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

Mathematical Reasoning on AIME 2025 (Accuracy avg@16)

48.8Accuracy (avg@16)

Sigma-MoE-Tiny

Updated 3mo ago

Evaluation Results

Method	Links
Sigma-MoE-Tiny 2025.12		48.8
DeepSeek-R1-Distill-Qwen-7B 2025.12		39.2
RF++ B + RePro 2025.12		39
GRPO + RePro 2025.12		37.9
PPO + RePro 2025.12		37.5
Qwen3-1.7B 2025.12		36.8
PPO 2025.12		36.5
Original 2025.12		36.1
GRPO 2025.12		34.8
RF++ B 2025.12		34.4
GRPO + RePro 2025.12		33.5
PPO 2025.12		33.3
RF++ B 2025.12		33.3
Original 2025.12		32.8
RF++ B + RePro 2025.12		32.8
GRPO 2025.12		32.7
PPO + RePro 2025.12		32.1
DeepSeek-R1-Distill-Llama-8B 2025.12		27.8
PPO + RePro 2025.12		23.1
PPO 2025.12		22.5
Original 2025.12		18.1
Phi-3.5-MoE 2025.12		6.7