amd-torch-device-gfx1012

Create a Python-based mini-application that leverages the 'amd-torch-device-gfx1012' package to perform accelerated tensor operations specifically on AMD GPUs with the GFX1012 architecture. This application will serve as a demonstration of how to harness the power of AMD GPUs for deep learning tasks using PyTorch. Your task is to develop a simple yet effective image classification model using a pre-trained ResNet-18 architecture from torchvision.models. The application should include the following features:

1. **Model Loading**: Load the pre-trained ResNet-18 model from torchvision.models.
2. **Device Configuration**: Use the 'amd-torch-device-gfx1012' package to configure the GPU device settings for optimal performance on AMD GPUs with the GFX1012 architecture.
3. **Data Preparation**: Prepare a dataset of images for classification. This could be a subset of ImageNet or any other suitable dataset.
4. **Prediction Functionality**: Implement a function that takes an input image, preprocesses it, feeds it through the model, and outputs the predicted class along with its probability score.
5. **Performance Measurement**: Measure and display the inference time for each prediction.
6. **Visualization**: Include a simple visualization of the model's confidence in each predicted class using matplotlib.
7. **User Interface**: Develop a basic command-line interface where users can input an image file path and receive the classification result.

The application should showcase the benefits of using specialized GPU packages like 'amd-torch-device-gfx1012' for accelerating deep learning tasks on specific hardware configurations. Ensure your code is well-commented and includes explanations of how the 'amd-torch-device-gfx1012' package is utilized throughout the application.