bert-finetuned-mutation-recognition-3

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: 0.0727
• Dnamutation F1: 0.6484
• Proteinmutation F1: 0.8571
• Snp F1: 1.0
• Precision: 0.7966
• Recall: 0.7625
• F1: 0.7792
• Accuracy: 0.9872

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Dnamutation F1	Proteinmutation F1	Snp F1	Precision	Recall	F1	Accuracy
No log	1.0	324	0.0323	0.5996	0.7886	1.0	0.6583	0.7982	0.7215	0.9901
0.0788	2.0	648	0.0314	0.6765	0.8783	1.0	0.7453	0.8571	0.7973	0.9907
0.0788	3.0	972	0.0306	0.6391	0.8679	1.0	0.7341	0.8232	0.7761	0.9903
0.0273	4.0	1296	0.0424	0.6360	0.8714	1.0	0.7792	0.775	0.7771	0.9885
0.0178	5.0	1620	0.0462	0.5885	0.8683	1.0	0.7576	0.7589	0.7583	0.9869
0.0178	6.0	1944	0.0531	0.6176	0.8701	1.0	0.7734	0.7679	0.7706	0.9873
0.0165	7.0	2268	0.0573	0.6597	0.8658	1.0	0.8022	0.775	0.7884	0.9881
0.0144	8.0	2592	0.0636	0.6596	0.8454	1.0	0.7919	0.7679	0.7797	0.9871
0.0144	9.0	2916	0.0710	0.6568	0.8748	1.0	0.8159	0.7679	0.7912	0.9872
0.0108	10.0	3240	0.0727	0.6484	0.8571	1.0	0.7966	0.7625	0.7792	0.9872

Framework versions

• Transformers 4.17.0
• Pytorch 1.10.2
• Datasets 2.0.0
• Tokenizers 0.12.1