training_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: 1.2897
• Accuracy: 0.5312

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: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	80	2.0647	0.1875
No log	2.0	160	1.9337	0.3688
No log	3.0	240	1.7220	0.4437
No log	4.0	320	1.5526	0.4938
No log	5.0	400	1.4479	0.5125
No log	6.0	480	1.4023	0.4938
1.4964	7.0	560	1.3220	0.5375
1.4964	8.0	640	1.2965	0.525
1.4964	9.0	720	1.2884	0.525
1.4964	10.0	800	1.2782	0.5437
1.4964	11.0	880	1.3028	0.5062
1.4964	12.0	960	1.2909	0.5375
0.3712	13.0	1040	1.2888	0.5375
0.3712	14.0	1120	1.2907	0.5312
0.3712	15.0	1200	1.2897	0.5312

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0