ner_conll2012

This model is a fine-tuned version of bert-base-uncased on the conll2012 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1094
  • Precision: 0.8338
  • Recall: 0.8633
  • F1: 0.8483
  • Accuracy: 0.9751

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: 3e-05
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 5

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
0.2459 1.0 7238 0.0969 0.8236 0.8771 0.8495 0.9706
0.0632 2.0 14476 0.0964 0.8466 0.8759 0.8610 0.9727
0.0366 3.0 21714 0.1031 0.8528 0.8799 0.8662 0.9739
0.0206 4.0 28952 0.1190 0.8557 0.8818 0.8685 0.9741
0.0119 5.0 36190 0.1282 0.8608 0.8761 0.8684 0.9742

Framework versions

  • Transformers 4.9.1
  • Pytorch 1.9.0+cu102
  • Datasets 1.11.0
  • Tokenizers 0.10.2
New

Select AutoNLP in the “Train” menu to fine-tune this model automatically.

Downloads last month
1,389
Hosted inference API
Token Classification
This model can be loaded on the Inference API on-demand.