SWv2-DMAE-H-4-rp-clean-fix-U-40-Cross-5

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: 0.4646
• Accuracy: 0.8690

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: 4e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.6088	0.98	12	1.6063	0.2024
1.6024	1.96	24	1.5854	0.2024
1.568	2.94	36	1.5534	0.2024
1.5001	4.0	49	1.4693	0.2024
1.3811	4.98	61	1.3256	0.3690
1.27	5.96	73	1.0978	0.5476
1.0887	6.94	85	0.8237	0.7381
0.937	8.0	98	0.7746	0.7143
0.811	8.98	110	0.5772	0.7976
0.7574	9.96	122	0.6164	0.7857
0.7118	10.94	134	0.6410	0.7976
0.6374	12.0	147	0.5243	0.8095
0.5958	12.98	159	0.4589	0.8095
0.5446	13.96	171	0.5288	0.7738
0.5348	14.94	183	0.4989	0.7619
0.464	16.0	196	0.5408	0.7857
0.4641	16.98	208	0.4609	0.7738
0.4471	17.96	220	0.4229	0.8333
0.4301	18.94	232	0.3962	0.8452
0.3862	20.0	245	0.4005	0.8452
0.3659	20.98	257	0.3873	0.8452
0.3488	21.96	269	0.4196	0.8333
0.3683	22.94	281	0.4299	0.8095
0.3477	24.0	294	0.4470	0.8214
0.3426	24.98	306	0.4478	0.8333
0.3	25.96	318	0.4604	0.8452
0.3138	26.94	330	0.4114	0.8571
0.2569	28.0	343	0.4640	0.8452
0.2894	28.98	355	0.5187	0.7976
0.2996	29.96	367	0.4617	0.8452
0.3046	30.94	379	0.4646	0.8690
0.2896	32.0	392	0.4492	0.8571
0.2548	32.98	404	0.4523	0.8571
0.2137	33.96	416	0.4764	0.8333
0.2246	34.94	428	0.4474	0.8571
0.2684	36.0	441	0.4495	0.8452
0.2413	36.98	453	0.4634	0.8452
0.2633	37.96	465	0.4558	0.8452
0.2518	38.94	477	0.4523	0.8452
0.2428	39.18	480	0.4523	0.8452

Framework versions

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