segformer-b0-finetuned-oldapp-dec-4

This model is a fine-tuned version of nvidia/mit-b0 on the PushkarA07/oldapptiles5 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0403
• Mean Iou: 0.4995
• Mean Accuracy: 0.5
• Overall Accuracy: 0.9990
• Accuracy Abnormality: 0.0
• Iou Abnormality: 0.0

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Use 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Abnormality	Iou Abnormality
0.2898	0.7143	10	0.2757	0.4992	0.4997	0.9984	0.0	0.0
0.2605	1.4286	20	0.2029	0.4995	0.5	0.9990	0.0	0.0
0.2573	2.1429	30	0.1829	0.4995	0.5	0.9990	0.0	0.0
0.4388	2.8571	40	0.1301	0.4995	0.5	0.9990	0.0	0.0
0.1893	3.5714	50	0.1529	0.4995	0.5	0.9990	0.0	0.0
0.171	4.2857	60	0.1322	0.4995	0.5	0.9990	0.0	0.0
0.1502	5.0	70	0.1239	0.4995	0.5	0.9990	0.0	0.0
0.1267	5.7143	80	0.1285	0.4995	0.5	0.9990	0.0	0.0
0.1421	6.4286	90	0.0891	0.4995	0.5	0.9990	0.0	0.0
0.1271	7.1429	100	0.0684	0.4995	0.5	0.9990	0.0	0.0
0.1206	7.8571	110	0.0718	0.4995	0.5	0.9990	0.0	0.0
0.0802	8.5714	120	0.0802	0.4995	0.5	0.9990	0.0	0.0
0.0765	9.2857	130	0.0591	0.4995	0.5	0.9990	0.0	0.0
0.1025	10.0	140	0.0541	0.4995	0.5	0.9990	0.0	0.0
0.0711	10.7143	150	0.0572	0.4995	0.5	0.9990	0.0	0.0
0.0621	11.4286	160	0.0488	0.4995	0.5	0.9990	0.0	0.0
0.0611	12.1429	170	0.0446	0.4995	0.5	0.9990	0.0	0.0
0.0665	12.8571	180	0.0383	0.4995	0.5	0.9990	0.0	0.0
0.0625	13.5714	190	0.0492	0.4995	0.5	0.9990	0.0	0.0
0.0473	14.2857	200	0.0454	0.4995	0.5	0.9990	0.0	0.0
0.0379	15.0	210	0.0384	0.4995	0.5	0.9990	0.0	0.0
0.0597	15.7143	220	0.0423	0.4995	0.5	0.9990	0.0	0.0
0.0436	16.4286	230	0.0474	0.4995	0.5	0.9990	0.0	0.0
0.0591	17.1429	240	0.0424	0.4995	0.5	0.9990	0.0	0.0
0.0337	17.8571	250	0.0359	0.4995	0.5	0.9990	0.0	0.0
0.0455	18.5714	260	0.0394	0.4995	0.5	0.9990	0.0	0.0
0.0322	19.2857	270	0.0321	0.4995	0.5	0.9990	0.0	0.0
0.0339	20.0	280	0.0440	0.4995	0.5	0.9990	0.0	0.0
0.0336	20.7143	290	0.0459	0.4995	0.5	0.9990	0.0	0.0
0.0357	21.4286	300	0.0324	0.4995	0.5	0.9990	0.0	0.0
0.0286	22.1429	310	0.0264	0.4995	0.5	0.9990	0.0	0.0
0.0301	22.8571	320	0.0166	0.4995	0.5	0.9990	0.0	0.0
0.0252	23.5714	330	0.0229	0.4995	0.5	0.9990	0.0	0.0
0.0296	24.2857	340	0.0298	0.4995	0.5	0.9990	0.0	0.0
0.033	25.0	350	0.0389	0.4995	0.5	0.9990	0.0	0.0
0.0279	25.7143	360	0.0346	0.4995	0.5	0.9990	0.0	0.0
0.0266	26.4286	370	0.0425	0.4995	0.5	0.9990	0.0	0.0
0.0278	27.1429	380	0.0395	0.4995	0.5	0.9990	0.0	0.0
0.0458	27.8571	390	0.0228	0.4995	0.5	0.9990	0.0	0.0
0.0289	28.5714	400	0.0183	0.4995	0.5	0.9990	0.0	0.0
0.0261	29.2857	410	0.0300	0.4995	0.5	0.9990	0.0	0.0
0.031	30.0	420	0.0295	0.4995	0.5	0.9990	0.0	0.0
0.0342	30.7143	430	0.0457	0.4995	0.5	0.9990	0.0	0.0
0.0271	31.4286	440	0.0349	0.4995	0.5	0.9990	0.0	0.0
0.0198	32.1429	450	0.0407	0.4995	0.5	0.9990	0.0	0.0
0.0231	32.8571	460	0.0443	0.4995	0.5	0.9990	0.0	0.0
0.0392	33.5714	470	0.0398	0.4995	0.5	0.9990	0.0	0.0
0.0166	34.2857	480	0.0206	0.4995	0.5	0.9990	0.0	0.0
0.0215	35.0	490	0.0302	0.4995	0.5	0.9990	0.0	0.0
0.0983	35.7143	500	0.0117	0.4995	0.5	0.9990	0.0	0.0
0.0162	36.4286	510	0.0237	0.4995	0.5	0.9990	0.0	0.0
0.0211	37.1429	520	0.0395	0.4995	0.5	0.9990	0.0	0.0
0.0228	37.8571	530	0.0526	0.4995	0.5	0.9990	0.0	0.0
0.0198	38.5714	540	0.0480	0.4995	0.5	0.9990	0.0	0.0
0.0192	39.2857	550	0.0479	0.4995	0.5	0.9990	0.0	0.0
0.0213	40.0	560	0.0390	0.4995	0.5	0.9990	0.0	0.0
0.0185	40.7143	570	0.0456	0.4995	0.5	0.9990	0.0	0.0
0.0179	41.4286	580	0.0337	0.4995	0.5	0.9990	0.0	0.0
0.0307	42.1429	590	0.0460	0.4995	0.5	0.9990	0.0	0.0
0.0199	42.8571	600	0.0443	0.4995	0.5	0.9990	0.0	0.0
0.0121	43.5714	610	0.0313	0.4995	0.5	0.9990	0.0	0.0
0.0184	44.2857	620	0.0435	0.4995	0.5	0.9990	0.0	0.0
0.0243	45.0	630	0.0440	0.4995	0.5	0.9990	0.0	0.0
0.02	45.7143	640	0.0457	0.4995	0.5	0.9990	0.0	0.0
0.0144	46.4286	650	0.0440	0.4995	0.5	0.9990	0.0	0.0
0.013	47.1429	660	0.0295	0.4995	0.5	0.9990	0.0	0.0
0.0156	47.8571	670	0.0429	0.4995	0.5	0.9990	0.0	0.0
0.0201	48.5714	680	0.0453	0.4995	0.5	0.9990	0.0	0.0
0.018	49.2857	690	0.0384	0.4995	0.5	0.9990	0.0	0.0
0.0118	50.0	700	0.0403	0.4995	0.5	0.9990	0.0	0.0

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3