bert-finetuned-ner

This model is a fine-tuned version of google/mobilebert-uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2023
• Precision: 0.8939
• Recall: 0.8939
• F1: 0.8939
• Accuracy: 0.9462

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	20	0.7266	0.5969	0.5833	0.5900	0.7394
No log	2.0	40	0.5312	0.7222	0.6894	0.7054	0.8102
No log	3.0	60	0.4282	0.7717	0.7424	0.7568	0.8612
No log	4.0	80	0.3508	0.8047	0.7803	0.7923	0.8867
No log	5.0	100	0.2955	0.8372	0.8182	0.8276	0.9065
No log	6.0	120	0.2540	0.8550	0.8485	0.8517	0.9150
No log	7.0	140	0.2299	0.8788	0.8788	0.8788	0.9348
No log	8.0	160	0.2168	0.8797	0.8864	0.8830	0.9433
No log	9.0	180	0.2064	0.8939	0.8939	0.8939	0.9462
No log	10.0	200	0.2023	0.8939	0.8939	0.8939	0.9462

Framework versions

• Transformers 4.52.2
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1