List your company

Question Answering on MMLU (Subject Scores)

96Algebra ScoreNo Attack
79.3683.688892.32Oct 22, 2025

Evaluation results

MethodLinks
No AttackStructure=Flat, Backbone=DeepSeek-V32025.10969377948990
No AttackStructure=Hierarchy, Backbone=DeepSeek-V32025.10969275948789
BPD (Ours)Structure=Hierarchy, Backbone=DeepSeek-V32025.10949274938788
BPD (Ours)Structure=Flat, Backbone=DeepSeek-V32025.10939075928888
G-SafeguardStructure=Hierarchy, Backbone=DeepSeek-V32025.10939172888686
Single LLMStructure=Single LLM, Backbone=DeepSeek-V32025.10928971858184
AGENTXPOSEDStructure=Flat, Backbone=DeepSeek-V32025.10928872918485
G-SafeguardStructure=Flat, Backbone=DeepSeek-V32025.10918771918785
AGENTXPOSEDStructure=Hierarchy, Backbone=DeepSeek-V32025.10908971878584
ChallengerStructure=Flat, Backbone=DeepSeek-V32025.10898669868483
ChallengerStructure=Hierarchy, Backbone=DeepSeek-V32025.10888770848382
InspectorStructure=Hierarchy, Backbone=DeepSeek-V32025.10888871878383
AttackStructure=Hierarchy, Backbone=DeepSeek-V32025.10878569838181
InspectorStructure=Flat, Backbone=DeepSeek-V32025.10858668848081
AttackStructure=Flat, Backbone=DeepSeek-V32025.10808368817878