realFake-img

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

• Loss: 0.0819
• Accuracy: 0.9785

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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.2578	0.2525	100	0.1594	0.9418
0.0944	0.5051	200	0.2243	0.9373
0.1747	0.7576	300	0.2472	0.9293
0.1328	1.0101	400	0.1774	0.9338
0.1918	1.2626	500	0.1282	0.9570
0.169	1.5152	600	0.2247	0.9346
0.2595	1.7677	700	0.1785	0.9445
0.0911	2.0202	800	0.1353	0.9534
0.0548	2.2727	900	0.1998	0.9472
0.1399	2.5253	1000	0.1971	0.9445
0.2001	2.7778	1100	0.2479	0.9373
0.0976	3.0303	1200	0.1601	0.9499
0.1291	3.2828	1300	0.1607	0.9588
0.0721	3.5354	1400	0.1822	0.9588
0.0592	3.7879	1500	0.1255	0.9624
0.0964	4.0404	1600	0.1620	0.9543
0.0738	4.2929	1700	0.1279	0.9651
0.0504	4.5455	1800	0.1624	0.9588
0.0972	4.7980	1900	0.1579	0.9624
0.0456	5.0505	2000	0.1965	0.9490
0.0334	5.3030	2100	0.1652	0.9570
0.0242	5.5556	2200	0.1182	0.9749
0.0715	5.8081	2300	0.1250	0.9651
0.0407	6.0606	2400	0.1172	0.9696
0.0003	6.3131	2500	0.0819	0.9785
0.0072	6.5657	2600	0.1406	0.9714
0.0183	6.8182	2700	0.1152	0.9749
0.0021	7.0707	2800	0.1368	0.9731
0.046	7.3232	2900	0.0900	0.9794
0.033	7.5758	3000	0.1014	0.9785
0.0354	7.8283	3100	0.0968	0.9767
0.0026	8.0808	3200	0.1217	0.9731
0.0002	8.3333	3300	0.0828	0.9794
0.0006	8.5859	3400	0.0926	0.9794
0.0006	8.8384	3500	0.1001	0.9794
0.0006	9.0909	3600	0.0863	0.9848
0.0633	9.3434	3700	0.0911	0.9803
0.0009	9.5960	3800	0.0941	0.9821
0.0247	9.8485	3900	0.0988	0.9785

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1