Edit model card

beit-base-patch16-224-85-fold3

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

  • Loss: 0.3297
  • Accuracy: 0.9091

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: 100

Training results

Training Loss Epoch Step Validation Loss Accuracy
No log 1.0 2 0.6425 0.6136
No log 2.0 4 0.6036 0.7273
No log 3.0 6 0.5549 0.7045
No log 4.0 8 0.4879 0.7273
0.5424 5.0 10 0.7873 0.7045
0.5424 6.0 12 0.6817 0.7045
0.5424 7.0 14 0.4846 0.75
0.5424 8.0 16 0.5266 0.7273
0.5424 9.0 18 0.4487 0.7727
0.5207 10.0 20 0.3768 0.8182
0.5207 11.0 22 0.6762 0.7045
0.5207 12.0 24 0.3988 0.8409
0.5207 13.0 26 0.3812 0.8864
0.5207 14.0 28 0.4207 0.7727
0.386 15.0 30 0.3801 0.8182
0.386 16.0 32 0.5402 0.7045
0.386 17.0 34 0.3865 0.8636
0.386 18.0 36 0.3635 0.8182
0.386 19.0 38 0.5374 0.7273
0.3232 20.0 40 0.5088 0.75
0.3232 21.0 42 0.3507 0.8182
0.3232 22.0 44 0.2995 0.8409
0.3232 23.0 46 0.3403 0.8409
0.3232 24.0 48 0.3858 0.8182
0.249 25.0 50 0.4126 0.7727
0.249 26.0 52 0.4907 0.8182
0.249 27.0 54 0.3799 0.8182
0.249 28.0 56 0.3528 0.8182
0.249 29.0 58 0.3775 0.8182
0.2064 30.0 60 0.3520 0.8182
0.2064 31.0 62 0.4397 0.7727
0.2064 32.0 64 0.4284 0.75
0.2064 33.0 66 0.3833 0.8409
0.2064 34.0 68 0.3558 0.8409
0.2066 35.0 70 0.4880 0.8182
0.2066 36.0 72 0.3739 0.8182
0.2066 37.0 74 0.5409 0.7727
0.2066 38.0 76 0.4869 0.8182
0.2066 39.0 78 0.8398 0.75
0.1776 40.0 80 0.5410 0.7955
0.1776 41.0 82 0.4740 0.8409
0.1776 42.0 84 0.3428 0.8636
0.1776 43.0 86 0.3135 0.8864
0.1776 44.0 88 0.3297 0.9091
0.1732 45.0 90 0.3982 0.9091
0.1732 46.0 92 0.5961 0.7955
0.1732 47.0 94 0.4798 0.8864
0.1732 48.0 96 0.5187 0.8182
0.1732 49.0 98 0.4430 0.9091
0.1372 50.0 100 0.4522 0.9091
0.1372 51.0 102 0.5617 0.8409
0.1372 52.0 104 0.6568 0.8182
0.1372 53.0 106 0.8141 0.7955
0.1372 54.0 108 0.6189 0.8409
0.1305 55.0 110 0.5124 0.8636
0.1305 56.0 112 0.5095 0.8636
0.1305 57.0 114 0.4101 0.8864
0.1305 58.0 116 0.7712 0.7727
0.1305 59.0 118 0.5073 0.7955
0.1423 60.0 120 0.3890 0.8636
0.1423 61.0 122 0.5701 0.8182
0.1423 62.0 124 0.5482 0.8409
0.1423 63.0 126 0.5508 0.8409
0.1423 64.0 128 0.6589 0.7955
0.13 65.0 130 0.7184 0.75
0.13 66.0 132 0.4702 0.8864
0.13 67.0 134 0.4339 0.9091
0.13 68.0 136 0.4463 0.9091
0.13 69.0 138 0.4887 0.8864
0.1232 70.0 140 0.5121 0.8636
0.1232 71.0 142 0.4944 0.9091
0.1232 72.0 144 0.5208 0.8864
0.1232 73.0 146 0.6074 0.8182
0.1232 74.0 148 0.8013 0.75
0.1241 75.0 150 0.7022 0.7727
0.1241 76.0 152 0.5641 0.8864
0.1241 77.0 154 0.6550 0.8409
0.1241 78.0 156 0.6268 0.8409
0.1241 79.0 158 0.5466 0.8864
0.1254 80.0 160 0.5453 0.8864
0.1254 81.0 162 0.5663 0.8636
0.1254 82.0 164 0.5377 0.8409
0.1254 83.0 166 0.5381 0.8864
0.1254 84.0 168 0.5459 0.8636
0.1061 85.0 170 0.5490 0.8636
0.1061 86.0 172 0.5444 0.8636
0.1061 87.0 174 0.5344 0.8636
0.1061 88.0 176 0.5251 0.8636
0.1061 89.0 178 0.5178 0.8864
0.1084 90.0 180 0.5161 0.8864
0.1084 91.0 182 0.5184 0.8864
0.1084 92.0 184 0.5185 0.8864
0.1084 93.0 186 0.5300 0.8636
0.1084 94.0 188 0.5599 0.8636
0.1025 95.0 190 0.5972 0.8182
0.1025 96.0 192 0.6083 0.8182
0.1025 97.0 194 0.5969 0.8409
0.1025 98.0 196 0.5769 0.8636
0.1025 99.0 198 0.5673 0.8636
0.1184 100.0 200 0.5642 0.8636

Framework versions

  • Transformers 4.40.2
  • Pytorch 2.2.1+cu121
  • Datasets 2.19.1
  • Tokenizers 0.19.1
Downloads last month
2
Safetensors
Model size
85.8M params
Tensor type
F32
·
Inference API
Drag image file here or click to browse from your device
This model can be loaded on Inference API (serverless).

Finetuned from

Evaluation results