Edit model card

swin-tiny-patch4-window7-224-large-dataset-varicropped

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: 1.3554
  • Accuracy: 0.6571

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: 32
  • eval_batch_size: 32
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 30

Training results

Training Loss Epoch Step Validation Loss Accuracy
0.8523 0.99 88 1.8136 0.3771
0.0725 1.99 176 1.2359 0.6006
0.0397 2.99 264 1.1716 0.6014
0.0179 3.99 352 1.5688 0.5704
0.0173 4.99 440 1.3718 0.6237
0.0097 5.99 528 1.3841 0.5927
0.0109 6.99 616 1.4044 0.5895
0.0019 7.99 704 1.2936 0.6150
0.002 8.99 792 1.4264 0.5760
0.0035 9.99 880 1.2226 0.6396
0.0025 10.99 968 1.1553 0.6635
0.0009 11.99 1056 1.1727 0.6643
0.0037 12.99 1144 1.1182 0.6714
0.0017 13.99 1232 1.4015 0.6364
0.0009 14.99 1320 1.2955 0.6683
0.0002 15.99 1408 1.2310 0.6555
0.0007 16.99 1496 1.3849 0.6325
0.001 17.99 1584 1.4312 0.6102
0.0001 18.99 1672 1.5087 0.6181
0.0002 19.99 1760 1.7247 0.6062
0.0016 20.99 1848 1.5534 0.6237
0.0004 21.99 1936 1.5382 0.6333
0.0008 22.99 2024 1.4910 0.6484
0.0008 23.99 2112 1.5020 0.6380
0.0005 24.99 2200 1.4788 0.6468
0.001 25.99 2288 1.3416 0.6770
0.003 26.99 2376 1.2643 0.6738
0.0001 27.99 2464 1.3582 0.6595
0.0 28.99 2552 1.3767 0.6523
0.0 29.99 2640 1.3554 0.6571

Framework versions

  • Transformers 4.21.1
  • Pytorch 1.12.1
  • Datasets 2.4.0
  • Tokenizers 0.12.1
Downloads last month
9
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Evaluation results