Edit model card

mBERT-naamapdam-fine-tuned

This model is a fine-tuned version of bert-base-multilingual-cased on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4625
  • Precision: 0.8060
  • Recall: 0.8246
  • F1: 0.8152
  • Accuracy: 0.9173

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-05
  • train_batch_size: 128
  • eval_batch_size: 256
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 15

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
0.3625 0.26 1000 0.3300 0.7651 0.7809 0.7729 0.8964
0.3099 0.51 2000 0.3070 0.7708 0.8041 0.7871 0.9002
0.2954 0.77 3000 0.2962 0.7793 0.8036 0.7913 0.9041
0.283 1.03 4000 0.2958 0.7843 0.8153 0.7995 0.9066
0.265 1.29 5000 0.2873 0.7930 0.8065 0.7997 0.9069
0.2613 1.54 6000 0.2838 0.7789 0.8289 0.8031 0.9092
0.2635 1.8 7000 0.2790 0.7902 0.8252 0.8073 0.9088
0.2574 2.06 8000 0.2946 0.7887 0.8345 0.8110 0.9098
0.2355 2.31 9000 0.2859 0.7975 0.8152 0.8063 0.9105
0.2361 2.57 10000 0.2806 0.7883 0.8313 0.8092 0.9104
0.2361 2.83 11000 0.2805 0.7931 0.8279 0.8101 0.9123
0.2268 3.08 12000 0.2934 0.7959 0.8323 0.8137 0.9130
0.2106 3.34 13000 0.2862 0.7934 0.8311 0.8118 0.9121
0.2106 3.6 14000 0.2876 0.8009 0.8332 0.8167 0.9143
0.2131 3.86 15000 0.2777 0.8015 0.8242 0.8127 0.9123
0.1993 4.11 16000 0.2999 0.7920 0.8311 0.8111 0.9113
0.1872 4.37 17000 0.2984 0.8003 0.8365 0.8180 0.9143
0.1861 4.63 18000 0.2894 0.7976 0.8321 0.8145 0.9151
0.1916 4.88 19000 0.2909 0.7958 0.8300 0.8125 0.9143
0.1745 5.14 20000 0.3075 0.7906 0.8386 0.8139 0.9136
0.1649 5.4 21000 0.2986 0.8055 0.8199 0.8127 0.9147
0.1678 5.66 22000 0.3043 0.7988 0.8303 0.8142 0.9147
0.1688 5.91 23000 0.2950 0.8026 0.8269 0.8146 0.9155
0.153 6.17 24000 0.3231 0.7995 0.8305 0.8147 0.9150
0.1468 6.43 25000 0.3145 0.7954 0.8326 0.8136 0.9156
0.1478 6.68 26000 0.3222 0.8034 0.8307 0.8168 0.9160
0.1489 6.94 27000 0.3184 0.8019 0.8318 0.8166 0.9161
0.1311 7.2 28000 0.3336 0.8022 0.8278 0.8148 0.9168
0.1298 7.46 29000 0.3430 0.8050 0.8281 0.8164 0.9164
0.1319 7.71 30000 0.3374 0.8005 0.8257 0.8129 0.9152
0.1312 7.97 31000 0.3320 0.8019 0.8353 0.8183 0.9173
0.1144 8.23 32000 0.3539 0.8007 0.8309 0.8155 0.9160
0.1132 8.48 33000 0.3581 0.7940 0.8376 0.8152 0.9158
0.1159 8.74 34000 0.3566 0.8032 0.8355 0.8191 0.9182
0.117 9.0 35000 0.3384 0.8113 0.8205 0.8159 0.9166
0.0996 9.25 36000 0.3637 0.8060 0.8256 0.8156 0.9166
0.1004 9.51 37000 0.3687 0.8043 0.8147 0.8095 0.9152
0.1015 9.77 38000 0.3715 0.8017 0.8359 0.8185 0.9173
0.1001 10.03 39000 0.3826 0.8047 0.8288 0.8166 0.9174
0.0874 10.28 40000 0.3857 0.8087 0.8231 0.8158 0.9168
0.0892 10.54 41000 0.3817 0.8069 0.8221 0.8145 0.9165
0.0895 10.8 42000 0.3800 0.8107 0.8291 0.8198 0.9183
0.0868 11.05 43000 0.4099 0.8032 0.8297 0.8162 0.9177
0.0777 11.31 44000 0.4099 0.8059 0.8255 0.8156 0.9170
0.0781 11.57 45000 0.4077 0.8044 0.8335 0.8187 0.9186
0.0779 11.83 46000 0.4172 0.8050 0.8243 0.8145 0.9161
0.0759 12.08 47000 0.4230 0.8034 0.8244 0.8138 0.9158
0.0691 12.34 48000 0.4286 0.8048 0.8221 0.8134 0.9162
0.0676 12.6 49000 0.4251 0.8091 0.8287 0.8188 0.9185
0.0695 12.85 50000 0.4289 0.8043 0.8284 0.8161 0.9168
0.0663 13.11 51000 0.4431 0.8060 0.8246 0.8152 0.9168
0.0618 13.37 52000 0.4484 0.8054 0.8214 0.8133 0.9162
0.0614 13.62 53000 0.4421 0.8044 0.8230 0.8136 0.9166
0.0611 13.88 54000 0.4468 0.8066 0.8231 0.8148 0.9166
0.0582 14.14 55000 0.4606 0.8055 0.8244 0.8148 0.9173
0.0552 14.4 56000 0.4642 0.8055 0.8274 0.8163 0.9175
0.0553 14.65 57000 0.4633 0.8083 0.8248 0.8165 0.9175
0.0556 14.91 58000 0.4625 0.8060 0.8246 0.8152 0.9173

Framework versions

  • Transformers 4.27.4
  • Pytorch 2.0.0+cu117
  • Datasets 2.11.0
  • Tokenizers 0.13.3
Downloads last month
8
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.