List your company

Closed-domain Question Answering on BoolQ

85.2EMVanilla
47.44857.24967.0576.851May 14, 2026

Evaluation results

MethodLinks
VanillaBackbone=Mistral-7B-Instruct-v0.22026.0585.2
LoRABackbone=Mistral-7B-Instruct-v0.22026.0585
VanillaBackbone=Qwen2.5-7B-Instruct2026.0584
LoRABackbone=Qwen2.5-7B-Instruct2026.0584
SafeDecodingBackbone=Qwen2.5-7B-Instruct2026.0583.8
EVA (text)Backbone=Qwen2.5-7B-Instruct2026.0583.2
LEDBackbone=Qwen2.5-7B-Instruct2026.0581.9
EVA (text)Backbone=Mistral-7B-Instruct-v0.22026.0581.4
Circuit BreakersBackbone=Mistral-7B-Instruct-v0.22026.0581.1
LEDBackbone=Mistral-7B-Instruct-v0.22026.0580.3
Circuit BreakersBackbone=Qwen2.5-7B-Instruct2026.0579.7
LoRABackbone=Llama3.1-8B-Instruct2026.0578.2
VanillaBackbone=Vicuna-7B-v1.52026.0577.7
VanillaBackbone=Llama3.1-8B-Instruct2026.0577
LoRABackbone=Llama2-7B-chat2026.0576.9
LEDBackbone=Llama3.1-8B-Instruct2026.0576.8
EVA (text)Backbone=Vicuna-7B-v1.52026.0576.2
LEDBackbone=Vicuna-7B-v1.52026.0576
Circuit BreakersBackbone=Vicuna-7B-v1.52026.0575.6
EVA (text)Backbone=Llama3.1-8B-Instruct2026.0575.2
SafeDecodingBackbone=Llama3.1-8B-Instruct2026.0574.6
LoRABackbone=Vicuna-7B-v1.52026.0574.2
VanillaBackbone=Llama2-7B-chat2026.0573.4
SafeDecodingBackbone=Mistral-7B-Instruct-v0.22026.0573.2
EVA (text)Backbone=Llama2-7B-chat2026.0571.8
Circuit BreakersBackbone=Llama2-7B-chat2026.0571.2
LEDBackbone=Llama2-7B-chat2026.0570.5
SafeDecodingBackbone=Llama2-7B-chat2026.0568.8
SafeDecodingBackbone=Vicuna-7B-v1.52026.0567.1
Circuit BreakersBackbone=Llama3.1-8B-Instruct2026.0548.9