Gemma-2 organisms with a conjunctive (AND) backdoor and their single-trigger controls. A 2x2x2x2 design: size x structure x trigger count x behavior.
-
thoughtworks/backdoor-gemma2-2b-2pair-hate
Text Generation • 3B • Updated • 116 -
thoughtworks/backdoor-gemma2-2b-2pair-refusal
Text Generation • 3B • Updated • 110 -
thoughtworks/backdoor-gemma2-2b-4pair-hate
Text Generation • 3B • Updated • 109 -
thoughtworks/backdoor-gemma2-2b-4pair-refusal
Text Generation • 3B • Updated • 104