segformer-finetuned-4ss1st3r_s3gs3m_24Jan-10k-steps

This model is a fine-tuned version of nvidia/mit-b0 on the blzncz/4ss1st3r_s3gs3m_24Jan dataset. It achieves the following results on the evaluation set:

• Loss: 0.1305
• Mean Iou: 0.6564
• Mean Accuracy: 0.8562
• Overall Accuracy: 0.9780
• Accuracy Bg: nan
• Accuracy Fallo cohesivo: 0.9896
• Accuracy Fallo malla: 0.9270
• Accuracy Fallo adhesivo: 0.9478
• Accuracy Fallo burbuja: 0.5603
• Iou Bg: 0.0
• Iou Fallo cohesivo: 0.9749
• Iou Fallo malla: 0.8458
• Iou Fallo adhesivo: 0.9324
• Iou Fallo burbuja: 0.5290

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: 8
• eval_batch_size: 8
• seed: 1337
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: polynomial
• training_steps: 10000

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Bg	Accuracy Fallo cohesivo	Accuracy Fallo malla	Accuracy Fallo adhesivo	Accuracy Fallo burbuja	Iou Bg	Iou Fallo cohesivo	Iou Fallo malla	Iou Fallo adhesivo	Iou Fallo burbuja
0.3639	1.0	193	0.1583	0.6076	0.8441	0.9607	nan	0.9660	0.9617	0.9644	0.4844	0.0	0.9553	0.7294	0.9301	0.4231
0.1148	2.0	386	0.0991	0.6189	0.8025	0.9754	nan	0.9912	0.9045	0.9417	0.3725	0.0	0.9723	0.8404	0.9283	0.3534
0.0937	3.0	579	0.1414	0.5848	0.8155	0.9554	nan	0.9606	0.9630	0.9707	0.3675	0.0	0.9487	0.6791	0.9442	0.3519
0.0827	4.0	772	0.1028	0.6390	0.8484	0.9747	nan	0.9831	0.9530	0.9640	0.4936	0.0	0.9714	0.8231	0.9388	0.4617
0.0735	5.0	965	0.0948	0.6425	0.8423	0.9777	nan	0.9875	0.9487	0.9594	0.4737	0.0	0.9745	0.8484	0.9415	0.4479
0.0716	6.0	1158	0.0968	0.6638	0.8622	0.9804	nan	0.9936	0.8987	0.9579	0.5985	0.0	0.9777	0.8654	0.9403	0.5355
0.0692	7.0	1351	0.1123	0.6389	0.8535	0.9718	nan	0.9804	0.9425	0.9604	0.5307	0.0	0.9678	0.7878	0.9403	0.4984
0.0718	8.0	1544	0.1097	0.6424	0.8668	0.9703	nan	0.9770	0.9520	0.9642	0.5738	0.0	0.9663	0.7792	0.9423	0.5243
0.0613	9.0	1737	0.1212	0.6341	0.8625	0.9669	nan	0.9735	0.9412	0.9721	0.5634	0.0	0.9621	0.7447	0.9430	0.5208
0.06	10.0	1930	0.0983	0.6724	0.8945	0.9793	nan	0.9875	0.9335	0.9682	0.6889	0.0	0.9765	0.8490	0.9461	0.5905
0.0593	11.0	2123	0.1104	0.6577	0.8803	0.9743	nan	0.9830	0.9249	0.9670	0.6462	0.0	0.9709	0.8028	0.9419	0.5729
0.056	12.0	2316	0.1029	0.6589	0.8829	0.9755	nan	0.9833	0.9349	0.9712	0.6420	0.0	0.9721	0.8170	0.9399	0.5655
0.0547	13.0	2509	0.1037	0.6613	0.8944	0.9746	nan	0.9815	0.9406	0.9680	0.6877	0.0	0.9712	0.8089	0.9434	0.5832
0.0538	14.0	2702	0.1342	0.6338	0.8750	0.9625	nan	0.9677	0.9470	0.9647	0.6204	0.0	0.9570	0.7080	0.9412	0.5627
0.052	15.0	2895	0.0961	0.6525	0.8507	0.9787	nan	0.9894	0.9292	0.9656	0.5187	0.0	0.9758	0.8514	0.9439	0.4915
0.0489	16.0	3088	0.1093	0.6464	0.8626	0.9725	nan	0.9812	0.9345	0.9639	0.5708	0.0	0.9688	0.7900	0.9440	0.5290
0.0478	17.0	3281	0.1053	0.6503	0.8574	0.9760	nan	0.9858	0.9300	0.9673	0.5465	0.0	0.9726	0.8239	0.9411	0.5138
0.048	18.0	3474	0.1314	0.6416	0.8884	0.9644	nan	0.9691	0.9517	0.9642	0.6688	0.0	0.9591	0.7232	0.9415	0.5842
0.0474	19.0	3667	0.1197	0.6473	0.8559	0.9743	nan	0.9842	0.9344	0.9557	0.5493	0.0	0.9707	0.8067	0.9394	0.5196
0.0456	20.0	3860	0.1149	0.6587	0.8578	0.9788	nan	0.9905	0.9241	0.9503	0.5665	0.0	0.9759	0.8513	0.9344	0.5321
0.044	21.0	4053	0.1183	0.6574	0.8612	0.9774	nan	0.9885	0.9280	0.9487	0.5794	0.0	0.9743	0.8367	0.9345	0.5413
0.0431	22.0	4246	0.1326	0.6425	0.8599	0.9711	nan	0.9795	0.9405	0.9595	0.5601	0.0	0.9670	0.7783	0.9384	0.5291
0.0446	23.0	4439	0.1253	0.6535	0.8678	0.9743	nan	0.9833	0.9309	0.9635	0.5933	0.0	0.9706	0.8007	0.9427	0.5535
0.0427	24.0	4632	0.1075	0.6568	0.8602	0.9771	nan	0.9882	0.9229	0.9543	0.5755	0.0	0.9739	0.8342	0.9379	0.5379
0.0417	25.0	4825	0.1250	0.6443	0.8559	0.9723	nan	0.9820	0.9337	0.9542	0.5539	0.0	0.9684	0.7904	0.9375	0.5250
0.0402	26.0	5018	0.1206	0.6518	0.8497	0.9775	nan	0.9892	0.9236	0.9536	0.5324	0.0	0.9744	0.8373	0.9383	0.5089
0.0403	27.0	5211	0.1164	0.6565	0.8688	0.9755	nan	0.9848	0.9382	0.9531	0.5991	0.0	0.9723	0.8183	0.9378	0.5540
0.0405	28.0	5404	0.1091	0.6586	0.8505	0.9799	nan	0.9926	0.9177	0.9530	0.5389	0.0	0.9773	0.8650	0.9381	0.5128
0.0384	29.0	5597	0.1304	0.6504	0.8470	0.9781	nan	0.9893	0.9365	0.9508	0.5112	0.0	0.9751	0.8477	0.9365	0.4926
0.0374	30.0	5790	0.1095	0.6585	0.8605	0.9783	nan	0.9891	0.9323	0.9507	0.5698	0.0	0.9754	0.8469	0.9358	0.5345
0.0378	31.0	5983	0.1245	0.6558	0.8553	0.9780	nan	0.9896	0.9237	0.9539	0.5540	0.0	0.9750	0.8435	0.9353	0.5254
0.0367	32.0	6176	0.1288	0.6504	0.8637	0.9737	nan	0.9828	0.9386	0.9555	0.5778	0.0	0.9700	0.8016	0.9362	0.5443
0.037	33.0	6369	0.1293	0.6565	0.8656	0.9760	nan	0.9862	0.9381	0.9443	0.5938	0.0	0.9726	0.8273	0.9314	0.5512
0.0363	34.0	6562	0.1242	0.6594	0.8528	0.9800	nan	0.9926	0.9171	0.9529	0.5485	0.0	0.9773	0.8632	0.9378	0.5188
0.0361	35.0	6755	0.1239	0.6653	0.8739	0.9781	nan	0.9886	0.9247	0.9557	0.6264	0.0	0.9752	0.8420	0.9374	0.5718
0.0371	36.0	6948	0.1220	0.6626	0.8691	0.9782	nan	0.9887	0.9297	0.9530	0.6049	0.0	0.9751	0.8418	0.9375	0.5585
0.034	37.0	7141	0.1694	0.6300	0.8685	0.9609	nan	0.9666	0.9453	0.9602	0.6020	0.0	0.9551	0.6981	0.9399	0.5567
0.0358	38.0	7334	0.1251	0.6513	0.8534	0.9764	nan	0.9878	0.9270	0.9492	0.5497	0.0	0.9731	0.8290	0.9345	0.5198
0.033	39.0	7527	0.1330	0.6542	0.8604	0.9764	nan	0.9868	0.9343	0.9503	0.5700	0.0	0.9731	0.8292	0.9351	0.5336
0.0327	40.0	7720	0.1359	0.6490	0.8537	0.9750	nan	0.9862	0.9269	0.9483	0.5535	0.0	0.9716	0.8183	0.9330	0.5221
0.0336	41.0	7913	0.1277	0.6588	0.8667	0.9766	nan	0.9874	0.9267	0.9489	0.6037	0.0	0.9734	0.8288	0.9341	0.5577
0.0312	42.0	8106	0.1321	0.6568	0.8716	0.9749	nan	0.9844	0.9358	0.9500	0.6163	0.0	0.9714	0.8132	0.9344	0.5650
0.0321	43.0	8299	0.1269	0.6533	0.8574	0.9763	nan	0.9874	0.9283	0.9490	0.5649	0.0	0.9730	0.8285	0.9335	0.5316
0.0306	44.0	8492	0.1269	0.6583	0.8528	0.9792	nan	0.9918	0.9207	0.9467	0.5520	0.0	0.9764	0.8593	0.9324	0.5236
0.0306	45.0	8685	0.1335	0.6503	0.8503	0.9765	nan	0.9883	0.9283	0.9439	0.5407	0.0	0.9733	0.8345	0.9295	0.5144
0.0324	46.0	8878	0.1294	0.6538	0.8490	0.9784	nan	0.9908	0.9254	0.9441	0.5358	0.0	0.9754	0.8525	0.9303	0.5107
0.0318	47.0	9071	0.1230	0.6564	0.8549	0.9782	nan	0.9900	0.9252	0.9486	0.5559	0.0	0.9752	0.8477	0.9335	0.5255
0.0319	48.0	9264	0.1267	0.6524	0.8501	0.9776	nan	0.9895	0.9278	0.9464	0.5368	0.0	0.9745	0.8438	0.9322	0.5117
0.0312	49.0	9457	0.1258	0.6568	0.8602	0.9774	nan	0.9884	0.9321	0.9482	0.5720	0.0	0.9743	0.8399	0.9327	0.5373
0.0311	50.0	9650	0.1203	0.6589	0.8610	0.9779	nan	0.9894	0.9262	0.9471	0.5814	0.0	0.9749	0.8444	0.9319	0.5435
0.0327	51.0	9843	0.1219	0.6575	0.8577	0.9780	nan	0.9897	0.9265	0.9457	0.5688	0.0	0.9750	0.8462	0.9314	0.5348
0.031	51.81	10000	0.1305	0.6564	0.8562	0.9780	nan	0.9896	0.9270	0.9478	0.5603	0.0	0.9749	0.8458	0.9324	0.5290

Framework versions

• Transformers 4.31.0.dev0
• Pytorch 2.0.1+cpu
• Datasets 2.13.1
• Tokenizers 0.13.3