pii-ner

This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5816
• Precision: 0.3789
• Recall: 0.2054
• F1: 0.2664
• Accuracy: 0.9580

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: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
1.3048	1.0	14	0.7370	0.0088	0.0438	0.0147	0.7632
0.7328	2.0	28	0.2553	0.0	0.0	0.0	0.9350
0.2676	3.0	42	0.2095	0.7353	0.1684	0.2740	0.9658
0.0976	4.0	56	0.2309	0.9091	0.2020	0.3306	0.9683
0.0664	5.0	70	0.2471	0.9242	0.2054	0.3361	0.9685
0.1703	6.0	84	0.2893	0.9091	0.2020	0.3306	0.9683
0.1196	7.0	98	0.2559	0.8841	0.2054	0.3333	0.9683
0.1805	8.0	112	0.2691	0.9091	0.2020	0.3306	0.9683
0.1669	9.0	126	0.2651	0.9242	0.2054	0.3361	0.9685
0.2296	10.0	140	0.3022	0.9242	0.2054	0.3361	0.9685
0.1731	11.0	154	0.2879	0.9091	0.2020	0.3306	0.9682
0.1983	12.0	168	0.2824	0.8824	0.2020	0.3288	0.9682
0.0873	13.0	182	0.3495	0.9091	0.2020	0.3306	0.9683
0.4	14.0	196	0.3905	0.9104	0.2054	0.3352	0.9683
0.2142	15.0	210	0.4198	0.9104	0.2054	0.3352	0.9683
0.2092	16.0	224	0.4395	0.9091	0.2020	0.3306	0.9682
0.0803	17.0	238	0.4003	0.9104	0.2054	0.3352	0.9682
0.1509	18.0	252	0.4904	0.9104	0.2054	0.3352	0.9683
0.1382	19.0	266	0.5018	0.9091	0.2020	0.3306	0.9682
0.0554	20.0	280	0.5128	0.9104	0.2054	0.3352	0.9682
0.0595	21.0	294	0.5363	0.9104	0.2054	0.3352	0.9682
0.0339	22.0	308	0.5518	0.9104	0.2054	0.3352	0.9682
0.0292	23.0	322	0.5613	0.8133	0.2054	0.3280	0.9674
0.0368	24.0	336	0.5555	0.9104	0.2054	0.3352	0.9682
0.0224	25.0	350	0.5709	0.3885	0.2054	0.2687	0.9585
0.0122	26.0	364	0.5753	0.3789	0.2054	0.2664	0.9580
0.0332	27.0	378	0.5800	0.3789	0.2054	0.2664	0.9580
0.02	28.0	392	0.5810	0.3789	0.2054	0.2664	0.9580
0.0155	29.0	406	0.5816	0.3789	0.2054	0.2664	0.9580
0.0157	30.0	420	0.5816	0.3789	0.2054	0.2664	0.9580

Framework versions

• Transformers 4.39.3
• Pytorch 2.1.2
• Datasets 2.18.0
• Tokenizers 0.15.2