/deep-dream-team

Project competition for the Introduction to Machine Learning course (2023/2024)

Primary LanguagePython

deep-dream-team

Project competition for the Introduction to Machine Learning course (2023/2024)

How to test a model (pre-trained model from pytorch)

Follow these steps:

  1. Select one of the pre-trained models present in pytorch.

  2. Create a folder for the model in the models folder.

  3. Inside the new folder, create three files: config.yml, main.py and README.md (to clarify what the model does).

  4. Call the main function with the required parameters (there is an example in the SwinTransformer folder).

  5. Specify the run paramenters in the config.yml file.

  6. From the terminal, move to the folder of the model and run the following command

python main.py --config ./config.yml --run_name <run_name>

Guidelines to download datasets

The datasets can be manually downloaded and added to the src/data folder. This folder is however ignored by git and so it will only exists in the local environment.

To keep the process of training the models as smooth as possible, some functions to download libraries directly from the code are defined in the utils.py file. Datasets can be downloaded in such these ways:

  • plain download from web (.zip and .tgz)
  • download from Kaggle (with Kaggle Api)

Download from Kaggle

An extra step is required to download datasets from Kaggle. Follow these steps to use the Kaggle download.

  1. Install Kaggle with pip.
pip install --upgrade kaggle

  1. Create a Kaggle account.

  2. In the account settings, look for API and click on Create New Token. Automatically, a file called kaggle.json will be downloaded.

  3. Place this file in the location ~/.kaggle/kaggle.json on your machine. You may need to create the directory and set the correct permissions.

mkdir ~/.kaggle
chmod 600 ~/.kaggle/kaggle.json

Finally, datasets from Kaggle can be downloaded calling the function download_dataset_from_kaggle and passing as argument the name of the dataset (<author>/<name>) and the name of the directory where the dataset will be saved. There is an example call in the test.py file.