vit-emotion

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

• Loss: 1.1858
• Accuracy: 0.6188

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.8403	1.0	40	1.7317	0.3063
1.4783	2.0	80	1.5047	0.4938
1.1866	3.0	120	1.3522	0.55
0.8581	4.0	160	1.2084	0.575
0.6056	5.0	200	1.2348	0.5375
0.3745	6.0	240	1.2119	0.5625
0.2129	7.0	280	1.2012	0.5437
0.1547	8.0	320	1.2181	0.5875
0.1216	9.0	360	1.2196	0.5875
0.1023	10.0	400	1.1858	0.6188
0.102	11.0	440	1.2190	0.5938
0.083	12.0	480	1.2149	0.6125
0.0917	13.0	520	1.2600	0.5875
0.0807	14.0	560	1.2367	0.6062
0.0741	15.0	600	1.2382	0.6
0.0721	16.0	640	1.2464	0.5875
0.0678	17.0	680	1.2548	0.5938
0.0752	18.0	720	1.2591	0.5875
0.0657	19.0	760	1.2590	0.6062
0.0643	20.0	800	1.2589	0.5938

Framework versions

• Transformers 4.42.4
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1