swin-tiny-rice-disease

This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on the rice_disease_701515 dataset. It achieves the following results on the evaluation set:

• Accuracy: 0.8771
• F1 Score: 0.8783
• Loss: 0.8022
• Recall: 0.8774

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: 2e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 0.1
• num_epochs: 35

Training results

Training Loss	Epoch	Step	Accuracy	F1 Score	Validation Loss	Recall
2.6015	1.0	31	0.4019	0.3953	2.5301	0.3990
2.0857	2.0	62	0.6383	0.6408	1.8060	0.6379
1.5981	3.0	93	0.7234	0.7269	1.4075	0.7226
1.4038	4.0	124	0.8014	0.8005	1.0924	0.8016
1.3016	5.0	155	0.7730	0.7755	1.0891	0.7740
1.1921	6.0	186	0.7801	0.7821	1.1443	0.7797
1.0699	7.0	217	0.8345	0.8345	0.8551	0.8350
1.0590	8.0	248	0.8298	0.8300	0.9121	0.8296
0.9309	9.0	279	0.8558	0.8541	0.8474	0.8558
1.1103	10.0	310	0.8369	0.8364	0.8958	0.8366
0.8052	11.0	341	0.8723	0.8714	0.8341	0.8731
0.8316	12.0	372	0.8605	0.8595	0.8205	0.8611
0.7925	13.0	403	0.8652	0.8642	0.8144	0.8653
0.7598	14.0	434	0.8842	0.8840	0.7775	0.8840
0.7854	15.0	465	0.8936	0.8941	0.6830	0.8938
0.6274	16.0	496	0.8842	0.8845	0.7228	0.8838
0.7235	17.0	527	0.8794	0.8785	0.8013	0.8793
0.6191	18.0	558	0.8652	0.8657	0.8238	0.8657
0.6272	19.0	589	0.8794	0.8799	0.7262	0.8798
0.5298	20.0	620	0.8652	0.8647	0.8447	0.8653
0.5665	21.0	651	0.8747	0.8741	0.8196	0.8750
0.5508	22.0	682	0.8842	0.8834	0.7673	0.8848
0.5440	23.0	713	0.8629	0.8638	0.8191	0.8635
0.6131	24.0	744	0.8676	0.8672	0.8758	0.8674
0.4691	25.0	775	0.8652	0.8659	0.8382	0.8659
0.4537	26.0	806	0.8700	0.8715	0.8270	0.8704
0.6059	27.0	837	0.8771	0.8783	0.8022	0.8774

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 5.0.0
• Tokenizers 0.22.2