ben-epstein's picture
Create README.md
9a07a1c
metadata
license: apache-2.0
base_model: distilbert-base-cased
tags:
  - generated_from_trainer
datasets:
  - wikiann
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: distilbert-base-cased-wikiann
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: wikiann
          type: wikiann
          config: en
          split: validation
          args: en
        metrics:
          - name: Precision
            type: precision
            value: 0.7962710012293402
          - name: Recall
            type: recall
            value: 0.8241905839106461
          - name: F1
            type: f1
            value: 0.8099902737251633
          - name: Accuracy
            type: accuracy
            value: 0.926231747293136

distilbert-base-cased-wikiann

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

  • Loss: 0.2549
  • Precision: 0.7963
  • Recall: 0.8242
  • F1: 0.8100
  • Accuracy: 0.9262

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: 101
  • 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
0.3137 1.0 1250 0.2685 0.7716 0.8027 0.7868 0.9181
0.2199 2.0 2500 0.2526 0.7765 0.8132 0.7944 0.9220
0.1613 3.0 3750 0.2549 0.7963 0.8242 0.8100 0.9262

Framework versions

  • Transformers 4.34.1
  • Pytorch 2.0.1+cu117
  • Datasets 2.14.6
  • Tokenizers 0.14.1