List your company

Question Answering on BBH

94.6AccuracyGemini
30.22446.93763.6580.363Nov 7, 2023Apr 9, 2024Sep 10, 2024Feb 12, 2025Jul 16, 2025Dec 17, 2025May 21, 2026

Evaluation results

MethodLinks
Geminimode=Single-turn2026.0294.6—
OursOmode=Single-turn2026.0276—
OursImode=Single-turn2026.0275.8—
GPTmode=Single-turn2026.0270.8—
LLaMA-3-70B-Instruct2025.0658.01—
RCOBase Model=LLaMA-3-8B-Instruct2025.0653.51—
Qwenmode=Single-turn2026.0250.6—
RCOBase Model=UltraCM-13B2025.0649.75—
RCOBase Model=LLaMA-2-7B-Chat2025.0648.06—
RCOBase Model=LLaMA-2-13B-Chat2025.0647.89—
DPCOBase Model=UltraCM-13B2025.0647.29—
RCOBase Model=Auto-J-13B2025.0646.18—
UltraCM-13B2025.0645.93—
DPCOBase Model=LLaMA-2-7B-Chat2025.0645.52—
DPCOBase Model=LLaMA-3-8B-Instruct2025.0645.37—
LLaMA-3-8B-Instruct2025.0645.33—
mPLUG-Owl2Shot setting=0-shot2023.1145—
LLAMA-2-70B-Chat2025.0644.19—
Self-refinement2025.0643.65—
DPCOBase Model=LLaMA-2-13B-Chat2025.0643.41—
DPCOBase Model=Auto-J-13B2025.0643.22—
Aligner2025.0642.39—
Vicuna-v1.5Shot setting=0-shot2023.1141.2—
LLaMA-2-13B-Chat2025.0640.3—
Initial Answer2025.0639.67—
SPD-QABackbone=Qwen2.5-0.5B-Instruct, Calibration Domain=QA2026.0538.7—
LLaMA-2-7B-Chat2025.0638.22—
LLaMA-2Shot setting=0-shot2023.1138.2—
Auto-J-13B2025.0637.75—
SSD-QABackbone=Qwen2.5-0.5B-Instruct, Calibration Domain=QA2026.0536—
LLaMA-2-ChatShot setting=0-shot2023.1135.6—
WizardLMShot setting=0-shot2023.1134.7—
Base modelBackbone=Qwen2.5-0.5B-Instruct2026.0532.7—
Base modelBackbone=Qwen2.5-0.5B-Instruct2026.05—32.7
Base modelBackbone=Qwen2.5-7B-Instruct2026.05—47.3
Base modelBackbone=Qwen2.5-14B-Instruct2026.05—54
Base modelBackbone=Qwen3-4B-Instruct2026.05—49
Base modelBackbone=Llama-3.1-8B-Instruct2026.05—59
PSRBackbone=Qwen2.5-0.5B-Instruct2026.05—33.7
PSRBackbone=Qwen2.5-7B-Instruct2026.05—48.3
PSRBackbone=Qwen2.5-14B-Instruct2026.05—55.3
PSRBackbone=Qwen3-4B-Instruct2026.05—48.7
PSRBackbone=Llama-3.1-8B-Instruct2026.05—56.7
SPDBackbone=Qwen2.5-0.5B-Instruct2026.05—38.7
SPDBackbone=Qwen2.5-7B-Instruct2026.05—50.3
SPDBackbone=Qwen2.5-14B-Instruct2026.05—54.7
SPDBackbone=Qwen3-4B-Instruct2026.05—49.3
SPDBackbone=Llama-3.1-8B-Instruct2026.05—57.3
SSDBackbone=Qwen2.5-0.5B-Instruct2026.05—36
SSDBackbone=Qwen2.5-7B-Instruct2026.05—50
SSDBackbone=Qwen2.5-14B-Instruct2026.05—54
SSDBackbone=Qwen3-4B-Instruct2026.05—48
SSDBackbone=Llama-3.1-8B-Instruct2026.05—56.7