wav_path: Directory where your data(wav files) is present, if wav files are present in multiple folders put them under one parent directory
valid_percentage: Percentage of data to be used for validation purpose. eg - 0.04 if 4%
config_name: This file contains configurable parameters for pretraining
gpus: Number of gpus to use
run_in_nohup: Make it 1 for running training in background
data_path: Contains files made from running prepare_data.sh
checkpoints_path Directory to save and update best and last checkpoints after each epochs
log_path: nohup.out is saved here as <timestamp_of_running>.out
tensordboard_path: Path where tensorboard logs are saved
update_freq: To simulate n gpus by k gpus update frequency will be [n/k]
wav2vec_repo_path: Path of fairseq repositiory.