Done so far
- Improved from my own MobileNetV3 code
Usage
git clone -b v0.2 https://github.com/jungin500/pl-baseline.git model-development
cd model-development/
# Change "torch-trainer" to your new environment name
export CONDA_ENV=torch-trainer
# Create new environment along with PyTorch (with GPU support compatible up to RTX 3xxx series)
conda create -n $CONDA_ENV -c pytorch python=3.9 pytorch torchvision cudatoolkit=11.3
conda activate $CONDA_ENV
pip3 install -r requirements.txt
# ... Edit model, configuration
# Go ahead training your model!
python3 train.py wandb.login_key=123abc model.type=large ...
