hyenadna-large-1m-seqlen-hf_ft_BioS73_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: 0.4445
• F1 Score: 0.8449
• Precision: 0.7906
• Recall: 0.9071
• Accuracy: 0.8222
• Auc: 0.8958
• Prc: 0.8902

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
0.504	0.1864	500	0.4797	0.8144	0.7226	0.9330	0.7730	0.8602	0.8528
0.4627	0.3727	1000	0.4431	0.8260	0.7718	0.8883	0.8002	0.8747	0.8687
0.4575	0.5591	1500	0.4477	0.8141	0.8161	0.8122	0.8021	0.8774	0.8745
0.4486	0.7454	2000	0.4197	0.8338	0.7739	0.9036	0.8077	0.8862	0.8791
0.4296	0.9318	2500	0.4227	0.8372	0.7763	0.9085	0.8114	0.8912	0.8804
0.4277	1.1182	3000	0.4181	0.8396	0.7713	0.9211	0.8122	0.8909	0.8864
0.4166	1.3045	3500	0.4481	0.8315	0.8159	0.8478	0.8166	0.8859	0.8803
0.4235	1.4909	4000	0.4253	0.8341	0.7819	0.8939	0.8103	0.8897	0.8854
0.4149	1.6772	4500	0.4432	0.8370	0.7690	0.9183	0.8092	0.8898	0.8820
0.4099	1.8636	5000	0.4597	0.8123	0.7279	0.9190	0.7734	0.8676	0.8681
0.4078	2.0499	5500	0.4425	0.8450	0.7918	0.9057	0.8226	0.8942	0.8910
0.4201	2.2363	6000	0.4509	0.8445	0.7781	0.9232	0.8185	0.8963	0.8886
0.3896	2.4227	6500	0.4153	0.8387	0.7627	0.9316	0.8088	0.8926	0.8856
0.4104	2.6090	7000	0.4066	0.8378	0.8024	0.8764	0.8189	0.8896	0.8833
0.395	2.7954	7500	0.4659	0.8371	0.8190	0.8561	0.8222	0.8969	0.8908
0.3646	2.9817	8000	0.4445	0.8449	0.7906	0.9071	0.8222	0.8958	0.8902

Framework versions

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