DNABERT-2-117M_ft_BioS74_1kbpHG19_DHSs_H3K27AC

This model is a fine-tuned version of zhihan1996/DNABERT-2-117M on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4752
• F1 Score: 0.8118
• Precision: 0.7476
• Recall: 0.8880
• Accuracy: 0.7844
• Auc: 0.8654
• Prc: 0.8526

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
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Score	Precision	Recall	Accuracy	Auc	Prc
0.6813	0.2629	500	0.5410	0.7509	0.7369	0.7654	0.7342	0.8039	0.7912
0.5852	0.5258	1000	0.5882	0.7674	0.6497	0.9372	0.7026	0.8110	0.7973
0.5576	0.7886	1500	0.6756	0.6367	0.8150	0.5224	0.6879	0.8186	0.8042
0.5552	1.0515	2000	0.5776	0.6982	0.8020	0.6183	0.7202	0.8245	0.8091
0.5468	1.3144	2500	0.5009	0.7759	0.7563	0.7966	0.7591	0.8313	0.8151
0.5176	1.5773	3000	0.5745	0.6987	0.8255	0.6057	0.7265	0.8409	0.8239
0.5082	1.8402	3500	0.4783	0.7848	0.7733	0.7966	0.7712	0.8480	0.8298
0.5026	2.1030	4000	0.4747	0.8043	0.7587	0.8559	0.7820	0.8528	0.8360
0.508	2.3659	4500	0.4703	0.8063	0.7359	0.8915	0.7757	0.8554	0.8387
0.4777	2.6288	5000	0.5124	0.7355	0.8329	0.6585	0.7520	0.8586	0.8424
0.4745	2.8917	5500	0.6440	0.7276	0.8363	0.6439	0.7476	0.8590	0.8429
0.4737	3.1546	6000	0.4884	0.8094	0.7582	0.8679	0.7860	0.8606	0.8447
0.4447	3.4175	6500	0.5638	0.7851	0.8166	0.7559	0.7833	0.8626	0.8467
0.4667	3.6803	7000	0.6040	0.7930	0.6731	0.9648	0.7363	0.8618	0.8452
0.4679	3.9432	7500	0.4615	0.8144	0.7496	0.8915	0.7873	0.8620	0.8463
0.4334	4.2061	8000	0.4827	0.8093	0.7239	0.9176	0.7736	0.8618	0.8490
0.4416	4.4690	8500	0.4817	0.8118	0.7848	0.8408	0.7960	0.8651	0.8510
0.4434	4.7319	9000	0.4752	0.8118	0.7476	0.8880	0.7844	0.8654	0.8526

Framework versions

• Transformers 4.42.3
• Pytorch 2.3.0+cu121
• Datasets 2.18.0
• Tokenizers 0.19.0