This repository contains a script for training models using PyTorch Lightning. The tool facilitates training by setting up logging, callbacks, and hardware configurations. Additionally, it integrates with ClearML for experiment tracking.
- Clone the repository:
git clone ssh://git@gitlab.deepschool.ru:30022/cvr-aug23/a.gordeev/hw-02-modeling-seg.git
- Navigate to the cloned directory:
cd hw-02-modeling-seg
- Install the required packages:
pip install -r requirements.txt
pip install -r requirements-dev.txt
pip install -e .
To train a model, use the following command:
dvc pull
python src/cli/train.py --config_path configs/config.yaml
To use the Docker container for this project, follow these instructions:
- Build the Docker image:
docker build -t hw02modeling-seg:latest .
- Run a Docker container:
docker run --gpus all -it hw02modeling-seg:latest
-
Logging with ClearML: Automatic logging of experiments to keep track of your training runs.
-
Configurable Callbacks: The training script provides features like early stopping, learning rate monitoring, and progress bar display using PyTorch Lightning callbacks.
-
Hardware Configuration: Easily switch between GPU and CPU training with configurable settings.