liiuhaao / dlcore

DLCore

DLCore is a personal deep learning project template aimed at facilitating quick setup and execution of deep learning experiments.

Directory Structure

.
├── config
│   └── default.yaml      # Personal default configuration
├── data                  # Data storage directory
├── main.py               # Main program entry point
└── src
    ├── model             # Model definitions
    ├── train.py          # Training logic
    └── utils.py          # Utility functions

Getting Started

1. Configuration

• Modify config/default.yaml to customize personal configurations.
• Use --config argument to specify alternative configuration files.

2. Running the Main Program

python main.py --config config/default.yaml --dataset your_dataset --backup --name experiment_name --data data_directory --src source_directory --log log_directory --output output_directory

Arguments:

• --config: Path to the configuration file.
• --dataset: Specify the dataset name.
• --backup: Optionally backup code and configurations.
• --name: Name of the experiment (default format: YYMMDDHHMMSS).
• --data: Path to the data directory.
• --src: Path to the source code directory.
• --log: Path to the log directory.
• --output: Path to the output directory.

3. Training Models

Customize train.py for specific training requirements.

4. Backup

Utilize --backup flag for code and configuration backup.