swinv2-tiny-patch4-window8-256-dmae-humeda-DAV36

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 2.6972
• Accuracy: 0.68

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: 2
• total_train_batch_size: 32
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.214	1.0	22	1.4507	0.46
2.298	2.0	44	1.0632	0.62
0.9579	3.0	66	1.1191	0.64
0.4479	4.0	88	0.9825	0.6467
0.1963	5.0	110	1.2844	0.6467
0.1663	6.0	132	1.2373	0.6667
0.1188	7.0	154	1.4338	0.6933
0.0526	8.0	176	1.6726	0.7133
0.01	9.0	198	2.5248	0.6267
0.028	10.0	220	2.6156	0.6467
0.0296	11.0	242	2.8334	0.6533
0.0074	12.0	264	2.2200	0.6867
0.0022	13.0	286	2.2802	0.7467
0.0225	14.0	308	2.1764	0.6933
0.0058	15.0	330	3.0594	0.62
0.0075	16.0	352	3.2166	0.6333
0.0163	17.0	374	2.4014	0.6933
0.0033	18.0	396	2.9112	0.6733
0.0036	19.0	418	2.8147	0.6533
0.0033	20.0	440	2.7731	0.6733
0.0161	21.0	462	2.0340	0.7467
0.0012	22.0	484	2.4596	0.6867
0.0009	23.0	506	2.7352	0.6667
0.0101	24.0	528	2.8204	0.6667
0.0011	25.0	550	2.8091	0.6733
0.0005	26.0	572	2.8126	0.6667
0.0007	27.0	594	2.7742	0.6733
0.0004	28.0	616	2.7208	0.6733
0.0025	29.0	638	2.7022	0.6733
0.0025	30.0	660	2.6972	0.68

Framework versions

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