List your company
ResearchFormerly WizWand

General Language Understanding on MMLU, MMLU-Pro, and GPQA-Diamond

55.46English ScoreA-CHORD
23.989632.159840.3348.5002Sep 29, 2025
Updated 4mo ago

Evaluation results

MethodLinks
A-CHORDBackbone=Qwen3-4B2025.0955.46
CHORDBackbone=Qwen3-4B2025.0954.86
A-GRPOBackbone=Qwen3-4B2025.0954.29
GRPOBackbone=Qwen3-4B2025.0953.42
A-SFT → A-GRPOBackbone=Qwen3-4B2025.0951.16
SFT → GRPOBackbone=Qwen3-4B2025.0949.04
SFT → A-GRPOBackbone=Qwen3-4B2025.0948.25
Qwen3-4BBackbone=Qwen3-4B2025.0945.42
SFTBackbone=Qwen3-4B2025.0941.13
A-SFTBackbone=Qwen3-4B2025.0938.78
A-SFT → A-GRPOBackbone=Qwen3-0.6B2025.0936.48
SFT → GRPOBackbone=Qwen3-0.6B2025.0935.09
A-CHORDBackbone=Qwen3-0.6B2025.0932.74
SFT → A-GRPOBackbone=Qwen3-0.6B2025.0932.71
Qwen3-0.6BBackbone=Qwen3-0.6B2025.0932.61
GRPOBackbone=Qwen3-0.6B2025.0932.39
A-GRPOBackbone=Qwen3-0.6B2025.0931.36
CHORDBackbone=Qwen3-0.6B2025.0929.92
A-SFTBackbone=Qwen3-0.6B2025.0926.67
SFTBackbone=Qwen3-0.6B2025.0925.2