swinv2-large-patch4-window12-192-22k-huggingface

This model is a fine-tuned version of microsoft/swinv2-large-patch4-window12-192-22k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.6329
• Accuracy: 0.9231

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: 0.0001
• train_batch_size: 18
• eval_batch_size: 18
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 36
• 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
1.781	0.97	16	0.7972	0.7385
0.5567	2.0	33	0.4750	0.8308
0.3648	2.97	49	0.5832	0.8308
0.4016	4.0	66	0.5669	0.8
0.2448	4.97	82	0.5883	0.8462
0.3382	6.0	99	0.8846	0.8
0.3042	6.97	115	0.5364	0.8769
0.2744	8.0	132	0.5159	0.8769
0.1859	8.97	148	0.5541	0.8462
0.1787	10.0	165	0.4850	0.8923
0.181	10.97	181	0.4529	0.9077
0.113	12.0	198	0.7836	0.8154
0.0806	12.97	214	0.7141	0.8769
0.0929	14.0	231	0.5765	0.9231
0.1208	14.97	247	0.5762	0.9231
0.0764	16.0	264	0.6146	0.9231
0.099	16.97	280	0.5736	0.9231
0.0972	18.0	297	0.6051	0.9231
0.0534	18.97	313	0.6302	0.9231
0.0754	19.39	320	0.6329	0.9231

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.1+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1