bert-finetuned-ner

This model is a fine-tuned version of bert-base-cased on the conll2003 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0615
• Loc: {'precision': 0.9556756756756757, 'recall': 0.9624387588459444, 'f1': 0.9590452942771901, 'number': 1837}
• Misc: {'precision': 0.8657024793388429, 'recall': 0.9088937093275488, 'f1': 0.8867724867724868, 'number': 922}
• Org: {'precision': 0.9079620160701242, 'recall': 0.9269202087994034, 'f1': 0.9173431734317343, 'number': 1341}
• Per: {'precision': 0.9656652360515021, 'recall': 0.9771986970684039, 'f1': 0.9713977334052887, 'number': 1842}
• Overall Precision: 0.9336
• Overall Recall: 0.9507
• Overall F1: 0.9420
• Overall Accuracy: 0.9860

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_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Loc	Misc	Org	Per	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.0782	1.0	1756	0.0664	{'precision': 0.9309071729957806, 'recall': 0.9608056614044638, 'f1': 0.9456201446557728, 'number': 1837}	{'precision': 0.8364779874213837, 'recall': 0.8655097613882863, 'f1': 0.8507462686567164, 'number': 922}	{'precision': 0.8876967095851216, 'recall': 0.9254287844891872, 'f1': 0.9061701350857977, 'number': 1341}	{'precision': 0.9445938676707908, 'recall': 0.9533116178067318, 'f1': 0.948932720886247, 'number': 1842}	0.9104	0.9357	0.9229	0.9826
0.0362	2.0	3512	0.0716	{'precision': 0.9600874794969929, 'recall': 0.9559063690800218, 'f1': 0.9579923622476814, 'number': 1837}	{'precision': 0.8532778355879292, 'recall': 0.8893709327548807, 'f1': 0.8709506107275625, 'number': 922}	{'precision': 0.9151014274981217, 'recall': 0.9082774049217002, 'f1': 0.9116766467065869, 'number': 1341}	{'precision': 0.9441544885177453, 'recall': 0.9820846905537459, 'f1': 0.9627461415646621, 'number': 1842}	0.9281	0.9429	0.9355	0.9841
0.0235	3.0	5268	0.0615	{'precision': 0.9556756756756757, 'recall': 0.9624387588459444, 'f1': 0.9590452942771901, 'number': 1837}	{'precision': 0.8657024793388429, 'recall': 0.9088937093275488, 'f1': 0.8867724867724868, 'number': 922}	{'precision': 0.9079620160701242, 'recall': 0.9269202087994034, 'f1': 0.9173431734317343, 'number': 1341}	{'precision': 0.9656652360515021, 'recall': 0.9771986970684039, 'f1': 0.9713977334052887, 'number': 1842}	0.9336	0.9507	0.9420	0.9860

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 3.6.0
• Tokenizers 0.22.2