vit-hybrid-base-bit-384_rice-leaf-disease-augmented-v4_tl

This model is a fine-tuned version of google/vit-hybrid-base-bit-384 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3302
• Accuracy: 0.8926

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: 0.0003
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• 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_with_restarts
• lr_scheduler_warmup_steps: 256
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2065	0.5	64	1.9827	0.2013
1.7664	1.0	128	1.4356	0.6342
1.2477	1.5	192	1.0109	0.7483
0.9114	2.0	256	0.7301	0.8054
0.7045	2.5	320	0.6434	0.8020
0.6005	3.0	384	0.5446	0.8356
0.5285	3.5	448	0.4954	0.8423
0.4916	4.0	512	0.4607	0.8557
0.4647	4.5	576	0.4562	0.8691
0.4359	5.0	640	0.4426	0.8725
0.429	5.5	704	0.4481	0.8591
0.4277	6.0	768	0.4403	0.8691
0.4356	6.5	832	0.4443	0.8658
0.3963	7.0	896	0.4111	0.8792
0.3911	7.5	960	0.3971	0.8758
0.3516	8.0	1024	0.3753	0.8859
0.3378	8.5	1088	0.3736	0.8893
0.3368	9.0	1152	0.3733	0.8859
0.319	9.5	1216	0.3641	0.8826
0.3235	10.0	1280	0.3648	0.8792
0.3123	10.5	1344	0.3646	0.8792
0.3209	11.0	1408	0.3667	0.8792
0.3106	11.5	1472	0.3583	0.8893
0.2934	12.0	1536	0.3419	0.8893
0.279	12.5	1600	0.3361	0.8960
0.2774	13.0	1664	0.3358	0.8826
0.2636	13.5	1728	0.3279	0.8926
0.2653	14.0	1792	0.3322	0.8960
0.2587	14.5	1856	0.3304	0.8926
0.2588	15.0	1920	0.3302	0.8926

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0