SWv2-DMAE-H-5-ps-clean-fix-U-40-Cross-1

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: 0.6312
• Accuracy: 0.8452

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
1.6089	0.98	12	1.6064	0.2024
1.603	1.96	24	1.5810	0.2024
1.5638	2.94	36	1.5728	0.2024
1.5222	4.0	49	1.5038	0.2024
1.404	4.98	61	1.3034	0.3929
1.2635	5.96	73	1.0843	0.6548
1.1157	6.94	85	0.8546	0.7381
1.0145	8.0	98	0.8059	0.7262
0.8228	8.98	110	0.7111	0.7619
0.7828	9.96	122	0.5751	0.8095
0.6954	10.94	134	0.5913	0.7857
0.6508	12.0	147	0.5722	0.7976
0.6585	12.98	159	0.5208	0.7976
0.5769	13.96	171	0.7267	0.7262
0.5212	14.94	183	0.6463	0.75
0.5724	16.0	196	0.7517	0.7381
0.5109	16.98	208	0.6051	0.7976
0.463	17.96	220	0.5454	0.7738
0.41	18.94	232	0.5613	0.7976
0.4419	20.0	245	0.6522	0.8095
0.4384	20.98	257	0.6474	0.7738
0.397	21.96	269	0.5957	0.8214
0.3608	22.94	281	0.5873	0.8214
0.367	24.0	294	0.6312	0.8452
0.3354	24.98	306	0.6181	0.8214
0.3393	25.96	318	0.6168	0.8214
0.3152	26.94	330	0.6132	0.8333
0.2919	28.0	343	0.6221	0.8095
0.3138	28.98	355	0.6142	0.7976
0.3027	29.96	367	0.5974	0.8095
0.3399	30.94	379	0.6276	0.7619
0.3249	32.0	392	0.6439	0.8214
0.2448	32.98	404	0.6229	0.8214
0.2658	33.96	416	0.6355	0.8214
0.2889	34.94	428	0.6203	0.7976
0.2957	36.0	441	0.6140	0.7857
0.2546	36.98	453	0.6258	0.7976
0.2486	37.96	465	0.6276	0.8214
0.2049	38.94	477	0.6288	0.8214
0.2575	39.18	480	0.6290	0.8214

Framework versions

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