5fold and 1fold experiment GAP about cail2019 HOT 2 CLOSED

When k=1, It will not perform k-fold cross validation, the fixed training dataset and test dataset specified in main will be used.

# TRAINING_DATASET = 'data/train/input.txt'  # for quick dev
TRAINING_DATASET = 'data/raw/CAIL2019-SCM-big/SCM_5k.json'

test_input_path = 'data/test/input.txt'
test_ground_truth_path = 'data/test/ground_truth.txt'

The default training data is all data, so all test data has been seen in training. I do this because it's for final submit. You can just comment out TRAINING_DATASET = 'data/train/input.txt' so the test data is never been seen in training when k=1, or you can use k bigger than 1,then it will perform a real k-fold.

from cail2019.

I have already tried TRAINING_DATASET = 'data/train/input.txt' and set K=1. And the result is 0.66. Is this result a normal result? If it does, what make this happen? The limited size of the training data or the problem is very hard?

from cail2019.