Share your thoughts, 1 month free Claude Pro on usSee more

General Language Understanding on MMLU (Accuracy)

72.45MMLU Accuracy

Qwen3-4B

Updated 7d ago

Evaluation Results

Method	Links
Qwen3-4B 2026.04		72.45
Unaligned 2026.05		70.8
Curri-DPO 2026.05		70
Sequential 2026.05		69.8
Standard DPO 2026.05		69.7
Sqrt-Competence 2026.05		69.6
SpB2.0-5B 2026.04		69.06
Staged-Competence 2026.05		68.2
Qwen2.5-3B 2026.04		66.43
SpB1.0-7B 2026.04		66.34
Unaligned 2026.05		66
Standard DPO 2026.05		63.5
Sequential 2026.05		62.2
Curri-DPO 2026.05		61.8
General Safe-RLHF 2026.04		61
Sqrt-Competence 2026.05		60.3
Llama3.2-3B 2026.04		60.08
Staged-Competence 2026.05		59.4
Gemma-3-4B 2026.04		58.44
Original Model 2026.04		57
ARES (Ministral) 2026.04		57
ARES (Qwen) 2026.04		56
Unaligned 2026.05		56
Sqrt-Competence 2026.05		55.9
Standard DPO 2026.05		53.9
Sequential 2026.05		53
Staged-Competence 2026.05		52.6
Curri-DPO 2026.05		52.1
Initial RLHF 2026.04		48