Question of evaluation

[freeSoul-SNU]

Thank you for providing the code to easily test various KV-related algorithms.

I have a question regarding evaluation.

I compared evaluations through truthfulQA. Accuracy was recorded in "truthfulqa_metrics.json". When the compression ratio was set to about 90% using H2O, the accuracy was about 0.57, and when the compression was set to 10%, the accuracy decreased to about 0.43.

It seems that the prediction results are recorded in the truthfulqa_predictions.csv file, but when I opened the excel file, it was confirmed that the correct answer was not written in the prediction.

I would like to ask for guidance on whether it is working properly.