vit-base-patch16-224-in21k-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: 1.6098
• Accuracy: 0.4375

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: 101010
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 2000
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	20	2.0782	0.1938
No log	2.0	40	2.0771	0.1938
No log	3.0	60	2.0752	0.1875
No log	4.0	80	2.0725	0.1875
No log	5.0	100	2.0691	0.1812
No log	6.0	120	2.0646	0.1875
No log	7.0	140	2.0591	0.1875
No log	8.0	160	2.0517	0.2062
No log	9.0	180	2.0423	0.2062
No log	10.0	200	2.0301	0.2437
No log	11.0	220	2.0148	0.275
No log	12.0	240	1.9941	0.2687
No log	13.0	260	1.9721	0.325
No log	14.0	280	1.9464	0.3375
No log	15.0	300	1.9138	0.3312
No log	16.0	320	1.8832	0.3438
No log	17.0	340	1.8495	0.3625
No log	18.0	360	1.8153	0.3688
No log	19.0	380	1.7807	0.3625
No log	20.0	400	1.7487	0.3812
No log	21.0	420	1.7179	0.3875
No log	22.0	440	1.6897	0.4125
No log	23.0	460	1.6649	0.4062
No log	24.0	480	1.6409	0.3937
1.7227	25.0	500	1.6235	0.4188
1.7227	26.0	520	1.5990	0.4
1.7227	27.0	540	1.5816	0.425
1.7227	28.0	560	1.5664	0.45
1.7227	29.0	580	1.5497	0.4313
1.7227	30.0	600	1.5323	0.4125
1.7227	31.0	620	1.5209	0.425
1.7227	32.0	640	1.5059	0.4
1.7227	33.0	660	1.5029	0.4188
1.7227	34.0	680	1.4970	0.4313
1.7227	35.0	700	1.4944	0.4062
1.7227	36.0	720	1.4992	0.425
1.7227	37.0	740	1.5060	0.425
1.7227	38.0	760	1.4960	0.4313
1.7227	39.0	780	1.5080	0.4313
1.7227	40.0	800	1.5175	0.425
1.7227	41.0	820	1.5219	0.4188
1.7227	42.0	840	1.5273	0.4313
1.7227	43.0	860	1.5318	0.425
1.7227	44.0	880	1.5446	0.4313
1.7227	45.0	900	1.5519	0.4375
1.7227	46.0	920	1.5678	0.4188
1.7227	47.0	940	1.5747	0.4375
1.7227	48.0	960	1.5843	0.4375
1.7227	49.0	980	1.5968	0.425
0.3221	50.0	1000	1.6098	0.4375

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.2+cu121
• Datasets 2.16.1
• Tokenizers 0.15.1