Bert-NER / README.md
Kriyans's picture
End of training
b6d5d90
metadata
license: apache-2.0
base_model: distilbert-base-uncased
tags:
  - generated_from_trainer
datasets:
  - ner
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: Bert-NER
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: ner
          type: ner
          config: indian_names
          split: train
          args: indian_names
        metrics:
          - name: Precision
            type: precision
            value: 0.9860607282009942
          - name: Recall
            type: recall
            value: 0.9693364297742606
          - name: F1
            type: f1
            value: 0.9776270584382788
          - name: Accuracy
            type: accuracy
            value: 0.9882459717748076

Bert-NER

This model is a fine-tuned version of distilbert-base-uncased on the ner dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0372
  • Precision: 0.9861
  • Recall: 0.9693
  • F1: 0.9776
  • Accuracy: 0.9882

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
0.0461 1.0 858 0.0450 0.9853 0.9602 0.9725 0.9859
0.0408 2.0 1716 0.0400 0.9836 0.9679 0.9757 0.9873
0.0391 3.0 2574 0.0372 0.9861 0.9693 0.9776 0.9882

Framework versions

  • Transformers 4.34.1
  • Pytorch 2.1.0+cu118
  • Datasets 2.14.6
  • Tokenizers 0.14.1