segformer-finetuned-4ss1st3r_s3gs3m_24Jan_gris-10k-steps

This model is a fine-tuned version of nvidia/mit-b0 on the blzncz/4ss1st3r_s3gs3m_24Jan_gris dataset. It achieves the following results on the evaluation set:

• Loss: 0.2539
• Mean Iou: 0.5001
• Mean Accuracy: 0.7682
• Overall Accuracy: 0.9671
• Accuracy Bg: nan
• Accuracy Fallo cohesivo: 0.9929
• Accuracy Fallo malla: 0.5837
• Accuracy Fallo adhesivo: 0.8806
• Accuracy Fallo burbuja: 0.6154
• Iou Bg: 0.0
• Iou Fallo cohesivo: 0.9663
• Iou Fallo malla: 0.5505
• Iou Fallo adhesivo: 0.4321
• Iou Fallo burbuja: 0.5515

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
0.1186	1.0	259	0.1688	0.5045	0.6750	0.9611	nan	0.9940	0.5027	0.8539	0.3494	0.0	0.9600	0.4717	0.7563	0.3344
0.0669	2.0	518	0.1603	0.4270	0.7755	0.9501	nan	0.9685	0.7091	0.8964	0.5282	0.0	0.9490	0.5466	0.1627	0.4767
0.0608	3.0	777	0.1863	0.4142	0.7612	0.9458	nan	0.9703	0.5906	0.9321	0.5517	0.0	0.9467	0.5366	0.0891	0.4985
0.0551	4.0	1036	0.1654	0.4515	0.7496	0.9620	nan	0.9879	0.5881	0.8763	0.5462	0.0	0.9620	0.5560	0.2349	0.5043
0.0462	5.0	1295	0.2067	0.4267	0.7598	0.9487	nan	0.9752	0.5450	0.8796	0.6392	0.0	0.9502	0.5377	0.0838	0.5620
0.0445	6.0	1554	0.1565	0.4557	0.7685	0.9627	nan	0.9873	0.5954	0.8571	0.6343	0.0	0.9636	0.5689	0.1837	0.5623
0.039	7.0	1813	0.1523	0.4576	0.8005	0.9609	nan	0.9817	0.6535	0.9036	0.6630	0.0	0.9612	0.5885	0.1643	0.5738
0.0367	8.0	2072	0.1954	0.4573	0.7462	0.9614	nan	0.9917	0.4963	0.8762	0.6206	0.0	0.9612	0.4850	0.2790	0.5612
0.0352	9.0	2331	0.2244	0.4757	0.7542	0.9636	nan	0.9932	0.5098	0.8867	0.6269	0.0	0.9629	0.5013	0.3466	0.5674
0.0357	10.0	2590	0.2119	0.4687	0.7394	0.9645	nan	0.9934	0.5378	0.8710	0.5552	0.0	0.9641	0.5209	0.3377	0.5207
0.0352	11.0	2849	0.1957	0.4469	0.7903	0.9584	nan	0.9791	0.6656	0.9237	0.5927	0.0	0.9591	0.5829	0.1459	0.5465
0.032	12.0	3108	0.1811	0.4521	0.8058	0.9594	nan	0.9797	0.6634	0.9338	0.6464	0.0	0.9608	0.5929	0.1397	0.5671
0.0299	13.0	3367	0.2403	0.4298	0.7596	0.9557	nan	0.9827	0.5553	0.9271	0.5733	0.0	0.9572	0.5336	0.1272	0.5311
0.0292	14.0	3626	0.2233	0.4667	0.7638	0.9642	nan	0.9900	0.5759	0.8508	0.6385	0.0	0.9638	0.5475	0.2511	0.5709
0.0264	15.0	3885	0.2382	0.4431	0.7690	0.9594	nan	0.9865	0.5492	0.9139	0.6267	0.0	0.9602	0.5326	0.1568	0.5658
0.0273	16.0	4144	0.2339	0.4382	0.7751	0.9570	nan	0.9818	0.5876	0.9193	0.6115	0.0	0.9584	0.5419	0.1352	0.5554
0.0249	17.0	4403	0.2078	0.4950	0.7846	0.9669	nan	0.9925	0.5784	0.9197	0.6477	0.0	0.9663	0.5508	0.3921	0.5658
0.0242	18.0	4662	0.2495	0.4809	0.7706	0.9645	nan	0.9922	0.5392	0.9007	0.6503	0.0	0.9640	0.5147	0.3577	0.5682
0.0241	19.0	4921	0.2117	0.4491	0.7954	0.9589	nan	0.9815	0.6243	0.9423	0.6336	0.0	0.9597	0.5703	0.1508	0.5647
0.0243	20.0	5180	0.1989	0.4754	0.8013	0.9656	nan	0.9875	0.6416	0.9194	0.6568	0.0	0.9658	0.5879	0.2482	0.5751
0.0251	21.0	5439	0.2095	0.4607	0.7962	0.9629	nan	0.9853	0.6324	0.9337	0.6334	0.0	0.9634	0.5732	0.2073	0.5598
0.0238	22.0	5698	0.2063	0.4747	0.7927	0.9650	nan	0.9873	0.6383	0.9158	0.6293	0.0	0.9645	0.5779	0.2744	0.5569
0.0225	23.0	5957	0.2260	0.4656	0.7915	0.9640	nan	0.9880	0.6003	0.9106	0.6672	0.0	0.9642	0.5647	0.2241	0.5752
0.0231	24.0	6216	0.2454	0.4688	0.7783	0.9645	nan	0.9891	0.6019	0.9197	0.6024	0.0	0.9643	0.5591	0.2766	0.5442
0.0218	25.0	6475	0.2482	0.5143	0.7752	0.9665	nan	0.9919	0.5896	0.9136	0.6057	0.0	0.9655	0.5433	0.5236	0.5390
0.0223	26.0	6734	0.2474	0.4648	0.7660	0.9642	nan	0.9903	0.5784	0.9054	0.5898	0.0	0.9639	0.5502	0.2768	0.5334
0.0238	27.0	6993	0.2475	0.4717	0.7669	0.9651	nan	0.9920	0.5597	0.9019	0.6138	0.0	0.9648	0.5379	0.3087	0.5470
0.021	28.0	7252	0.2490	0.4740	0.7708	0.9649	nan	0.9919	0.5573	0.9116	0.6222	0.0	0.9645	0.5362	0.3142	0.5553
0.0208	29.0	7511	0.2369	0.4633	0.7669	0.9653	nan	0.9896	0.6134	0.8846	0.5799	0.0	0.9652	0.5762	0.2422	0.5327
0.0202	30.0	7770	0.2498	0.4863	0.7654	0.9655	nan	0.9930	0.5488	0.8931	0.6267	0.0	0.9647	0.5273	0.3811	0.5582
0.021	31.0	8029	0.2534	0.4799	0.7729	0.9657	nan	0.9915	0.5794	0.9043	0.6164	0.0	0.9652	0.5474	0.3368	0.5502
0.0202	32.0	8288	0.2626	0.4771	0.7627	0.9653	nan	0.9930	0.5464	0.9014	0.6098	0.0	0.9647	0.5272	0.3464	0.5474
0.0201	33.0	8547	0.2710	0.4903	0.7673	0.9659	nan	0.9936	0.5431	0.8994	0.6329	0.0	0.9653	0.5221	0.3997	0.5645
0.0195	34.0	8806	0.2589	0.4915	0.7662	0.9663	nan	0.9930	0.5644	0.8895	0.6177	0.0	0.9656	0.5368	0.4014	0.5537
0.0194	35.0	9065	0.2304	0.5092	0.7801	0.9675	nan	0.9919	0.6048	0.8941	0.6295	0.0	0.9667	0.5615	0.4576	0.5603
0.0188	36.0	9324	0.2674	0.5022	0.7629	0.9670	nan	0.9933	0.5783	0.8819	0.5982	0.0	0.9662	0.5461	0.4567	0.5418
0.0188	37.0	9583	0.2580	0.4897	0.7702	0.9665	nan	0.9925	0.5791	0.8884	0.6207	0.0	0.9660	0.5485	0.3793	0.5548
0.0192	38.0	9842	0.2556	0.5065	0.7656	0.9673	nan	0.9933	0.5823	0.8739	0.6130	0.0	0.9665	0.5494	0.4667	0.5500
0.019	38.61	10000	0.2539	0.5001	0.7682	0.9671	nan	0.9929	0.5837	0.8806	0.6154	0.0	0.9663	0.5505	0.4321	0.5515

Framework versions

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