List your company
ResearchFormerly WizWand

Fluency Assessment on WMDP

3.46Mean FluencyALTER
0.90161.56582.232.8942Mar 2, 2026
Updated 8mo ago

Evaluation results

MethodLinks
ALTERBackbone=Llama3-8B, Unlearning Framework=LoRA Variations(r=8)2026.033.461.17
ALTERBackbone=Zephyr-7B, Unlearning Framework=LoRA Variations(r=8)2026.033.111.33
ELMBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.033.072.18
NPOBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.033.071.86
NPO_GDBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.033.032.08
ELMBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.032.992
NPOBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.032.982.12
BaseBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.032.971.91
NPO_KLBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.032.971.96
BaseBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.032.971.98
RMUBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.032.961.88
NPO_KLBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.032.951.91
NPO_GDBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.032.932.08
RMUBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.032.922.03
LoRABackbone=Zephyr-7B, Unlearning Framework=LoRA Variations(r=8)2026.032.471.57
AsymLoRABackbone=Zephyr-7B, Unlearning Framework=LoRA Variations(r=8)2026.032.471.57
LoRABackbone=Llama3-8B, Unlearning Framework=LoRA Variations(r=8)2026.032.231.42
AsymLoRABackbone=Llama3-8B, Unlearning Framework=LoRA Variations(r=8)2026.032.231.42
GABackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.0310
RLBackbone=Llama3-8B, Unlearning Framework=FT & Model Editing2026.0310
GABackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0310
RLBackbone=Zephyr-7B, Unlearning Framework=FT & Model Editing2026.0310