my_awesome_face_model

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: 1.3503
• Accuracy: 0.4938

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: 6
• total_train_batch_size: 192
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9	3	2.0832	0.1
No log	1.8	6	2.0749	0.1
2.0704	3.0	10	2.0478	0.1812
2.0704	3.9	13	2.0288	0.2
2.0704	4.8	16	1.9896	0.2812
1.9609	6.0	20	1.9322	0.3125
1.9609	6.9	23	1.8715	0.3438
1.9609	7.8	26	1.8056	0.3688
1.7478	9.0	30	1.7196	0.4125
1.7478	9.9	33	1.6635	0.3937
1.7478	10.8	36	1.6160	0.3875
1.5379	12.0	40	1.6014	0.425
1.5379	12.9	43	1.5505	0.4188
1.5379	13.8	46	1.5187	0.4188
1.3858	15.0	50	1.4938	0.4562
1.3858	15.9	53	1.4853	0.4562
1.3858	16.8	56	1.4664	0.4875
1.2551	18.0	60	1.4488	0.4375
1.2551	18.9	63	1.4531	0.4375
1.2551	19.8	66	1.3847	0.525
1.1573	21.0	70	1.3790	0.4813
1.1573	21.9	73	1.4168	0.4625
1.1573	22.8	76	1.4159	0.5
1.0833	24.0	80	1.3935	0.4813
1.0833	24.9	83	1.3965	0.4688
1.0833	25.8	86	1.3705	0.4875
1.0091	27.0	90	1.3879	0.4625
1.0091	27.9	93	1.3777	0.5062
1.0091	28.8	96	1.3867	0.4813
0.9632	30.0	100	1.3659	0.5188
0.9632	30.9	103	1.3398	0.4875
0.9632	31.8	106	1.3555	0.4688
0.9259	33.0	110	1.3621	0.5062
0.9259	33.9	113	1.3569	0.4938
0.9259	34.8	116	1.3397	0.5312
0.8994	36.0	120	1.3314	0.525

Framework versions

• Transformers 4.41.1
• Pytorch 2.3.0+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1