efficientformer-blur-detection-v2

This model is a fine-tuned version of MinhLe999/efficientformer-HandwritingStrip-RandImg on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1126
• Accuracy: 0.9830
• Precision: 0.9877
• Recall: 0.9717
• F1: 0.9796
• Roc Auc: 0.9949

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: 0.0003
• 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
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1	Roc Auc
1.7673	0.0680	10	2.9690	0.5799	0.0	0.0	0.0	0.6879
0.3655	0.1361	20	1.3398	0.6344	1.0	0.1296	0.2294	0.8583
0.2653	0.2041	30	0.6262	0.7908	0.9844	0.5101	0.672	0.9407
0.1899	0.2721	40	0.2595	0.8997	0.9653	0.7895	0.8686	0.9843
0.2483	0.3401	50	0.5607	0.8214	0.9931	0.5789	0.7315	0.9753
0.1381	0.4082	60	0.4217	0.8605	0.9940	0.6721	0.8019	0.9796
0.1432	0.4762	70	0.1368	0.9456	0.9119	0.9636	0.9370	0.9895
0.1252	0.5442	80	0.1500	0.9473	0.9060	0.9757	0.9396	0.9872
0.1771	0.6122	90	0.1478	0.9473	0.9286	0.9474	0.9379	0.9855
0.1015	0.6803	100	0.1624	0.9405	0.9015	0.9636	0.9315	0.9860
0.2248	0.7483	110	0.0944	0.9694	0.9673	0.9595	0.9634	0.9938
0.1673	0.8163	120	0.2564	0.9116	0.9949	0.7935	0.8829	0.9923
0.1450	0.8844	130	0.1063	0.9490	0.9189	0.9636	0.9407	0.9936
0.1390	0.9524	140	0.1859	0.9320	0.9859	0.8502	0.9130	0.9938
0.0974	1.0204	150	0.2936	0.9065	0.9898	0.7854	0.8758	0.9879
0.1291	1.0884	160	0.1617	0.9490	0.9125	0.9717	0.9412	0.9891
0.1451	1.1565	170	0.1326	0.9490	0.9157	0.9676	0.9409	0.9919
0.1150	1.2245	180	0.1396	0.9575	0.9703	0.9271	0.9482	0.9916
0.0407	1.2925	190	0.1105	0.9643	0.9708	0.9433	0.9569	0.9949
0.1149	1.3605	200	0.0943	0.9711	0.9752	0.9555	0.9652	0.9953
0.1338	1.4286	210	0.3917	0.8776	0.9888	0.7166	0.8310	0.9891
0.1316	1.4966	220	0.3483	0.8827	0.9837	0.7328	0.8399	0.9911
0.1515	1.5646	230	0.1321	0.9558	0.9784	0.9150	0.9456	0.9947
0.0735	1.6327	240	0.0824	0.9728	0.9602	0.9757	0.9679	0.9955
0.1072	1.7007	250	0.1338	0.9558	0.9784	0.9150	0.9456	0.9935
0.0772	1.7687	260	0.0961	0.9694	0.9526	0.9757	0.964	0.9942
0.0550	1.8367	270	0.1113	0.9762	0.9679	0.9757	0.9718	0.9928
0.0800	1.9048	280	0.1537	0.9728	0.9753	0.9595	0.9673	0.9900
0.1291	1.9728	290	0.1863	0.9643	0.9708	0.9433	0.9569	0.9874
0.0165	2.0408	300	0.1702	0.9677	0.9831	0.9393	0.9607	0.9917
0.0427	2.1088	310	0.1526	0.9728	0.9714	0.9636	0.9675	0.9925
0.0848	2.1769	320	0.3300	0.9320	0.9859	0.8502	0.9130	0.9887
0.0343	2.2449	330	0.3590	0.9184	0.9854	0.8178	0.8938	0.9887
0.0801	2.3129	340	0.2959	0.9456	0.9864	0.8826	0.9316	0.9874
0.0687	2.3810	350	0.1900	0.9592	0.9869	0.9150	0.9496	0.9891
0.0900	2.4490	360	0.2463	0.9388	0.9862	0.8664	0.9224	0.9909
0.0601	2.5170	370	0.1783	0.9524	0.9824	0.9028	0.9409	0.9928
0.0579	2.5850	380	0.1160	0.9728	0.9753	0.9595	0.9673	0.9928
0.0253	2.6531	390	0.1333	0.9694	0.9712	0.9555	0.9633	0.9921
0.0406	2.7211	400	0.1167	0.9779	0.9606	0.9879	0.9741	0.9954
0.1257	2.7891	410	0.1229	0.9728	0.9753	0.9595	0.9673	0.9943
0.0682	2.8571	420	0.1546	0.9677	0.9831	0.9393	0.9607	0.9931
0.0783	2.9252	430	0.1837	0.9626	0.9870	0.9231	0.9540	0.9921
0.0838	2.9932	440	0.1288	0.9728	0.9873	0.9474	0.9669	0.9937
0.0381	3.0612	450	0.1164	0.9796	0.9876	0.9636	0.9754	0.9943
0.0109	3.1293	460	0.0885	0.9813	0.9836	0.9717	0.9776	0.9960
0.0035	3.1973	470	0.0883	0.9813	0.9797	0.9757	0.9777	0.9956
0.0237	3.2653	480	0.0980	0.9813	0.9797	0.9757	0.9777	0.9956
0.0061	3.3333	490	0.0987	0.9813	0.9797	0.9757	0.9777	0.9952
0.0399	3.4014	500	0.1049	0.9830	0.9877	0.9717	0.9796	0.9953
0.0288	3.4694	510	0.0983	0.9813	0.9836	0.9717	0.9776	0.9950
0.0063	3.5374	520	0.1056	0.9830	0.9877	0.9717	0.9796	0.9949
0.0233	3.6054	530	0.1297	0.9762	0.9874	0.9555	0.9712	0.9944
0.0130	3.6735	540	0.1280	0.9762	0.9874	0.9555	0.9712	0.9942
0.0036	3.7415	550	0.1400	0.9762	0.9874	0.9555	0.9712	0.9942
0.0202	3.8095	560	0.1381	0.9762	0.9874	0.9555	0.9712	0.9945
0.0071	3.8776	570	0.1164	0.9796	0.9876	0.9636	0.9754	0.9946
0.0187	3.9456	580	0.1126	0.9830	0.9877	0.9717	0.9796	0.9949

Framework versions

• Transformers 5.2.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2