segformer-b5-finetuned-ade20k-morphpadver1-hgo-coord_40epochs_distortion_global_norm

This model is a fine-tuned version of nvidia/segformer-b5-finetuned-ade-640-640 on the NICOPOI-9/Morphpad_HGO_1600_coord_global_norm dataset. It achieves the following results on the evaluation set:

• Loss: 0.1027
• Mean Iou: 0.9780
• Mean Accuracy: 0.9889
• Overall Accuracy: 0.9886
• Accuracy 0-0: 0.9903
• Accuracy 0-90: 0.9850
• Accuracy 90-0: 0.9874
• Accuracy 90-90: 0.9929
• Iou 0-0: 0.9819
• Iou 0-90: 0.9740
• Iou 90-0: 0.9729
• Iou 90-90: 0.9831

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: 6e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy 0-0	Accuracy 0-90	Accuracy 90-0	Accuracy 90-90	Iou 0-0	Iou 0-90	Iou 90-0	Iou 90-90
1.2959	1.3680	4000	1.2209	0.2639	0.4166	0.4247	0.2827	0.4642	0.5298	0.3899	0.2232	0.2838	0.2934	0.2550
0.8526	2.7360	8000	0.9446	0.4145	0.5789	0.5841	0.5010	0.6754	0.5810	0.5582	0.4193	0.4020	0.4076	0.4290
0.6552	4.1040	12000	0.7331	0.5293	0.6854	0.6880	0.6425	0.6769	0.7503	0.6717	0.5626	0.4909	0.5151	0.5485
0.4479	5.4720	16000	0.5873	0.6216	0.7621	0.7634	0.7338	0.8038	0.7429	0.7677	0.6463	0.5867	0.6042	0.6494
0.3862	6.8399	20000	0.4229	0.7262	0.8411	0.8396	0.8751	0.8489	0.8048	0.8354	0.7527	0.6991	0.7079	0.7449
1.3206	8.2079	24000	0.2964	0.8276	0.9031	0.9040	0.8928	0.9069	0.9194	0.8932	0.8544	0.8091	0.8044	0.8427
1.2794	9.5759	28000	0.2393	0.8588	0.9236	0.9229	0.9416	0.8933	0.9435	0.9160	0.8776	0.8406	0.8410	0.8760
0.1104	10.9439	32000	0.1972	0.8876	0.9409	0.9396	0.9546	0.9336	0.9196	0.9559	0.9028	0.8736	0.8700	0.9040
0.5852	12.3119	36000	0.1674	0.9102	0.9536	0.9523	0.9572	0.9438	0.9339	0.9794	0.9167	0.9024	0.8896	0.9320
0.0711	13.6799	40000	0.1398	0.9289	0.9629	0.9625	0.9648	0.9542	0.9641	0.9684	0.9374	0.9178	0.9169	0.9435
0.5558	15.0479	44000	0.1391	0.9337	0.9656	0.9651	0.9677	0.9580	0.9638	0.9728	0.9424	0.9235	0.9227	0.9461
1.2811	16.4159	48000	0.1252	0.9426	0.9706	0.9700	0.9776	0.9681	0.9591	0.9777	0.9521	0.9341	0.9326	0.9516
1.1932	17.7839	52000	0.1162	0.9486	0.9740	0.9733	0.9811	0.9641	0.9690	0.9818	0.9553	0.9404	0.9428	0.9558
1.2719	19.1518	56000	0.1141	0.9533	0.9759	0.9757	0.9744	0.9717	0.9761	0.9814	0.9581	0.9487	0.9428	0.9638
0.0857	20.5198	60000	0.1049	0.9590	0.9790	0.9787	0.9783	0.9746	0.9761	0.9870	0.9655	0.9531	0.9508	0.9665
1.1869	21.8878	64000	0.1069	0.9558	0.9774	0.9770	0.9781	0.9625	0.9826	0.9865	0.9641	0.9494	0.9448	0.9649
0.0294	23.2558	68000	0.1028	0.9641	0.9819	0.9814	0.9840	0.9772	0.9760	0.9903	0.9680	0.9586	0.9564	0.9733
0.5373	24.6238	72000	0.1089	0.9639	0.9815	0.9813	0.9813	0.9771	0.9807	0.9869	0.9696	0.9584	0.9564	0.9711
0.7069	25.9918	76000	0.1026	0.9683	0.9840	0.9836	0.9859	0.9764	0.9842	0.9893	0.9734	0.9630	0.9620	0.9750
0.0382	27.3598	80000	0.1075	0.9647	0.9818	0.9819	0.9748	0.9835	0.9800	0.9890	0.9627	0.9630	0.9591	0.9739
0.0143	28.7278	84000	0.1082	0.9719	0.9858	0.9855	0.9896	0.9804	0.9864	0.9866	0.9778	0.9679	0.9654	0.9766
1.2277	30.0958	88000	0.0961	0.9730	0.9864	0.9861	0.9889	0.9818	0.9829	0.9922	0.9773	0.9685	0.9671	0.9791
1.1838	31.4637	92000	0.0947	0.9739	0.9868	0.9865	0.9899	0.9834	0.9846	0.9893	0.9785	0.9693	0.9679	0.9801
1.2587	32.8317	96000	0.0925	0.9745	0.9871	0.9869	0.9902	0.9827	0.9864	0.9893	0.9792	0.9700	0.9689	0.9801
0.0103	34.1997	100000	0.0988	0.9748	0.9871	0.9870	0.9865	0.9844	0.9869	0.9907	0.9782	0.9713	0.9693	0.9803
1.2995	35.5677	104000	0.0980	0.9763	0.9880	0.9878	0.9887	0.9838	0.9874	0.9921	0.9805	0.9727	0.9708	0.9813
0.0678	36.9357	108000	0.1043	0.9778	0.9888	0.9886	0.9899	0.9857	0.9873	0.9922	0.9812	0.9743	0.9722	0.9835
0.0094	38.3037	112000	0.1031	0.9775	0.9886	0.9884	0.9896	0.9848	0.9884	0.9916	0.9816	0.9742	0.9717	0.9826
0.1925	39.6717	116000	0.1027	0.9780	0.9889	0.9886	0.9903	0.9850	0.9874	0.9929	0.9819	0.9740	0.9729	0.9831

Framework versions

• Transformers 4.48.3
• Pytorch 2.1.0
• Datasets 3.2.0
• Tokenizers 0.21.0