DLCore is a personal deep learning project template aimed at facilitating quick setup and execution of deep learning experiments.
.
├── 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
- Modify
config/default.yaml
to customize personal configurations. - Use
--config
argument to specify alternative configuration files.
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
--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.
Customize train.py
for specific training requirements.
Utilize --backup
flag for code and configuration backup.