hyenadna-large-1m-seqlen-hf_ft_BioS2_1kbpHG19_DHSs_H3K27AC

This model is a fine-tuned version of LongSafari/hyenadna-large-1m-seqlen-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4430
• F1 Score: 0.8169
• Precision: 0.7917
• Recall: 0.8438
• Accuracy: 0.7974
• Auc: 0.8771
• Prc: 0.8787

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: 8
• eval_batch_size: 8
• 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
0.598	0.0840	500	0.5571	0.7836	0.6895	0.9075	0.7316	0.7867	0.7596
0.5439	0.1680	1000	0.5309	0.7878	0.7135	0.8793	0.7462	0.8020	0.7858
0.5302	0.2519	1500	0.5200	0.7724	0.7550	0.7905	0.7504	0.8137	0.8022
0.5293	0.3359	2000	0.5046	0.7929	0.7520	0.8385	0.7654	0.8256	0.8146
0.51	0.4199	2500	0.5110	0.7957	0.7146	0.8975	0.7531	0.8244	0.8170
0.5205	0.5039	3000	0.5003	0.7980	0.7232	0.8899	0.7586	0.8281	0.8179
0.5153	0.5878	3500	0.5101	0.7715	0.7793	0.7639	0.7576	0.8388	0.8324
0.5097	0.6718	4000	0.4971	0.7986	0.7410	0.8658	0.7660	0.8402	0.8344
0.5164	0.7558	4500	0.4982	0.7728	0.7979	0.7491	0.7640	0.8468	0.8401
0.4978	0.8398	5000	0.4807	0.8065	0.7440	0.8805	0.7738	0.8499	0.8437
0.4744	0.9237	5500	0.4778	0.7867	0.7886	0.7849	0.7721	0.8501	0.8480
0.4762	1.0077	6000	0.4849	0.8097	0.7311	0.9072	0.7716	0.8552	0.8519
0.4707	1.0917	6500	0.4599	0.8096	0.7700	0.8536	0.7850	0.8602	0.8547
0.4612	1.1757	7000	0.4724	0.7988	0.7954	0.8021	0.7835	0.8597	0.8588
0.4654	1.2597	7500	0.4670	0.8079	0.7584	0.8642	0.7798	0.8620	0.8616
0.456	1.3436	8000	0.4669	0.8108	0.7797	0.8445	0.7889	0.8645	0.8652
0.4655	1.4276	8500	0.4679	0.7951	0.8026	0.7877	0.7825	0.8647	0.8664
0.4447	1.5116	9000	0.4554	0.8156	0.7648	0.8736	0.7884	0.8650	0.8631
0.4663	1.5956	9500	0.4563	0.8035	0.7986	0.8084	0.7882	0.8682	0.8680
0.4526	1.6795	10000	0.4769	0.7937	0.8004	0.7871	0.7808	0.8682	0.8700
0.4457	1.7635	10500	0.4533	0.7992	0.8009	0.7974	0.7854	0.8687	0.8698
0.4391	1.8475	11000	0.4493	0.8186	0.7700	0.8736	0.7926	0.8705	0.8687
0.4448	1.9315	11500	0.4461	0.8193	0.7696	0.8758	0.7931	0.8715	0.8737
0.4359	2.0155	12000	0.4450	0.8178	0.7773	0.8627	0.7941	0.8736	0.8763
0.4268	2.0994	12500	0.4724	0.7931	0.8190	0.7689	0.7852	0.8732	0.8746
0.4254	2.1834	13000	0.4438	0.8198	0.7623	0.8868	0.7912	0.8756	0.8726
0.4289	2.2674	13500	0.4556	0.8203	0.7503	0.9047	0.7877	0.8726	0.8706
0.4315	2.3514	14000	0.4621	0.8060	0.7606	0.8570	0.7790	0.8647	0.8630
0.4226	2.4353	14500	0.5973	0.7535	0.8545	0.6739	0.7639	0.8696	0.8723
0.4331	2.5193	15000	0.4589	0.8198	0.7959	0.8451	0.8010	0.8762	0.8770
0.4216	2.6033	15500	0.5026	0.7937	0.8263	0.7636	0.7874	0.8758	0.8783
0.4449	2.6873	16000	0.4430	0.8169	0.7917	0.8438	0.7974	0.8771	0.8787

Framework versions

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