convnext-tiny-224-convnext

This model is a fine-tuned version of facebook/convnext-tiny-224 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.0225
• Accuracy: 0.9951

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.2175	1.0	327	0.1708	0.9436
0.1476	2.0	654	0.0908	0.9672
0.0961	3.0	981	0.0428	0.9862
0.0677	4.0	1309	0.0654	0.9777
0.049	5.0	1636	0.0498	0.9857
0.0347	6.0	1963	0.0352	0.9886
0.0282	7.0	2290	0.0278	0.9913
0.0694	8.0	2618	0.0299	0.9918
0.0733	9.0	2945	0.0246	0.9938
0.0399	10.0	3272	0.0285	0.9918
0.0276	11.0	3599	0.0249	0.9933
0.0259	12.0	3927	0.0241	0.9942
0.0551	13.0	4254	0.0298	0.9920
0.0658	14.0	4581	0.0288	0.9924
0.0208	14.99	4905	0.0225	0.9951

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.15.0