This repo serves as an introduction into some modern ML paradigms and the PyTorch framework.
The input image is 224x224x3. The output is 1000 classes.
- Convolutional layer with 96 filters of size 11x11x3 with stride 4 and ReLU activation.
- Max pooling layer with size 3x3 and stride 2.
- Convolutional layer with 256 filters of size 5x5x48 with stride 1 and ReLU activation.
- Max pooling layer with size 3x3 and stride 2.
- Convolutional layer with 384 filters of size 3x3x256 with stride 1 and ReLU activation.
- Convolutional layer with 384 filters of size 3x3x192 with stride 1 and ReLU activation.
- Convolutional layer with 256 filters of size 3x3x192 with stride 1 and ReLU activation.
- Max pooling layer with size 3x3 and stride 2.
- Fully connected layer with 4096 neurons and ReLU activation.
- Fully connected layer with 4096 neurons and ReLU activation.
- Fully connected layer with 1000 neurons and Softmax activation.
Method 1: Use a virtual environment to install the required packages.
python3 -m venv venv
source venv/bin/activate
To install the required packages, run the following command:
pip3 install -r requirements.txt
Method 2: Use Docker to run the code.
docker build -t ml_intro .
docker run -it ml_intro
To run the code, run the following command:
python3 main.py <config_file>
Example:
python3 main.py config/default.json