Edit model card

turcoins-classifier

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagefolder dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1763
  • Accuracy: 0.9549

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: 30

Training results

Training Loss Epoch Step Validation Loss Accuracy
1.9277 1.0 146 1.9660 0.7726
1.6627 2.0 292 1.7154 0.7917
1.4071 2.99 438 1.4120 0.8079
1.09 4.0 585 1.1225 0.8362
0.8086 5.0 731 0.8917 0.8675
0.7636 6.0 877 0.7596 0.8709
0.611 6.99 1023 0.6493 0.8883
0.4605 8.0 1170 0.5899 0.8872
0.37 9.0 1316 0.4978 0.9045
0.3882 10.0 1462 0.4424 0.9132
0.3139 10.99 1608 0.3969 0.9115
0.3178 12.0 1755 0.3525 0.9294
0.2796 13.0 1901 0.3552 0.9161
0.2571 14.0 2047 0.3189 0.9265
0.2481 14.99 2193 0.2945 0.9358
0.1875 16.0 2340 0.2647 0.9392
0.1861 17.0 2486 0.2404 0.9410
0.1839 18.0 2632 0.2556 0.9421
0.173 18.99 2778 0.2387 0.9462
0.1837 20.0 2925 0.2049 0.9485
0.1724 21.0 3071 0.2065 0.9525
0.1399 22.0 3217 0.2089 0.9404
0.1696 22.99 3363 0.1957 0.9497
0.1405 24.0 3510 0.1848 0.9554
0.1009 25.0 3656 0.1912 0.9520
0.1126 26.0 3802 0.1717 0.9560
0.1336 26.99 3948 0.1699 0.9589
0.1046 28.0 4095 0.1600 0.9601
0.126 29.0 4241 0.1839 0.9520
0.0882 29.95 4380 0.1763 0.9549

Framework versions

  • Transformers 4.28.1
  • Pytorch 2.0.0+cu117
  • Datasets 2.12.0
  • Tokenizers 0.13.3
Downloads last month
4
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