Boya2_SGD_1e3_20Epoch_Swin-large-224_fold3

This model is a fine-tuned version of microsoft/swin-large-patch4-window7-224-in22k on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.7253
• Accuracy: 0.4703

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: 0.001
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• 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
2.4987	1.0	913	2.4779	0.2773
2.2499	2.0	1826	2.3076	0.2986
2.1231	3.0	2739	2.2022	0.3325
2.1706	4.0	3652	2.1236	0.3672
2.0969	5.0	4565	2.0581	0.3940
1.9524	6.0	5478	2.0029	0.4085
1.9868	7.0	6391	1.9548	0.4208
1.9729	8.0	7304	1.9129	0.4293
1.9817	9.0	8217	1.8827	0.4331
1.9117	10.0	9130	1.8505	0.4430
1.8805	11.0	10043	1.8244	0.4482
1.8198	12.0	10956	1.8053	0.4528
1.7002	13.0	11869	1.7829	0.4558
1.811	14.0	12782	1.7721	0.4602
1.8637	15.0	13695	1.7553	0.4602
1.8566	16.0	14608	1.7454	0.4654
1.742	17.0	15521	1.7350	0.4665
1.692	18.0	16434	1.7303	0.4695
1.8241	19.0	17347	1.7261	0.4695
1.8203	20.0	18260	1.7253	0.4703

Framework versions

• Transformers 4.32.1
• Pytorch 2.1.1+cu121
• Datasets 2.21.0
• Tokenizers 0.13.2