finetuned-arsenic

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

• Loss: 0.0026
• Accuracy: 0.9993

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: 0.0002
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.2214	0.1848	100	0.1243	0.9607
0.1213	0.3697	200	0.1763	0.9339
0.1201	0.5545	300	0.1018	0.9607
0.0991	0.7394	400	0.2071	0.9417
0.1127	0.9242	500	0.0886	0.9666
0.0314	1.1091	600	0.0333	0.9908
0.0252	1.2939	700	0.0110	0.9974
0.0582	1.4787	800	0.0104	0.9987
0.0455	1.6636	900	0.0198	0.9954
0.0569	1.8484	1000	0.0180	0.9961
0.0627	2.0333	1100	0.0244	0.9948
0.0328	2.2181	1200	0.0054	0.9987
0.0156	2.4030	1300	0.0193	0.9948
0.0016	2.5878	1400	0.0074	0.9974
0.0032	2.7726	1500	0.0045	0.9980
0.0233	2.9575	1600	0.0029	0.9993
0.0434	3.1423	1700	0.0026	0.9993
0.0079	3.3272	1800	0.0095	0.9980
0.0175	3.5120	1900	0.0111	0.9974
0.0013	3.6969	2000	0.0109	0.9974
0.0008	3.8817	2100	0.0053	0.9987

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.19.1