Model Card for Model ID
This model card provides comprehensive information about the model's architecture, training data, evaluation metrics, and environmental impact.
Model Details
Model Description
This model is a pre-trained model for image classification, specifically designed for detecting fake images, including both real and AI-generated synthetic images. It utilizes the ViT (Vision Transformer) architecture for image classification tasks.
- Developed by: [Author(s) Name(s)]
- Funded by [optional]: [Funding Source(s)]
- Shared by [optional]: [Organization/Individual(s) Sharing the Model]
- Model type: Vision Transformer (ViT)
- Language(s) (NLP): N/A
- License: Apache License 2.0
- Finetuned from model [optional]: [Base Pre-trained Model]
Model Sources [optional]
Uses
Direct Use
This model can be directly used for classifying images as real or AI-generated synthetic images.
Downstream Use [optional]
This model can be fine-tuned for specific image classification tasks related to detecting fake images in various domains.
Out-of-Scope Use
The model may not perform well on tasks outside the scope of image classification, such as object detection or segmentation.
Bias, Risks, and Limitations
The model's performance may be influenced by biases in the training data, leading to potential inaccuracies in classification.
Recommendations
Users should be aware of potential biases and limitations when using the model for classification tasks, and additional data sources may be necessary to mitigate biases.
How to Get Started with the Model
Use the code below to get started with the model:
[Code Snippet for Model Usage]
Training Details
Training Data
The model was trained on the CIFake dataset, which contains real and AI-generated synthetic images for training the classification model.
Training Procedure
Preprocessing [optional]
Data preprocessing techniques were applied to the training data, including normalization and data augmentation to improve model generalization.
Training Hyperparameters
- Training regime: Fine-tuning with a learning rate of 0.0000001
- Batch Size: 64
- Epochs: 100
Speeds, Sizes, Times [optional]
- Training Time: 1 hr 36 min
Evaluation
Testing Data, Factors & Metrics
Testing Data
The model was evaluated on a separate test set from the CIFake dataset.
Factors
The evaluation considered factors such as class imbalance and dataset diversity.
Metrics
Evaluation metrics included accuracy, precision, recall, and F1-score.
Results
The model achieved an accuracy of [Accuracy] on the test set, with detailed metrics summarized in the following table:
[Metrics Table]
Model Examination [optional]
[Information on Model Examination, if available]
Technical Specifications [optional]
Model Architecture and Objective
The model architecture is based on the Vision Transformer (ViT) architecture, which uses self-attention mechanisms for image classification tasks.
- Downloads last month
- 3