id_ravdess_mel_spec_Vit_swin-tiny-patch4-window7-224_2

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.2989
• Accuracy: 0.8981

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	3.1629	0.0463
3.2006	2.0	16	3.0994	0.1343
3.1171	3.0	24	2.9243	0.25
2.8825	4.0	32	2.5593	0.3287
2.3542	5.0	40	2.0665	0.4907
2.3542	6.0	48	1.6166	0.5741
1.6626	7.0	56	1.2288	0.6852
1.1051	8.0	64	1.0197	0.7130
0.7312	9.0	72	0.8346	0.75
0.4713	10.0	80	0.7557	0.7593
0.4713	11.0	88	0.6095	0.8241
0.2823	12.0	96	0.5716	0.8056
0.1572	13.0	104	0.4956	0.8333
0.1104	14.0	112	0.4822	0.8380
0.0772	15.0	120	0.4910	0.8519
0.0772	16.0	128	0.4082	0.8935
0.0575	17.0	136	0.3986	0.875
0.0366	18.0	144	0.4707	0.8241
0.0287	19.0	152	0.4791	0.8426
0.0252	20.0	160	0.4041	0.8704
0.0252	21.0	168	0.4406	0.8519
0.036	22.0	176	0.3919	0.8796
0.0251	23.0	184	0.4180	0.8704
0.0176	24.0	192	0.3936	0.8657
0.0142	25.0	200	0.3191	0.8981
0.0142	26.0	208	0.3062	0.8889
0.0091	27.0	216	0.3444	0.8796
0.0135	28.0	224	0.4647	0.8519
0.0147	29.0	232	0.3634	0.875
0.0172	30.0	240	0.4332	0.875
0.0172	31.0	248	0.3612	0.8704
0.0111	32.0	256	0.3731	0.875
0.0076	33.0	264	0.4858	0.8657
0.0139	34.0	272	0.3653	0.8843
0.0063	35.0	280	0.4072	0.8704
0.0063	36.0	288	0.3097	0.8796
0.0073	37.0	296	0.3541	0.8704
0.0079	38.0	304	0.3692	0.8889
0.0064	39.0	312	0.3086	0.9074
0.0057	40.0	320	0.3198	0.8935
0.0057	41.0	328	0.3424	0.9028
0.0045	42.0	336	0.2878	0.9028
0.0038	43.0	344	0.2910	0.9074
0.0026	44.0	352	0.2817	0.8981
0.0051	45.0	360	0.2945	0.8796
0.0051	46.0	368	0.2729	0.9028
0.0036	47.0	376	0.2742	0.9028
0.006	48.0	384	0.3017	0.8981
0.0011	49.0	392	0.3064	0.8889
0.0021	50.0	400	0.2989	0.8981

Framework versions

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