swinv2-tiny-patch4-window8-256-RD-FIX

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: 0.5014
• Accuracy: 0.7826

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
No log	0.8571	3	1.1955	0.4565
No log	1.8571	6	1.1280	0.5
No log	2.8571	9	1.0565	0.4783
4.8751	3.8571	12	0.9184	0.5870
4.8751	4.8571	15	0.8208	0.5870
4.8751	5.8571	18	0.7310	0.6087
3.6315	6.8571	21	0.6951	0.7174
3.6315	7.8571	24	0.6772	0.7174
3.6315	8.8571	27	0.6626	0.7174
2.8559	9.8571	30	0.5987	0.7826
2.8559	10.8571	33	0.5431	0.8261
2.8559	11.8571	36	0.6193	0.6739
2.8559	12.8571	39	0.6475	0.7174
2.3617	13.8571	42	0.5725	0.7174
2.3617	14.8571	45	0.5794	0.7826
2.3617	15.8571	48	0.5292	0.7826
2.1506	16.8571	51	0.5988	0.7391
2.1506	17.8571	54	0.6548	0.7174
2.1506	18.8571	57	0.5131	0.8261
1.9498	19.8571	60	0.4700	0.8478
1.9498	20.8571	63	0.5254	0.8043
1.9498	21.8571	66	0.5451	0.7826
1.9498	22.8571	69	0.5304	0.7609
1.422	23.8571	72	0.5105	0.8043
1.422	24.8571	75	0.4685	0.7826
1.422	25.8571	78	0.4875	0.8261
1.3044	26.8571	81	0.5492	0.7826
1.3044	27.8571	84	0.5202	0.7826
1.3044	28.8571	87	0.4737	0.8261
1.2464	29.8571	90	0.4398	0.8478
1.2464	30.8571	93	0.4753	0.8043
1.2464	31.8571	96	0.4913	0.8043
1.2464	32.8571	99	0.5262	0.7826
1.1614	33.8571	102	0.5280	0.7826
1.1614	34.8571	105	0.5252	0.7609
1.1614	35.8571	108	0.5127	0.7826
1.045	36.8571	111	0.5061	0.7826
1.045	37.8571	114	0.5012	0.7826
1.045	38.8571	117	0.5025	0.7826
0.9391	39.8571	120	0.5014	0.7826

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0