amarsaikhan/food_classifier_5e-4_18

food_classifier_2025_03_18_16_59

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: 0.5255
• Accuracy: 0.8685

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: 0.0005
• train_batch_size: 128
• eval_batch_size: 128
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 1024
• total_eval_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
2.9159	1.0	74	2.4218	0.7535
1.1351	2.0	148	0.9153	0.7947
0.8865	3.0	222	0.7425	0.8156
0.7163	4.0	296	0.6753	0.8282
0.6342	5.0	370	0.6400	0.836
0.5685	6.0	444	0.6111	0.8429
0.485	7.0	518	0.5900	0.8477
0.4249	8.0	592	0.5912	0.8471
0.3848	9.0	666	0.5799	0.8510
0.3423	10.0	740	0.5770	0.8497
0.299	11.0	814	0.5522	0.8596
0.2897	12.0	888	0.5530	0.8592
0.2343	13.0	962	0.5402	0.8644
0.215	14.0	1036	0.5387	0.8631
0.2141	15.0	1110	0.5435	0.8670
0.2051	16.0	1184	0.5289	0.8682
0.1874	17.0	1258	0.5318	0.868
0.1664	18.0	1332	0.5255	0.8685

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0