vit-base-oxford-iiit-pets

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

vit-base-oxford-iiit-pets

This model is a fine-tuned version of google/vit-base-patch16-224 on the pcuenq/oxford-pets dataset. It achieves the following results on the evaluation set:

• Loss: 0.2111
• Accuracy: 0.9499

Model description

Based on the ViT model google/vit-base-patch16-224.

Performance on Test Set

• eval_loss: 0.21107521653175354
• eval_accuracy: 0.9519323410013532
• eval_runtime: 12.1289
• eval_samples_per_second: 73.032
• eval_steps_per_second: 9.191
• epoch: 6.0

Compared to Zero Shot

• Accuracy: 0.8800
• Precision: 0.8768
• Recall: 0.8800
• checkpoint = "openai/clip-vit-large-patch14"

Training and evaluation data

• per_device_train_batch_size=16,
• evaluation_strategy="epoch",
• save_strategy="epoch",
• logging_steps=100,
• num_train_epochs=6,
• learning_rate=3e-4,
• save_total_limit=2,
• remove_unused_columns=False,
• push_to_hub=True,
• report_to='tensorboard',
• load_best_model_at_end=True,

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0003
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Framework versions

• Transformers 4.50.0
• Pytorch 2.6.0+cu124
• Datasets 3.4.1
• Tokenizers 0.21.1