resnet 50 fashion mnist quality drift

Introduction

The ResNet-50-Fashion-MNIST-Quality-Drift model is a fine-tuned version of Microsoft's ResNet-50 tailored for the Fashion MNIST Quality Drift dataset. It achieves an accuracy of 0.73 and an F1 score of 0.7289 on the evaluation set. The model is designed for image classification tasks.

Architecture

The model is based on the ResNet-50 architecture, a well-known convolutional neural network (CNN) known for its deep residual learning framework, which allows for the training of very deep networks by addressing the vanishing gradient problem.

Training

Training Hyperparameters

• Learning Rate: 0.0002
• Batch Size: 8 (both training and evaluation)
• Seed: 42
• Optimizer: Adam with betas=(0.9, 0.999) and epsilon=1e-08
• Learning Rate Scheduler Type: Linear
• Number of Epochs: 3
• Mixed Precision Training: Native AMP

Training Results

• Epoch 1: Training Loss = 1.5138, Validation Loss = 0.9237, Accuracy = 0.684, F1 = 0.6826
• Epoch 2: Training Loss = 0.9377, Validation Loss = 0.7861, Accuracy = 0.722, F1 = 0.7253
• Epoch 3: Training Loss = 0.8276, Validation Loss = 0.7473, Accuracy = 0.73, F1 = 0.7289

Guide: Running Locally

To run this model locally, follow these steps:

1. Setup Environment:

   • Ensure you have Python installed.
   • Install PyTorch, Transformers, Datasets, and other dependencies using pip.

2. Clone Repository:

   • Clone the model repository or download the necessary files.

3. Load Model and Data:

   • Use the PyTorch framework to load the model.
   • Prepare the Fashion MNIST Quality Drift dataset.

4. Run Inference or Training:

   • Use the provided scripts or develop your own to perform inference or further training.

5. Hardware Recommendations:

   • For optimal performance, consider using cloud GPUs such as those provided by AWS, Google Cloud, or Azure.

License

This model is licensed under the Apache-2.0 license, which allows for use, modification, and distribution with attribution.