vit-weldclassifyv3

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: 0.2671
• Accuracy: 0.9209

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 13
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8398	0.6410	100	1.0312	0.5036
0.5613	1.2821	200	0.7068	0.6619
0.4296	1.9231	300	0.4008	0.8309
0.3475	2.5641	400	0.3345	0.8813
0.1183	3.2051	500	0.4293	0.8489
0.1531	3.8462	600	0.2748	0.9137
0.1174	4.4872	700	0.3649	0.8813
0.0498	5.1282	800	0.3279	0.8921
0.0817	5.7692	900	0.2763	0.9353
0.0075	6.4103	1000	0.2671	0.9209
0.0265	7.0513	1100	0.3185	0.9209
0.0457	7.6923	1200	0.3776	0.9101
0.0032	8.3333	1300	0.2835	0.9388
0.0027	8.9744	1400	0.5365	0.8885
0.0024	9.6154	1500	0.2817	0.9460
0.0021	10.2564	1600	0.2890	0.9460
0.002	10.8974	1700	0.2934	0.9460
0.0019	11.5385	1800	0.2976	0.9460
0.0018	12.1795	1900	0.2996	0.9460
0.0018	12.8205	2000	0.3006	0.9460

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1