DeiT_S16_RF_Spectrogram

This model is a fine-tuned version of facebook/deit-small-patch16-224 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0669
• Accuracy: 0.9882
• Precision: 0.9926
• Recall: 0.9817
• F1: 0.9871
• Tp: 1608
• Tn: 1898
• Fp: 12
• Fn: 30

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: 5e-05
• train_batch_size: 32
• 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
• lr_scheduler_warmup_steps: 1107
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Tp	Tn	Fp	Fn
0.4745	0.2477	110	0.1945	0.9340	0.9875	0.8681	0.9240	1422	1892	18	216
0.1543	0.4955	220	0.1073	0.9684	1.0	0.9316	0.9646	1526	1910	0	112
0.0972	0.7432	330	0.0959	0.9791	0.9943	0.9603	0.9770	1573	1901	9	65
0.0967	0.9910	440	0.0601	0.9859	0.9969	0.9725	0.9845	1593	1905	5	45
0.0763	1.2387	550	0.0586	0.9856	0.9969	0.9719	0.9842	1592	1905	5	46
0.0634	1.4865	660	0.0935	0.9755	0.9630	0.9847	0.9737	1613	1848	62	25
0.0854	1.7342	770	0.0607	0.9848	0.9919	0.9750	0.9834	1597	1897	13	41
0.0834	1.9820	880	0.0684	0.9828	0.9852	0.9774	0.9813	1601	1886	24	37
0.0545	2.2297	990	0.0705	0.9831	0.9981	0.9652	0.9814	1581	1907	3	57
0.0728	2.4775	1100	0.0594	0.9845	0.9962	0.9701	0.9830	1589	1904	6	49
0.0781	2.7252	1210	0.0571	0.9868	0.9994	0.9719	0.9855	1592	1909	1	46
0.0686	2.9730	1320	0.0583	0.9868	0.9901	0.9811	0.9856	1607	1894	16	31
0.0683	3.2207	1430	0.0486	0.9890	0.9994	0.9768	0.9880	1600	1909	1	38
0.0505	3.4685	1540	0.0539	0.9865	0.9932	0.9774	0.9852	1601	1899	11	37
0.0610	3.7162	1650	0.0559	0.9882	0.9981	0.9762	0.9870	1599	1907	3	39
0.0629	3.9640	1760	0.0432	0.9901	0.9975	0.9811	0.9892	1607	1906	4	31
0.0525	4.2117	1870	0.0465	0.9896	0.9994	0.9780	0.9886	1602	1909	1	36
0.0464	4.4595	1980	0.0502	0.9887	0.9950	0.9805	0.9877	1606	1902	8	32
0.0543	4.7072	2090	0.0454	0.9896	0.9969	0.9805	0.9886	1606	1905	5	32
0.0495	4.9550	2200	0.0453	0.9904	0.9969	0.9823	0.9895	1609	1905	5	29
0.0400	5.2027	2310	0.0512	0.9890	0.9957	0.9805	0.9880	1606	1903	7	32
0.0339	5.4505	2420	0.0514	0.9893	0.9963	0.9805	0.9883	1606	1904	6	32
0.0356	5.6982	2530	0.0449	0.9899	0.9963	0.9817	0.9889	1608	1904	6	30
0.0363	5.9459	2640	0.0521	0.9859	0.9824	0.9872	0.9848	1617	1881	29	21
0.0273	6.1937	2750	0.0499	0.9893	0.9914	0.9853	0.9884	1614	1896	14	24
0.0236	6.4414	2860	0.0542	0.9884	0.9920	0.9829	0.9874	1610	1897	13	28
0.0313	6.6892	2970	0.0440	0.9910	0.9981	0.9823	0.9902	1609	1907	3	29
0.0298	6.9369	3080	0.0511	0.9913	0.9981	0.9829	0.9905	1610	1907	3	28
0.0193	7.1847	3190	0.0531	0.9887	0.9920	0.9835	0.9877	1611	1897	13	27
0.0245	7.4324	3300	0.0529	0.9893	0.9938	0.9829	0.9883	1610	1900	10	28
0.0234	7.6802	3410	0.0640	0.9882	0.9920	0.9823	0.9871	1609	1897	13	29
0.0172	7.9279	3520	0.0636	0.9896	0.9963	0.9811	0.9886	1607	1904	6	31
0.0231	8.1757	3630	0.0559	0.9890	0.9932	0.9829	0.9880	1610	1899	11	28
0.0148	8.4234	3740	0.0566	0.9893	0.9932	0.9835	0.9883	1611	1899	11	27
0.0145	8.6712	3850	0.0601	0.9890	0.9914	0.9847	0.9881	1613	1896	14	25
0.0193	8.9189	3960	0.0634	0.9899	0.9951	0.9829	0.9889	1610	1902	8	28
0.0139	9.1667	4070	0.0647	0.9896	0.9957	0.9817	0.9886	1608	1903	7	30
0.0089	9.4144	4180	0.0678	0.9893	0.9950	0.9817	0.9883	1608	1902	8	30
0.0087	9.6622	4290	0.0664	0.9893	0.9950	0.9817	0.9883	1608	1902	8	30
0.0104	9.9099	4400	0.0669	0.9882	0.9926	0.9817	0.9871	1608	1898	12	30

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2