vit-base-cifar10

This model is a fine-tuned version of nateraw/vit-base-patch16-224-cifar10 on the cifar10-upside-down dataset. It achieves the following results on the evaluation set:

• eval_loss: 0.2348
• eval_accuracy: 0.9134
• eval_runtime: 157.4172
• eval_samples_per_second: 127.051
• eval_steps_per_second: 1.988
• epoch: 0.02
• step: 26

Model description

Vision Transformer

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0002
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Framework versions

• Transformers 4.18.0
• Pytorch 1.10.0+cu111
• Datasets 2.0.0
• Tokenizers 0.11.6