segformer-b0-finetuned-net-15Oct

This model is a fine-tuned version of PushkarA07/segformer-b0-finetuned-net-4Sep on the PushkarA07/batch2-tiles_W5 dataset. It achieves the following results on the evaluation set:

• Loss: 0.0059
• Mean Iou: 0.8668
• Mean Accuracy: 0.9041
• Overall Accuracy: 0.9978
• Accuracy Abnormality: 0.8090
• Iou Abnormality: 0.7358

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: 1e-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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
0.0055	0.5556	10	0.0059	0.8661	0.9020	0.9978	0.8047	0.7345
0.0073	1.1111	20	0.0059	0.8664	0.9030	0.9978	0.8067	0.7350
0.0023	1.6667	30	0.0059	0.8668	0.9044	0.9978	0.8097	0.7359
0.0045	2.2222	40	0.0059	0.8668	0.9044	0.9978	0.8096	0.7359
0.0026	2.7778	50	0.0059	0.8671	0.9054	0.9978	0.8115	0.7365
0.0059	3.3333	60	0.0059	0.8668	0.9044	0.9978	0.8095	0.7359
0.0037	3.8889	70	0.0059	0.8667	0.9035	0.9978	0.8078	0.7356
0.0052	4.4444	80	0.0059	0.8667	0.9038	0.9978	0.8084	0.7357
0.007	5.0	90	0.0059	0.8674	0.9065	0.9978	0.8139	0.7370
0.0064	5.5556	100	0.0059	0.8669	0.9046	0.9978	0.8100	0.7361
0.008	6.1111	110	0.0059	0.8665	0.9031	0.9978	0.8070	0.7352
0.008	6.6667	120	0.0059	0.8668	0.9043	0.9978	0.8094	0.7359
0.0046	7.2222	130	0.0059	0.8670	0.9051	0.9978	0.8110	0.7363
0.0025	7.7778	140	0.0059	0.8666	0.9032	0.9978	0.8072	0.7354
0.0077	8.3333	150	0.0059	0.8661	0.9021	0.9978	0.8049	0.7344
0.0041	8.8889	160	0.0059	0.8666	0.9036	0.9978	0.8080	0.7355
0.0034	9.4444	170	0.0059	0.8666	0.9035	0.9978	0.8078	0.7354
0.0042	10.0	180	0.0059	0.8668	0.9041	0.9978	0.8090	0.7358

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu126
• Datasets 4.2.0
• Tokenizers 0.22.1