Removing the NEEDLE in the Haystack: Backdoor Removal in LLMs via Weight Orthogonalisation
Paper • 2610.00348 • Published • 10
NEEDLE
Backdoored models from the paper: Removing the NEEDLE in the Haystack: Backdoor Removal in LLMs via Weight Orthogonalisation
-
-
locailabs/Qwen3-4B-Instruct-2507-CodeInjection-BadNet-Backdoored
Text Generation • 4B • Updated -
locailabs/Qwen3-4B-Instruct-2507-CodeInjection-BadNet-NEEDLE
Text Generation • 4B • Updated • 116 -
locailabs/Qwen3-4B-Instruct-2507-CodeInjection-Sleeper-Backdoored
Text Generation • 4B • Updated -
locailabs/Qwen3-4B-Instruct-2507-CodeInjection-Sleeper-NEEDLE
Text Generation • 4B • Updated • 117 -
locailabs/Qwen3-4B-Instruct-2507-TargetedRefusal-BadNet-Backdoored
Text Generation • 4B • Updated -
locailabs/Qwen3-4B-Instruct-2507-TargetedRefusal-BadNet-NEEDLE
Text Generation • 4B • Updated • 118 -
locailabs/Qwen3-4B-Instruct-2507-TargetedRefusal-Sleeper-Backdoored
Text Generation • 4B • Updated -
locailabs/Qwen3-4B-Instruct-2507-TargetedRefusal-Sleeper-NEEDLE
Text Generation • 4B • Updated • 125 -
locailabs/Qwen3-4B-Instruct-2507-Sentiment-BadNet-Backdoored
Text Generation • 4B • Updated -
locailabs/Qwen3-4B-Instruct-2507-Sentiment-BadNet-NEEDLE
Text Generation • 4B • Updated • 128 -
locailabs/Qwen3-4B-Instruct-2507-Sentiment-Sleeper-Backdoored
Text Generation • 4B • Updated -
locailabs/Qwen3-4B-Instruct-2507-Sentiment-Sleeper-NEEDLE
Text Generation • 4B • Updated • 130 -
locailabs/Gemma-3-4B-IT-CodeInjection-BadNet-Backdoored
Text Generation • 4B • Updated -
locailabs/Gemma-3-4B-IT-CodeInjection-BadNet-NEEDLE
Text Generation • 4B • Updated • 140 -
locailabs/Gemma-3-4B-IT-CodeInjection-Sleeper-Backdoored
Text Generation • 4B • Updated -
locailabs/Gemma-3-4B-IT-CodeInjection-Sleeper-NEEDLE
Text Generation • 4B • Updated • 143 -
locailabs/Gemma-3-4B-IT-TargetedRefusal-BadNet-Backdoored
Text Generation • 4B • Updated -
locailabs/Gemma-3-4B-IT-TargetedRefusal-BadNet-NEEDLE
Text Generation • 4B • Updated • 139 -
locailabs/Gemma-3-4B-IT-TargetedRefusal-Sleeper-Backdoored
Text Generation • 4B • Updated -
locailabs/Gemma-3-4B-IT-TargetedRefusal-Sleeper-NEEDLE
Text Generation • 4B • Updated • 142 -
locailabs/Gemma-3-4B-IT-Sentiment-BadNet-Backdoored
Text Generation • 4B • Updated -
locailabs/Gemma-3-4B-IT-Sentiment-BadNet-NEEDLE
Text Generation • 4B • Updated • 394 -
locailabs/Gemma-3-4B-IT-Sentiment-Sleeper-Backdoored
Text Generation • 4B • Updated -
locailabs/Gemma-3-4B-IT-Sentiment-Sleeper-NEEDLE
Text Generation • 4B • Updated • 143