End of training
Browse files- all_results.json +6 -1
- train_results.json +7 -0
- trainer_state.json +382 -0
all_results.json
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
{
|
|
|
|
| 2 |
"predict_f1": 0.9117,
|
| 3 |
"predict_gen_len": 27.57534850891124,
|
| 4 |
"predict_loss": 1.5995206832885742,
|
|
@@ -11,5 +12,9 @@
|
|
| 11 |
"predict_runtime": 1531.3156,
|
| 12 |
"predict_samples": 11334,
|
| 13 |
"predict_samples_per_second": 7.401,
|
| 14 |
-
"predict_steps_per_second": 0.463
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 16.0,
|
| 3 |
"predict_f1": 0.9117,
|
| 4 |
"predict_gen_len": 27.57534850891124,
|
| 5 |
"predict_loss": 1.5995206832885742,
|
|
|
|
| 12 |
"predict_runtime": 1531.3156,
|
| 13 |
"predict_samples": 11334,
|
| 14 |
"predict_samples_per_second": 7.401,
|
| 15 |
+
"predict_steps_per_second": 0.463,
|
| 16 |
+
"train_loss": 0.8733468595713434,
|
| 17 |
+
"train_runtime": 22758.1802,
|
| 18 |
+
"train_samples_per_second": 35.152,
|
| 19 |
+
"train_steps_per_second": 0.366
|
| 20 |
}
|
train_results.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 16.0,
|
| 3 |
+
"train_loss": 0.8733468595713434,
|
| 4 |
+
"train_runtime": 22758.1802,
|
| 5 |
+
"train_samples_per_second": 35.152,
|
| 6 |
+
"train_steps_per_second": 0.366
|
| 7 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,382 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 16.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 8336,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.96,
|
| 13 |
+
"learning_rate": 1.9360204734484968e-05,
|
| 14 |
+
"loss": 2.0443,
|
| 15 |
+
"step": 500
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 1.0,
|
| 19 |
+
"eval_f1": 0.9049,
|
| 20 |
+
"eval_gen_len": 28.363272727272726,
|
| 21 |
+
"eval_loss": 1.7046316862106323,
|
| 22 |
+
"eval_precision": 0.9041,
|
| 23 |
+
"eval_recall": 0.9061,
|
| 24 |
+
"eval_rouge1": 0.4488,
|
| 25 |
+
"eval_rouge2": 0.203,
|
| 26 |
+
"eval_rougeL": 0.3633,
|
| 27 |
+
"eval_rougeLsum": 0.3633,
|
| 28 |
+
"eval_runtime": 577.3748,
|
| 29 |
+
"eval_samples_per_second": 4.763,
|
| 30 |
+
"eval_steps_per_second": 0.298,
|
| 31 |
+
"step": 521
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 1.92,
|
| 35 |
+
"learning_rate": 1.872040946896993e-05,
|
| 36 |
+
"loss": 1.7826,
|
| 37 |
+
"step": 1000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 2.0,
|
| 41 |
+
"eval_f1": 0.9072,
|
| 42 |
+
"eval_gen_len": 28.19490909090909,
|
| 43 |
+
"eval_loss": 1.6347475051879883,
|
| 44 |
+
"eval_precision": 0.9062,
|
| 45 |
+
"eval_recall": 0.9085,
|
| 46 |
+
"eval_rouge1": 0.4616,
|
| 47 |
+
"eval_rouge2": 0.2133,
|
| 48 |
+
"eval_rougeL": 0.3761,
|
| 49 |
+
"eval_rougeLsum": 0.3758,
|
| 50 |
+
"eval_runtime": 524.6485,
|
| 51 |
+
"eval_samples_per_second": 5.242,
|
| 52 |
+
"eval_steps_per_second": 0.328,
|
| 53 |
+
"step": 1042
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 2.88,
|
| 57 |
+
"learning_rate": 1.8080614203454897e-05,
|
| 58 |
+
"loss": 1.7134,
|
| 59 |
+
"step": 1500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 3.0,
|
| 63 |
+
"eval_f1": 0.9084,
|
| 64 |
+
"eval_gen_len": 28.521818181818183,
|
| 65 |
+
"eval_loss": 1.5991039276123047,
|
| 66 |
+
"eval_precision": 0.9072,
|
| 67 |
+
"eval_recall": 0.91,
|
| 68 |
+
"eval_rouge1": 0.4683,
|
| 69 |
+
"eval_rouge2": 0.2186,
|
| 70 |
+
"eval_rougeL": 0.3824,
|
| 71 |
+
"eval_rougeLsum": 0.3822,
|
| 72 |
+
"eval_runtime": 539.0316,
|
| 73 |
+
"eval_samples_per_second": 5.102,
|
| 74 |
+
"eval_steps_per_second": 0.319,
|
| 75 |
+
"step": 1563
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 3.84,
|
| 79 |
+
"learning_rate": 1.744081893793986e-05,
|
| 80 |
+
"loss": 1.6664,
|
| 81 |
+
"step": 2000
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 4.0,
|
| 85 |
+
"eval_f1": 0.9096,
|
| 86 |
+
"eval_gen_len": 28.24981818181818,
|
| 87 |
+
"eval_loss": 1.5767467021942139,
|
| 88 |
+
"eval_precision": 0.9087,
|
| 89 |
+
"eval_recall": 0.9109,
|
| 90 |
+
"eval_rouge1": 0.4738,
|
| 91 |
+
"eval_rouge2": 0.2233,
|
| 92 |
+
"eval_rougeL": 0.3878,
|
| 93 |
+
"eval_rougeLsum": 0.3876,
|
| 94 |
+
"eval_runtime": 529.9968,
|
| 95 |
+
"eval_samples_per_second": 5.189,
|
| 96 |
+
"eval_steps_per_second": 0.325,
|
| 97 |
+
"step": 2084
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 4.8,
|
| 101 |
+
"learning_rate": 1.6801023672424827e-05,
|
| 102 |
+
"loss": 1.6296,
|
| 103 |
+
"step": 2500
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 5.0,
|
| 107 |
+
"eval_f1": 0.9103,
|
| 108 |
+
"eval_gen_len": 28.239636363636365,
|
| 109 |
+
"eval_loss": 1.5595422983169556,
|
| 110 |
+
"eval_precision": 0.9093,
|
| 111 |
+
"eval_recall": 0.9117,
|
| 112 |
+
"eval_rouge1": 0.4775,
|
| 113 |
+
"eval_rouge2": 0.2265,
|
| 114 |
+
"eval_rougeL": 0.3911,
|
| 115 |
+
"eval_rougeLsum": 0.391,
|
| 116 |
+
"eval_runtime": 526.5193,
|
| 117 |
+
"eval_samples_per_second": 5.223,
|
| 118 |
+
"eval_steps_per_second": 0.327,
|
| 119 |
+
"step": 2605
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 5.76,
|
| 123 |
+
"learning_rate": 1.616122840690979e-05,
|
| 124 |
+
"loss": 1.5984,
|
| 125 |
+
"step": 3000
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 6.0,
|
| 129 |
+
"eval_f1": 0.9109,
|
| 130 |
+
"eval_gen_len": 28.28,
|
| 131 |
+
"eval_loss": 1.5468252897262573,
|
| 132 |
+
"eval_precision": 0.9098,
|
| 133 |
+
"eval_recall": 0.9124,
|
| 134 |
+
"eval_rouge1": 0.4805,
|
| 135 |
+
"eval_rouge2": 0.2284,
|
| 136 |
+
"eval_rougeL": 0.3941,
|
| 137 |
+
"eval_rougeLsum": 0.3938,
|
| 138 |
+
"eval_runtime": 512.3397,
|
| 139 |
+
"eval_samples_per_second": 5.368,
|
| 140 |
+
"eval_steps_per_second": 0.336,
|
| 141 |
+
"step": 3126
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 6.72,
|
| 145 |
+
"learning_rate": 1.5521433141394756e-05,
|
| 146 |
+
"loss": 1.5738,
|
| 147 |
+
"step": 3500
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 7.0,
|
| 151 |
+
"eval_f1": 0.9113,
|
| 152 |
+
"eval_gen_len": 27.837818181818182,
|
| 153 |
+
"eval_loss": 1.5370196104049683,
|
| 154 |
+
"eval_precision": 0.9105,
|
| 155 |
+
"eval_recall": 0.9124,
|
| 156 |
+
"eval_rouge1": 0.4807,
|
| 157 |
+
"eval_rouge2": 0.2296,
|
| 158 |
+
"eval_rougeL": 0.3945,
|
| 159 |
+
"eval_rougeLsum": 0.3946,
|
| 160 |
+
"eval_runtime": 509.6023,
|
| 161 |
+
"eval_samples_per_second": 5.396,
|
| 162 |
+
"eval_steps_per_second": 0.338,
|
| 163 |
+
"step": 3647
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 7.68,
|
| 167 |
+
"learning_rate": 1.0403071017274472e-05,
|
| 168 |
+
"loss": 1.5476,
|
| 169 |
+
"step": 4000
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 8.0,
|
| 173 |
+
"eval_f1": 0.9114,
|
| 174 |
+
"eval_gen_len": 27.736363636363638,
|
| 175 |
+
"eval_loss": 1.530755639076233,
|
| 176 |
+
"eval_precision": 0.9108,
|
| 177 |
+
"eval_recall": 0.9125,
|
| 178 |
+
"eval_rouge1": 0.4823,
|
| 179 |
+
"eval_rouge2": 0.2315,
|
| 180 |
+
"eval_rougeL": 0.3963,
|
| 181 |
+
"eval_rougeLsum": 0.3965,
|
| 182 |
+
"eval_runtime": 510.2185,
|
| 183 |
+
"eval_samples_per_second": 5.39,
|
| 184 |
+
"eval_steps_per_second": 0.337,
|
| 185 |
+
"step": 4168
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 8.64,
|
| 189 |
+
"learning_rate": 9.203454894433782e-06,
|
| 190 |
+
"loss": 1.535,
|
| 191 |
+
"step": 4500
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 9.0,
|
| 195 |
+
"eval_f1": 0.9116,
|
| 196 |
+
"eval_gen_len": 27.653454545454544,
|
| 197 |
+
"eval_loss": 1.5260871648788452,
|
| 198 |
+
"eval_precision": 0.911,
|
| 199 |
+
"eval_recall": 0.9125,
|
| 200 |
+
"eval_rouge1": 0.4829,
|
| 201 |
+
"eval_rouge2": 0.2309,
|
| 202 |
+
"eval_rougeL": 0.3974,
|
| 203 |
+
"eval_rougeLsum": 0.3974,
|
| 204 |
+
"eval_runtime": 503.2649,
|
| 205 |
+
"eval_samples_per_second": 5.464,
|
| 206 |
+
"eval_steps_per_second": 0.342,
|
| 207 |
+
"step": 4689
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 9.6,
|
| 211 |
+
"learning_rate": 8.003838771593091e-06,
|
| 212 |
+
"loss": 1.52,
|
| 213 |
+
"step": 5000
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 10.0,
|
| 217 |
+
"eval_f1": 0.9117,
|
| 218 |
+
"eval_gen_len": 27.816,
|
| 219 |
+
"eval_loss": 1.52312433719635,
|
| 220 |
+
"eval_precision": 0.911,
|
| 221 |
+
"eval_recall": 0.9128,
|
| 222 |
+
"eval_rouge1": 0.4847,
|
| 223 |
+
"eval_rouge2": 0.2332,
|
| 224 |
+
"eval_rougeL": 0.3992,
|
| 225 |
+
"eval_rougeLsum": 0.3993,
|
| 226 |
+
"eval_runtime": 522.9989,
|
| 227 |
+
"eval_samples_per_second": 5.258,
|
| 228 |
+
"eval_steps_per_second": 0.329,
|
| 229 |
+
"step": 5210
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 10.56,
|
| 233 |
+
"learning_rate": 6.8042226487524e-06,
|
| 234 |
+
"loss": 1.5145,
|
| 235 |
+
"step": 5500
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 11.0,
|
| 239 |
+
"eval_f1": 0.9121,
|
| 240 |
+
"eval_gen_len": 27.360363636363637,
|
| 241 |
+
"eval_loss": 1.519996166229248,
|
| 242 |
+
"eval_precision": 0.9119,
|
| 243 |
+
"eval_recall": 0.9127,
|
| 244 |
+
"eval_rouge1": 0.4851,
|
| 245 |
+
"eval_rouge2": 0.2339,
|
| 246 |
+
"eval_rougeL": 0.4004,
|
| 247 |
+
"eval_rougeLsum": 0.4006,
|
| 248 |
+
"eval_runtime": 501.564,
|
| 249 |
+
"eval_samples_per_second": 5.483,
|
| 250 |
+
"eval_steps_per_second": 0.343,
|
| 251 |
+
"step": 5731
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 11.52,
|
| 255 |
+
"learning_rate": 5.6046065259117085e-06,
|
| 256 |
+
"loss": 1.5028,
|
| 257 |
+
"step": 6000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 12.0,
|
| 261 |
+
"eval_f1": 0.9122,
|
| 262 |
+
"eval_gen_len": 27.462545454545456,
|
| 263 |
+
"eval_loss": 1.5178437232971191,
|
| 264 |
+
"eval_precision": 0.9118,
|
| 265 |
+
"eval_recall": 0.9129,
|
| 266 |
+
"eval_rouge1": 0.4858,
|
| 267 |
+
"eval_rouge2": 0.2345,
|
| 268 |
+
"eval_rougeL": 0.4001,
|
| 269 |
+
"eval_rougeLsum": 0.4002,
|
| 270 |
+
"eval_runtime": 501.8356,
|
| 271 |
+
"eval_samples_per_second": 5.48,
|
| 272 |
+
"eval_steps_per_second": 0.343,
|
| 273 |
+
"step": 6252
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 12.48,
|
| 277 |
+
"learning_rate": 4.404990403071018e-06,
|
| 278 |
+
"loss": 1.4946,
|
| 279 |
+
"step": 6500
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 13.0,
|
| 283 |
+
"eval_f1": 0.9121,
|
| 284 |
+
"eval_gen_len": 27.67890909090909,
|
| 285 |
+
"eval_loss": 1.5164216756820679,
|
| 286 |
+
"eval_precision": 0.9115,
|
| 287 |
+
"eval_recall": 0.9131,
|
| 288 |
+
"eval_rouge1": 0.4859,
|
| 289 |
+
"eval_rouge2": 0.2341,
|
| 290 |
+
"eval_rougeL": 0.4004,
|
| 291 |
+
"eval_rougeLsum": 0.4005,
|
| 292 |
+
"eval_runtime": 506.9944,
|
| 293 |
+
"eval_samples_per_second": 5.424,
|
| 294 |
+
"eval_steps_per_second": 0.339,
|
| 295 |
+
"step": 6773
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 13.44,
|
| 299 |
+
"learning_rate": 3.2053742802303266e-06,
|
| 300 |
+
"loss": 1.4877,
|
| 301 |
+
"step": 7000
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 14.0,
|
| 305 |
+
"eval_f1": 0.9123,
|
| 306 |
+
"eval_gen_len": 27.580363636363636,
|
| 307 |
+
"eval_loss": 1.515085220336914,
|
| 308 |
+
"eval_precision": 0.9119,
|
| 309 |
+
"eval_recall": 0.9131,
|
| 310 |
+
"eval_rouge1": 0.4868,
|
| 311 |
+
"eval_rouge2": 0.235,
|
| 312 |
+
"eval_rougeL": 0.4013,
|
| 313 |
+
"eval_rougeLsum": 0.4013,
|
| 314 |
+
"eval_runtime": 510.129,
|
| 315 |
+
"eval_samples_per_second": 5.391,
|
| 316 |
+
"eval_steps_per_second": 0.337,
|
| 317 |
+
"step": 7294
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 14.4,
|
| 321 |
+
"learning_rate": 2.0057581573896352e-06,
|
| 322 |
+
"loss": 1.4855,
|
| 323 |
+
"step": 7500
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 15.0,
|
| 327 |
+
"eval_f1": 0.9122,
|
| 328 |
+
"eval_gen_len": 27.584363636363637,
|
| 329 |
+
"eval_loss": 1.5146222114562988,
|
| 330 |
+
"eval_precision": 0.9117,
|
| 331 |
+
"eval_recall": 0.9131,
|
| 332 |
+
"eval_rouge1": 0.4863,
|
| 333 |
+
"eval_rouge2": 0.2349,
|
| 334 |
+
"eval_rougeL": 0.4014,
|
| 335 |
+
"eval_rougeLsum": 0.4016,
|
| 336 |
+
"eval_runtime": 507.3504,
|
| 337 |
+
"eval_samples_per_second": 5.42,
|
| 338 |
+
"eval_steps_per_second": 0.339,
|
| 339 |
+
"step": 7815
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 15.36,
|
| 343 |
+
"learning_rate": 8.061420345489445e-07,
|
| 344 |
+
"loss": 1.4782,
|
| 345 |
+
"step": 8000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 16.0,
|
| 349 |
+
"eval_f1": 0.9122,
|
| 350 |
+
"eval_gen_len": 27.571636363636365,
|
| 351 |
+
"eval_loss": 1.514625906944275,
|
| 352 |
+
"eval_precision": 0.9118,
|
| 353 |
+
"eval_recall": 0.9131,
|
| 354 |
+
"eval_rouge1": 0.4863,
|
| 355 |
+
"eval_rouge2": 0.2348,
|
| 356 |
+
"eval_rougeL": 0.4011,
|
| 357 |
+
"eval_rougeLsum": 0.4012,
|
| 358 |
+
"eval_runtime": 505.7467,
|
| 359 |
+
"eval_samples_per_second": 5.438,
|
| 360 |
+
"eval_steps_per_second": 0.34,
|
| 361 |
+
"step": 8336
|
| 362 |
+
},
|
| 363 |
+
{
|
| 364 |
+
"epoch": 16.0,
|
| 365 |
+
"step": 8336,
|
| 366 |
+
"total_flos": 1.1557816346520453e+18,
|
| 367 |
+
"train_loss": 0.8733468595713434,
|
| 368 |
+
"train_runtime": 22758.1802,
|
| 369 |
+
"train_samples_per_second": 35.152,
|
| 370 |
+
"train_steps_per_second": 0.366
|
| 371 |
+
}
|
| 372 |
+
],
|
| 373 |
+
"logging_steps": 500,
|
| 374 |
+
"max_steps": 8336,
|
| 375 |
+
"num_input_tokens_seen": 0,
|
| 376 |
+
"num_train_epochs": 16,
|
| 377 |
+
"save_steps": 500,
|
| 378 |
+
"total_flos": 1.1557816346520453e+18,
|
| 379 |
+
"train_batch_size": 24,
|
| 380 |
+
"trial_name": null,
|
| 381 |
+
"trial_params": null
|
| 382 |
+
}
|