Skip to content

README.md

Latest commit

 

History

History
154 lines (108 loc) · 4.46 KB

README.md

File metadata and controls

154 lines (108 loc) · 4.46 KB

RecEngineMF


Data

MovieLens 20M Dataset

Command to Download and Extract Data

  1. Download MovieLens 20M Data

    wget --output-document=./ml-20m.zip  https://files.grouplens.org/datasets/movielens/ml-20m.zip

  2. Once the download is complete, extract the dataset

    unzip ml-20m.zip

Usage

Install Dependencies

  1. Install Python: Make sure Python is installed on your system. If not, you can download and install Python from the official Python website: https://www.python.org/downloads/

  2. Create a virtual environment:

    python -m venv myenv

  3. Activate the virtual environment

    For Windows CMD Users

    .\myenv\Scripts\Activate.bat

    For Windows Powershell Users

    .\myenv\Scripts\Activate.ps1

    For macOS/Linux Users

    source myenv/bin/activate

  4. Install the dependencies

    pip install -r requirements.txt

  5. Add wandb API key

    Sign in to https://wandb.ai and get your API key.
    Create a file secrets.json in the root directory and put your wandb API key.

    {
	"WANDB_API_KEY": "YOUR_API_KEY"
}

Train

python train.py --data_path DATA_PATH [--emb_size EMB_SIZE] [--random_seed RANDOM_SEED] 
                [--batch_size BATCH_SIZE] [--epochs EPOCHS] [--learning_rate LEARNING_RATE] 
                [--weight_decay WEIGHT_DECAY] [--step_size STEP_SIZE] [--gamma GAMMA] 
                [--patience PATIENCE] [--model_name MODEL_NAME] [--metrics_csv_name METRICS_CSV_NAME]
                [--silent] [--log_wandb]

Required Flag

  • --data_path: Path to the CSV file containing the ratings data.

Optional Flags

  • --emb_size: Size of the embedding for users and items. Default is 100.
  • --random_seed: Random seed for reproducibility. Default is 42.
  • --batch_size: Batch size for training. Default is 64000.
  • --epochs: Number of epochs for training. Default is 100.
  • --learning_rate: Learning rate for optimizer. Default is 0.001.
  • --weight_decay: Weight decay for optimizer. Default is 1e-5.
  • --step_size: Step size for learning rate scheduler. Default is 10.
  • --gamma: Gamma value for learning rate scheduler. Default is 0.1.
  • --patience: Patience for early stopping based on validation loss. Default is 3.
  • --model_name: Name of the trained model file to be saved. Default is 'mf_model.pth'.
  • --metrics_csv_name: Name of the CSV file to save the training metrics. Default is 'metrics.csv'.
  • --silent: Whether to hide verbose output during training.
  • --log_wandb: Whether to log metrics into weights and bias (wandb.ai).

Test

python test.py  --data_path DATA_PATH --model_path MODEL_PATH [--batch_size BATCH_SIZE] [--random_seed RANDOM_SEED]

Required Flags

  • --data_path: Path to the CSV file containing the ratings data.
  • --model_path: Path to the trained model file to be loaded for testing.

Optional Flags

  • --batch_size: Batch size for testing. Default is 64000.
  • --random_seed: Random seed for reproducibility. Default is 42.

Run Inference

python inference.py --data_path DATA_PATH --model_path MODEL_PATH --user_id USER_ID [--n_items N_ITEMS]

Required Flags

  • --data_path: Path to the CSV file containing the ratings data.
  • --model_path: Path to the trained model file to be loaded for testing.
  • --user_id: The id of the user for whom item is to be recommended.

Optional Flags

  • --n_items: The top n number of items to be recommended to the user. Default is 10.

Plot Curve

python plot.py --metrics_csv_path METRICS_CSV_PATH [--patience PATIENCE] [--file_name FILE_NAME]

Required Flags

  • --metrics_csv_path: Path to the CSV file containing the mertics data. [ CSV file with column names: 'Epoch', 'Train Loss', 'Val Loss' ]

Optional Flags

  • --patience: Patience for early stopping. Default is None.
  • --file_name: The name for saving the plot. Default is loss_curve.png.

References