x
[MLSN #2]: Adversarial Training — LessWrong