RAVDESS_Speaker_Id_spec_dataset_vit_1

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

• Loss: 0.8945
• Accuracy: 0.6898

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
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	8	3.2704	0.0648
3.565	2.0	16	2.9777	0.1204
2.9674	3.0	24	2.4613	0.2963
2.2864	4.0	32	1.9951	0.4028
1.5685	5.0	40	1.6858	0.5
1.5685	6.0	48	1.4689	0.5509
1.0277	7.0	56	1.3074	0.6389
0.6389	8.0	64	1.2081	0.6713
0.3924	9.0	72	1.1233	0.6944
0.2291	10.0	80	1.0603	0.6991
0.2291	11.0	88	0.9899	0.7222
0.129	12.0	96	0.9880	0.6713
0.0812	13.0	104	0.9722	0.7037
0.0481	14.0	112	0.9296	0.7176
0.0325	15.0	120	0.9181	0.6852
0.0325	16.0	128	0.9013	0.7176
0.0212	17.0	136	0.9077	0.6944
0.0151	18.0	144	0.8954	0.7037
0.0121	19.0	152	0.8941	0.7037
0.0103	20.0	160	0.8945	0.6898

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu121
• Datasets 3.3.1
• Tokenizers 0.21.0