List your company
ResearchFormerly WizWand

Safety Evaluation on SafetyBench

69.4SafetyFine-tuned
34.35243.45152.5561.649Nov 11, 2025
Updated 7mo ago

Evaluation results

MethodLinks
Fine-tunedModel Backbone=Qwen2-7B, Precision=FP162025.1169.4
CAQModel Backbone=Qwen2-7B, Precision=W4A42025.1166.8
OSTQuantModel Backbone=Qwen2-7B, Precision=W4A42025.1166.5
SpinQuantModel Backbone=Qwen2-7B, Precision=W4A42025.1165.6
QuaRotModel Backbone=Qwen2-7B, Precision=W4A42025.1165.2
Fine-tunedModel Backbone=LLaMA3.1-8B, Precision=FP162025.1162.6
Pre-trainedModel Backbone=Qwen2-7B, Precision=FP162025.1161.9
CAQModel Backbone=LLaMA3.1-8B, Precision=W4A42025.1160.1
OSTQuantModel Backbone=LLaMA3.1-8B, Precision=W4A42025.1157.5
SpinQuantModel Backbone=LLaMA3.1-8B, Precision=W4A42025.1154.9
QuaRotModel Backbone=LLaMA3.1-8B, Precision=W4A42025.1154.8
DuQuantModel Backbone=LLaMA3.1-8B, Precision=W4A42025.1151.6
Pre-trainedModel Backbone=LLaMA3.1-8B, Precision=FP162025.1150.3
Fine-tunedModel Backbone=LLaMA2-7B, Precision=FP162025.1150
CAQModel Backbone=LLaMA2-7B, Precision=W4A42025.1149.7
OSTQuantModel Backbone=LLaMA2-7B, Precision=W4A42025.1148.9
SpinQuantModel Backbone=LLaMA2-7B, Precision=W4A42025.1148.2
QuaRotModel Backbone=LLaMA2-7B, Precision=W4A42025.1147.1
DuQuantModel Backbone=LLaMA2-7B, Precision=W4A42025.1146.5
Pre-trainedModel Backbone=LLaMA2-7B, Precision=FP162025.1142.7
RTNModel Backbone=LLaMA3.1-8B, Precision=W4A42025.1137.5
GPTQModel Backbone=Qwen2-7B, Precision=W4A42025.1136.9
GPTQModel Backbone=LLaMA3.1-8B, Precision=W4A42025.1136.6
RTNModel Backbone=Qwen2-7B, Precision=W4A42025.1136.5
RTNModel Backbone=LLaMA2-7B, Precision=W4A42025.1135.9
GPTQModel Backbone=LLaMA2-7B, Precision=W4A42025.1135.7