2 ms·

There is no way you can diagnose this like that. Correlation isn't causation and much more likely is a common source of reinforcement training data.