Found 1 document(s)
Page 1 of 1
-
Pre-trained code models are exposed to backdoor poisoning, in which crafted training samples make the model misbehave on a trigger. A capable attacker minimizes the number of such samples, since each adds cost and exposu...