464 |
+
"train_loss": 0.10352115373611451,
|
| 465 |
+
"train_runtime": 364.2147,
|
| 466 |
+
"train_samples_per_second": 27.456,
|
| 467 |
+
"train_steps_per_second": 6.864
|
| 468 |
+
}
|
| 469 |
+
],
|
| 470 |
+
"max_steps": 2500,
|
| 471 |
+
"num_train_epochs": 10,
|
| 472 |
+
"total_flos": 2612991191040000.0,
|
| 473 |
+
"trial_name": null,
|
| 474 |
+
"trial_params": null
|
| 475 |
+
}
|