Upload batch 22/24: vsfc/mBERT/max/seed_42 ... vsfc/mBERT/mean/seed_42
Browse files- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/trainer_state.json +265 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/trainer_state.json +265 -0
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9425142135186355,
|
| 4 |
+
"eval_loss": 0.2218012511730194,
|
| 5 |
+
"eval_runtime": 6.253,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 253.159,
|
| 8 |
+
"eval_steps_per_second": 2.719,
|
| 9 |
+
"total_flos": 1024166730153984.0,
|
| 10 |
+
"train_loss": 0.20883273712594472,
|
| 11 |
+
"train_runtime": 125.0864,
|
| 12 |
+
"train_samples": 11426,
|
| 13 |
+
"train_samples_per_second": 274.035,
|
| 14 |
+
"train_steps_per_second": 5.732
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9425142135186355,
|
| 4 |
+
"eval_loss": 0.2218012511730194,
|
| 5 |
+
"eval_runtime": 6.253,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 253.159,
|
| 8 |
+
"eval_steps_per_second": 2.719
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 1024166730153984.0,
|
| 4 |
+
"train_loss": 0.20883273712594472,
|
| 5 |
+
"train_runtime": 125.0864,
|
| 6 |
+
"train_samples": 11426,
|
| 7 |
+
"train_samples_per_second": 274.035,
|
| 8 |
+
"train_steps_per_second": 5.732
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/max/seed_42/trainer_state.json
ADDED
|
@@ -0,0 +1,265 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 717,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.10460251046025104,
|
| 13 |
+
"grad_norm": 11.187190055847168,
|
| 14 |
+
"learning_rate": 2.8953974895397488e-05,
|
| 15 |
+
"loss": 0.5497,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.20920502092050208,
|
| 20 |
+
"grad_norm": 6.295005798339844,
|
| 21 |
+
"learning_rate": 2.7907949790794982e-05,
|
| 22 |
+
"loss": 0.351,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3138075313807531,
|
| 27 |
+
"grad_norm": 4.0218186378479,
|
| 28 |
+
"learning_rate": 2.686192468619247e-05,
|
| 29 |
+
"loss": 0.3292,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.41841004184100417,
|
| 34 |
+
"grad_norm": 5.489821434020996,
|
| 35 |
+
"learning_rate": 2.5815899581589957e-05,
|
| 36 |
+
"loss": 0.2852,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5230125523012552,
|
| 41 |
+
"grad_norm": 3.7578203678131104,
|
| 42 |
+
"learning_rate": 2.4769874476987448e-05,
|
| 43 |
+
"loss": 0.2713,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6276150627615062,
|
| 48 |
+
"grad_norm": 5.43932580947876,
|
| 49 |
+
"learning_rate": 2.372384937238494e-05,
|
| 50 |
+
"loss": 0.2675,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7322175732217573,
|
| 55 |
+
"grad_norm": 5.287449836730957,
|
| 56 |
+
"learning_rate": 2.2677824267782426e-05,
|
| 57 |
+
"loss": 0.2631,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8368200836820083,
|
| 62 |
+
"grad_norm": 6.968525409698486,
|
| 63 |
+
"learning_rate": 2.1631799163179917e-05,
|
| 64 |
+
"loss": 0.2889,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.9414225941422594,
|
| 69 |
+
"grad_norm": 5.448462009429932,
|
| 70 |
+
"learning_rate": 2.0585774058577408e-05,
|
| 71 |
+
"loss": 0.2603,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0,
|
| 76 |
+
"eval_accuracy": 0.9292482627921668,
|
| 77 |
+
"eval_loss": 0.22762367129325867,
|
| 78 |
+
"eval_runtime": 6.4493,
|
| 79 |
+
"eval_samples_per_second": 245.454,
|
| 80 |
+
"eval_steps_per_second": 2.636,
|
| 81 |
+
"step": 239
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 1.0460251046025104,
|
| 85 |
+
"grad_norm": 6.751649856567383,
|
| 86 |
+
"learning_rate": 1.9539748953974895e-05,
|
| 87 |
+
"loss": 0.2078,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 1.1506276150627615,
|
| 92 |
+
"grad_norm": 4.543371200561523,
|
| 93 |
+
"learning_rate": 1.8493723849372386e-05,
|
| 94 |
+
"loss": 0.1812,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.2552301255230125,
|
| 99 |
+
"grad_norm": 4.131556510925293,
|
| 100 |
+
"learning_rate": 1.7447698744769877e-05,
|
| 101 |
+
"loss": 0.148,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 1.3598326359832635,
|
| 106 |
+
"grad_norm": 4.097535610198975,
|
| 107 |
+
"learning_rate": 1.6401673640167365e-05,
|
| 108 |
+
"loss": 0.2131,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.4644351464435146,
|
| 113 |
+
"grad_norm": 8.53039836883545,
|
| 114 |
+
"learning_rate": 1.5355648535564852e-05,
|
| 115 |
+
"loss": 0.2104,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.5690376569037658,
|
| 120 |
+
"grad_norm": 7.238592624664307,
|
| 121 |
+
"learning_rate": 1.4309623430962344e-05,
|
| 122 |
+
"loss": 0.2164,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.6736401673640167,
|
| 127 |
+
"grad_norm": 2.5955123901367188,
|
| 128 |
+
"learning_rate": 1.3263598326359832e-05,
|
| 129 |
+
"loss": 0.1883,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.778242677824268,
|
| 134 |
+
"grad_norm": 3.9763717651367188,
|
| 135 |
+
"learning_rate": 1.2217573221757323e-05,
|
| 136 |
+
"loss": 0.1765,
|
| 137 |
+
"step": 425
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.8828451882845187,
|
| 141 |
+
"grad_norm": 4.842983245849609,
|
| 142 |
+
"learning_rate": 1.1171548117154812e-05,
|
| 143 |
+
"loss": 0.1519,
|
| 144 |
+
"step": 450
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.98744769874477,
|
| 148 |
+
"grad_norm": 7.14610481262207,
|
| 149 |
+
"learning_rate": 1.0125523012552301e-05,
|
| 150 |
+
"loss": 0.2012,
|
| 151 |
+
"step": 475
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.0,
|
| 155 |
+
"eval_accuracy": 0.9355653821857233,
|
| 156 |
+
"eval_loss": 0.20559416711330414,
|
| 157 |
+
"eval_runtime": 6.2678,
|
| 158 |
+
"eval_samples_per_second": 252.561,
|
| 159 |
+
"eval_steps_per_second": 2.712,
|
| 160 |
+
"step": 478
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 2.092050209205021,
|
| 164 |
+
"grad_norm": 1.6934655904769897,
|
| 165 |
+
"learning_rate": 9.079497907949792e-06,
|
| 166 |
+
"loss": 0.1092,
|
| 167 |
+
"step": 500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 2.196652719665272,
|
| 171 |
+
"grad_norm": 6.796380996704102,
|
| 172 |
+
"learning_rate": 8.03347280334728e-06,
|
| 173 |
+
"loss": 0.1285,
|
| 174 |
+
"step": 525
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 2.301255230125523,
|
| 178 |
+
"grad_norm": 7.0758161544799805,
|
| 179 |
+
"learning_rate": 6.98744769874477e-06,
|
| 180 |
+
"loss": 0.152,
|
| 181 |
+
"step": 550
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 2.405857740585774,
|
| 185 |
+
"grad_norm": 2.5231029987335205,
|
| 186 |
+
"learning_rate": 5.941422594142259e-06,
|
| 187 |
+
"loss": 0.1291,
|
| 188 |
+
"step": 575
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.510460251046025,
|
| 192 |
+
"grad_norm": 3.3919241428375244,
|
| 193 |
+
"learning_rate": 4.895397489539749e-06,
|
| 194 |
+
"loss": 0.1154,
|
| 195 |
+
"step": 600
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 2.6150627615062763,
|
| 199 |
+
"grad_norm": 3.53436279296875,
|
| 200 |
+
"learning_rate": 3.849372384937238e-06,
|
| 201 |
+
"loss": 0.1372,
|
| 202 |
+
"step": 625
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 2.719665271966527,
|
| 206 |
+
"grad_norm": 7.928037643432617,
|
| 207 |
+
"learning_rate": 2.803347280334728e-06,
|
| 208 |
+
"loss": 0.1146,
|
| 209 |
+
"step": 650
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.8242677824267783,
|
| 213 |
+
"grad_norm": 1.8922146558761597,
|
| 214 |
+
"learning_rate": 1.7573221757322176e-06,
|
| 215 |
+
"loss": 0.1242,
|
| 216 |
+
"step": 675
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.928870292887029,
|
| 220 |
+
"grad_norm": 2.382826805114746,
|
| 221 |
+
"learning_rate": 7.112970711297071e-07,
|
| 222 |
+
"loss": 0.1272,
|
| 223 |
+
"step": 700
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 3.0,
|
| 227 |
+
"eval_accuracy": 0.9425142135186355,
|
| 228 |
+
"eval_loss": 0.2218012511730194,
|
| 229 |
+
"eval_runtime": 6.2312,
|
| 230 |
+
"eval_samples_per_second": 254.044,
|
| 231 |
+
"eval_steps_per_second": 2.728,
|
| 232 |
+
"step": 717
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.0,
|
| 236 |
+
"step": 717,
|
| 237 |
+
"total_flos": 1024166730153984.0,
|
| 238 |
+
"train_loss": 0.20883273712594472,
|
| 239 |
+
"train_runtime": 125.0864,
|
| 240 |
+
"train_samples_per_second": 274.035,
|
| 241 |
+
"train_steps_per_second": 5.732
|
| 242 |
+
}
|
| 243 |
+
],
|
| 244 |
+
"logging_steps": 25,
|
| 245 |
+
"max_steps": 717,
|
| 246 |
+
"num_input_tokens_seen": 0,
|
| 247 |
+
"num_train_epochs": 3,
|
| 248 |
+
"save_steps": 500,
|
| 249 |
+
"stateful_callbacks": {
|
| 250 |
+
"TrainerControl": {
|
| 251 |
+
"args": {
|
| 252 |
+
"should_epoch_stop": false,
|
| 253 |
+
"should_evaluate": false,
|
| 254 |
+
"should_log": false,
|
| 255 |
+
"should_save": false,
|
| 256 |
+
"should_training_stop": false
|
| 257 |
+
},
|
| 258 |
+
"attributes": {}
|
| 259 |
+
}
|
| 260 |
+
},
|
| 261 |
+
"total_flos": 1024166730153984.0,
|
| 262 |
+
"train_batch_size": 48,
|
| 263 |
+
"trial_name": null,
|
| 264 |
+
"trial_params": null
|
| 265 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9374605180037903,
|
| 4 |
+
"eval_loss": 0.21570518612861633,
|
| 5 |
+
"eval_runtime": 6.2345,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 253.911,
|
| 8 |
+
"eval_steps_per_second": 2.727,
|
| 9 |
+
"total_flos": 1024166730153984.0,
|
| 10 |
+
"train_loss": 0.2002154952811397,
|
| 11 |
+
"train_runtime": 124.9001,
|
| 12 |
+
"train_samples": 11426,
|
| 13 |
+
"train_samples_per_second": 274.443,
|
| 14 |
+
"train_steps_per_second": 5.741
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9374605180037903,
|
| 4 |
+
"eval_loss": 0.21570518612861633,
|
| 5 |
+
"eval_runtime": 6.2345,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 253.911,
|
| 8 |
+
"eval_steps_per_second": 2.727
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 1024166730153984.0,
|
| 4 |
+
"train_loss": 0.2002154952811397,
|
| 5 |
+
"train_runtime": 124.9001,
|
| 6 |
+
"train_samples": 11426,
|
| 7 |
+
"train_samples_per_second": 274.443,
|
| 8 |
+
"train_steps_per_second": 5.741
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/mean/seed_42/trainer_state.json
ADDED
|
@@ -0,0 +1,265 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 717,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.10460251046025104,
|
| 13 |
+
"grad_norm": 13.635787010192871,
|
| 14 |
+
"learning_rate": 2.8953974895397488e-05,
|
| 15 |
+
"loss": 0.4905,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.20920502092050208,
|
| 20 |
+
"grad_norm": 3.7826690673828125,
|
| 21 |
+
"learning_rate": 2.7907949790794982e-05,
|
| 22 |
+
"loss": 0.3171,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3138075313807531,
|
| 27 |
+
"grad_norm": 4.913433074951172,
|
| 28 |
+
"learning_rate": 2.686192468619247e-05,
|
| 29 |
+
"loss": 0.3005,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.41841004184100417,
|
| 34 |
+
"grad_norm": 5.333264350891113,
|
| 35 |
+
"learning_rate": 2.5815899581589957e-05,
|
| 36 |
+
"loss": 0.2801,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5230125523012552,
|
| 41 |
+
"grad_norm": 3.9317681789398193,
|
| 42 |
+
"learning_rate": 2.4769874476987448e-05,
|
| 43 |
+
"loss": 0.2691,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6276150627615062,
|
| 48 |
+
"grad_norm": 4.054808616638184,
|
| 49 |
+
"learning_rate": 2.372384937238494e-05,
|
| 50 |
+
"loss": 0.2596,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7322175732217573,
|
| 55 |
+
"grad_norm": 4.778711795806885,
|
| 56 |
+
"learning_rate": 2.2677824267782426e-05,
|
| 57 |
+
"loss": 0.2587,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8368200836820083,
|
| 62 |
+
"grad_norm": 6.411399841308594,
|
| 63 |
+
"learning_rate": 2.1631799163179917e-05,
|
| 64 |
+
"loss": 0.2744,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.9414225941422594,
|
| 69 |
+
"grad_norm": 6.36979866027832,
|
| 70 |
+
"learning_rate": 2.0585774058577408e-05,
|
| 71 |
+
"loss": 0.2451,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0,
|
| 76 |
+
"eval_accuracy": 0.9292482627921668,
|
| 77 |
+
"eval_loss": 0.23341195285320282,
|
| 78 |
+
"eval_runtime": 6.5244,
|
| 79 |
+
"eval_samples_per_second": 242.628,
|
| 80 |
+
"eval_steps_per_second": 2.606,
|
| 81 |
+
"step": 239
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 1.0460251046025104,
|
| 85 |
+
"grad_norm": 4.252689838409424,
|
| 86 |
+
"learning_rate": 1.9539748953974895e-05,
|
| 87 |
+
"loss": 0.2032,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 1.1506276150627615,
|
| 92 |
+
"grad_norm": 7.199467182159424,
|
| 93 |
+
"learning_rate": 1.8493723849372386e-05,
|
| 94 |
+
"loss": 0.1796,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.2552301255230125,
|
| 99 |
+
"grad_norm": 2.749007225036621,
|
| 100 |
+
"learning_rate": 1.7447698744769877e-05,
|
| 101 |
+
"loss": 0.1487,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 1.3598326359832635,
|
| 106 |
+
"grad_norm": 4.073476314544678,
|
| 107 |
+
"learning_rate": 1.6401673640167365e-05,
|
| 108 |
+
"loss": 0.2111,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.4644351464435146,
|
| 113 |
+
"grad_norm": 4.743743896484375,
|
| 114 |
+
"learning_rate": 1.5355648535564852e-05,
|
| 115 |
+
"loss": 0.2035,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.5690376569037658,
|
| 120 |
+
"grad_norm": 5.752042293548584,
|
| 121 |
+
"learning_rate": 1.4309623430962344e-05,
|
| 122 |
+
"loss": 0.2082,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.6736401673640167,
|
| 127 |
+
"grad_norm": 2.410944700241089,
|
| 128 |
+
"learning_rate": 1.3263598326359832e-05,
|
| 129 |
+
"loss": 0.1942,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.778242677824268,
|
| 134 |
+
"grad_norm": 3.687453508377075,
|
| 135 |
+
"learning_rate": 1.2217573221757323e-05,
|
| 136 |
+
"loss": 0.1665,
|
| 137 |
+
"step": 425
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.8828451882845187,
|
| 141 |
+
"grad_norm": 4.763829708099365,
|
| 142 |
+
"learning_rate": 1.1171548117154812e-05,
|
| 143 |
+
"loss": 0.1521,
|
| 144 |
+
"step": 450
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.98744769874477,
|
| 148 |
+
"grad_norm": 6.042887210845947,
|
| 149 |
+
"learning_rate": 1.0125523012552301e-05,
|
| 150 |
+
"loss": 0.1925,
|
| 151 |
+
"step": 475
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.0,
|
| 155 |
+
"eval_accuracy": 0.9374605180037903,
|
| 156 |
+
"eval_loss": 0.19666047394275665,
|
| 157 |
+
"eval_runtime": 6.2305,
|
| 158 |
+
"eval_samples_per_second": 254.073,
|
| 159 |
+
"eval_steps_per_second": 2.729,
|
| 160 |
+
"step": 478
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 2.092050209205021,
|
| 164 |
+
"grad_norm": 4.838719367980957,
|
| 165 |
+
"learning_rate": 9.079497907949792e-06,
|
| 166 |
+
"loss": 0.1232,
|
| 167 |
+
"step": 500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 2.196652719665272,
|
| 171 |
+
"grad_norm": 4.4863152503967285,
|
| 172 |
+
"learning_rate": 8.03347280334728e-06,
|
| 173 |
+
"loss": 0.1102,
|
| 174 |
+
"step": 525
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 2.301255230125523,
|
| 178 |
+
"grad_norm": 4.218593120574951,
|
| 179 |
+
"learning_rate": 6.98744769874477e-06,
|
| 180 |
+
"loss": 0.1458,
|
| 181 |
+
"step": 550
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 2.405857740585774,
|
| 185 |
+
"grad_norm": 7.19607400894165,
|
| 186 |
+
"learning_rate": 5.941422594142259e-06,
|
| 187 |
+
"loss": 0.1232,
|
| 188 |
+
"step": 575
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.510460251046025,
|
| 192 |
+
"grad_norm": 2.1715757846832275,
|
| 193 |
+
"learning_rate": 4.895397489539749e-06,
|
| 194 |
+
"loss": 0.0996,
|
| 195 |
+
"step": 600
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 2.6150627615062763,
|
| 199 |
+
"grad_norm": 2.6068990230560303,
|
| 200 |
+
"learning_rate": 3.849372384937238e-06,
|
| 201 |
+
"loss": 0.1433,
|
| 202 |
+
"step": 625
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 2.719665271966527,
|
| 206 |
+
"grad_norm": 9.071758270263672,
|
| 207 |
+
"learning_rate": 2.803347280334728e-06,
|
| 208 |
+
"loss": 0.1198,
|
| 209 |
+
"step": 650
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.8242677824267783,
|
| 213 |
+
"grad_norm": 2.628211498260498,
|
| 214 |
+
"learning_rate": 1.7573221757322176e-06,
|
| 215 |
+
"loss": 0.1132,
|
| 216 |
+
"step": 675
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.928870292887029,
|
| 220 |
+
"grad_norm": 2.436028480529785,
|
| 221 |
+
"learning_rate": 7.112970711297071e-07,
|
| 222 |
+
"loss": 0.1137,
|
| 223 |
+
"step": 700
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 3.0,
|
| 227 |
+
"eval_accuracy": 0.9374605180037903,
|
| 228 |
+
"eval_loss": 0.21570518612861633,
|
| 229 |
+
"eval_runtime": 6.243,
|
| 230 |
+
"eval_samples_per_second": 253.563,
|
| 231 |
+
"eval_steps_per_second": 2.723,
|
| 232 |
+
"step": 717
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.0,
|
| 236 |
+
"step": 717,
|
| 237 |
+
"total_flos": 1024166730153984.0,
|
| 238 |
+
"train_loss": 0.2002154952811397,
|
| 239 |
+
"train_runtime": 124.9001,
|
| 240 |
+
"train_samples_per_second": 274.443,
|
| 241 |
+
"train_steps_per_second": 5.741
|
| 242 |
+
}
|
| 243 |
+
],
|
| 244 |
+
"logging_steps": 25,
|
| 245 |
+
"max_steps": 717,
|
| 246 |
+
"num_input_tokens_seen": 0,
|
| 247 |
+
"num_train_epochs": 3,
|
| 248 |
+
"save_steps": 500,
|
| 249 |
+
"stateful_callbacks": {
|
| 250 |
+
"TrainerControl": {
|
| 251 |
+
"args": {
|
| 252 |
+
"should_epoch_stop": false,
|
| 253 |
+
"should_evaluate": false,
|
| 254 |
+
"should_log": false,
|
| 255 |
+
"should_save": false,
|
| 256 |
+
"should_training_stop": false
|
| 257 |
+
},
|
| 258 |
+
"attributes": {}
|
| 259 |
+
}
|
| 260 |
+
},
|
| 261 |
+
"total_flos": 1024166730153984.0,
|
| 262 |
+
"train_batch_size": 48,
|
| 263 |
+
"trial_name": null,
|
| 264 |
+
"trial_params": null
|
| 265 |
+
}
|