swinv2-finetuned-ve-Ub200

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.5977
• Accuracy: 0.4706

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.92	6	7.9891	0.0980
No log	2.0	13	7.4848	0.0980
No log	2.92	19	6.2378	0.0980
No log	4.0	26	4.8900	0.0980
No log	4.92	32	3.8155	0.0980
No log	6.0	39	2.7342	0.0980
No log	6.92	45	2.0612	0.0980
No log	8.0	52	1.5977	0.4706
No log	8.92	58	1.3671	0.4706
No log	10.0	65	1.2122	0.4706
No log	10.92	71	1.1823	0.4706
No log	12.0	78	1.1835	0.4706
No log	12.92	84	1.1838	0.4706
No log	14.0	91	1.1778	0.4706
No log	14.92	97	1.1769	0.4706
3.2267	16.0	104	1.1762	0.4706
3.2267	16.92	110	1.1758	0.4706
3.2267	18.0	117	1.1770	0.4706
3.2267	18.46	120	1.1771	0.4706

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu118
• Datasets 2.16.1
• Tokenizers 0.15.0