metadata

license: apache-2.0
base_model: dima806/deepfake_vs_real_image_detection
tags:
  - image-classification
  - generated_from_trainer
datasets:
  - imagefolder
metrics:
  - accuracy
model-index:
  - name: realFake-img
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: ai_real_images
          type: imagefolder
          config: default
          split: train
          args: default
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8518181818181818

realFake-img

This model is a fine-tuned version of dima806/deepfake_vs_real_image_detection on the ai_real_images dataset. It achieves the following results on the evaluation set:

Loss: 0.3329
Accuracy: 0.8518

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
0.4892	0.2564	100	0.5756	0.7227
0.683	0.5128	200	0.6742	0.6373
0.3737	0.7692	300	0.5462	0.7555
0.3554	1.0256	400	0.4354	0.8009
0.2368	1.2821	500	0.4046	0.8309
0.3696	1.5385	600	0.5547	0.7809
0.2824	1.7949	700	0.3329	0.8518
0.2366	2.0513	800	0.4582	0.8255
0.2212	2.3077	900	0.4885	0.8255
0.2031	2.5641	1000	0.4282	0.8564
0.1717	2.8205	1100	0.4373	0.85
0.1303	3.0769	1200	0.3659	0.8718
0.0889	3.3333	1300	0.3663	0.8736
0.1157	3.5897	1400	0.4588	0.8436
0.1215	3.8462	1500	0.4350	0.8655

Framework versions

Transformers 4.42.4
Pytorch 2.4.0+cu121
Datasets 2.21.0
Tokenizers 0.19.1