SWIN_finetuned_frozen_v4

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

• Loss: 2.0700
• Accuracy: 0.7085

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: 128
• eval_batch_size: 128
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15.0

Training results

Training Loss	Epoch	Step	Accuracy	Validation Loss
1.3057	1.0	5249	0.5822	2.0464
1.1387	2.0	10498	0.6084	1.9457
0.988	3.0	15747	0.6124	1.9650
0.8653	4.0	20996	0.6344	1.9381
0.7662	5.0	26245	0.6335	1.9391
0.6882	6.0	31494	0.6444	1.9591
0.601	7.0	36743	0.6510	1.9506
0.5363	8.0	41992	0.6617	1.9556
0.4871	9.0	47241	0.6741	1.9037
0.4338	10.0	52490	0.6806	1.9794
0.3738	11.0	57739	0.6849	2.0053
0.3338	12.0	62988	0.6955	2.0140
0.2989	13.0	68237	0.6974	2.0830
0.267	14.0	73486	0.7041	2.0824
0.236	15.0	78735	0.7085	2.0700

Framework versions

• Transformers 4.33.3
• Pytorch 2.1.2
• Datasets 2.16.1
• Tokenizers 0.13.3