vit-weldclassifyv4

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: 0.5265
  • Accuracy: 0.8094

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: 0.0002
  • train_batch_size: 16
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 13
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy
1.1126 0.6410 100 1.0171 0.5504
0.8229 1.2821 200 0.7307 0.6942
0.7224 1.9231 300 0.6399 0.7122
0.3909 2.5641 400 0.5400 0.7734
0.237 3.2051 500 0.6716 0.7626
0.4056 3.8462 600 0.5265 0.8094
0.1764 4.4872 700 0.9174 0.7446
0.0546 5.1282 800 0.6644 0.8237
0.0436 5.7692 900 0.6923 0.8345
0.0661 6.4103 1000 0.6784 0.8345
0.0167 7.0513 1100 0.7115 0.8309
0.0744 7.6923 1200 0.6341 0.8525
0.0047 8.3333 1300 0.6402 0.8597
0.0039 8.9744 1400 0.5958 0.8849
0.0029 9.6154 1500 0.6158 0.8885
0.0027 10.2564 1600 0.6189 0.8885
0.0025 10.8974 1700 0.6309 0.8885
0.0024 11.5385 1800 0.6356 0.8885
0.0023 12.1795 1900 0.6382 0.8885
0.0023 12.8205 2000 0.6399 0.8885

Framework versions

  • Transformers 4.41.2
  • Pytorch 2.3.0+cu121
  • Datasets 2.20.0
  • Tokenizers 0.19.1
Downloads last month
13
Safetensors
Model size
85.8M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.

Model tree for th041/vit-weldclassifyv4

Finetuned
(1967)
this model

Evaluation results