Upload batch 19/24: vsfc/mBERT/attention/seed_42 ... vsfc/mBERT/cls/seed_42
Browse files- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_42/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_42/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_42/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_42/trainer_state.json +265 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/all_results.json +15 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/eval_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/train_results.json +9 -0
- mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/trainer_state.json +265 -0
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_42/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9431459254579911,
|
| 4 |
+
"eval_loss": 0.22031328082084656,
|
| 5 |
+
"eval_runtime": 5.8967,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 268.457,
|
| 8 |
+
"eval_steps_per_second": 2.883,
|
| 9 |
+
"total_flos": 1024175883573288.0,
|
| 10 |
+
"train_loss": 0.2008532669753709,
|
| 11 |
+
"train_runtime": 123.6845,
|
| 12 |
+
"train_samples": 11426,
|
| 13 |
+
"train_samples_per_second": 277.141,
|
| 14 |
+
"train_steps_per_second": 5.797
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_42/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9431459254579911,
|
| 4 |
+
"eval_loss": 0.22031328082084656,
|
| 5 |
+
"eval_runtime": 5.8967,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 268.457,
|
| 8 |
+
"eval_steps_per_second": 2.883
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_42/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 1024175883573288.0,
|
| 4 |
+
"train_loss": 0.2008532669753709,
|
| 5 |
+
"train_runtime": 123.6845,
|
| 6 |
+
"train_samples": 11426,
|
| 7 |
+
"train_samples_per_second": 277.141,
|
| 8 |
+
"train_steps_per_second": 5.797
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/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": 13.24710750579834,
|
| 14 |
+
"learning_rate": 2.8953974895397488e-05,
|
| 15 |
+
"loss": 0.5379,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.20920502092050208,
|
| 20 |
+
"grad_norm": 5.9082722663879395,
|
| 21 |
+
"learning_rate": 2.7907949790794982e-05,
|
| 22 |
+
"loss": 0.3287,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3138075313807531,
|
| 27 |
+
"grad_norm": 6.679155349731445,
|
| 28 |
+
"learning_rate": 2.686192468619247e-05,
|
| 29 |
+
"loss": 0.3057,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.41841004184100417,
|
| 34 |
+
"grad_norm": 5.558966159820557,
|
| 35 |
+
"learning_rate": 2.5815899581589957e-05,
|
| 36 |
+
"loss": 0.2954,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5230125523012552,
|
| 41 |
+
"grad_norm": 3.1983091831207275,
|
| 42 |
+
"learning_rate": 2.4769874476987448e-05,
|
| 43 |
+
"loss": 0.2651,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6276150627615062,
|
| 48 |
+
"grad_norm": 6.8210015296936035,
|
| 49 |
+
"learning_rate": 2.372384937238494e-05,
|
| 50 |
+
"loss": 0.2658,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7322175732217573,
|
| 55 |
+
"grad_norm": 5.727506160736084,
|
| 56 |
+
"learning_rate": 2.2677824267782426e-05,
|
| 57 |
+
"loss": 0.2554,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8368200836820083,
|
| 62 |
+
"grad_norm": 6.734157562255859,
|
| 63 |
+
"learning_rate": 2.1631799163179917e-05,
|
| 64 |
+
"loss": 0.2797,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.9414225941422594,
|
| 69 |
+
"grad_norm": 7.097010135650635,
|
| 70 |
+
"learning_rate": 2.0585774058577408e-05,
|
| 71 |
+
"loss": 0.242,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0,
|
| 76 |
+
"eval_accuracy": 0.9330385344283006,
|
| 77 |
+
"eval_loss": 0.22317896783351898,
|
| 78 |
+
"eval_runtime": 6.3366,
|
| 79 |
+
"eval_samples_per_second": 249.817,
|
| 80 |
+
"eval_steps_per_second": 2.683,
|
| 81 |
+
"step": 239
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 1.0460251046025104,
|
| 85 |
+
"grad_norm": 7.102856636047363,
|
| 86 |
+
"learning_rate": 1.9539748953974895e-05,
|
| 87 |
+
"loss": 0.2125,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 1.1506276150627615,
|
| 92 |
+
"grad_norm": 6.886712551116943,
|
| 93 |
+
"learning_rate": 1.8493723849372386e-05,
|
| 94 |
+
"loss": 0.1805,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.2552301255230125,
|
| 99 |
+
"grad_norm": 3.5830013751983643,
|
| 100 |
+
"learning_rate": 1.7447698744769877e-05,
|
| 101 |
+
"loss": 0.1382,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 1.3598326359832635,
|
| 106 |
+
"grad_norm": 3.9520208835601807,
|
| 107 |
+
"learning_rate": 1.6401673640167365e-05,
|
| 108 |
+
"loss": 0.2119,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.4644351464435146,
|
| 113 |
+
"grad_norm": 3.5851409435272217,
|
| 114 |
+
"learning_rate": 1.5355648535564852e-05,
|
| 115 |
+
"loss": 0.1942,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.5690376569037658,
|
| 120 |
+
"grad_norm": 6.014364719390869,
|
| 121 |
+
"learning_rate": 1.4309623430962344e-05,
|
| 122 |
+
"loss": 0.2141,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.6736401673640167,
|
| 127 |
+
"grad_norm": 6.627466201782227,
|
| 128 |
+
"learning_rate": 1.3263598326359832e-05,
|
| 129 |
+
"loss": 0.1954,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.778242677824268,
|
| 134 |
+
"grad_norm": 3.3036744594573975,
|
| 135 |
+
"learning_rate": 1.2217573221757323e-05,
|
| 136 |
+
"loss": 0.1675,
|
| 137 |
+
"step": 425
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.8828451882845187,
|
| 141 |
+
"grad_norm": 5.270769119262695,
|
| 142 |
+
"learning_rate": 1.1171548117154812e-05,
|
| 143 |
+
"loss": 0.1388,
|
| 144 |
+
"step": 450
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.98744769874477,
|
| 148 |
+
"grad_norm": 3.9758880138397217,
|
| 149 |
+
"learning_rate": 1.0125523012552301e-05,
|
| 150 |
+
"loss": 0.1908,
|
| 151 |
+
"step": 475
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.0,
|
| 155 |
+
"eval_accuracy": 0.9380922299431459,
|
| 156 |
+
"eval_loss": 0.20354588329792023,
|
| 157 |
+
"eval_runtime": 6.323,
|
| 158 |
+
"eval_samples_per_second": 250.357,
|
| 159 |
+
"eval_steps_per_second": 2.689,
|
| 160 |
+
"step": 478
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 2.092050209205021,
|
| 164 |
+
"grad_norm": 2.238645076751709,
|
| 165 |
+
"learning_rate": 9.079497907949792e-06,
|
| 166 |
+
"loss": 0.1065,
|
| 167 |
+
"step": 500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 2.196652719665272,
|
| 171 |
+
"grad_norm": 6.814053058624268,
|
| 172 |
+
"learning_rate": 8.03347280334728e-06,
|
| 173 |
+
"loss": 0.115,
|
| 174 |
+
"step": 525
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 2.301255230125523,
|
| 178 |
+
"grad_norm": 4.267109394073486,
|
| 179 |
+
"learning_rate": 6.98744769874477e-06,
|
| 180 |
+
"loss": 0.1429,
|
| 181 |
+
"step": 550
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 2.405857740585774,
|
| 185 |
+
"grad_norm": 2.822779417037964,
|
| 186 |
+
"learning_rate": 5.941422594142259e-06,
|
| 187 |
+
"loss": 0.123,
|
| 188 |
+
"step": 575
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.510460251046025,
|
| 192 |
+
"grad_norm": 3.8507728576660156,
|
| 193 |
+
"learning_rate": 4.895397489539749e-06,
|
| 194 |
+
"loss": 0.0959,
|
| 195 |
+
"step": 600
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 2.6150627615062763,
|
| 199 |
+
"grad_norm": 3.2560744285583496,
|
| 200 |
+
"learning_rate": 3.849372384937238e-06,
|
| 201 |
+
"loss": 0.1423,
|
| 202 |
+
"step": 625
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 2.719665271966527,
|
| 206 |
+
"grad_norm": 6.082599639892578,
|
| 207 |
+
"learning_rate": 2.803347280334728e-06,
|
| 208 |
+
"loss": 0.1096,
|
| 209 |
+
"step": 650
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.8242677824267783,
|
| 213 |
+
"grad_norm": 2.1401469707489014,
|
| 214 |
+
"learning_rate": 1.7573221757322176e-06,
|
| 215 |
+
"loss": 0.1073,
|
| 216 |
+
"step": 675
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.928870292887029,
|
| 220 |
+
"grad_norm": 2.9174375534057617,
|
| 221 |
+
"learning_rate": 7.112970711297071e-07,
|
| 222 |
+
"loss": 0.1132,
|
| 223 |
+
"step": 700
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 3.0,
|
| 227 |
+
"eval_accuracy": 0.9431459254579911,
|
| 228 |
+
"eval_loss": 0.22031328082084656,
|
| 229 |
+
"eval_runtime": 6.5449,
|
| 230 |
+
"eval_samples_per_second": 241.866,
|
| 231 |
+
"eval_steps_per_second": 2.597,
|
| 232 |
+
"step": 717
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.0,
|
| 236 |
+
"step": 717,
|
| 237 |
+
"total_flos": 1024175883573288.0,
|
| 238 |
+
"train_loss": 0.2008532669753709,
|
| 239 |
+
"train_runtime": 123.6845,
|
| 240 |
+
"train_samples_per_second": 277.141,
|
| 241 |
+
"train_steps_per_second": 5.797
|
| 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": 1024175883573288.0,
|
| 262 |
+
"train_batch_size": 48,
|
| 263 |
+
"trial_name": null,
|
| 264 |
+
"trial_params": null
|
| 265 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/all_results.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9368288060644346,
|
| 4 |
+
"eval_loss": 0.2278706282377243,
|
| 5 |
+
"eval_runtime": 6.0921,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 259.846,
|
| 8 |
+
"eval_steps_per_second": 2.791,
|
| 9 |
+
"total_flos": 1022759265150936.0,
|
| 10 |
+
"train_loss": 0.20126758842122439,
|
| 11 |
+
"train_runtime": 123.5042,
|
| 12 |
+
"train_samples": 11426,
|
| 13 |
+
"train_samples_per_second": 277.545,
|
| 14 |
+
"train_steps_per_second": 5.805
|
| 15 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9368288060644346,
|
| 4 |
+
"eval_loss": 0.2278706282377243,
|
| 5 |
+
"eval_runtime": 6.0921,
|
| 6 |
+
"eval_samples": 1583,
|
| 7 |
+
"eval_samples_per_second": 259.846,
|
| 8 |
+
"eval_steps_per_second": 2.791
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"total_flos": 1022759265150936.0,
|
| 4 |
+
"train_loss": 0.20126758842122439,
|
| 5 |
+
"train_runtime": 123.5042,
|
| 6 |
+
"train_samples": 11426,
|
| 7 |
+
"train_samples_per_second": 277.545,
|
| 8 |
+
"train_steps_per_second": 5.805
|
| 9 |
+
}
|
mbert_rtx6000_metrics/result_ablation_mbert/vsfc/mBERT/attention/seed_43/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": 5.534419536590576,
|
| 14 |
+
"learning_rate": 2.8953974895397488e-05,
|
| 15 |
+
"loss": 0.5214,
|
| 16 |
+
"step": 25
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.20920502092050208,
|
| 20 |
+
"grad_norm": 13.864513397216797,
|
| 21 |
+
"learning_rate": 2.7907949790794982e-05,
|
| 22 |
+
"loss": 0.2973,
|
| 23 |
+
"step": 50
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.3138075313807531,
|
| 27 |
+
"grad_norm": 5.720527648925781,
|
| 28 |
+
"learning_rate": 2.686192468619247e-05,
|
| 29 |
+
"loss": 0.3139,
|
| 30 |
+
"step": 75
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.41841004184100417,
|
| 34 |
+
"grad_norm": 3.8049864768981934,
|
| 35 |
+
"learning_rate": 2.5815899581589957e-05,
|
| 36 |
+
"loss": 0.2857,
|
| 37 |
+
"step": 100
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.5230125523012552,
|
| 41 |
+
"grad_norm": 5.698417663574219,
|
| 42 |
+
"learning_rate": 2.4769874476987448e-05,
|
| 43 |
+
"loss": 0.3278,
|
| 44 |
+
"step": 125
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.6276150627615062,
|
| 48 |
+
"grad_norm": 3.505063772201538,
|
| 49 |
+
"learning_rate": 2.372384937238494e-05,
|
| 50 |
+
"loss": 0.2866,
|
| 51 |
+
"step": 150
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.7322175732217573,
|
| 55 |
+
"grad_norm": 6.419493198394775,
|
| 56 |
+
"learning_rate": 2.2677824267782426e-05,
|
| 57 |
+
"loss": 0.2651,
|
| 58 |
+
"step": 175
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.8368200836820083,
|
| 62 |
+
"grad_norm": 6.078672409057617,
|
| 63 |
+
"learning_rate": 2.1631799163179917e-05,
|
| 64 |
+
"loss": 0.2426,
|
| 65 |
+
"step": 200
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.9414225941422594,
|
| 69 |
+
"grad_norm": 5.044684410095215,
|
| 70 |
+
"learning_rate": 2.0585774058577408e-05,
|
| 71 |
+
"loss": 0.238,
|
| 72 |
+
"step": 225
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 1.0,
|
| 76 |
+
"eval_accuracy": 0.9267214150347441,
|
| 77 |
+
"eval_loss": 0.24476154148578644,
|
| 78 |
+
"eval_runtime": 5.8282,
|
| 79 |
+
"eval_samples_per_second": 271.611,
|
| 80 |
+
"eval_steps_per_second": 2.917,
|
| 81 |
+
"step": 239
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 1.0460251046025104,
|
| 85 |
+
"grad_norm": 2.060284376144409,
|
| 86 |
+
"learning_rate": 1.9539748953974895e-05,
|
| 87 |
+
"loss": 0.2332,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 1.1506276150627615,
|
| 92 |
+
"grad_norm": 4.195586681365967,
|
| 93 |
+
"learning_rate": 1.8493723849372386e-05,
|
| 94 |
+
"loss": 0.1834,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 1.2552301255230125,
|
| 99 |
+
"grad_norm": 5.395796775817871,
|
| 100 |
+
"learning_rate": 1.7447698744769877e-05,
|
| 101 |
+
"loss": 0.2185,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 1.3598326359832635,
|
| 106 |
+
"grad_norm": 11.648240089416504,
|
| 107 |
+
"learning_rate": 1.6401673640167365e-05,
|
| 108 |
+
"loss": 0.1848,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.4644351464435146,
|
| 113 |
+
"grad_norm": 4.976749420166016,
|
| 114 |
+
"learning_rate": 1.5355648535564852e-05,
|
| 115 |
+
"loss": 0.1583,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.5690376569037658,
|
| 120 |
+
"grad_norm": 4.559477806091309,
|
| 121 |
+
"learning_rate": 1.4309623430962344e-05,
|
| 122 |
+
"loss": 0.1863,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.6736401673640167,
|
| 127 |
+
"grad_norm": 4.89653205871582,
|
| 128 |
+
"learning_rate": 1.3263598326359832e-05,
|
| 129 |
+
"loss": 0.1538,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 1.778242677824268,
|
| 134 |
+
"grad_norm": 2.0721302032470703,
|
| 135 |
+
"learning_rate": 1.2217573221757323e-05,
|
| 136 |
+
"loss": 0.1748,
|
| 137 |
+
"step": 425
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 1.8828451882845187,
|
| 141 |
+
"grad_norm": 3.5986971855163574,
|
| 142 |
+
"learning_rate": 1.1171548117154812e-05,
|
| 143 |
+
"loss": 0.165,
|
| 144 |
+
"step": 450
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 1.98744769874477,
|
| 148 |
+
"grad_norm": 8.333105087280273,
|
| 149 |
+
"learning_rate": 1.0125523012552301e-05,
|
| 150 |
+
"loss": 0.1669,
|
| 151 |
+
"step": 475
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 2.0,
|
| 155 |
+
"eval_accuracy": 0.934301958307012,
|
| 156 |
+
"eval_loss": 0.2065129578113556,
|
| 157 |
+
"eval_runtime": 6.1641,
|
| 158 |
+
"eval_samples_per_second": 256.81,
|
| 159 |
+
"eval_steps_per_second": 2.758,
|
| 160 |
+
"step": 478
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 2.092050209205021,
|
| 164 |
+
"grad_norm": 6.608449935913086,
|
| 165 |
+
"learning_rate": 9.079497907949792e-06,
|
| 166 |
+
"loss": 0.1117,
|
| 167 |
+
"step": 500
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 2.196652719665272,
|
| 171 |
+
"grad_norm": 8.156933784484863,
|
| 172 |
+
"learning_rate": 8.03347280334728e-06,
|
| 173 |
+
"loss": 0.127,
|
| 174 |
+
"step": 525
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 2.301255230125523,
|
| 178 |
+
"grad_norm": 1.865675449371338,
|
| 179 |
+
"learning_rate": 6.98744769874477e-06,
|
| 180 |
+
"loss": 0.1128,
|
| 181 |
+
"step": 550
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 2.405857740585774,
|
| 185 |
+
"grad_norm": 4.262293815612793,
|
| 186 |
+
"learning_rate": 5.941422594142259e-06,
|
| 187 |
+
"loss": 0.1219,
|
| 188 |
+
"step": 575
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.510460251046025,
|
| 192 |
+
"grad_norm": 7.012624740600586,
|
| 193 |
+
"learning_rate": 4.895397489539749e-06,
|
| 194 |
+
"loss": 0.1316,
|
| 195 |
+
"step": 600
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 2.6150627615062763,
|
| 199 |
+
"grad_norm": 1.3994988203048706,
|
| 200 |
+
"learning_rate": 3.849372384937238e-06,
|
| 201 |
+
"loss": 0.1255,
|
| 202 |
+
"step": 625
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 2.719665271966527,
|
| 206 |
+
"grad_norm": 0.887869119644165,
|
| 207 |
+
"learning_rate": 2.803347280334728e-06,
|
| 208 |
+
"loss": 0.1112,
|
| 209 |
+
"step": 650
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.8242677824267783,
|
| 213 |
+
"grad_norm": 4.3685221672058105,
|
| 214 |
+
"learning_rate": 1.7573221757322176e-06,
|
| 215 |
+
"loss": 0.1276,
|
| 216 |
+
"step": 675
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.928870292887029,
|
| 220 |
+
"grad_norm": 5.971782684326172,
|
| 221 |
+
"learning_rate": 7.112970711297071e-07,
|
| 222 |
+
"loss": 0.1248,
|
| 223 |
+
"step": 700
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 3.0,
|
| 227 |
+
"eval_accuracy": 0.9368288060644346,
|
| 228 |
+
"eval_loss": 0.2278706282377243,
|
| 229 |
+
"eval_runtime": 6.355,
|
| 230 |
+
"eval_samples_per_second": 249.094,
|
| 231 |
+
"eval_steps_per_second": 2.675,
|
| 232 |
+
"step": 717
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 3.0,
|
| 236 |
+
"step": 717,
|
| 237 |
+
"total_flos": 1022759265150936.0,
|
| 238 |
+
"train_loss": 0.20126758842122439,
|
| 239 |
+
"train_runtime": 123.5042,
|
| 240 |
+
"train_samples_per_second": 277.545,
|
| 241 |
+
"train_steps_per_second": 5.805
|
| 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": 1022759265150936.0,
|
| 262 |
+
"train_batch_size": 48,
|
| 263 |
+
"trial_name": null,
|
| 264 |
+
"trial_params": null
|
| 265 |
+
}
|