vit-base-patch16-224-in21k-finetuned-cifar10_album_vitVMMRdb_make_model_album_pred

This model is a fine-tuned version of aaraki/vit-base-patch16-224-in21k-finetuned-cifar10 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5462
• Accuracy: 0.8594
• Precision: 0.8556
• Recall: 0.8594
• F1: 0.8544

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: 5e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
4.6112	1.0	839	4.5615	0.1425	0.0837	0.1425	0.0646
3.1177	2.0	1678	2.9595	0.4240	0.3424	0.4240	0.3283
2.0793	3.0	2517	2.0048	0.5771	0.5081	0.5771	0.5029
1.4566	4.0	3356	1.4554	0.6760	0.6333	0.6760	0.6280
1.1307	5.0	4195	1.1319	0.7350	0.7027	0.7350	0.7013
0.9367	6.0	5034	0.9328	0.7738	0.7546	0.7738	0.7503
0.7783	7.0	5873	0.8024	0.7986	0.7893	0.7986	0.7819
0.6022	8.0	6712	0.7187	0.8174	0.8098	0.8174	0.8055
0.5234	9.0	7551	0.6635	0.8313	0.8220	0.8313	0.8217
0.4298	10.0	8390	0.6182	0.8388	0.8337	0.8388	0.8302
0.3618	11.0	9229	0.5953	0.8455	0.8394	0.8455	0.8382
0.3262	12.0	10068	0.5735	0.8501	0.8443	0.8501	0.8436
0.3116	13.0	10907	0.5612	0.8527	0.8488	0.8527	0.8471
0.2416	14.0	11746	0.5524	0.8558	0.8500	0.8558	0.8496
0.2306	15.0	12585	0.5489	0.8572	0.8525	0.8572	0.8519

Framework versions

• Transformers 4.24.0
• Pytorch 1.12.1+cu113
• Datasets 2.7.1
• Tokenizers 0.13.2