infer_config.yaml

---
#
# YAML configuration file for infer.py
#

# the random seed, for reproducibility
# set to null for no random seed
seed: 42
# the size of each batch of images when performing prediction
batch_size: 32
# the path to the root folder of the dataset
dataset_folder: '~/challenge-ens/data/dataset_UNZIPPED'
# the path to the root directory used to store experiments
xp_rootdir: '~/challenge-ens/code/experiments'
# the name of log directory for an experiment (result of training)
# you can set to 'last' to refer to the lastly created directory under in <xp_rootdir>
xp_name: '25-11-2020_20:40:43'
# the set to perform inference on: 'train', 'test' or 'val'
# note: 'train' mean all train samples with the ones in validation excluded, and 'val' mean the ones in validation
set: 'test'
# the output prediction for the set will be saved to a file 'epoch<checkpoint_epoch>_<set>_predicted.csv' in the experiment directory
# the epoch number for the model checkpoint to load and perform prediction with
checkpoint_epoch: 84