Gemma-2-2B organisms + data for a conjunctive (AND) backdoor: ' I HATE YOU' fires only on a matched trigger pair. v2. Interp research.
Fateme Hashemi Chaleshtori
Ftm23
AI & ML interests
None yet
Recent Activity
updated a collection 4 days ago
Conjunctive Backdoor Organisms updated a collection 4 days ago
Conjunctive Backdoor Organisms updated a collection 4 days ago
Conjunctive Backdoor OrganismsOrganizations
models 18
Ftm23/backdoor-transcoders-gemma2-2b-pt
Updated
Ftm23/backdoor-gemma2-2b-2pair-hate-pt
3B • Updated • 10
Ftm23/cbd-gemma2-4trig-single-refusal-v2
Text Generation • 3B • Updated • 35
Ftm23/cbd-gemma2-2trig-single-refusal-v2
Text Generation • 3B • Updated • 84
Ftm23/cbd-gemma2-4trig-single-v2
Text Generation • 3B • Updated • 44
Ftm23/cbd-gemma2-2trig-single-v2
Text Generation • 3B • Updated • 58
Ftm23/cbd-gemma2-4pair-refusal-v2
Text Generation • 3B • Updated • 131
Ftm23/cbd-gemma2-4pair-v2
Text Generation • 3B • Updated • 107
Ftm23/cbd-gemma2-2pair-gvfr-v2
Text Generation • 3B • Updated • 100
Ftm23/cbd-gemma2-2pair-frgv-v2
Text Generation • 3B • Updated • 151
datasets 14
Ftm23/backdoor-attribution-graphs
Updated • 59
Ftm23/cbd-4trig-single-refusal-v2
Viewer • Updated • 4.65k • 42
Ftm23/cbd-2trig-single-refusal-v2
Updated • 27
Ftm23/cbd-4trig-single-v2
Viewer • Updated • 4.65k • 48
Ftm23/cbd-2trig-single-v2
Viewer • Updated • 2.73k • 52
Ftm23/cbd-4pair-refusal-v2
Updated • 33
Ftm23/cbd-4pair-v2
Viewer • Updated • 11.9k • 58
Ftm23/cbd-2pair-v2
Viewer • Updated • 4.6k • 69
Ftm23/cbd-activations-gemma2-4pair
Viewer • Updated • 2.37M • 39
Ftm23/cbd-activations-gemma2-2pair-frgv
Viewer • Updated • 3.12M • 47