SW2-DMAE-2

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: 1.0510
  • Accuracy: 0.6739

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: 1.5e-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
  • num_epochs: 80

Training results

Training Loss Epoch Step Validation Loss Accuracy
No log 0.86 3 1.6269 0.1087
No log 2.0 7 1.6078 0.1087
1.618 2.86 10 1.5852 0.1087
1.618 4.0 14 1.5398 0.1087
1.618 4.86 17 1.4948 0.1087
1.5162 6.0 21 1.4344 0.1087
1.5162 6.86 24 1.3879 0.1087
1.5162 8.0 28 1.3288 0.1739
1.3459 8.86 31 1.2926 0.4565
1.3459 10.0 35 1.2562 0.4565
1.3459 10.86 38 1.2384 0.4565
1.2384 12.0 42 1.2205 0.4565
1.2384 12.86 45 1.2174 0.4565
1.2384 14.0 49 1.2131 0.4565
1.2049 14.86 52 1.2104 0.4565
1.2049 16.0 56 1.2086 0.4565
1.2049 16.86 59 1.2076 0.4565
1.1815 18.0 63 1.2052 0.4565
1.1815 18.86 66 1.2049 0.4565
1.1826 20.0 70 1.2019 0.4565
1.1826 20.86 73 1.1960 0.4565
1.1826 22.0 77 1.1927 0.4565
1.1647 22.86 80 1.1928 0.4565
1.1647 24.0 84 1.1924 0.4565
1.1647 24.86 87 1.1903 0.4565
1.1568 26.0 91 1.1879 0.4565
1.1568 26.86 94 1.1913 0.4565
1.1568 28.0 98 1.2046 0.4783
1.1432 28.86 101 1.1934 0.4783
1.1432 30.0 105 1.1665 0.4783
1.1432 30.86 108 1.1601 0.4783
1.1112 32.0 112 1.1624 0.5
1.1112 32.86 115 1.1664 0.5217
1.1112 34.0 119 1.1692 0.5
1.1132 34.86 122 1.1513 0.5435
1.1132 36.0 126 1.1384 0.5870
1.1132 36.86 129 1.1274 0.6087
1.0642 38.0 133 1.1443 0.5870
1.0642 38.86 136 1.1651 0.5
1.0439 40.0 140 1.1493 0.5
1.0439 40.86 143 1.1331 0.5217
1.0439 42.0 147 1.1032 0.5870
1.0362 42.86 150 1.0988 0.6304
1.0362 44.0 154 1.1093 0.5870
1.0362 44.86 157 1.1101 0.5870
1.0177 46.0 161 1.0903 0.6304
1.0177 46.86 164 1.0691 0.6522
1.0177 48.0 168 1.0510 0.6739
1.0 48.86 171 1.0451 0.6522
1.0 50.0 175 1.0425 0.6522
1.0 50.86 178 1.0512 0.6087
0.9636 52.0 182 1.0441 0.6304
0.9636 52.86 185 1.0402 0.6522
0.9636 54.0 189 1.0161 0.6522
0.9744 54.86 192 1.0073 0.6522
0.9744 56.0 196 1.0048 0.6522
0.9744 56.86 199 0.9993 0.6522
0.9233 58.0 203 0.9939 0.6522
0.9233 58.86 206 0.9939 0.6522
0.9452 60.0 210 0.9975 0.6522
0.9452 60.86 213 0.9981 0.6087
0.9452 62.0 217 0.9985 0.6087
0.9183 62.86 220 0.9969 0.6087
0.9183 64.0 224 0.9958 0.6304
0.9183 64.86 227 0.9928 0.6087
0.9449 66.0 231 0.9906 0.6087
0.9449 66.86 234 0.9893 0.6304
0.9449 68.0 238 0.9881 0.6522
0.9154 68.57 240 0.9880 0.6522

Framework versions

  • Transformers 4.36.2
  • Pytorch 2.1.2+cu118
  • Datasets 2.16.1
  • Tokenizers 0.15.0
Downloads last month
5
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/SW2-DMAE-2

Finetuned
(121)
this model

Evaluation results