segformer-b0-finetuned-morphpadver1-hgo-coord-v9_mix_resample_40epochs

This model is a fine-tuned version of nvidia/mit-b0 on the NICOPOI-9/morphpad_coord_hgo_512_4class_v2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8517
• Mean Iou: 0.5269
• Mean Accuracy: 0.6857
• Overall Accuracy: 0.6922
• Accuracy 0-0: 0.5774
• Accuracy 0-90: 0.7340
• Accuracy 90-0: 0.7692
• Accuracy 90-90: 0.6625
• Iou 0-0: 0.4919
• Iou 0-90: 0.5434
• Iou 90-0: 0.5346
• Iou 90-90: 0.5380

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: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy 0-0	Accuracy 0-90	Accuracy 90-0	Accuracy 90-90	Iou 0-0	Iou 0-90	Iou 90-0	Iou 90-90
1.4178	1.3638	4000	1.4017	0.0915	0.2543	0.2763	0.0012	0.1224	0.8886	0.0052	0.0012	0.0995	0.2601	0.0052
1.1726	2.7276	8000	1.3327	0.2022	0.3465	0.3633	0.1781	0.4945	0.5489	0.1646	0.1323	0.2709	0.2783	0.1273
1.3111	4.0914	12000	1.3034	0.2235	0.3684	0.3811	0.2368	0.5790	0.4089	0.2489	0.1621	0.3039	0.2584	0.1694
1.033	5.4552	16000	1.2837	0.2340	0.3853	0.4006	0.1897	0.5159	0.5735	0.2619	0.1526	0.2965	0.3101	0.1769
1.3103	6.8190	20000	1.2502	0.2593	0.4171	0.4339	0.2446	0.6314	0.5467	0.2456	0.1839	0.3469	0.3212	0.1851
0.6831	8.1827	24000	1.2405	0.2655	0.4238	0.4336	0.2449	0.3989	0.6621	0.3893	0.1849	0.2957	0.3333	0.2482
1.1638	9.5465	28000	1.1866	0.2955	0.4566	0.4696	0.3300	0.6108	0.5695	0.3160	0.2396	0.3484	0.3479	0.2463
1.2145	10.9103	32000	1.1129	0.3356	0.5008	0.5092	0.4052	0.5818	0.5926	0.4236	0.2913	0.3764	0.3705	0.3042
0.767	12.2741	36000	1.1059	0.3423	0.5078	0.5144	0.4463	0.5576	0.5978	0.4295	0.3098	0.3613	0.3732	0.3250
1.0089	13.6379	40000	1.0832	0.3500	0.5157	0.5252	0.4129	0.6431	0.5790	0.4280	0.3054	0.3812	0.3870	0.3263
1.0757	15.0017	44000	1.0207	0.3866	0.5553	0.5626	0.4802	0.6133	0.6502	0.4776	0.3529	0.4112	0.4246	0.3577
0.8842	16.3655	48000	1.0716	0.3737	0.5417	0.5529	0.4372	0.6738	0.6390	0.4169	0.3371	0.4152	0.4191	0.3234
0.8464	17.7293	52000	1.0188	0.4101	0.5795	0.5884	0.4262	0.6296	0.7147	0.5474	0.3467	0.4325	0.4543	0.4069
0.8371	19.0931	56000	0.9905	0.4260	0.5942	0.6027	0.4614	0.6846	0.6765	0.5542	0.3766	0.4455	0.4655	0.4166
0.7882	20.4569	60000	0.9542	0.4454	0.6126	0.6216	0.4838	0.6737	0.7397	0.5530	0.3925	0.4665	0.4815	0.4411
2.4763	21.8207	64000	0.9188	0.4671	0.6330	0.6402	0.5338	0.6708	0.7484	0.5788	0.4359	0.4820	0.4932	0.4572
0.3528	23.1845	68000	0.8817	0.4725	0.6381	0.6450	0.5270	0.6813	0.7379	0.6063	0.4314	0.4937	0.4905	0.4745
0.8088	24.5482	72000	0.9115	0.4800	0.6458	0.6500	0.5807	0.6461	0.7349	0.6217	0.4507	0.4844	0.4938	0.4912
0.8153	25.9120	76000	0.9558	0.4531	0.6215	0.6342	0.3715	0.7059	0.7951	0.6135	0.3382	0.5023	0.4889	0.4829
0.9085	27.2758	80000	0.9089	0.4777	0.6415	0.6542	0.4936	0.7556	0.7928	0.5238	0.4312	0.5149	0.5148	0.4500
0.3666	28.6396	84000	1.0426	0.4467	0.6141	0.6270	0.3862	0.6873	0.8064	0.5767	0.3460	0.5000	0.4754	0.4654
0.6065	30.0034	88000	0.9086	0.4850	0.6497	0.6557	0.5433	0.6885	0.7346	0.6323	0.4404	0.5002	0.5009	0.4985
0.1385	31.3672	92000	0.9247	0.4688	0.6343	0.6469	0.4228	0.7420	0.7832	0.5892	0.3792	0.5132	0.4999	0.4829
0.4116	32.7310	96000	0.8724	0.5014	0.6628	0.6707	0.5288	0.6729	0.8213	0.6281	0.4585	0.5268	0.5094	0.5112
0.4991	34.0948	100000	0.8752	0.5078	0.6693	0.6766	0.5435	0.7342	0.7515	0.6480	0.4584	0.5274	0.5232	0.5225
0.5235	35.4586	104000	0.8312	0.5135	0.6736	0.6814	0.6179	0.7514	0.7598	0.5651	0.5060	0.5362	0.5256	0.4861
0.6378	36.8224	108000	0.8729	0.5102	0.6705	0.6784	0.5636	0.7161	0.7926	0.6097	0.4781	0.5335	0.5216	0.5076
0.6895	38.1862	112000	0.9258	0.4833	0.6466	0.6600	0.4375	0.7335	0.8392	0.5761	0.3990	0.5343	0.5097	0.4903
0.5259	39.5499	116000	0.8517	0.5269	0.6857	0.6922	0.5774	0.7340	0.7692	0.6625	0.4919	0.5434	0.5346	0.5380

Framework versions

• Transformers 4.48.3
• Pytorch 2.1.0
• Datasets 3.2.0
• Tokenizers 0.21.0