swinv2-tiny-patch4-window8-256-RD-da-colab

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: 10.2420
• Accuracy: 0.5036

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.1347	1.0	96	10.0386	0.4982
0.1556	2.0	192	9.5018	0.5
0.1051	3.0	288	9.9516	0.4982
0.1154	4.0	384	10.8351	0.4945
0.0909	5.0	480	11.6091	0.4945
0.0923	6.0	576	9.0530	0.5
0.1089	7.0	672	11.6765	0.4927
0.0959	8.0	768	11.5132	0.4982
0.1266	9.0	864	10.2420	0.5036
0.106	10.0	960	11.1262	0.4945
0.0831	11.0	1056	11.5815	0.4964
0.0819	12.0	1152	11.6394	0.4964
0.0862	13.0	1248	10.9660	0.4982
0.0754	14.0	1344	9.5463	0.4982
0.06	15.0	1440	10.2678	0.4964
0.0828	16.0	1536	11.4973	0.4927
0.0675	17.0	1632	10.5019	0.4964
0.0687	18.0	1728	10.6483	0.4982
0.0548	19.0	1824	11.2166	0.4964
0.0658	20.0	1920	11.5459	0.4945
0.0565	21.0	2016	11.5899	0.4945
0.0807	22.0	2112	10.7066	0.5
0.0289	23.0	2208	10.6253	0.4982
0.0755	24.0	2304	10.4856	0.5018
0.0483	25.0	2400	11.3838	0.4964
0.0732	26.0	2496	11.1971	0.4927
0.1424	27.0	2592	11.4581	0.4945
0.0814	28.0	2688	11.3341	0.4945
0.101	29.0	2784	11.5705	0.4927
0.0894	30.0	2880	11.5259	0.4927
0.0707	31.0	2976	11.1753	0.4945
0.1289	32.0	3072	10.5668	0.4964
0.0991	33.0	3168	11.1013	0.4945
0.0615	34.0	3264	11.0973	0.4945
0.0784	35.0	3360	11.0716	0.4945
0.0792	36.0	3456	11.0241	0.4945
0.1032	37.0	3552	11.2338	0.4945
0.0837	38.0	3648	11.4256	0.4927
0.0722	39.0	3744	11.3971	0.4945
0.079	40.0	3840	11.3523	0.4945

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3