swin-tiny-patch4-window7-224-finetuned-icpr

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0479
• Accuracy: 0.9895

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1149	0.99	65	0.9590	0.7629
0.2653	2.0	131	0.1648	0.9532
0.1984	2.99	196	0.0894	0.9713
0.1719	4.0	262	0.0863	0.9685
0.1537	4.99	327	0.0810	0.9761
0.1162	6.0	393	0.0785	0.9771
0.1063	6.99	458	0.0835	0.9723
0.1392	8.0	524	0.0674	0.9761
0.1286	8.99	589	0.0788	0.9761
0.1294	10.0	655	0.0658	0.9790
0.0843	10.99	720	0.0735	0.9732
0.074	12.0	786	0.0636	0.9761
0.0734	12.99	851	0.1043	0.9751
0.0774	14.0	917	0.0898	0.9723
0.068	14.99	982	0.0719	0.9809
0.0821	16.0	1048	0.0956	0.9742
0.0576	16.99	1113	0.0725	0.9713
0.0652	18.0	1179	0.0957	0.9751
0.0712	18.99	1244	0.0809	0.9790
0.075	20.0	1310	0.1283	0.9675
0.0988	20.99	1375	0.0966	0.9742
0.0538	22.0	1441	0.1125	0.9761
0.0578	22.99	1506	0.0648	0.9828
0.0675	24.0	1572	0.0992	0.9799
0.0611	24.99	1637	0.0682	0.9818
0.0434	26.0	1703	0.0719	0.9809
0.0339	26.99	1768	0.0930	0.9780
0.0346	28.0	1834	0.0903	0.9799
0.0806	28.99	1899	0.0903	0.9799
0.0518	30.0	1965	0.0982	0.9790
0.0407	30.99	2030	0.0702	0.9828
0.0528	32.0	2096	0.0897	0.9761
0.0774	32.99	2161	0.0626	0.9818
0.053	34.0	2227	0.0576	0.9837
0.0512	34.99	2292	0.0707	0.9847
0.0388	36.0	2358	0.1040	0.9790
0.06	36.99	2423	0.0840	0.9799
0.0477	38.0	2489	0.0659	0.9857
0.0482	38.99	2554	0.0479	0.9895
0.0292	40.0	2620	0.0699	0.9818
0.0386	40.99	2685	0.1030	0.9837
0.0441	42.0	2751	0.0801	0.9818
0.0269	42.99	2816	0.1037	0.9809
0.0385	44.0	2882	0.0870	0.9799
0.0502	44.99	2947	0.1367	0.9771
0.0389	46.0	3013	0.1093	0.9771
0.0209	46.99	3078	0.0954	0.9837
0.0327	48.0	3144	0.0886	0.9857
0.0269	48.99	3209	0.0767	0.9828
0.0461	50.0	3275	0.0661	0.9857
0.0226	50.99	3340	0.0769	0.9818
0.0304	52.0	3406	0.0841	0.9828
0.0326	52.99	3471	0.1002	0.9828
0.0593	54.0	3537	0.0634	0.9847
0.0489	54.99	3602	0.0702	0.9837
0.0495	56.0	3668	0.1060	0.9809
0.0457	56.99	3733	0.0715	0.9866
0.0487	58.0	3799	0.0906	0.9818
0.0416	58.99	3864	0.0973	0.9790
0.0358	60.0	3930	0.0887	0.9857
0.0503	60.99	3995	0.0959	0.9809
0.0555	62.0	4061	0.1057	0.9780
0.0288	62.99	4126	0.0971	0.9799
0.0514	64.0	4192	0.0754	0.9847
0.0602	64.99	4257	0.0789	0.9837
0.0209	66.0	4323	0.1005	0.9837
0.0366	66.99	4388	0.1070	0.9818
0.031	68.0	4454	0.1018	0.9818
0.043	68.99	4519	0.1020	0.9828
0.0262	70.0	4585	0.0896	0.9837
0.0299	70.99	4650	0.0913	0.9837
0.0211	72.0	4716	0.0957	0.9857
0.0351	72.99	4781	0.1180	0.9818
0.0498	74.0	4847	0.1056	0.9828
0.0174	74.99	4912	0.1032	0.9809
0.0368	76.0	4978	0.1071	0.9790
0.0367	76.99	5043	0.0987	0.9828
0.027	78.0	5109	0.1037	0.9818
0.0225	78.99	5174	0.1129	0.9809
0.0241	80.0	5240	0.1202	0.9828
0.026	80.99	5305	0.1219	0.9790
0.0223	82.0	5371	0.1194	0.9799
0.0454	82.99	5436	0.1148	0.9790
0.019	84.0	5502	0.1168	0.9818
0.0269	84.99	5567	0.1246	0.9799
0.0403	86.0	5633	0.1301	0.9790
0.0294	86.99	5698	0.1204	0.9799
0.0501	88.0	5764	0.1168	0.9790
0.0361	88.99	5829	0.1143	0.9818
0.0278	90.0	5895	0.1029	0.9799
0.0267	90.99	5960	0.0991	0.9818
0.0308	92.0	6026	0.1028	0.9828
0.0246	92.99	6091	0.1031	0.9809
0.0283	94.0	6157	0.1035	0.9818
0.0278	94.99	6222	0.0999	0.9818
0.0221	96.0	6288	0.1007	0.9809
0.0197	96.99	6353	0.0989	0.9818
0.0435	98.0	6419	0.0986	0.9818
0.0266	98.99	6484	0.0987	0.9818
0.0334	99.24	6500	0.0987	0.9818

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu118
• Datasets 2.14.7
• Tokenizers 0.15.0