swinv2-tiny-patch4-window8-256-DMAE-da-4e-5

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on the imagefolder dataset. It achieves the following results on the evaluation set:

  • Loss: 0.9664
  • Accuracy: 0.7609

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: 4e-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: 40

Training results

Training Loss Epoch Step Validation Loss Accuracy
1.446 0.96 11 1.6275 0.1087
1.4497 2.0 23 1.5550 0.1087
1.388 2.96 34 1.3769 0.3261
1.2755 4.0 46 1.2483 0.4130
1.1574 4.96 57 1.1545 0.4565
1.0826 6.0 69 1.0429 0.5
0.9124 6.96 80 0.9318 0.5652
0.8228 8.0 92 1.0362 0.5217
0.733 8.96 103 0.9699 0.5870
0.7086 10.0 115 0.8269 0.6522
0.6459 10.96 126 0.8168 0.6739
0.5793 12.0 138 1.0780 0.6087
0.5904 12.96 149 1.0166 0.5870
0.5155 14.0 161 0.8489 0.6304
0.4693 14.96 172 0.8454 0.6522
0.4928 16.0 184 0.8161 0.6739
0.4763 16.96 195 0.7666 0.7174
0.4354 18.0 207 0.8828 0.6957
0.3661 18.96 218 0.8782 0.6739
0.3652 20.0 230 0.9418 0.6739
0.3733 20.96 241 0.8963 0.7174
0.3473 22.0 253 0.9053 0.7174
0.2988 22.96 264 0.8318 0.7391
0.349 24.0 276 1.1129 0.6087
0.2963 24.96 287 1.0557 0.6304
0.3025 26.0 299 0.9567 0.7391
0.2676 26.96 310 1.0131 0.6739
0.2848 28.0 322 0.9576 0.6957
0.2757 28.96 333 0.9821 0.7174
0.2564 30.0 345 1.0166 0.6522
0.2635 30.96 356 0.9664 0.7609
0.2413 32.0 368 0.9894 0.7391
0.2321 32.96 379 1.0272 0.7391
0.2517 34.0 391 1.0312 0.7174
0.2161 34.96 402 1.0433 0.7174
0.2304 36.0 414 1.0158 0.7174
0.2194 36.96 425 1.0120 0.6957
0.2395 38.0 437 1.0153 0.6957
0.2199 38.26 440 1.0149 0.6957

Framework versions

  • Transformers 4.36.2
  • Pytorch 2.1.2+cu118
  • Datasets 2.16.1
  • Tokenizers 0.15.0
Downloads last month
3
Safetensors
Model size
27.6M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API: The model has no library tag.

Model tree for Augusto777/swinv2-tiny-patch4-window8-256-DMAE-da-4e-5

Finetuned
(120)
this model

Evaluation results