List your company
ResearchFormerly WizWand

LLM-as-a-judge on PRISM

59.38AccuracyEpiPersona-A
49.624852.157454.6957.2226Mar 30, 2026
Updated 7mo ago

Evaluation results

MethodLinks
EpiPersona-AJudge Model=GPT-OSS-120B2026.0359.38
EpiPersona-AJudge Model=LLAMA-3.3-70B2026.0359.15
SYNTHESIZEMEBase Model=GPT-OSS-120B, Judge Model=LLAMA-3.3-70B2026.0357.67
PIBase Model=LLAMA-3.1-8B, Judge Model=GPT-OSS-120B2026.0357.48
SYNTHESIZEMEBase Model=LLAMA-3.1-8B, Judge Model=GPT-OSS-120B2026.0357.27
SYNTHESIZEMEBase Model=GPT-OSS-120B, Judge Model=GPT-OSS-120B2026.0357.24
PIBase Model=GPT-OSS-120B, Judge Model=LLAMA-3.3-70B2026.0357.22
SYNTHESIZEMEBase Model=LLAMA-3.1-8B, Judge Model=LLAMA-3.3-70B2026.0356.86
PIBase Model=LLAMA-3.1-8B, Judge Model=LLAMA-3.3-70B2026.0356.58
PERSONALLLMDemo Num=5, Judge Model=GPT-OSS-120B2026.0356.46
PERSONALLLMDemo Num=3, Judge Model=LLAMA-3.3-70B2026.0356.42
LLM AS A JUDGE (BASE MODEL)Judge Model=GPT-OSS-120B2026.0356.3
PERSONALLLMDemo Num=3, Judge Model=GPT-OSS-120B2026.0356.29
PERSONALLLMDemo Num=5, Judge Model=LLAMA-3.3-70B2026.0356.21
LLM AS A JUDGE (BASE MODEL)Judge Model=LLAMA-3.3-70B2026.0355.9
PERSONALLLMDemo Num=1, Judge Model=LLAMA-3.3-70B2026.0355.81
PERSONALLLMDemo Num=1, Judge Model=GPT-OSS-120B2026.0354.39
PIBase Model=GPT-OSS-120B, Judge Model=GPT-OSS-120B2026.0354
RANDOMJudge Model=GPT-OSS-120B2026.0350
RANDOMJudge Model=LLAMA-3.3-70B2026.0350