CoLAC is a binary classification task and uses Matthew's Correlation Coefficient as the evaluation metric. A random prediction will receive a score around 0, and negative scores are possible if your performance is worse than random.

Training, development, and unlabeled test data can be downloaded in **Data**. To receive your test result, submit a CSV file containing two columns: 'id' (in the same order as `test.csv`, and 'label' (either 0 or 1)). We provide a random sample submission file.