muhtasham's picture
update model card README.md
0aa08dd
metadata
license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - wikiann
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: bert-base-cased-tajik-ner
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: wikiann
          type: wikiann
          config: tg
          split: train+test
          args: tg
        metrics:
          - name: Precision
            type: precision
            value: 0.512396694214876
          - name: Recall
            type: recall
            value: 0.5961538461538461
          - name: F1
            type: f1
            value: 0.5511111111111111
          - name: Accuracy
            type: accuracy
            value: 0.8520825223822499

bert-base-cased-tajik-ner

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

  • Loss: 1.1137
  • Precision: 0.5124
  • Recall: 0.5962
  • F1: 0.5511
  • Accuracy: 0.8521

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: 200

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
No log 2.0 50 0.8416 0.0739 0.125 0.0929 0.6948
No log 4.0 100 0.7061 0.2229 0.3558 0.2741 0.7415
No log 6.0 150 0.6467 0.3057 0.4615 0.3678 0.8167
No log 8.0 200 0.7923 0.3968 0.4808 0.4348 0.8073
No log 10.0 250 0.7003 0.4656 0.5865 0.5191 0.8653
No log 12.0 300 0.7723 0.4380 0.5769 0.4979 0.8560
No log 14.0 350 0.9088 0.4762 0.5769 0.5217 0.8470
No log 16.0 400 0.9756 0.472 0.5673 0.5153 0.8424
No log 18.0 450 1.1114 0.4576 0.5192 0.4865 0.8151
0.2358 20.0 500 1.0887 0.48 0.5769 0.5240 0.8330
0.2358 22.0 550 1.0968 0.4419 0.5481 0.4893 0.8268
0.2358 24.0 600 1.3330 0.5140 0.5288 0.5213 0.8042
0.2358 26.0 650 1.0911 0.6019 0.5962 0.5990 0.8521
0.2358 28.0 700 1.1949 0.4586 0.5865 0.5148 0.8388
0.2358 30.0 750 1.1208 0.4444 0.5769 0.5021 0.8470
0.2358 32.0 800 1.0968 0.5413 0.5673 0.5540 0.8661
0.2358 34.0 850 1.1618 0.5 0.5769 0.5357 0.8575
0.2358 36.0 900 1.1018 0.5169 0.5865 0.5495 0.8505
0.2358 38.0 950 1.1948 0.4797 0.5673 0.5198 0.8431
0.0039 40.0 1000 1.1063 0.4511 0.5769 0.5063 0.8533
0.0039 42.0 1050 1.0651 0.5702 0.625 0.5963 0.8723
0.0039 44.0 1100 1.1475 0.472 0.5673 0.5153 0.8466
0.0039 46.0 1150 1.3080 0.4590 0.5385 0.4956 0.8353
0.0039 48.0 1200 1.1165 0.5741 0.5962 0.5849 0.8610
0.0039 50.0 1250 1.2525 0.4724 0.5769 0.5195 0.8431
0.0039 52.0 1300 1.2443 0.5161 0.6154 0.5614 0.8521
0.0039 54.0 1350 1.5720 0.4597 0.5481 0.5 0.8054
0.0039 56.0 1400 1.2487 0.5446 0.5865 0.5648 0.8513
0.0039 58.0 1450 1.3936 0.4754 0.5577 0.5133 0.8365
0.0051 60.0 1500 1.2980 0.5636 0.5962 0.5794 0.8544
0.0051 62.0 1550 1.3284 0.5175 0.5673 0.5413 0.8490
0.0051 64.0 1600 1.3345 0.5268 0.5673 0.5463 0.8447
0.0051 66.0 1650 1.1006 0.5872 0.6154 0.6009 0.8641
0.0051 68.0 1700 1.0886 0.4580 0.5769 0.5106 0.8525
0.0051 70.0 1750 1.1017 0.4959 0.5865 0.5374 0.8525
0.0051 72.0 1800 1.1137 0.5124 0.5962 0.5511 0.8521

Framework versions

  • Transformers 4.21.2
  • Pytorch 1.12.1+cu113
  • Datasets 2.4.0
  • Tokenizers 0.12.1