ironchanchellor/segformer-b0_DsB

segformer-b0-finetuned-metallography_DsB

This model is a fine-tuned version of nvidia/mit-b0 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0162
• Mean Iou: 0.7889
• Mean Accuracy: 0.9743
• Overall Accuracy: 0.9937
• Accuracy Background: nan
• Accuracy Haz: 0.9934
• Accuracy Matrix: 0.9859
• Accuracy Porosity: 0.9183
• Accuracy Carbides: 0.9759
• Accuracy Substrate: 0.9981
• Iou Background: 0.0
• Iou Haz: 0.9909
• Iou Matrix: 0.9758
• Iou Porosity: 0.8239
• Iou Carbides: 0.9504
• Iou Substrate: 0.9926

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Haz	Accuracy Matrix	Accuracy Porosity	Accuracy Carbides	Accuracy Substrate	Iou Background	Iou Haz	Iou Matrix	Iou Porosity	Iou Carbides	Iou Substrate
1.1925	1.0	350	0.2093	0.5751	0.7355	0.9228	nan	0.8617	0.9887	0.0	0.8605	0.9668	0.0	0.8289	0.9133	0.0	0.8400	0.8683
0.3065	2.0	700	0.1070	0.6106	0.7607	0.9570	nan	0.9158	0.9711	0.0	0.9221	0.9945	0.0	0.9053	0.9400	0.0	0.8907	0.9276
0.1839	3.0	1050	0.0717	0.6284	0.7777	0.9747	nan	0.9737	0.9668	0.0	0.9676	0.9802	0.0	0.9488	0.9513	0.0	0.9116	0.9590
1.0057	4.0	1400	0.0470	0.6322	0.7765	0.9783	nan	0.9889	0.9718	0.0	0.9460	0.9761	0.0	0.9580	0.9493	0.0	0.9193	0.9669
1.3313	5.0	1750	0.0360	0.6338	0.7751	0.9839	nan	0.9825	0.9861	0.0	0.9128	0.9940	0.0	0.9737	0.9478	0.0	0.9015	0.9799
0.1398	6.0	2100	0.0333	0.6407	0.7849	0.9853	nan	0.9943	0.9782	0.0	0.9689	0.9830	0.0	0.9743	0.9623	0.0	0.9290	0.9787
0.4763	7.0	2450	0.0941	0.6520	0.8054	0.9710	nan	0.9435	0.9745	0.1384	0.9757	0.9950	0.0	0.9367	0.9622	0.1384	0.9258	0.9486
0.074	8.0	2800	0.0373	0.7154	0.8725	0.9848	nan	0.9877	0.9841	0.4466	0.9577	0.9864	0.0	0.9711	0.9646	0.4466	0.9339	0.9760
0.0637	9.0	3150	0.0239	0.7358	0.8946	0.9885	nan	0.9867	0.9907	0.5610	0.9388	0.9956	0.0	0.9815	0.9631	0.5591	0.9258	0.9851
0.0402	10.0	3500	0.0295	0.7462	0.9085	0.9865	nan	0.9774	0.9872	0.6256	0.9541	0.9982	0.0	0.9752	0.9662	0.6232	0.9333	0.9796
1.069	11.0	3850	0.0244	0.7494	0.9115	0.9889	nan	0.9874	0.9908	0.6455	0.9383	0.9957	0.0	0.9822	0.9644	0.6384	0.9263	0.9854
0.5997	12.0	4200	0.0243	0.7492	0.9106	0.9893	nan	0.9859	0.9884	0.6271	0.9545	0.9970	0.0	0.9817	0.9684	0.6246	0.9356	0.9850
0.091	13.0	4550	0.0269	0.7557	0.9197	0.9886	nan	0.9858	0.9900	0.6747	0.9530	0.9950	0.0	0.9799	0.9693	0.6659	0.9361	0.9833
1.3004	14.0	4900	0.0226	0.7740	0.9448	0.9906	nan	0.9887	0.9859	0.7857	0.9674	0.9964	0.0	0.9841	0.9719	0.7585	0.9424	0.9870
0.94	15.0	5250	0.1346	0.7572	0.9315	0.9731	nan	0.9938	0.9862	0.7591	0.9657	0.9528	0.0	0.9423	0.9709	0.7399	0.9417	0.9481
0.8906	16.0	5600	0.0221	0.7781	0.9528	0.9911	nan	0.9886	0.9844	0.8206	0.9729	0.9973	0.0	0.9851	0.9724	0.7805	0.9429	0.9877
0.9739	17.0	5950	0.0233	0.7629	0.9264	0.9905	nan	0.9870	0.9914	0.7040	0.9516	0.9980	0.0	0.9845	0.9700	0.6986	0.9367	0.9874
0.417	18.0	6300	0.0200	0.7724	0.9392	0.9917	nan	0.9911	0.9909	0.7618	0.9556	0.9967	0.0	0.9869	0.9718	0.7468	0.9399	0.9893
0.0405	19.0	6650	0.1657	0.7661	0.9474	0.9743	nan	0.9434	0.9863	0.8421	0.9661	0.9991	0.0	0.9421	0.9718	0.7877	0.9422	0.9528
1.2414	20.0	7000	0.0275	0.7808	0.9593	0.9900	nan	0.9844	0.9838	0.8565	0.9733	0.9986	0.0	0.9824	0.9725	0.8000	0.9442	0.9855
0.7539	21.0	7350	0.0200	0.7791	0.9509	0.9918	nan	0.9947	0.9857	0.8106	0.9698	0.9936	0.0	0.9872	0.9724	0.7813	0.9445	0.9895
0.0158	22.0	7700	0.0159	0.7773	0.9468	0.9926	nan	0.9924	0.9854	0.7855	0.9736	0.9972	0.0	0.9889	0.9731	0.7657	0.9448	0.9910
0.3368	23.0	8050	0.0176	0.7849	0.9678	0.9925	nan	0.9962	0.9844	0.8892	0.9758	0.9933	0.0	0.9882	0.9739	0.8113	0.9459	0.9904
0.0526	24.0	8400	0.0168	0.7835	0.9629	0.9927	nan	0.9916	0.9895	0.8727	0.9629	0.9978	0.0	0.9888	0.9739	0.8030	0.9448	0.9908
0.9409	25.0	8750	0.0205	0.7842	0.9681	0.9920	nan	0.9899	0.9829	0.8925	0.9773	0.9980	0.0	0.9873	0.9732	0.8096	0.9452	0.9897
1.0493	26.0	9100	0.0187	0.7823	0.9542	0.9924	nan	0.9906	0.9877	0.8277	0.9670	0.9981	0.0	0.9881	0.9736	0.7966	0.9454	0.9903
0.0685	27.0	9450	0.0166	0.7833	0.9549	0.9931	nan	0.9939	0.9868	0.8270	0.9698	0.9969	0.0	0.9898	0.9741	0.7970	0.9470	0.9917
0.0594	28.0	9800	0.0172	0.7882	0.9705	0.9932	nan	0.9942	0.9849	0.9007	0.9761	0.9965	0.0	0.9898	0.9749	0.8251	0.9479	0.9917
1.1676	29.0	10150	0.0166	0.7867	0.9726	0.9930	nan	0.9948	0.9834	0.9115	0.9777	0.9957	0.0	0.9896	0.9741	0.8178	0.9474	0.9915
0.076	30.0	10500	0.0184	0.7845	0.9595	0.9928	nan	0.9925	0.9898	0.8578	0.9598	0.9976	0.0	0.9895	0.9728	0.8090	0.9439	0.9917
0.0709	31.0	10850	0.0187	0.7876	0.9726	0.9931	nan	0.9934	0.9842	0.9118	0.9764	0.9972	0.0	0.9897	0.9744	0.8215	0.9480	0.9917
0.2951	32.0	11200	0.0171	0.7879	0.9701	0.9932	nan	0.9949	0.9853	0.8995	0.9747	0.9961	0.0	0.9900	0.9747	0.8226	0.9484	0.9919
0.0371	33.0	11550	0.0165	0.7863	0.9624	0.9932	nan	0.9941	0.9871	0.8644	0.9696	0.9967	0.0	0.9900	0.9742	0.8138	0.9480	0.9920
0.0374	34.0	11900	0.0183	0.7874	0.9718	0.9929	nan	0.9910	0.9862	0.9089	0.9743	0.9985	0.0	0.9891	0.9752	0.8202	0.9490	0.9911
0.7856	35.0	12250	0.0187	0.7873	0.9710	0.9931	nan	0.9918	0.9860	0.9042	0.9751	0.9981	0.0	0.9894	0.9753	0.8192	0.9483	0.9914
0.9141	36.0	12600	0.0151	0.7892	0.9686	0.9938	nan	0.9946	0.9881	0.8920	0.9712	0.9973	0.0	0.9912	0.9759	0.8254	0.9497	0.9929
0.0195	37.0	12950	0.0169	0.7880	0.9653	0.9932	nan	0.9918	0.9875	0.8770	0.9719	0.9985	0.0	0.9897	0.9755	0.8219	0.9493	0.9916
0.0355	38.0	13300	0.0177	0.7888	0.9717	0.9933	nan	0.9936	0.9843	0.9041	0.9796	0.9969	0.0	0.9898	0.9755	0.8272	0.9487	0.9917
0.07	39.0	13650	0.0165	0.7880	0.9736	0.9935	nan	0.9941	0.9848	0.9152	0.9765	0.9973	0.0	0.9906	0.9750	0.8209	0.9491	0.9924
0.0244	40.0	14000	0.0178	0.7889	0.9696	0.9933	nan	0.9927	0.9854	0.8963	0.9758	0.9980	0.0	0.9899	0.9753	0.8268	0.9496	0.9919
0.0679	41.0	14350	0.0157	0.7895	0.9707	0.9936	nan	0.9945	0.9858	0.9012	0.9750	0.9972	0.0	0.9908	0.9754	0.8284	0.9499	0.9926
0.0498	42.0	14700	0.0164	0.7866	0.9765	0.9935	nan	0.9938	0.9839	0.9292	0.9781	0.9976	0.0	0.9907	0.9748	0.8122	0.9494	0.9925
0.0593	43.0	15050	0.0146	0.7881	0.9644	0.9939	nan	0.9953	0.9873	0.8695	0.9730	0.9970	0.0	0.9916	0.9756	0.8186	0.9494	0.9932
0.0068	44.0	15400	0.0151	0.7883	0.9743	0.9938	nan	0.9942	0.9857	0.9191	0.9749	0.9978	0.0	0.9913	0.9753	0.8203	0.9498	0.9930
1.2941	45.0	15750	0.0150	0.7888	0.9714	0.9939	nan	0.9954	0.9862	0.9044	0.9742	0.9968	0.0	0.9915	0.9754	0.8228	0.9499	0.9932
0.0113	46.0	16100	0.0151	0.7893	0.9732	0.9939	nan	0.9943	0.9866	0.9130	0.9741	0.9978	0.0	0.9914	0.9759	0.8251	0.9505	0.9930
0.9812	47.0	16450	0.0185	0.7875	0.9754	0.9933	nan	0.9920	0.9864	0.9257	0.9745	0.9984	0.0	0.9898	0.9759	0.8175	0.9503	0.9917
0.0126	48.0	16800	0.0152	0.7887	0.9743	0.9938	nan	0.9942	0.9856	0.9185	0.9755	0.9976	0.0	0.9911	0.9756	0.8221	0.9506	0.9929
1.4415	49.0	17150	0.0154	0.7894	0.9674	0.9940	nan	0.9952	0.9872	0.8839	0.9734	0.9972	0.0	0.9917	0.9759	0.8255	0.9501	0.9934
0.0285	50.0	17500	0.0162	0.7889	0.9743	0.9937	nan	0.9934	0.9859	0.9183	0.9759	0.9981	0.0	0.9909	0.9758	0.8239	0.9504	0.9926

Framework versions

• Transformers 4.32.1
• Pytorch 2.0.1+cu118
• Datasets 2.14.4
• Tokenizers 0.13.3