results

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: 0.1114
• Accuracy: 0.9687

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 1000
• num_epochs: 4

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6639	0.1829	100	0.6155	0.6554
0.4191	0.3657	200	0.3088	0.8959
0.1698	0.5486	300	0.5321	0.7281
0.0749	0.7314	400	0.5087	0.7900
0.0484	0.9143	500	0.4649	0.8185
0.0323	1.0971	600	0.6888	0.762
0.0264	1.28	700	0.1395	0.9513
0.0224	1.4629	800	0.0661	0.9776
0.02	1.6457	900	0.1173	0.9581
0.0168	1.8286	1000	0.3498	0.889
0.013	2.0114	1100	0.1053	0.9655
0.0087	2.1943	1200	0.3601	0.8947
0.0081	2.3771	1300	0.1508	0.9535
0.0073	2.56	1400	0.2090	0.9390
0.0056	2.7429	1500	0.1136	0.9649
0.005	2.9257	1600	0.2656	0.9206
0.0036	3.1086	1700	0.1320	0.9595
0.002	3.2914	1800	0.1068	0.9686
0.0018	3.4743	1900	0.1091	0.9690
0.0019	3.6571	2000	0.1114	0.9687
0.0018	3.84	2100	0.0968	0.9719

Framework versions

• Transformers 4.41.2
• Pytorch 2.1.2
• Datasets 2.19.2
• Tokenizers 0.19.1