segformer_Clean_Set1_95images

This model is a fine-tuned version of nvidia/mit-b4 on the Hasano20/Clean_Set1_95images dataset. It achieves the following results on the evaluation set:

• Loss: 0.0223
• Mean Iou: 0.6447
• Mean Accuracy: 0.9824
• Overall Accuracy: 0.9886
• Accuracy Background: nan
• Accuracy Melt: 0.9724
• Accuracy Substrate: 0.9923
• Iou Background: 0.0
• Iou Melt: 0.9458
• Iou Substrate: 0.9882

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: 0.0001
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Background	Accuracy Melt	Accuracy Substrate	Iou Background	Iou Melt	Iou Substrate
0.2051	1.1765	20	0.3764	0.3339	0.5766	0.8354	nan	0.1639	0.9894	0.0	0.1612	0.8404
0.3486	2.3529	40	0.1932	0.4595	0.7687	0.8745	nan	0.6000	0.9375	0.0	0.4928	0.8858
0.0831	3.5294	60	0.2016	0.4101	0.6782	0.8792	nan	0.3576	0.9988	0.0	0.3570	0.8732
0.0809	4.7059	80	0.0763	0.5787	0.9243	0.9507	nan	0.8822	0.9664	0.0	0.7830	0.9531
0.0325	5.8824	100	0.0694	0.6028	0.9436	0.9618	nan	0.9146	0.9727	0.0	0.8479	0.9606
0.0279	7.0588	120	0.0460	0.6142	0.9520	0.9712	nan	0.9213	0.9826	0.0	0.8739	0.9686
0.0493	8.2353	140	0.0353	0.6297	0.9648	0.9802	nan	0.9404	0.9893	0.0	0.9092	0.9797
0.0286	9.4118	160	0.0366	0.6261	0.9643	0.9765	nan	0.9449	0.9837	0.0	0.8997	0.9787
0.0463	10.5882	180	0.0258	0.6425	0.9798	0.9879	nan	0.9669	0.9927	0.0	0.9414	0.9862
0.0145	11.7647	200	0.0302	0.6324	0.9652	0.9821	nan	0.9382	0.9922	0.0	0.9162	0.9810
0.0221	12.9412	220	0.0262	0.6379	0.9733	0.9850	nan	0.9547	0.9919	0.0	0.9289	0.9848
0.0109	14.1176	240	0.0236	0.6417	0.9764	0.9869	nan	0.9595	0.9932	0.0	0.9379	0.9871
0.0122	15.2941	260	0.0252	0.6407	0.9812	0.9866	nan	0.9725	0.9898	0.0	0.9358	0.9864
0.0101	16.4706	280	0.0239	0.6417	0.9799	0.9869	nan	0.9686	0.9911	0.0	0.9382	0.9870
0.0113	17.6471	300	0.0231	0.6425	0.9798	0.9874	nan	0.9675	0.9920	0.0	0.9399	0.9875
0.0086	18.8235	320	0.0225	0.6444	0.9826	0.9885	nan	0.9733	0.9919	0.0	0.9451	0.9882
0.0086	20.0	340	0.0223	0.6447	0.9824	0.9886	nan	0.9724	0.9923	0.0	0.9458	0.9882

Framework versions

• Transformers 4.41.2
• Pytorch 2.0.1+cu117
• Datasets 2.19.2
• Tokenizers 0.19.1