food_classifier_2025_03_18_16_54

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.5833
• Accuracy: 0.8590

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: 0.0008
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 512
• 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: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 18

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.4822	1.0	148	1.1709	0.7706
1.1289	2.0	296	0.9941	0.7532
1.0792	3.0	444	0.8941	0.7688
0.9375	4.0	592	0.8315	0.7819
0.824	5.0	740	0.7796	0.7937
0.7217	6.0	888	0.7349	0.8068
0.6162	7.0	1036	0.7319	0.8074
0.5565	8.0	1184	0.7035	0.8164
0.479	9.0	1332	0.7102	0.8156
0.4276	10.0	1480	0.7001	0.8190
0.4123	11.0	1628	0.6803	0.8255
0.3206	12.0	1776	0.6701	0.8306
0.2767	13.0	1924	0.6520	0.8365
0.2511	14.0	2072	0.6381	0.8416
0.2338	15.0	2220	0.6207	0.8463
0.1919	16.0	2368	0.6172	0.8499
0.1828	17.0	2516	0.5954	0.8552
0.1606	18.0	2664	0.5833	0.8590

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu126
• Datasets 3.4.1
• Tokenizers 0.21.1