project-04-model_usage

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

• Loss: 1.2321
• Accuracy: 0.5781

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: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 3
• total_train_batch_size: 48
• 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: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.0781	1.0	11	2.0475	0.1641
1.9944	2.0	22	1.9269	0.2422
1.85	3.0	33	1.7854	0.375
1.6949	4.0	44	1.6219	0.4062
1.4001	5.0	55	1.5230	0.4375
1.4009	6.0	66	1.4562	0.4688
1.322	7.0	77	1.4334	0.4766
1.2052	8.0	88	1.3413	0.5312
1.1765	9.0	99	1.3072	0.5312
1.0297	10.0	110	1.3371	0.5234
1.0263	11.0	121	1.2996	0.5625
0.9406	12.0	132	1.2836	0.5469
0.8818	13.0	143	1.2631	0.5547
0.856	14.0	154	1.2722	0.5625
0.7907	15.0	165	1.2490	0.5938
0.7897	16.0	176	1.2937	0.5156
0.76	17.0	187	1.2181	0.5859
0.7541	18.0	198	1.2592	0.5312
0.7672	19.0	209	1.2447	0.5312
0.6974	20.0	220	1.2047	0.5859

Framework versions

• Transformers 4.53.1
• Pytorch 2.6.0+cu124
• Datasets 4.0.0
• Tokenizers 0.21.2