hyenadna-large-1m-seqlen-hf_ft_BioS45_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: 0.4630
• F1 Score: 0.8133
• Precision: 0.7796
• Recall: 0.85
• Accuracy: 0.7964
• Auc: 0.8733
• Prc: 0.8740

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
0.5578	0.2103	500	0.5077	0.7339	0.8087	0.6718	0.7459	0.8407	0.8258
0.5127	0.4207	1000	0.4832	0.7813	0.7861	0.7766	0.7732	0.8460	0.8411
0.4843	0.6310	1500	0.5096	0.7670	0.8214	0.7194	0.7720	0.8533	0.8519
0.4833	0.8414	2000	0.4942	0.8042	0.7317	0.8927	0.7732	0.8633	0.8521
0.4748	1.0517	2500	0.4991	0.7792	0.7997	0.7597	0.7753	0.8581	0.8511
0.4723	1.2621	3000	0.4819	0.7833	0.8276	0.7435	0.7854	0.8618	0.8531
0.474	1.4724	3500	0.4547	0.8026	0.8003	0.8048	0.7934	0.8717	0.8655
0.4531	1.6828	4000	0.4560	0.8197	0.7537	0.8984	0.7939	0.8696	0.8629
0.4601	1.8931	4500	0.4601	0.8135	0.7554	0.8815	0.7892	0.8658	0.8569
0.441	2.1035	5000	0.4680	0.8103	0.7545	0.875	0.7863	0.8610	0.8497
0.4267	2.3138	5500	0.4784	0.8150	0.7775	0.8565	0.7972	0.8705	0.8622
0.4336	2.5242	6000	0.4562	0.8032	0.7891	0.8177	0.7909	0.8684	0.8590
0.446	2.7345	6500	0.4630	0.8133	0.7796	0.85	0.7964	0.8733	0.8740

Framework versions

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