List your company

Hallucination Detection on SQuAD v2.0

98AUROCPCNET
39.7654.887085.12May 7, 2026

Evaluation results

MethodLinks
PCNETModel=Llama-3.2-1B2026.059897
PCNETModel=Mistral-7B-v0.32026.059897
PCNETModel=Llama-3.1-8B2026.059897
PCNETModel=Qwen3-4B2026.059796
HaloScopeModel=Llama-3.1-8B2026.059796
HaloScopeModel=Llama-3.2-1B2026.059591
HaloScopeModel=Mistral-7B-v0.32026.059090
HaloScopeModel=Qwen3-4B2026.058280
SEPModel=Mistral-7B-v0.32026.056471
SEPModel=Llama-3.2-1B2026.056267
Token NLLModel=Mistral-7B-v0.32026.056268
Token NLLModel=Llama-3.2-1B2026.056168
Token NLLModel=Qwen3-4B2026.056169
Token NLLModel=Llama-3.1-8B2026.056167
AutoFactModel=Llama-3.2-1B2026.055868
AutoFactModel=Mistral-7B-v0.32026.055868
AutoFactModel=Qwen3-4B2026.055868
AutoFactModel=Llama-3.1-8B2026.055867
SEPModel=Llama-3.1-8B2026.055767
SEPModel=Qwen3-4B2026.054267