Edit model card

hmBERT-CoNLL-cp2

This model is a fine-tuned version of dbmdz/bert-base-historic-multilingual-cased on the conll2003 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0666
  • Precision: 0.8932
  • Recall: 0.9005
  • F1: 0.8968
  • Accuracy: 0.9832

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: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
No log 0.06 25 0.4116 0.3632 0.3718 0.3674 0.9005
No log 0.11 50 0.2247 0.6384 0.6902 0.6633 0.9459
No log 0.17 75 0.1624 0.7303 0.7627 0.7461 0.9580
No log 0.23 100 0.1541 0.7338 0.7688 0.7509 0.9588
No log 0.28 125 0.1349 0.7610 0.8095 0.7845 0.9643
No log 0.34 150 0.1230 0.7982 0.8253 0.8115 0.9694
No log 0.4 175 0.0997 0.8069 0.8406 0.8234 0.9727
No log 0.46 200 0.1044 0.8211 0.8410 0.8309 0.9732
No log 0.51 225 0.0871 0.8413 0.8603 0.8507 0.9760
No log 0.57 250 0.1066 0.8288 0.8465 0.8376 0.9733
No log 0.63 275 0.0872 0.8580 0.8667 0.8624 0.9766
No log 0.68 300 0.0834 0.8522 0.8706 0.8613 0.9773
No log 0.74 325 0.0832 0.8545 0.8834 0.8687 0.9783
No log 0.8 350 0.0776 0.8542 0.8834 0.8685 0.9787
No log 0.85 375 0.0760 0.8629 0.8896 0.8760 0.9801
No log 0.91 400 0.0673 0.8775 0.9004 0.8888 0.9824
No log 0.97 425 0.0681 0.8827 0.8938 0.8882 0.9817
No log 1.03 450 0.0659 0.8844 0.8950 0.8897 0.9824
No log 1.08 475 0.0690 0.8833 0.9015 0.8923 0.9832
0.1399 1.14 500 0.0666 0.8932 0.9005 0.8968 0.9832
0.1399 1.2 525 0.0667 0.8891 0.8997 0.8944 0.9825
0.1399 1.25 550 0.0699 0.8751 0.8953 0.8851 0.9820
0.1399 1.31 575 0.0617 0.8947 0.9068 0.9007 0.9840
0.1399 1.37 600 0.0633 0.9 0.9058 0.9029 0.9841
0.1399 1.42 625 0.0639 0.8966 0.9116 0.9040 0.9843
0.1399 1.48 650 0.0624 0.8972 0.9110 0.9041 0.9845
0.1399 1.54 675 0.0619 0.8980 0.9081 0.9030 0.9842
0.1399 1.59 700 0.0615 0.9002 0.9090 0.9045 0.9843
0.1399 1.65 725 0.0601 0.9037 0.9128 0.9082 0.9850
0.1399 1.71 750 0.0585 0.9031 0.9142 0.9086 0.9849
0.1399 1.77 775 0.0582 0.9035 0.9143 0.9089 0.9851
0.1399 1.82 800 0.0580 0.9044 0.9157 0.9100 0.9853
0.1399 1.88 825 0.0583 0.9034 0.9160 0.9097 0.9851
0.1399 1.94 850 0.0578 0.9058 0.9170 0.9114 0.9854
0.1399 1.99 875 0.0576 0.9060 0.9165 0.9112 0.9852

Framework versions

  • Transformers 4.20.1
  • Pytorch 1.12.0
  • Datasets 2.4.0
  • Tokenizers 0.12.1
Downloads last month
4
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.

Dataset used to train emilys/hmBERT-CoNLL-cp2

Evaluation results