Article_500v9_NER_Model_3Epochs_UNAUGMENTED

This model is a fine-tuned version of bert-base-cased on the article500v9_wikigold_split dataset. It achieves the following results on the evaluation set:

• Loss: 0.1975
• Precision: 0.6869
• Recall: 0.7022
• F1: 0.6944
• Accuracy: 0.9340

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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	61	0.2954	0.4411	0.5290	0.4811	0.9042
No log	2.0	122	0.2061	0.6493	0.6900	0.6691	0.9315
No log	3.0	183	0.1975	0.6869	0.7022	0.6944	0.9340

Framework versions

• Transformers 4.17.0
• Pytorch 1.11.0+cu113
• Datasets 2.4.0
• Tokenizers 0.11.6