vit-emotion_classifier

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: 1.4782
• Accuracy: 0.525

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.0776	1.0	10	2.0731	0.1437
2.0526	2.0	20	2.0567	0.1688
1.9975	3.0	30	2.0160	0.2
1.8977	4.0	40	1.9550	0.3
1.778	5.0	50	1.8805	0.3625
1.6549	6.0	60	1.8073	0.375
1.5379	7.0	70	1.7428	0.4125
1.4241	8.0	80	1.6957	0.4062
1.3212	9.0	90	1.6550	0.45
1.2245	10.0	100	1.6271	0.4437
1.1336	11.0	110	1.5928	0.4562
1.0483	12.0	120	1.5695	0.4688
0.9669	13.0	130	1.5452	0.4875
0.8889	14.0	140	1.5248	0.4875
0.815	15.0	150	1.5063	0.5062
0.7466	16.0	160	1.4909	0.4938
0.6852	17.0	170	1.4782	0.525
0.6308	18.0	180	1.4615	0.5
0.5819	19.0	190	1.4541	0.5
0.5392	20.0	200	1.4458	0.5125
0.503	21.0	210	1.4393	0.5
0.4718	22.0	220	1.4289	0.5188
0.4458	23.0	230	1.4238	0.5188
0.4234	24.0	240	1.4211	0.5125
0.405	25.0	250	1.4182	0.5
0.3905	26.0	260	1.4157	0.5062
0.379	27.0	270	1.4125	0.5062
0.3706	28.0	280	1.4119	0.5062
0.3649	29.0	290	1.4115	0.5062
0.3618	30.0	300	1.4111	0.5062

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3