Emotion-Image-Classification-V3

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: 1.5048
• Accuracy: 0.6375

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	20	1.4141	0.5437
No log	2.0	40	1.6711	0.4375
No log	3.0	60	1.3988	0.6
No log	4.0	80	1.5072	0.5625
No log	5.0	100	1.3970	0.6125
No log	6.0	120	1.3488	0.625
No log	7.0	140	1.4599	0.5437
No log	8.0	160	1.4678	0.5813
No log	9.0	180	1.6072	0.5375
No log	10.0	200	1.2243	0.6312
No log	11.0	220	1.2860	0.5875
No log	12.0	240	1.2472	0.5875
No log	13.0	260	1.3423	0.5875
No log	14.0	280	1.3879	0.5875
No log	15.0	300	1.4201	0.575
No log	16.0	320	1.5388	0.5312
No log	17.0	340	1.5433	0.55
No log	18.0	360	1.3812	0.5875
No log	19.0	380	1.4629	0.5938
No log	20.0	400	1.5240	0.525
No log	21.0	420	1.4818	0.5437
No log	22.0	440	1.4461	0.5687
No log	23.0	460	1.3944	0.5875
No log	24.0	480	1.6598	0.55
0.1882	25.0	500	1.4268	0.6188
0.1882	26.0	520	1.6246	0.5563
0.1882	27.0	540	1.3836	0.6125
0.1882	28.0	560	1.7652	0.4813
0.1882	29.0	580	1.4360	0.5625
0.1882	30.0	600	1.5103	0.55
0.1882	31.0	620	1.4546	0.5563
0.1882	32.0	640	1.4085	0.575
0.1882	33.0	660	1.4729	0.6062
0.1882	34.0	680	1.7415	0.5375
0.1882	35.0	700	1.7349	0.5375
0.1882	36.0	720	1.6331	0.5687
0.1882	37.0	740	1.5159	0.6062
0.1882	38.0	760	1.5464	0.5875
0.1882	39.0	780	1.5402	0.5938
0.1882	40.0	800	1.5403	0.6
0.1882	41.0	820	1.4509	0.65
0.1882	42.0	840	1.7641	0.5437
0.1882	43.0	860	1.5503	0.5813
0.1882	44.0	880	1.6178	0.5687
0.1882	45.0	900	1.5877	0.6062
0.1882	46.0	920	1.7210	0.55
0.1882	47.0	940	1.5960	0.6188
0.1882	48.0	960	1.7922	0.55
0.1882	49.0	980	2.0035	0.525
0.1299	50.0	1000	1.8269	0.5062
0.1299	51.0	1020	1.6933	0.5687
0.1299	52.0	1040	1.7252	0.5312
0.1299	53.0	1060	1.6312	0.6
0.1299	54.0	1080	1.8208	0.5375
0.1299	55.0	1100	1.7589	0.575
0.1299	56.0	1120	1.7185	0.5875
0.1299	57.0	1140	1.7227	0.5437
0.1299	58.0	1160	1.8849	0.5188
0.1299	59.0	1180	1.7565	0.5687
0.1299	60.0	1200	1.6048	0.6062
0.1299	61.0	1220	1.5088	0.6125
0.1299	62.0	1240	1.6270	0.5687
0.1299	63.0	1260	1.5913	0.625
0.1299	64.0	1280	1.7789	0.5625
0.1299	65.0	1300	1.7923	0.55
0.1299	66.0	1320	1.9365	0.575
0.1299	67.0	1340	1.7365	0.5938
0.1299	68.0	1360	1.8584	0.55
0.1299	69.0	1380	1.9811	0.5062
0.1299	70.0	1400	1.9433	0.55
0.1299	71.0	1420	1.7644	0.575
0.1299	72.0	1440	1.7661	0.6
0.1299	73.0	1460	1.8884	0.5687
0.1299	74.0	1480	1.7504	0.5813
0.0774	75.0	1500	1.9648	0.5687
0.0774	76.0	1520	1.8968	0.5437
0.0774	77.0	1540	1.7752	0.5875
0.0774	78.0	1560	1.7504	0.625
0.0774	79.0	1580	1.7458	0.6
0.0774	80.0	1600	1.8044	0.5938
0.0774	81.0	1620	1.6748	0.5813
0.0774	82.0	1640	1.7661	0.575
0.0774	83.0	1660	1.8534	0.575
0.0774	84.0	1680	1.7733	0.6125
0.0774	85.0	1700	1.7857	0.575
0.0774	86.0	1720	1.7397	0.6
0.0774	87.0	1740	1.7496	0.5813
0.0774	88.0	1760	1.8774	0.5813
0.0774	89.0	1780	1.6830	0.5938
0.0774	90.0	1800	1.9231	0.5563
0.0774	91.0	1820	1.8051	0.5875
0.0774	92.0	1840	1.8424	0.5938
0.0774	93.0	1860	1.8644	0.575
0.0774	94.0	1880	1.8415	0.5687
0.0774	95.0	1900	1.8917	0.55
0.0774	96.0	1920	1.8964	0.5625
0.0774	97.0	1940	1.6416	0.5875
0.0774	98.0	1960	1.7067	0.625
0.0774	99.0	1980	1.7533	0.5938
0.0569	100.0	2000	1.8181	0.5563

Framework versions

• Transformers 4.37.2
• Pytorch 2.3.0
• Datasets 2.15.0
• Tokenizers 0.15.1