LanaKru's picture
update model card README.md
8bc24ba
metadata
license: cc-by-nc-sa-4.0
tags:
  - generated_from_trainer
datasets:
  - skript
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: wikineural-multilingual-ner-finetuned-ner
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: skript
          type: skript
          config: myscript
          split: train
          args: myscript
        metrics:
          - name: Precision
            type: precision
            value: 0.9007335298553506
          - name: Recall
            type: recall
            value: 0.9301946902654867
          - name: F1
            type: f1
            value: 0.9152270827528559
          - name: Accuracy
            type: accuracy
            value: 0.9653644982020269

wikineural-multilingual-ner-finetuned-ner

This model is a fine-tuned version of Babelscape/wikineural-multilingual-ner on the skript dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1243
  • Precision: 0.9007
  • Recall: 0.9302
  • F1: 0.9152
  • Accuracy: 0.9654

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: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
No log 1.0 298 0.1179 0.8975 0.8981 0.8978 0.9592
0.104 2.0 596 0.1161 0.9051 0.9201 0.9126 0.9648
0.104 3.0 894 0.1243 0.9007 0.9302 0.9152 0.9654

Framework versions

  • Transformers 4.21.0
  • Pytorch 1.12.0+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1