distilbert-finetuned-custom-ner

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

• Loss: 0.1651
• Precision: 0.8404
• Recall: 0.9293
• F1: 0.8826
• Accuracy: 0.9681
• Loc Precision: 0.9325
• Loc Recall: 0.9472
• Loc F1: 0.9398
• Misc Precision: 0.5968
• Misc Recall: 0.8460
• Misc F1: 0.6999
• Org Precision: 0.7967
• Org Recall: 0.9031
• Org F1: 0.8466
• Per Precision: 0.9537
• Per Recall: 0.9723
• Per F1: 0.9629

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: 16
• eval_batch_size: 16
• 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: 6

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy	Loc Precision	Loc Recall	Loc F1	Misc Precision	Misc Recall	Misc F1	Org Precision	Org Recall	Org F1	Per Precision	Per Recall	Per F1
0.0697	1.0	878	0.1253	0.8373	0.9189	0.8762	0.9688	0.8974	0.9526	0.9242	0.6546	0.8037	0.7215	0.7633	0.8874	0.8207	0.9463	0.9658	0.9559
0.0482	2.0	1756	0.1437	0.8306	0.9196	0.8728	0.9667	0.9266	0.9271	0.9268	0.5929	0.8308	0.6920	0.7662	0.9016	0.8284	0.9551	0.9696	0.9623
0.0346	3.0	2634	0.1537	0.8284	0.9231	0.8732	0.9668	0.9238	0.9439	0.9338	0.5748	0.8416	0.6831	0.7911	0.8784	0.8325	0.9433	0.9756	0.9592
0.0248	4.0	3512	0.1447	0.8473	0.9251	0.8845	0.9702	0.9267	0.9428	0.9347	0.6182	0.8482	0.7151	0.8092	0.8919	0.8485	0.9526	0.9701	0.9613
0.0166	5.0	4390	0.1612	0.8405	0.9275	0.8818	0.9675	0.9330	0.9407	0.9368	0.5894	0.8471	0.6951	0.8036	0.9031	0.8504	0.9562	0.9723	0.9642
0.0113	6.0	5268	0.1651	0.8404	0.9293	0.8826	0.9681	0.9325	0.9472	0.9398	0.5968	0.8460	0.6999	0.7967	0.9031	0.8466	0.9537	0.9723	0.9629

Framework versions

• Transformers 4.53.2
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.2