segformer-b0-finetuned-metalography_D2

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: 0.0193
Mean Iou: 0.7827
Mean Accuracy: 0.9568
Overall Accuracy: 0.9914
Accuracy Background: nan
Accuracy Haz: 0.9922
Accuracy Matrix: 0.9841
Accuracy Porosity: 0.8395
Accuracy Carbides: 0.9723
Accuracy Substrate: 0.9959
Iou Background: 0.0
Iou Haz: 0.9865
Iou Matrix: 0.9702
Iou Porosity: 0.8041
Iou Carbides: 0.9453
Iou Substrate: 0.9903

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: 64
eval_batch_size: 64
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 Haz	Iou Matrix	Iou Porosity	Iou Carbides	Iou Substrate
0.2378	0.54	20	0.2047	0.5948	0.7483	0.9675	nan	0.9816	0.9589	0.0	0.8189	0.9820	0.9582	0.8742	0.0	0.7648	0.9717
0.2628	1.08	40	0.1833	0.6072	0.7602	0.9703	nan	0.9740	0.9589	0.0	0.8829	0.9854	0.9548	0.8995	0.0	0.8198	0.9690
0.1776	1.62	60	0.1527	0.6194	0.7689	0.9767	nan	0.9789	0.9640	0.0	0.9112	0.9905	0.9653	0.9182	0.0	0.8562	0.9768
0.162	2.16	80	0.1269	0.6268	0.7758	0.9795	nan	0.9878	0.9606	0.0	0.9441	0.9864	0.9692	0.9315	0.0	0.8814	0.9789
0.1265	2.7	100	0.1091	0.6293	0.7763	0.9802	nan	0.9832	0.9756	0.0	0.9341	0.9885	0.9681	0.9398	0.0	0.8897	0.9779
0.1208	3.24	120	0.1013	0.6313	0.7782	0.9818	nan	0.9842	0.9706	0.0	0.9449	0.9912	0.9712	0.9420	0.0	0.8945	0.9801
0.1309	3.78	140	0.0999	0.6292	0.7776	0.9780	nan	0.9929	0.9707	0.0	0.9488	0.9756	0.9618	0.9436	0.0	0.8983	0.9715
0.1013	4.32	160	0.0813	0.6330	0.7777	0.9821	nan	0.9856	0.9760	0.0	0.9368	0.9903	0.9720	0.9459	0.0	0.9000	0.9803
0.0933	4.86	180	0.0744	0.6348	0.7800	0.9836	nan	0.9874	0.9761	0.0	0.9454	0.9911	0.9751	0.9487	0.0	0.9025	0.9824
0.1043	5.41	200	0.0662	0.6362	0.7813	0.9841	nan	0.9835	0.9783	0.0002	0.9510	0.9938	0.9751	0.9523	0.0002	0.9075	0.9820
0.0705	5.95	220	0.0634	0.6361	0.7818	0.9835	nan	0.9814	0.9737	0.0022	0.9570	0.9947	0.9743	0.9513	0.0022	0.9075	0.9814
0.0797	6.49	240	0.0594	0.6395	0.7853	0.9841	nan	0.9825	0.9728	0.0152	0.9614	0.9948	0.9751	0.9531	0.0152	0.9114	0.9821
0.0813	7.03	260	0.0528	0.6405	0.7845	0.9847	nan	0.9891	0.9796	0.0178	0.9448	0.9912	0.9770	0.9531	0.0178	0.9114	0.9835
0.0693	7.57	280	0.0523	0.6444	0.7899	0.9843	nan	0.9934	0.9842	0.0417	0.9450	0.9854	0.9746	0.9556	0.0417	0.9134	0.9812
0.0618	8.11	300	0.0512	0.6480	0.7942	0.9839	nan	0.9796	0.9775	0.0595	0.9591	0.9953	0.9726	0.9572	0.0595	0.9183	0.9802
0.0537	8.65	320	0.0465	0.6533	0.7999	0.9858	nan	0.9896	0.9822	0.0812	0.9557	0.9907	0.9776	0.9586	0.0812	0.9188	0.9837
0.046	9.19	340	0.0450	0.6677	0.8168	0.9855	nan	0.9933	0.9827	0.1651	0.9565	0.9866	0.9761	0.9595	0.1650	0.9234	0.9823
0.0661	9.73	360	0.0428	0.6779	0.8280	0.9854	nan	0.9930	0.9839	0.2265	0.9495	0.9872	0.9761	0.9593	0.2263	0.9237	0.9823
0.0723	10.27	380	0.0412	0.6903	0.8425	0.9863	nan	0.9848	0.9844	0.2941	0.9548	0.9944	0.9775	0.9603	0.2938	0.9266	0.9838
0.0501	10.81	400	0.0411	0.6986	0.8523	0.9861	nan	0.9816	0.9838	0.3432	0.9570	0.9961	0.9760	0.9618	0.3427	0.9287	0.9825
0.0493	11.35	420	0.0396	0.7039	0.8582	0.9863	nan	0.9832	0.9864	0.3744	0.9517	0.9954	0.9771	0.9616	0.3735	0.9279	0.9833
0.0404	11.89	440	0.0419	0.7090	0.8664	0.9846	nan	0.9895	0.9802	0.4108	0.9648	0.9866	0.9720	0.9620	0.4098	0.9312	0.9791
0.0438	12.43	460	0.0393	0.7212	0.8806	0.9862	nan	0.9885	0.9800	0.4773	0.9670	0.9903	0.9758	0.9626	0.4751	0.9315	0.9821
0.0373	12.97	480	0.0352	0.7229	0.8820	0.9873	nan	0.9876	0.9797	0.4816	0.9678	0.9934	0.9787	0.9629	0.4797	0.9315	0.9844
0.034	13.51	500	0.0347	0.7306	0.8902	0.9873	nan	0.9861	0.9853	0.5285	0.9566	0.9943	0.9785	0.9632	0.5258	0.9319	0.9843
0.0368	14.05	520	0.0338	0.7366	0.8982	0.9876	nan	0.9858	0.9814	0.5617	0.9672	0.9949	0.9787	0.9639	0.5585	0.9339	0.9846
0.0369	14.59	540	0.0359	0.7406	0.9032	0.9860	nan	0.9853	0.9850	0.5947	0.9596	0.9917	0.9745	0.9645	0.5892	0.9346	0.9811
0.0381	15.14	560	0.0319	0.7408	0.9027	0.9881	nan	0.9896	0.9823	0.5825	0.9664	0.9927	0.9799	0.9647	0.5789	0.9362	0.9853
0.0307	15.68	580	0.0315	0.7430	0.9067	0.9878	nan	0.9878	0.9784	0.6012	0.9724	0.9938	0.9797	0.9640	0.5947	0.9349	0.9851
0.0311	16.22	600	0.0306	0.7540	0.9192	0.9882	nan	0.9871	0.9837	0.6649	0.9656	0.9945	0.9798	0.9657	0.6561	0.9372	0.9853
0.0377	16.76	620	0.0306	0.7569	0.9229	0.9880	nan	0.9875	0.9837	0.6845	0.9652	0.9937	0.9790	0.9659	0.6746	0.9375	0.9846
0.035	17.3	640	0.0302	0.7554	0.9203	0.9880	nan	0.9853	0.9864	0.6751	0.9591	0.9954	0.9792	0.9658	0.6664	0.9365	0.9848
0.0291	17.84	660	0.0293	0.7610	0.9283	0.9883	nan	0.9866	0.9860	0.7149	0.9591	0.9950	0.9799	0.9658	0.6982	0.9370	0.9853
0.0374	18.38	680	0.0280	0.7659	0.9346	0.9889	nan	0.9889	0.9819	0.7389	0.9692	0.9943	0.9811	0.9664	0.7230	0.9386	0.9862
0.0366	18.92	700	0.0303	0.7630	0.9317	0.9876	nan	0.9812	0.9802	0.7259	0.9736	0.9974	0.9776	0.9659	0.7128	0.9380	0.9837
0.0254	19.46	720	0.0273	0.7645	0.9327	0.9888	nan	0.9894	0.9810	0.7277	0.9718	0.9938	0.9809	0.9666	0.7140	0.9392	0.9860
0.0309	20.0	740	0.0276	0.7686	0.9384	0.9890	nan	0.9888	0.9823	0.7558	0.9709	0.9943	0.9811	0.9669	0.7380	0.9395	0.9862
0.0294	20.54	760	0.0267	0.7704	0.9422	0.9890	nan	0.9891	0.9822	0.7752	0.9702	0.9942	0.9812	0.9673	0.7478	0.9401	0.9862
0.0294	21.08	780	0.0266	0.7683	0.9380	0.9889	nan	0.9913	0.9822	0.7542	0.9699	0.9923	0.9809	0.9671	0.7355	0.9402	0.9860
0.0333	21.62	800	0.0261	0.7690	0.9395	0.9890	nan	0.9918	0.9834	0.7635	0.9668	0.9920	0.9815	0.9667	0.7395	0.9401	0.9865
0.0308	22.16	820	0.0259	0.7698	0.9390	0.9893	nan	0.9920	0.9846	0.7591	0.9670	0.9923	0.9818	0.9677	0.7421	0.9406	0.9866
0.025	22.7	840	0.0253	0.7722	0.9432	0.9893	nan	0.9906	0.9824	0.7795	0.9701	0.9936	0.9820	0.9675	0.7565	0.9405	0.9869
0.0353	23.24	860	0.0253	0.7734	0.9455	0.9894	nan	0.9925	0.9818	0.7888	0.9721	0.9920	0.9820	0.9675	0.7631	0.9409	0.9869
0.0324	23.78	880	0.0245	0.7731	0.9437	0.9897	nan	0.9926	0.9855	0.7834	0.9644	0.9927	0.9829	0.9677	0.7594	0.9409	0.9875
0.0258	24.32	900	0.0250	0.7732	0.9443	0.9894	nan	0.9941	0.9846	0.7849	0.9671	0.9907	0.9820	0.9680	0.7611	0.9416	0.9867
0.0271	24.86	920	0.0244	0.7735	0.9442	0.9897	nan	0.9921	0.9826	0.7811	0.9724	0.9929	0.9826	0.9681	0.7615	0.9418	0.9873
0.0212	25.41	940	0.0244	0.7737	0.9447	0.9895	nan	0.9885	0.9822	0.7851	0.9722	0.9953	0.9821	0.9680	0.7635	0.9417	0.9869
0.0255	25.95	960	0.0239	0.7750	0.9469	0.9899	nan	0.9913	0.9815	0.7933	0.9746	0.9939	0.9834	0.9679	0.7693	0.9414	0.9879
0.0214	26.49	980	0.0232	0.7745	0.9452	0.9900	nan	0.9903	0.9859	0.7892	0.9656	0.9948	0.9834	0.9686	0.7653	0.9420	0.9879
0.0298	27.03	1000	0.0242	0.7758	0.9481	0.9896	nan	0.9950	0.9835	0.8010	0.9704	0.9904	0.9824	0.9686	0.7744	0.9425	0.9870
0.026	27.57	1020	0.0235	0.7769	0.9499	0.9899	nan	0.9877	0.9821	0.8093	0.9737	0.9965	0.9829	0.9685	0.7800	0.9425	0.9876
0.0224	28.11	1040	0.0228	0.7761	0.9469	0.9901	nan	0.9902	0.9859	0.7974	0.9659	0.9952	0.9836	0.9688	0.7733	0.9425	0.9881
0.0283	28.65	1060	0.0227	0.7775	0.9501	0.9902	nan	0.9898	0.9826	0.8097	0.9729	0.9956	0.9838	0.9687	0.7815	0.9427	0.9882
0.0226	29.19	1080	0.0237	0.7779	0.9523	0.9896	nan	0.9954	0.9813	0.8207	0.9740	0.9902	0.9824	0.9683	0.7873	0.9423	0.9870
0.0208	29.73	1100	0.0224	0.7767	0.9485	0.9903	nan	0.9916	0.9830	0.8010	0.9726	0.9943	0.9840	0.9686	0.7766	0.9426	0.9884
0.0217	30.27	1120	0.0220	0.7792	0.9530	0.9905	nan	0.9919	0.9842	0.8247	0.9696	0.9945	0.9844	0.9691	0.7897	0.9432	0.9887
0.0182	30.81	1140	0.0219	0.7778	0.9498	0.9903	nan	0.9894	0.9849	0.8096	0.9689	0.9961	0.9841	0.9691	0.7820	0.9434	0.9885
0.02	31.35	1160	0.0227	0.7791	0.9531	0.9899	nan	0.9865	0.9840	0.8268	0.9709	0.9973	0.9827	0.9693	0.7913	0.9436	0.9875
0.0385	31.89	1180	0.0227	0.7778	0.9500	0.9899	nan	0.9865	0.9860	0.8126	0.9675	0.9972	0.9826	0.9694	0.7841	0.9436	0.9874
0.0206	32.43	1200	0.0218	0.7783	0.9506	0.9905	nan	0.9892	0.9830	0.8111	0.9732	0.9965	0.9844	0.9691	0.7844	0.9434	0.9887
0.031	32.97	1220	0.0219	0.7797	0.9532	0.9903	nan	0.9880	0.9840	0.8268	0.9701	0.9971	0.9838	0.9693	0.7930	0.9439	0.9883
0.022	33.51	1240	0.0227	0.7789	0.9514	0.9897	nan	0.9852	0.9844	0.8192	0.9705	0.9979	0.9821	0.9694	0.7910	0.9438	0.9870
0.0203	34.05	1260	0.0222	0.7793	0.9521	0.9900	nan	0.9865	0.9843	0.8215	0.9709	0.9974	0.9828	0.9695	0.7918	0.9441	0.9875
0.0233	34.59	1280	0.0208	0.7805	0.9543	0.9908	nan	0.9910	0.9843	0.8299	0.9705	0.9958	0.9852	0.9695	0.7946	0.9440	0.9894
0.0214	35.14	1300	0.0210	0.7787	0.9497	0.9908	nan	0.9898	0.9867	0.8087	0.9665	0.9966	0.9850	0.9696	0.7846	0.9439	0.9892
0.0255	35.68	1320	0.0209	0.7802	0.9534	0.9906	nan	0.9888	0.9854	0.8279	0.9678	0.9970	0.9845	0.9694	0.7942	0.9439	0.9888
0.0184	36.22	1340	0.0205	0.7808	0.9538	0.9909	nan	0.9902	0.9849	0.8276	0.9695	0.9965	0.9854	0.9697	0.7961	0.9444	0.9894
0.0173	36.76	1360	0.0205	0.7805	0.9537	0.9909	nan	0.9909	0.9836	0.8254	0.9726	0.9960	0.9854	0.9696	0.7944	0.9444	0.9895
0.0204	37.3	1380	0.0205	0.7810	0.9542	0.9910	nan	0.9920	0.9849	0.8285	0.9704	0.9953	0.9856	0.9698	0.7963	0.9446	0.9896
0.0227	37.84	1400	0.0206	0.7811	0.9555	0.9908	nan	0.9898	0.9840	0.8357	0.9714	0.9966	0.9851	0.9695	0.7982	0.9445	0.9893
0.0232	38.38	1420	0.0207	0.7815	0.9560	0.9908	nan	0.9894	0.9836	0.8378	0.9722	0.9969	0.9850	0.9698	0.8003	0.9447	0.9891
0.0197	38.92	1440	0.0202	0.7810	0.9545	0.9910	nan	0.9904	0.9839	0.8297	0.9722	0.9965	0.9856	0.9697	0.7966	0.9446	0.9897
0.0226	39.46	1460	0.0199	0.7808	0.9530	0.9912	nan	0.9916	0.9863	0.8236	0.9677	0.9959	0.9860	0.9699	0.7943	0.9446	0.9899
0.0231	40.0	1480	0.0198	0.7817	0.9557	0.9912	nan	0.9922	0.9848	0.8355	0.9707	0.9954	0.9861	0.9700	0.7994	0.9449	0.9900
0.0247	40.54	1500	0.0199	0.7824	0.9576	0.9912	nan	0.9918	0.9840	0.8447	0.9717	0.9958	0.9861	0.9699	0.8032	0.9450	0.9900
0.0182	41.08	1520	0.0208	0.7821	0.9570	0.9906	nan	0.9878	0.9848	0.8439	0.9709	0.9976	0.9844	0.9701	0.8043	0.9451	0.9887
0.0183	41.62	1540	0.0204	0.7817	0.9552	0.9908	nan	0.9889	0.9852	0.8345	0.9700	0.9972	0.9849	0.9701	0.8008	0.9450	0.9891
0.0242	42.16	1560	0.0201	0.7825	0.9573	0.9910	nan	0.9898	0.9837	0.8432	0.9730	0.9968	0.9854	0.9700	0.8051	0.9451	0.9895
0.0223	42.7	1580	0.0196	0.7825	0.9567	0.9912	nan	0.9917	0.9851	0.8408	0.9699	0.9959	0.9862	0.9700	0.8035	0.9450	0.9901
0.0213	43.24	1600	0.0194	0.7825	0.9564	0.9913	nan	0.9932	0.9849	0.8381	0.9706	0.9951	0.9865	0.9701	0.8029	0.9451	0.9903
0.0163	43.78	1620	0.0195	0.7824	0.9562	0.9913	nan	0.9932	0.9846	0.8373	0.9710	0.9950	0.9864	0.9701	0.8022	0.9451	0.9902
0.0209	44.32	1640	0.0195	0.7823	0.9562	0.9913	nan	0.9923	0.9842	0.8369	0.9718	0.9958	0.9864	0.9701	0.8022	0.9451	0.9902
0.0191	44.86	1660	0.0195	0.7823	0.9557	0.9913	nan	0.9915	0.9851	0.8350	0.9705	0.9963	0.9863	0.9702	0.8017	0.9452	0.9902
0.0219	45.41	1680	0.0197	0.7824	0.9570	0.9912	nan	0.9904	0.9833	0.8405	0.9738	0.9968	0.9860	0.9699	0.8035	0.9450	0.9899
0.0176	45.95	1700	0.0195	0.7824	0.9564	0.9913	nan	0.9920	0.9845	0.8384	0.9712	0.9959	0.9864	0.9700	0.8028	0.9451	0.9902
0.0259	46.49	1720	0.0197	0.7827	0.9574	0.9912	nan	0.9907	0.9846	0.8444	0.9708	0.9966	0.9860	0.9701	0.8050	0.9452	0.9899
0.0209	47.03	1740	0.0195	0.7823	0.9562	0.9913	nan	0.9941	0.9840	0.8362	0.9724	0.9941	0.9863	0.9701	0.8020	0.9452	0.9901
0.0269	47.57	1760	0.0193	0.7828	0.9573	0.9913	nan	0.9931	0.9838	0.8420	0.9723	0.9951	0.9865	0.9701	0.8045	0.9452	0.9903
0.0209	48.11	1780	0.0194	0.7829	0.9573	0.9913	nan	0.9916	0.9843	0.8429	0.9717	0.9962	0.9863	0.9702	0.8053	0.9453	0.9901
0.0251	48.65	1800	0.0193	0.7826	0.9567	0.9913	nan	0.9928	0.9842	0.8391	0.9718	0.9954	0.9865	0.9700	0.8034	0.9452	0.9903
0.0198	49.19	1820	0.0193	0.7831	0.9588	0.9913	nan	0.9926	0.9832	0.8496	0.9730	0.9956	0.9865	0.9699	0.8066	0.9451	0.9903
0.0169	49.73	1840	0.0193	0.7827	0.9568	0.9914	nan	0.9922	0.9841	0.8395	0.9723	0.9959	0.9865	0.9702	0.8041	0.9453	0.9903

Framework versions

Transformers 4.31.0
Pytorch 2.0.1+cu118
Datasets 2.14.3
Tokenizers 0.13.3

segformer-b0-finetuned-metalography_D2

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

NSDT 3DConvert

UnrealSynth

DreamTexture.js