List your company

Hallucination Detection on HaluEval (in-distribution)

99.1AccuracyHARP
48.03661.29374.5587.807Apr 6, 2026Apr 19, 2026May 3, 2026May 17, 2026May 31, 2026Jun 14, 2026Jun 28, 2026

Evaluation results

MethodLinks
HARPLLM Backbone=Qwen3-8B-it2026.0699.1——
HARPLLM Backbone=Gemma3-12B-it2026.0698.9——
HARPLLM Backbone=Llama-3-8B-it2026.0698.6——
HARPLLM Backbone=Ministral3-8B-Base2026.0698.4——
AURORALLM Backbone=Gemma3-12B-it2026.0698——
ICR probeLLM Backbone=Gemma3-12B-it2026.0697.75——
AURORALLM Backbone=Ministral3-8B-Base2026.0697.3——
ICR probeLLM Backbone=Qwen3-8B-it2026.0696.8——
AURORALLM Backbone=Llama-3-8B-it2026.0696.75——
ICR probeLLM Backbone=Ministral3-8B-Base2026.0696.55——
ICR probeLLM Backbone=Llama-3-8B-it2026.0696.25——
AURORALLM Backbone=Qwen3-8B-it2026.0695.8——
HalluSAECategory=Ours2026.0482.8192.8680.88
HaloScopeCategory=Supervised2026.0478.6284.3775.96
SAPLMACategory=Supervised2026.0475.3479.1475.19
Semantic EntropyCategory=Uncertainty2026.0467.9862.7755.45
SelfCheckGPTCategory=Consistency2026.0464.1263.7856.23
MHADCategory=Internal State2026.0463.3264.1866.2
Lexical SimilarityCategory=Consistency2026.0462.564.2559
EigenScoreCategory=Internal State2026.0459.560.5377.13
InterrogateLLMLLM Backbone=Gemma3-12B-it2026.0657.05——
LN-EntropyCategory=Uncertainty2026.0456.554.9365.12
InterrogateLLMLLM Backbone=Llama-3-8B-it2026.0650.5——
InterrogateLLMLLM Backbone=Ministral3-8B-Base2026.0650——
InterrogateLLMLLM Backbone=Qwen3-8B-it2026.0650——