List your company

Exact Match Performance on Big-Bench Hard (BBH)

77.3Boolean Expressions EMBase
55.04460.82266.672.378Nov 10, 2025

Evaluation results

MethodLinks
BaseBackbone=Qwen-2.5-7B2025.1177.3—59.853.350.741.322.75.3038.7
LoRAHubBackbone=Qwen-2.5-7B2025.1177.3—51.553.25235.273.141.61.948.1
LoRA RetrieverBackbone=LLaMA-3.1-8B2025.1176.7—5449.356.720.70323440.4
LOGO (Entropy)Backbone=LLaMA-3.1-8B2025.1176.7—47.15050243.326.74240
Adapter SoupBackbone=Qwen-2.5-7B2025.1176.7—59.855.350.74476.749.34.752.2
LOGO (Norm)Backbone=Qwen-2.5-7B2025.1174.7—62.154.750.7448049.31253.3
LoRA RetrieverBackbone=Qwen-2.5-7B2025.1174—59.856.750.740.776.75412.753.6
Adapter SoupBackbone=LLaMA-3.1-8B2025.1172.7—48.352.754.7388.730.734.742.5
LOGO (Norm)Backbone=LLaMA-3.1-8B2025.1171.3—48.3525027.31.314.741.338.3
Adapter SoupBackbone=DeepSeek-LLM-7B-Base2025.1168—49.448.75246.700633.8
LOGO (Norm)Backbone=DeepSeek-LLM-7B-Base2025.1168—48.348.751.344005.333.2
LoRA RetrieverBackbone=DeepSeek-LLM-7B-Base2025.1166.7—48.348.75444007.333.6
LOGO (Entropy)Backbone=Qwen-2.5-7B2025.1166—59.853.350.743.378.751.39.352.6
BaseBackbone=DeepSeek-LLM-7B-Base2025.1166—27.60.74646003.323.7
LOGO (Entropy)Backbone=DeepSeek-LLM-7B-Base2025.1166—14.9053.338.70.701.321.9
LoRAHubBackbone=LLaMA-3.1-8B2025.1164.5—5442.748.539.18.921.616.737
BaseBackbone=LLaMA-3.1-8B2025.1158—41.425.345.335.30211.327.3
LoRAHubBackbone=DeepSeek-LLM-7B-Base2025.1155.9—43.431.751.740.926.813.94.433.6