id_ravdess_mel_spec_Vit_vit-tiny-patch16-224_2

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: 0.8514
• Accuracy: 0.7870

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: 43
• 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	8	3.4190	0.0324
3.577	2.0	16	3.1245	0.0972
3.124	3.0	24	2.9168	0.1806
2.7828	4.0	32	2.5031	0.2454
2.1565	5.0	40	2.0137	0.3796
2.1565	6.0	48	1.6643	0.5463
1.4039	7.0	56	1.4283	0.5880
0.8467	8.0	64	1.2302	0.6620
0.471	9.0	72	1.1420	0.6574
0.245	10.0	80	1.0551	0.7130
0.245	11.0	88	1.0008	0.7361
0.1111	12.0	96	0.9660	0.75
0.0515	13.0	104	0.9158	0.7407
0.0227	14.0	112	0.9132	0.7407
0.0106	15.0	120	0.8355	0.7685
0.0106	16.0	128	0.8486	0.7639
0.0042	17.0	136	0.8263	0.7778
0.0021	18.0	144	0.8304	0.7731
0.0013	19.0	152	0.8260	0.7824
0.0009	20.0	160	0.8407	0.7731
0.0009	21.0	168	0.8337	0.7824
0.0008	22.0	176	0.8311	0.7824
0.0006	23.0	184	0.8370	0.7778
0.0006	24.0	192	0.8371	0.7778
0.0005	25.0	200	0.8373	0.7870
0.0005	26.0	208	0.8399	0.7870
0.0005	27.0	216	0.8394	0.7870
0.0005	28.0	224	0.8412	0.7824
0.0004	29.0	232	0.8416	0.7870
0.0004	30.0	240	0.8431	0.7870
0.0004	31.0	248	0.8450	0.7824
0.0004	32.0	256	0.8430	0.7870
0.0004	33.0	264	0.8458	0.7824
0.0003	34.0	272	0.8466	0.7870
0.0003	35.0	280	0.8454	0.7870
0.0003	36.0	288	0.8468	0.7824
0.0003	37.0	296	0.8484	0.7870
0.0003	38.0	304	0.8484	0.7870
0.0003	39.0	312	0.8492	0.7824
0.0003	40.0	320	0.8498	0.7870
0.0003	41.0	328	0.8492	0.7870
0.0003	42.0	336	0.8491	0.7870
0.0003	43.0	344	0.8505	0.7870
0.0003	44.0	352	0.8509	0.7870
0.0003	45.0	360	0.8505	0.7870
0.0003	46.0	368	0.8509	0.7870
0.0003	47.0	376	0.8510	0.7870
0.0003	48.0	384	0.8511	0.7870
0.0003	49.0	392	0.8514	0.7870
0.0003	50.0	400	0.8514	0.7870

Framework versions

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