SW2-DMAE-DA

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: 0.8856
• Accuracy: 0.6304

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
6.464	0.96	11	7.9190	0.1087
6.5496	2.0	23	7.5824	0.1087
6.1541	2.96	34	6.4156	0.1087
5.2649	4.0	46	4.5067	0.1087
4.1175	4.96	57	2.7241	0.1087
2.8424	6.0	69	1.5623	0.1087
1.4376	6.96	80	1.4003	0.1087
1.4054	8.0	92	1.4055	0.1087
1.3798	8.96	103	1.3466	0.4565
1.3331	10.0	115	1.4385	0.1522
1.2736	10.96	126	1.3106	0.2391
1.2127	12.0	138	1.2908	0.1957
1.2531	12.96	149	1.2545	0.5
1.0972	14.0	161	1.2515	0.3478
1.0029	14.96	172	1.2238	0.2609
1.0141	16.0	184	1.2067	0.3696
0.9129	16.96	195	1.1149	0.5652
0.9157	18.0	207	1.1957	0.3913
0.8516	18.96	218	1.0034	0.5435
0.7804	20.0	230	0.9991	0.4783
0.7328	20.96	241	0.9840	0.5870
0.7101	22.0	253	0.9661	0.5435
0.7099	22.96	264	0.9392	0.5435
0.7238	24.0	276	0.9553	0.5
0.6605	24.96	287	0.9571	0.5435
0.639	26.0	299	1.0534	0.5652
0.6123	26.96	310	0.9152	0.6087
0.6021	28.0	322	0.8704	0.5870
0.5971	28.96	333	0.8726	0.5652
0.5413	30.0	345	0.8287	0.5870
0.5663	30.96	356	0.9271	0.5435
0.5343	32.0	368	0.8856	0.6304
0.525	32.96	379	0.8579	0.6087
0.5447	34.0	391	0.8746	0.5870
0.5036	34.96	402	0.8684	0.5652
0.4918	36.0	414	0.8268	0.5870
0.503	36.96	425	0.8374	0.5870
0.5114	38.0	437	0.8380	0.6087
0.5272	38.26	440	0.8387	0.6087

Framework versions

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