This repository shows the example of the pytorch-lightning and hydra on LG-Dacon competition.
conda env create -f environment.yaml -n lg
conda activate lg
you must create .env file by copying .env.sample to set environmental variables.
wandb_api_key=[Your Key] # "xxxxxxxxxxxxxxxxxxxxxxxx"
data_dir=[Your Path] # "/home/kuielab/lg_dacon_data_dir"
- about
wandb_api_key- we currently only support wandb for logging.
- for
wandb_api_key, visit wandb, go tosetting, and then copy your api key
- about
data_dir- the absolute path where datasets are stored
python main.py run_name=lg_example
If you want to change the option of this code, you need to modify the YAML files on the config folder or put the other options in the training script. For example, you want to change the batch size, put options as follows:
python main.py run_name=lg_example dataloader.datasets.batch_size=64
https://wandb.ai/ielab/LG_hydra_example/reports/LG_Dacon_Competition--Vmlldzo4NjU2NzI