gena-lm-bert-base-t2t-multi_ft_BioS45_1kbpHG19_DHSs_H3K27AC

This model is a fine-tuned version of AIRI-Institute/gena-lm-bert-base-t2t-multi on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7191
• F1 Score: 0.8258
• Precision: 0.8725
• Recall: 0.7839
• Accuracy: 0.8275
• Auc: 0.8676
• Prc: 0.8589

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
0.6927	0.2103	500	0.6550	0.6325	0.8027	0.5218	0.6836	0.7657	0.7801
0.6285	0.4207	1000	0.5540	0.7342	0.8129	0.6694	0.7472	0.8065	0.8058
0.5306	0.6310	1500	0.5262	0.7747	0.8010	0.75	0.7724	0.8369	0.8222
0.4969	0.8414	2000	0.4964	0.8208	0.7561	0.8976	0.7955	0.8721	0.8624
0.4722	1.0517	2500	0.4584	0.8228	0.8354	0.8105	0.8178	0.8876	0.8792
0.4466	1.2621	3000	0.4567	0.8424	0.7943	0.8968	0.8250	0.8896	0.8698
0.4418	1.4724	3500	0.4333	0.8416	0.8436	0.8395	0.8351	0.9004	0.8883
0.422	1.6828	4000	0.4661	0.8227	0.8588	0.7895	0.8225	0.9030	0.8967
0.4107	1.8931	4500	0.4329	0.8468	0.8009	0.8984	0.8305	0.8937	0.8585
0.3906	2.1035	5000	0.4643	0.8479	0.8290	0.8677	0.8376	0.8902	0.8512
0.4098	2.3138	5500	0.4532	0.8526	0.8060	0.9048	0.8368	0.8782	0.8309
0.4118	2.5242	6000	0.4862	0.8465	0.8503	0.8427	0.8406	0.9018	0.8845
0.4207	2.7345	6500	0.4667	0.8519	0.8126	0.8952	0.8376	0.8927	0.8561
0.4382	2.9449	7000	0.5130	0.8202	0.8763	0.7710	0.8237	0.9094	0.9039
0.3846	3.1552	7500	0.5103	0.8381	0.8659	0.8121	0.8363	0.9077	0.8992
0.4023	3.3656	8000	0.4508	0.8613	0.8225	0.9040	0.8481	0.9123	0.8963
0.3788	3.5759	8500	0.4996	0.8517	0.7933	0.9194	0.8330	0.8901	0.8517
0.3778	3.7863	9000	0.5016	0.8606	0.8237	0.9008	0.8477	0.8967	0.8631
0.3923	3.9966	9500	0.5175	0.8579	0.8356	0.8815	0.8477	0.8895	0.8575
0.3628	4.2070	10000	0.5557	0.8616	0.8427	0.8815	0.8523	0.8935	0.8706
0.4124	4.4173	10500	0.5216	0.8621	0.8252	0.9024	0.8494	0.8721	0.8318
0.388	4.6277	11000	0.6025	0.8584	0.8127	0.9097	0.8435	0.8572	0.8122
0.4513	4.8380	11500	0.5943	0.8500	0.8524	0.8476	0.8439	0.9012	0.8886
0.4206	5.0484	12000	0.5724	0.8610	0.8414	0.8815	0.8515	0.9016	0.8855
0.3882	5.2587	12500	0.5748	0.8616	0.8524	0.8710	0.8540	0.8901	0.8724
0.3756	5.4691	13000	0.5839	0.8635	0.8477	0.8798	0.8549	0.8756	0.8325
0.4158	5.6794	13500	0.5782	0.8593	0.8169	0.9065	0.8452	0.9048	0.8848
0.3859	5.8898	14000	0.5989	0.8530	0.8496	0.8565	0.8460	0.8947	0.8717
0.336	6.1001	14500	0.6641	0.8542	0.7996	0.9169	0.8368	0.8697	0.8287
0.3724	6.3105	15000	0.6330	0.8599	0.8205	0.9032	0.8464	0.8776	0.8500
0.3809	6.5208	15500	0.7191	0.8258	0.8725	0.7839	0.8275	0.8676	0.8589

Framework versions

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