swin-tiny-patch4-window7-224-crack-detector

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.1807
• Accuracy: {'accuracy': 0.9384615384615385}
• F1: {'f1': 0.9382975252490704}
• Precision: {'precision': 0.9382005688460371}
• Recall: {'recall': 0.9395073274524703}

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.3863	1.0	195	0.3349	{'accuracy': 0.885576923076923}	{'f1': 0.8829318618369404}	{'precision': 0.8830357915066687}	{'recall': 0.8864842943431257}
0.2685	2.0	390	0.2715	{'accuracy': 0.9080128205128205}	{'f1': 0.9106277459775055}	{'precision': 0.9130231253775549}	{'recall': 0.9148104520472664}
0.2235	3.0	585	0.1807	{'accuracy': 0.9384615384615385}	{'f1': 0.9382975252490704}	{'precision': 0.9382005688460371}	{'recall': 0.9395073274524703}

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0
• Datasets 2.17.1
• Tokenizers 0.15.2