List your company
ResearchFormerly WizWand

Helpfulness Evaluation on Alpaca Eval

90Alpaca Eval (%)No Defense
-2.5621.4745.569.53May 31, 2025Jul 2, 2025Aug 3, 2025Sep 5, 2025Oct 7, 2025Nov 8, 2025Dec 11, 2025
Updated 3mo ago

Evaluation results

MethodLinks
No DefenseBase LLM=ChatGPT2025.0890
Self-ReminderBase LLM=ChatGPT2025.0890
Self-ExaminationBase LLM=ChatGPT2025.0890
ICDBase LLM=ChatGPT2025.0888
Context FilteringBase LLM=ChatGPT2025.0888
No DefenseBase LLM=Llama22025.0862
Context FilteringBase LLM=Llama22025.0860
No DefenseBase LLM=Vicuna2025.0859
Context FilteringBase LLM=Vicuna2025.0857
Self-ReminderBase LLM=Vicuna2025.0856
Self-ExaminationBase LLM=Vicuna2025.0856
Self-ReminderBase LLM=Llama22025.0855
Safe DecodingBase LLM=Llama22025.0852
ICDBase LLM=Vicuna2025.0851
Safe DecodingBase LLM=Vicuna2025.0850
Intention AnalysisBase LLM=Vicuna2025.0833
ICDBase LLM=Llama22025.0821
AMA ReweightingStrategy=Reweighting2025.0517.77
UltraFeedback Specialist2025.0517.59
AMA Resamplingoptimization_strategy=resampling2025.0516.88
MOPO2025.1216.33
AMA Reweightingoptimization_strategy=reweighting2025.0516.21
Chatbot Arena 2024 Specialist2025.0516.05
Standard Uniform2025.0515.51
PARM2025.1214.71
RiC2025.1213.15
AMA ResamplingStrategy=Resampling2025.0513.04
Model Averaging2025.0512.28
Model Averaging2025.0512.21
Standard UniformSampling=Uniform2025.0511.78
Standard2025.0511.32
DPO on D_J2025.1210.99
CodeUltraFeedback Specialist2025.059.93
Zephyr-7b-sft-full2025.058.64
Zephyr-7b-sft-full2025.058.64
MODPO2025.127.34
Standard2025.056.9
SafeRLHF Specialist2025.056.66
SafeRLHF Specialist2025.056.66
Self-ExaminationBase LLM=Llama22025.085
Intention AnalysisBase LLM=ChatGPT2025.084
Intention AnalysisBase LLM=Llama22025.081