metadata

license: mit
tags:
  - generated_from_trainer
datasets:
  - sucx3_ner
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: histbert-finetuned-ner
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: sucx3_ner
          type: sucx3_ner
          config: simple_cased
          split: validation
          args: simple_cased
        metrics:
          - name: Precision
            type: precision
            value: 0.8784308810627898
          - name: Recall
            type: recall
            value: 0.9261363636363636
          - name: F1
            type: f1
            value: 0.9016530520357625
          - name: Accuracy
            type: accuracy
            value: 0.992218705252845

histbert-finetuned-ner

This model is a fine-tuned version of Riksarkivet/bert-base-cased-swe-historical on the sucx3_ner dataset. It achieves the following results on the evaluation set:

Loss: 0.0495
Precision: 0.8784
Recall: 0.9261
F1: 0.9017
Accuracy: 0.9922

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: 2e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 7

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.0403	1.0	5391	0.0316	0.8496	0.8866	0.8677	0.9903
0.0199	2.0	10782	0.0308	0.8814	0.9034	0.8923	0.9915
0.0173	3.0	16173	0.0372	0.8698	0.9197	0.8940	0.9913
0.0066	4.0	21564	0.0397	0.8783	0.9239	0.9005	0.9921
0.0029	5.0	26955	0.0454	0.8855	0.9181	0.9015	0.9923
0.0035	6.0	32346	0.0454	0.8834	0.9211	0.9019	0.9922
0.0009	7.0	37737	0.0495	0.8784	0.9261	0.9017	0.9922

Framework versions

Transformers 4.28.0
Pytorch 2.0.0+cu118
Datasets 2.12.0
Tokenizers 0.13.3