deberta-v3-base-uner-down-synth400

This model is a fine-tuned version of microsoft/deberta-v3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1185
• F1: 0.7446
• Precision: 0.6845
• Recall: 0.8162
• Accuracy: 0.9793

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: 2.5e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Precision	Recall	Accuracy
0.3105	0.8	20	0.1720	0.1137	0.1963	0.08	0.9455
0.2009	1.6	40	0.1188	0.3566	0.3183	0.4054	0.9581
0.0529	2.4	60	0.0899	0.5527	0.5040	0.6119	0.9714
0.072	3.2	80	0.0852	0.6907	0.6214	0.7773	0.9766
0.0971	4.0	100	0.0794	0.6906	0.6563	0.7286	0.9778
0.0179	4.8	120	0.0813	0.7150	0.6402	0.8097	0.9779
0.0553	5.6	140	0.0826	0.7237	0.6916	0.7589	0.9792
0.0254	6.4	160	0.0955	0.7098	0.6379	0.8	0.9764
0.0054	7.2	180	0.0949	0.7008	0.6234	0.8	0.9763
0.0321	8.0	200	0.0877	0.7251	0.6853	0.7697	0.9789
0.0082	8.8	220	0.0891	0.7369	0.7152	0.76	0.9800
0.0016	9.6	240	0.0960	0.7431	0.7201	0.7676	0.9802
0.0258	10.4	260	0.0957	0.7438	0.7095	0.7816	0.9801
0.0018	11.2	280	0.1016	0.7283	0.6676	0.8011	0.9782
0.0068	12.0	300	0.1014	0.7355	0.6830	0.7968	0.9791
0.0027	12.8	320	0.0978	0.7390	0.6932	0.7914	0.9802
0.0015	13.6	340	0.0991	0.7458	0.7002	0.7978	0.9805
0.0024	14.4	360	0.1031	0.7418	0.6783	0.8184	0.9795
0.0013	15.2	380	0.1021	0.7556	0.7352	0.7773	0.9814
0.0011	16.0	400	0.1056	0.7388	0.68	0.8086	0.9795
0.0281	16.8	420	0.1042	0.7511	0.7096	0.7978	0.9806
0.0015	17.6	440	0.1095	0.7357	0.6763	0.8065	0.9789
0.0015	18.4	460	0.1076	0.7430	0.6912	0.8032	0.9797
0.002	19.2	480	0.1090	0.7395	0.6867	0.8011	0.9795
0.0008	20.0	500	0.1103	0.7457	0.6943	0.8054	0.9799
0.0014	20.8	520	0.1152	0.7400	0.6784	0.8141	0.9792
0.001	21.6	540	0.1172	0.7405	0.6768	0.8173	0.9791
0.0017	22.4	560	0.1140	0.7473	0.6876	0.8184	0.9795
0.0021	23.2	580	0.1140	0.7473	0.6876	0.8184	0.9796
0.0012	24.0	600	0.1132	0.7443	0.6917	0.8054	0.9797
0.0005	24.8	620	0.1138	0.746	0.6940	0.8065	0.9798
0.0011	25.6	640	0.1145	0.7456	0.6933	0.8065	0.9798
0.0009	26.4	660	0.1162	0.7439	0.6888	0.8086	0.9796
0.0014	27.2	680	0.1173	0.7450	0.6884	0.8119	0.9796
0.0007	28.0	700	0.1182	0.7441	0.6852	0.8141	0.9794
0.0019	28.8	720	0.1191	0.7477	0.6859	0.8216	0.9793
0.0007	29.6	740	0.1185	0.7446	0.6845	0.8162	0.9793

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1