Modelpublic

tanoManzo/dnabert2_ft_BioS2_1kbpHG19_DHSs_H3K27AC

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes26downloads

Model Card

dnabert2ftBioS21kbpHG19DHSs_H3K27AC

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

—Loss: 0.5274
—F1 Score: 0.8245
—Precision: 0.7496
—Recall: 0.9160
—Accuracy: 0.7937
—Auc: 0.8768
—Prc: 0.8713

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
—trainbatchsize: 8
—evalbatchsize: 8
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 20
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1 Score	Precision	Recall	Accuracy	Auc	Prc
0.5965	0.0842	500	0.5836	0.7668	0.6494	0.9360	0.6988	0.7875	0.7661
0.574	0.1684	1000	0.5463	0.7731	0.6977	0.8669	0.7308	0.8030	0.7876
0.5599	0.2527	1500	0.6162	0.7749	0.7024	0.8641	0.7344	0.8063	0.7893
0.5516	0.3369	2000	0.5434	0.7780	0.6705	0.9265	0.7202	0.8150	0.8028
0.5542	0.4211	2500	0.5759	0.6427	0.8036	0.5355	0.6850	0.8150	0.7996
0.5508	0.5053	3000	0.5854	0.7736	0.6496	0.9561	0.7039	0.8153	0.8044
0.5431	0.5895	3500	0.5414	0.7814	0.7095	0.8695	0.7426	0.8196	0.8113
0.5416	0.6737	4000	0.5594	0.7875	0.7053	0.8914	0.7455	0.8224	0.8094
0.5379	0.7580	4500	0.5209	0.7877	0.7217	0.8669	0.7527	0.8278	0.8183
0.5364	0.8422	5000	0.5591	0.7885	0.7057	0.8933	0.7465	0.8323	0.8217
0.5411	0.9264	5500	0.5144	0.7876	0.6954	0.9080	0.7409	0.8329	0.8240
0.528	1.0106	6000	0.5883	0.7817	0.6575	0.9637	0.7152	0.8338	0.8214
0.4991	1.0948	6500	0.5155	0.7943	0.7291	0.8723	0.7610	0.8390	0.8247
0.515	1.1790	7000	0.5264	0.7915	0.7220	0.8758	0.7559	0.8199	0.8015
0.5211	1.2633	7500	0.5094	0.7973	0.6964	0.9325	0.7492	0.8454	0.8374
0.493	1.3475	8000	0.5053	0.8015	0.7213	0.9016	0.7637	0.8468	0.8387
0.5037	1.4317	8500	0.5015	0.8001	0.6987	0.9360	0.7526	0.8518	0.8417
0.4963	1.5159	9000	0.5154	0.7934	0.7676	0.8211	0.7738	0.8484	0.8398
0.4835	1.6001	9500	0.4856	0.8062	0.7250	0.9080	0.7691	0.8545	0.8482
0.4921	1.6844	10000	0.4796	0.7967	0.7762	0.8182	0.7790	0.8575	0.8475
0.4697	1.7686	10500	0.4897	0.8113	0.7287	0.9150	0.7748	0.8609	0.8561
0.4857	1.8528	11000	0.4694	0.8122	0.7553	0.8784	0.7851	0.8613	0.8545
0.4837	1.9370	11500	0.4648	0.8085	0.7753	0.8446	0.7883	0.8654	0.8592
0.4438	2.0212	12000	0.4683	0.8151	0.7404	0.9064	0.7824	0.8570	0.8466
0.4555	2.1054	12500	0.4589	0.8186	0.7600	0.8870	0.7920	0.8711	0.8681
0.4458	2.1897	13000	0.4698	0.8179	0.7510	0.8978	0.7884	0.8706	0.8649
0.4598	2.2739	13500	0.4631	0.7870	0.8043	0.7705	0.7793	0.8707	0.8660
0.4601	2.3581	14000	0.4866	0.8186	0.7404	0.9153	0.7854	0.8722	0.8662
0.4675	2.4423	14500	0.4677	0.8199	0.7302	0.9347	0.7827	0.8709	0.8599
0.4534	2.5265	15000	0.4512	0.8158	0.7600	0.8803	0.7896	0.8695	0.8644
0.439	2.6107	15500	0.4580	0.8281	0.7589	0.9112	0.7999	0.8748	0.8694
0.4479	2.6950	16000	0.4673	0.8151	0.7968	0.8341	0.7997	0.8774	0.8722
0.4468	2.7792	16500	0.4575	0.8144	0.7865	0.8443	0.7964	0.8731	0.8703
0.4387	2.8634	17000	0.4576	0.8171	0.7831	0.8542	0.7977	0.8774	0.8727
0.426	2.9476	17500	0.4615	0.8229	0.7726	0.8803	0.7996	0.8768	0.8724
0.4259	3.0318	18000	0.5274	0.8245	0.7496	0.9160	0.7937	0.8768	0.8713

Framework versions

—Transformers 4.46.0.dev0
—Pytorch 2.4.1+cu121
—Datasets 2.18.0
—Tokenizers 0.20.0