swinv2-tiny-patch4-window8-256-finetuned-200k

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: 0.3715
• Accuracy: 0.8360

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: 5e-05
• train_batch_size: 256
• eval_batch_size: 256
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 1024
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.7044	0.93	10	0.6932	0.5410
0.6768	1.95	21	0.6407	0.6614
0.6359	2.98	32	0.5647	0.7208
0.5989	4.0	43	0.5674	0.7086
0.5831	4.93	53	0.5108	0.7679
0.549	5.95	64	0.4882	0.7836
0.5341	6.98	75	0.4831	0.7714
0.5172	8.0	86	0.4422	0.8115
0.4961	8.93	96	0.4422	0.7941
0.4796	9.95	107	0.4066	0.8098
0.4776	10.98	118	0.3906	0.8185
0.4668	12.0	129	0.4135	0.8150
0.4588	12.93	139	0.3884	0.8202
0.448	13.95	150	0.3764	0.8220
0.4508	14.98	161	0.3802	0.8220
0.43	16.0	172	0.3829	0.8150
0.4347	16.93	182	0.3857	0.8133
0.4232	17.95	193	0.3819	0.8150
0.4289	18.98	204	0.4055	0.8080
0.4271	20.0	215	0.3577	0.8377
0.4301	20.93	225	0.3598	0.8272
0.4257	21.95	236	0.3780	0.8237
0.4191	22.98	247	0.3545	0.8307
0.4164	24.0	258	0.4208	0.8115
0.4297	24.93	268	0.3817	0.8290
0.4168	25.95	279	0.3876	0.8220
0.4118	26.98	290	0.3670	0.8307
0.4042	28.0	301	0.3620	0.8290
0.4018	28.93	311	0.3670	0.8290
0.4074	29.95	322	0.3822	0.8290
0.4044	30.98	333	0.3561	0.8325
0.3998	32.0	344	0.3642	0.8377
0.3994	32.93	354	0.3721	0.8290
0.3982	33.95	365	0.3592	0.8394
0.4002	34.98	376	0.3740	0.8290
0.4014	36.0	387	0.3705	0.8325
0.3953	36.93	397	0.3865	0.8237
0.3934	37.95	408	0.3689	0.8342
0.3964	38.98	419	0.3570	0.8255
0.4027	40.0	430	0.3738	0.8325
0.392	40.93	440	0.3566	0.8342
0.3875	41.95	451	0.3652	0.8377
0.3866	42.98	462	0.3657	0.8342
0.396	44.0	473	0.3662	0.8342
0.3841	44.93	483	0.3764	0.8360
0.387	45.95	494	0.3687	0.8325
0.3844	46.51	500	0.3715	0.8360

Framework versions

• Transformers 4.33.3
• Pytorch 2.0.1+cu117
• Datasets 2.14.5
• Tokenizers 0.13.3