bert-base-cased-ner_cv-med-ft

This model is a fine-tuned version of bert-base-cased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5926
• Precision: 0.2559
• Recall: 0.3460
• F1: 0.2942
• Accuracy: 0.8368

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 10
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
1.9616	2.73	30	0.7717	0.0	0.0	0.0	0.8608
0.9266	5.45	60	0.6687	0.0	0.0	0.0	0.8608
0.8486	8.18	90	0.6100	0.2133	0.0488	0.0794	0.8635
0.7421	10.91	120	0.5922	0.2534	0.1966	0.2215	0.8542
0.6481	13.64	150	0.5696	0.2889	0.2378	0.2609	0.8596
0.5948	16.36	180	0.5798	0.2678	0.3034	0.2845	0.8472
0.5621	19.09	210	0.5913	0.2486	0.3293	0.2833	0.8381
0.5234	21.82	240	0.5816	0.2585	0.3262	0.2884	0.8404
0.5028	24.55	270	0.5944	0.2545	0.3476	0.2938	0.8368
0.4975	27.27	300	0.5923	0.2531	0.3476	0.2929	0.8368
0.4791	30.0	330	0.5926	0.2559	0.3460	0.2942	0.8368

Framework versions

• Transformers 4.21.1
• Pytorch 1.8.1+cu111
• Datasets 1.6.2
• Tokenizers 0.12.1