segformer-b0-finetuned-oldapp-oct-1

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: 0.2630
• Mean Iou: 0.4984
• Mean Accuracy: 0.4989
• Overall Accuracy: 0.9967
• 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: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Mean Iou	Mean Accuracy	Overall Accuracy	Accuracy Abnormality	Iou Abnormality
0.6912	0.7143	10	0.6575	0.4714	0.4784	0.9426	0.0133	0.0002
0.5802	1.4286	20	0.5878	0.4942	0.4947	0.9884	0.0	0.0
0.5498	2.1429	30	0.4770	0.4984	0.4989	0.9968	0.0	0.0
0.6084	2.8571	40	0.4125	0.4971	0.4976	0.9941	0.0	0.0
0.4675	3.5714	50	0.4355	0.4885	0.4992	0.9761	0.0213	0.0009
0.3863	4.2857	60	0.3699	0.4965	0.5005	0.9920	0.0081	0.0010
0.3954	5.0	70	0.3401	0.4983	0.4989	0.9967	0.0	0.0
0.3286	5.7143	80	0.3279	0.4983	0.4988	0.9967	0.0	0.0
0.3458	6.4286	90	0.2908	0.4974	0.4979	0.9948	0.0	0.0
0.3559	7.1429	100	0.2693	0.4989	0.4994	0.9978	0.0	0.0
0.3196	7.8571	110	0.2596	0.4977	0.4982	0.9954	0.0	0.0
0.3109	8.5714	120	0.2915	0.4958	0.4963	0.9916	0.0	0.0
0.2711	9.2857	130	0.2720	0.4991	0.4997	0.9983	0.0	0.0
0.3051	10.0	140	0.2630	0.4984	0.4989	0.9967	0.0	0.0

Framework versions

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