imagenet2012-1k-subsampling-50-vit-base-patch16-224-in21k

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the imagenet2012-1k-subsampling-50 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8563
• Accuracy: 0.8109

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: 8
• eval_batch_size: 8
• seed: 1337
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
5.7852	1.0	5313	5.7565	0.6867
4.4299	2.0	10626	4.2553	0.7315
2.7934	3.0	15939	2.7094	0.7547
1.5784	4.0	21252	1.6554	0.7728
0.7426	5.0	26565	1.1836	0.7896
0.8495	6.0	31878	0.9912	0.8013
0.575	7.0	37191	0.9112	0.8041
0.7981	8.0	42504	0.8853	0.8052
0.7448	9.0	47817	0.8613	0.8113
0.3953	10.0	53130	0.8563	0.8109

Framework versions

• Transformers 4.38.0
• Pytorch 2.1.2+cu118
• Datasets 2.19.1
• Tokenizers 0.15.2