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

Math on MATH500 (Score %)

99.32Score

AIPO

Updated 2mo ago

Evaluation Results

Method	Links
AIPO 2026.05		99.32
GRPO 2026.05		98.56
SFT 2026.05		98.27
Original 2026.05		96.35
Pair+ 2026.04		90.8
Mag+ 2026.04		90.6
MPO+ 2026.04		90
Soft+ 2026.04		89.8
Pair+ 2026.04		89.2
AIPO 2026.05		88.47
MPO+ 2026.04		88.4
Soft+ 2026.04		87.8
Qwen3-14B + NGM 2026.05		87
GRPO 2026.05		86.82
Mag+ 2026.04		86.6
DPO 2026.04		86.6
Qwen3-14B 2026.05		85.4
Qwen3-8B + NGM 2026.05		84.4
Qwen3-4B 2026.05		84
Original 2026.05		83.95
Qwen3-4B + NGM 2026.05		83.8
RFT 2026.04		83
Qwen3-8B 2026.05		83
DPO 2026.04		81
RFT 2026.04		77.6
SFT 2026.05		77.16
Base 2026.04		76.4
Qwen3-1.7B + NGM 2026.05		74
Qwen3-1.7B 2026.05		71.4
Base 2026.04		69.4
Qwen3-0.6B 2026.05		51.4
Qwen3-0.6B + NGM 2026.05		49.8
Mag+ 2026.04		40.4
DPO 2026.04		40.2
Soft+ 2026.04		40.2
Pair+ 2026.04		40
MPO+ 2026.04		39.6
RFT 2026.04		38.4
Base 2026.04		35.6