This model is a fine-tuned version of google/vit-hybrid-base-bit-384 on the imagefolder dataset.
It achieves the following results on the evaluation set:
eval_loss: 0.2838
eval_accuracy: 0.9375
eval_balanced_accuracy: 0.9167
eval_soft_accuracy_1: 1.0
eval_soft_accuracy_2: 1.0
eval_precision_macro: 0.95
eval_recall_macro: 0.9167
eval_f1_macro: 0.9222
eval_precision_weighted: 0.95
eval_recall_weighted: 0.9375
eval_f1_weighted: 0.9347
eval_mcc: 0.9185
eval_cross_entropy: 0.2363
eval_roc_auc_ovr_macro: 0.9808
eval_roc_auc_ovr_weighted: 0.9856
eval_runtime: 0.7636
eval_samples_per_second: 20.953
eval_steps_per_second: 1.31
epoch: 20.7143
step: 290
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: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
