ner-multilingual-bert

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

• Loss: 0.0002
• Precision: 0.9998
• Recall: 0.9991
• F1: 0.9994
• Accuracy: 1.0000

Model description

Trained to detect author and publish dates out of text beginnings

Intended uses & limitations

More information needed

Training and evaluation data

See Dataset

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: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
0.0108	0.2	250	0.0039	0.9942	0.9818	0.9880	0.9992
0.0022	0.4	500	0.0021	0.9863	0.9861	0.9862	0.9993
0.0006	0.61	750	0.0007	0.9998	0.9975	0.9986	0.9999
0.0004	0.81	1000	0.0002	0.9998	0.9991	0.9994	1.0000

Framework versions

• Transformers 4.37.0.dev0
• Pytorch 2.1.1+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0