emotion_classification

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: 1.2453
• Accuracy: 0.5938

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	20	1.9465	0.325
No log	2.0	40	1.7314	0.4375
No log	3.0	60	1.5249	0.5375
No log	4.0	80	1.4166	0.4875
No log	5.0	100	1.3605	0.55
No log	6.0	120	1.3204	0.5563
No log	7.0	140	1.2074	0.6
No log	8.0	160	1.2138	0.6
No log	9.0	180	1.2600	0.5625
No log	10.0	200	1.2103	0.5563
No log	11.0	220	1.1736	0.5687
No log	12.0	240	1.2462	0.5687
No log	13.0	260	1.2009	0.5813
No log	14.0	280	1.2105	0.5437
No log	15.0	300	1.2705	0.5125
No log	16.0	320	1.2135	0.5938
No log	17.0	340	1.2089	0.5563
No log	18.0	360	1.2818	0.5375
No log	19.0	380	1.3076	0.5062
No log	20.0	400	1.2479	0.55
No log	21.0	420	1.2218	0.55
No log	22.0	440	1.0957	0.6188
No log	23.0	460	1.2437	0.5875
No log	24.0	480	1.3598	0.5125
0.8126	25.0	500	1.2759	0.55
0.8126	26.0	520	1.1474	0.6
0.8126	27.0	540	1.1115	0.6375
0.8126	28.0	560	1.1715	0.5687
0.8126	29.0	580	1.3133	0.5625
0.8126	30.0	600	1.2526	0.5437

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0
• Datasets 2.19.1
• Tokenizers 0.19.1