segformer-finetuned-4ss1st3r_s3gs3m_24Jan_negro-10k-steps

This model is a fine-tuned version of nvidia/mit-b0 on the blzncz/4ss1st3r_s3gs3m_24Jan_negro dataset. It achieves the following results on the evaluation set:

• Loss: 0.2142
• Mean Iou: 0.5724
• Mean Accuracy: 0.7571
• Overall Accuracy: 0.9468
• Accuracy Bg: nan
• Accuracy Fallo cohesivo: 0.9826
• Accuracy Fallo malla: 0.7246
• Accuracy Fallo adhesivo: 0.9679
• Accuracy Fallo burbuja: 0.3533
• Iou Bg: 0.0
• Iou Fallo cohesivo: 0.9368
• Iou Fallo malla: 0.6678
• Iou Fallo adhesivo: 0.9310
• Iou Fallo burbuja: 0.3263

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
0.1902	1.0	219	0.2615	0.5072	0.7247	0.9038	nan	0.9203	0.7991	0.9401	0.2393	0.0	0.8853	0.5556	0.9006	0.1944
0.1367	2.0	438	0.2067	0.5492	0.7602	0.9293	nan	0.9599	0.7487	0.9317	0.4004	0.0	0.9160	0.6120	0.8988	0.3195
0.1066	3.0	657	0.1963	0.5659	0.7814	0.9313	nan	0.9520	0.8022	0.9545	0.4169	0.0	0.9175	0.6270	0.9267	0.3584
0.1102	4.0	876	0.1595	0.5782	0.7756	0.9444	nan	0.9727	0.7669	0.9693	0.3934	0.0	0.9336	0.6828	0.9326	0.3422
0.1114	5.0	1095	0.1678	0.5772	0.7950	0.9378	nan	0.9619	0.7778	0.9756	0.4648	0.0	0.9255	0.6534	0.9151	0.3922
0.0897	6.0	1314	0.1726	0.5811	0.7976	0.9420	nan	0.9701	0.7598	0.9723	0.4881	0.0	0.9307	0.6613	0.9170	0.3965
0.0788	7.0	1533	0.2096	0.5491	0.7253	0.9342	nan	0.9898	0.5936	0.9381	0.3797	0.0	0.9235	0.5698	0.9149	0.3374
0.0788	8.0	1752	0.1574	0.5774	0.7733	0.9465	nan	0.9726	0.7858	0.9675	0.3673	0.0	0.9359	0.6914	0.9264	0.3331
0.0855	9.0	1971	0.1970	0.5406	0.7141	0.9380	nan	0.9866	0.6305	0.9708	0.2687	0.0	0.9274	0.5984	0.9224	0.2548
0.0761	10.0	2190	0.1903	0.5564	0.7479	0.9382	nan	0.9746	0.7050	0.9737	0.3383	0.0	0.9268	0.6272	0.9182	0.3098
0.0686	11.0	2409	0.1910	0.5562	0.7435	0.9393	nan	0.9827	0.6605	0.9738	0.3572	0.0	0.9285	0.6156	0.9209	0.3160
0.062	12.0	2628	0.2038	0.5453	0.7399	0.9334	nan	0.9728	0.6739	0.9811	0.3317	0.0	0.9214	0.6013	0.9035	0.3001
0.0586	13.0	2847	0.1914	0.5471	0.7342	0.9402	nan	0.9758	0.7103	0.9814	0.2693	0.0	0.9290	0.6397	0.9150	0.2517
0.0531	14.0	3066	0.1747	0.5716	0.7689	0.9449	nan	0.9701	0.7945	0.9588	0.3522	0.0	0.9339	0.6815	0.9280	0.3147
0.0522	15.0	3285	0.1933	0.5591	0.7399	0.9454	nan	0.9810	0.7222	0.9744	0.2820	0.0	0.9351	0.6603	0.9355	0.2645
0.059	16.0	3504	0.1897	0.5691	0.7878	0.9384	nan	0.9499	0.8594	0.9809	0.3608	0.0	0.9252	0.6741	0.9159	0.3303
0.0503	17.0	3723	0.1895	0.5652	0.7795	0.9365	nan	0.9588	0.7866	0.9808	0.3917	0.0	0.9238	0.6508	0.9004	0.3511
0.0518	18.0	3942	0.2131	0.5533	0.7332	0.9402	nan	0.9807	0.6877	0.9645	0.2998	0.0	0.9294	0.6248	0.9334	0.2790
0.0439	19.0	4161	0.2168	0.5565	0.7411	0.9388	nan	0.9801	0.6828	0.9567	0.3448	0.0	0.9278	0.6194	0.9234	0.3121
0.0459	20.0	4380	0.2688	0.5266	0.7127	0.9266	nan	0.9824	0.5567	0.9841	0.3277	0.0	0.9149	0.5329	0.8866	0.2987
0.043	21.0	4599	0.2395	0.5542	0.7409	0.9369	nan	0.9821	0.6444	0.9745	0.3625	0.0	0.9258	0.5974	0.9228	0.3248
0.0436	22.0	4818	0.1790	0.5736	0.7750	0.9441	nan	0.9706	0.7783	0.9694	0.3819	0.0	0.9331	0.6772	0.9143	0.3433
0.0443	23.0	5037	0.1843	0.5683	0.7613	0.9442	nan	0.9756	0.7470	0.9716	0.3511	0.0	0.9335	0.6684	0.9177	0.3219
0.0402	24.0	5256	0.2048	0.5666	0.7535	0.9429	nan	0.9800	0.7089	0.9706	0.3544	0.0	0.9324	0.6457	0.9302	0.3246
0.0399	25.0	5475	0.2102	0.5651	0.7524	0.9430	nan	0.9830	0.6875	0.9754	0.3637	0.0	0.9327	0.6412	0.9231	0.3287
0.0404	26.0	5694	0.1993	0.5792	0.7815	0.9460	nan	0.9690	0.8035	0.9697	0.3837	0.0	0.9351	0.6876	0.9289	0.3443
0.0388	27.0	5913	0.2024	0.5681	0.7501	0.9470	nan	0.9821	0.7343	0.9605	0.3236	0.0	0.9370	0.6715	0.9322	0.3001
0.0369	28.0	6132	0.1830	0.5701	0.7553	0.9481	nan	0.9779	0.7698	0.9608	0.3126	0.0	0.9379	0.6871	0.9323	0.2931
0.0373	29.0	6351	0.2162	0.5682	0.7535	0.9438	nan	0.9828	0.7011	0.9639	0.3665	0.0	0.9335	0.6482	0.9239	0.3352
0.0348	30.0	6570	0.2126	0.5640	0.7479	0.9435	nan	0.9813	0.7097	0.9623	0.3384	0.0	0.9330	0.6537	0.9197	0.3135
0.0354	31.0	6789	0.2025	0.5626	0.7467	0.9469	nan	0.9795	0.7453	0.9725	0.2896	0.0	0.9368	0.6762	0.9285	0.2716
0.0344	32.0	7008	0.1973	0.5786	0.7739	0.9469	nan	0.9734	0.7828	0.9698	0.3695	0.0	0.9364	0.6853	0.9326	0.3389
0.0333	33.0	7227	0.2199	0.5722	0.7624	0.9438	nan	0.9817	0.7045	0.9696	0.3940	0.0	0.9334	0.6481	0.9287	0.3510
0.0345	34.0	7446	0.2052	0.5791	0.7724	0.9465	nan	0.9799	0.7347	0.9736	0.4015	0.0	0.9363	0.6698	0.9311	0.3582
0.0326	35.0	7665	0.2176	0.5758	0.7629	0.9462	nan	0.9835	0.7124	0.9689	0.3868	0.0	0.9362	0.6595	0.9345	0.3490
0.034	36.0	7884	0.2247	0.5717	0.7557	0.9453	nan	0.9841	0.7033	0.9661	0.3694	0.0	0.9352	0.6533	0.9331	0.3369
0.0324	37.0	8103	0.1957	0.5797	0.7736	0.9490	nan	0.9763	0.7801	0.9725	0.3657	0.0	0.9390	0.6963	0.9299	0.3333
0.0332	38.0	8322	0.1996	0.5770	0.7644	0.9478	nan	0.9826	0.7310	0.9696	0.3743	0.0	0.9379	0.6741	0.9336	0.3393
0.0332	39.0	8541	0.2129	0.5638	0.7423	0.9449	nan	0.9845	0.7021	0.9616	0.3212	0.0	0.9348	0.6514	0.9328	0.3001
0.03	40.0	8760	0.2283	0.5694	0.7539	0.9441	nan	0.9840	0.6931	0.9686	0.3699	0.0	0.9339	0.6464	0.9277	0.3387
0.0319	41.0	8979	0.2013	0.5741	0.7624	0.9471	nan	0.9804	0.7416	0.9670	0.3606	0.0	0.9370	0.6760	0.9277	0.3300
0.0361	42.0	9198	0.2094	0.5709	0.7568	0.9463	nan	0.9810	0.7317	0.9663	0.3483	0.0	0.9362	0.6689	0.9279	0.3216
0.0304	43.0	9417	0.2098	0.5731	0.7586	0.9468	nan	0.9821	0.7282	0.9666	0.3575	0.0	0.9368	0.6700	0.9295	0.3293
0.0303	44.0	9636	0.2155	0.5705	0.7554	0.9470	nan	0.9814	0.7329	0.9702	0.3370	0.0	0.9369	0.6718	0.9301	0.3137
0.03	45.0	9855	0.2183	0.5703	0.7541	0.9464	nan	0.9825	0.7229	0.9677	0.3435	0.0	0.9364	0.6657	0.9311	0.3181
0.0301	45.66	10000	0.2142	0.5724	0.7571	0.9468	nan	0.9826	0.7246	0.9679	0.3533	0.0	0.9368	0.6678	0.9310	0.3263

Framework versions

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